434

.>Kt^^°^^o,

1 y' ' ' r

NOAA Technical Report NMFS Circular 434

Osteology, Phylogeny, and
Higher Classification of
the Fishes of the
Order Plectognathi
(Tetraodontiformes)

James C. Tyler

OCTOBER 1980

; Marina Biological Laboratory ]

libra:iy
OCT 141992

Woods Hoio. iViuSS.

U.S. DEPARTMENT OF COMMERCE

National Oceanic and Atmospheric Administration

National Marine Fisheries Service

NOAA TECHNICAL REPORTS
National Marine Fisheries Service, Circulars

The maj<ir resp.msibiluies of ihe National Marine Fisheries Service (NMFS) are to monitor and .,
distribution of fishery resources, to understand and predict fluctuations in the quantity and ili m h i :i ! i
for optimum use of the resources. NMFS is also charged with the development and impU nu i m i ■, . !
grounds, development and enforcement of domestic fisheries regulations, surveillance of Ion. ,,

development and enforcement of international fishery agreements and policies. NMFS also :i-.-;-t -: f, :i L
and economic analvsis programs, and mortgage insurance and vessel construction subsidies. It collects,
various phases of the industry.

The NOAA Technical Report N.MFS Circular series continues a series that has been in existence sir
publications of general interest intended to aid conservation and management. Publications that review
technical level certain broad areas of research appear in this series. Technical papers originating in econorr
vestigations appear in the Circular series.

NOAA Technical Report NMFS Circulars are available free in limited numbers to governmental agencies,
also available in exchange for other scientific and technical publicatiims in the marine sciences Individual
otherwise noted) from D82:J, User Services Branch. Environmental .Science Inlorniation Center. NOAA, Ri

nii-d States coastal waters, and the
iiidustrv through marketing service
lalyzes, and publishes statistics on

■ 19-tl. The Ciriulars are technical
1 considerable detail and at a high
s studies and from management in-

s. both Federal and State. Thev are

al copies mav be obtained (unless

ille. MD 20852. Recent Cir-

396. Whales, dolphins, and porpoises of the western North Atlantic. A
guide to their identification. By Stephen Leatherwood, David K. Cald-
well, and Howard E. Winn. August 1976, iv + 176 p., 168 figs., 1 table,
14 app. figs. For sale by the Superintendent of Documents, U.S. Gov-
ernment Printing Office, Washington. DC 20402; Stock No. 003-020-
00119-0.

397. Marine flora and fauna of the northeastern United States.
Cnidaria: Scyphozoa. By Flonald J. Larson. August 1976, iii + 18 p.,
28 figs. For sale by the Superintendent of Documents, U.S. Govern-
ment Printing Office, Washington, DC 20402; Stock No. 003-017-
00387.

398. Marine flora and fauna of the northeastern United States. Higher
fungi: Ascomycetes, Deuteromycetes, and Basidiomycetes. By A. R.
Cavaliere. March 1977, iii -I- 49 p., 24 figs. For sale by the Super-
intendent of Documents, U.S. Government Printing Office, Washington,
DC 20402; Stock No. 003-020-00126-2.

399. Marine flora and fauna of the northeastern United States.
Copepoda: Harpacticoida. By Bruce C. Coull. March 1977, iii -t- 48 p.,
100 figs. For sale by the Superintendent of Documents, U.S. Govern-
ment Priming Office, Washington, DC 20402; Stock No. 003-020-00125 -

400. Fishery publication index
Lee C. Thorson. March 1977,
intendent of Documents, U.S. Gi
DC 20402; Stock No. 003-020-0012'

By Mary Ellen Engett and
p. For sale by the Super-
Printing Office, Washington,

401. Fisheries and fishery resources of New York Bight. By J. L.
McHugh. March 1977, v + 50 p.. 43 figs., 49 tables. For sale by the
Superintendent of Documents, LI.S. Government Printing Office, Wash-
ington, DC 20402; Stock No. 003-020-00129-7.

402. Guide to the identification of scorpionfish larvae (Family Scor-
paenidae) in the eastern Pacific with comparative notes on species of
Sebastes and Helicolenus from other oceans. By H. Geoffrey Moser,
Elbert H. Ahlstrom, and Elaine M. Sandknop. April 1977, v -I- 71 p., 40
figs., 38 tables. For sale by the Superintendent of Documents, U.S. Gov-
ernment Printing Office, Washington, DC 20402; Stock No. 003-020-
00128-9.

0040-4.

ling Office, Washington, DC 20402; Stock No. 003-017

406. Marine flora and fauna of the northeastern United States.
Copepoda: Lcrnaeopodidae and Sphyriidae. By Ju-Shey Ho. De-
cember 1977, iii -I- 14 p., 16 figs. For sale by the Superintendent of
Documents, U.S. Government Printing Office, Washington, DC 20402;
Stock No. 003-017-00412-1.

407. Distribution of decapod Crustacea off northeastern United States
based on specimens at the Northeast Fisheries Center, Woods Hole.
Massachusetts. By Austin B. Williams and Roland L. Wigley. De-
cember 1977, iii -I- 44 p., 2 figs., 1 table, 57 charts.

408. Collecti.

409. Marine flora and fauna
Copepoda: Cyclopoids parasitic t
1978. iii + 12 p.. 17 figs.

By Richard S.

ortheastern I'nited St<
By .Ju-Shey Ho. Febri

\m tripos bloom in the New York Bight: Causes
Thomas C. Malone. May 1978. iv -^ 14 p.. 17

411. Systematics and biology of the tilefishes (Perciformes: Branchio-
stegidae and Malacanthidae). with descriptions of two new species. By
•James K. Dooley. April 1978, v -I- 78 p.. 44 figs., 26 tables.

412. Synopsis of biological data on the red porgy, Pagrus pagrus (Lin-
naeus). By Charles S. Manooch III and William W. Hassler. May
1978. iii -)- 19 p.. 12 figs., 7 tables. For sale by the Superintendent of
Documents, U.S. Government Printing Office, Washington, DC 20402;
Stock No. 003-017-00418-0.

413. Marine flora and fauna of the northeastern United States.
Crustacea: Branchiura. By Roger F. Cressey. May 1978, iii -I- 10 p., 15
figs. For sale by the Superintendent of Documents, U.S. Government
Printing Office, Washington, DC 20402; Stock No. 003-017-00419-8.

414

Synopsis of biological d:
tes amencanus (Walbauml
8, iii + 43 p., 21 figs., 28 tables.

for the winter flounder, Pseudopleuro-
By Grace Klein-MacPhee. November

403. Marine flora and fauna of the northeastern United States. Sipun-
cula. By Edward B. Cutler. July 1977, iii -(- 7 p., 6 figs. For sale by
the Superintendent of Documents, U.S. Government Printing Office,
Washington, DC 20402; Stock No. 003-012-00404-0.

404. Revision of the sea basses of the genus Oiptectrum (Pisces: Ser-
ranidae). By Stephen A. Bortone. September 1977, v -I- 49 p., 15 figs.,
9 tables.

405. Marine flora and fauna of the northeastern United States. Echino-
dermata: Holothuroidea. By David L. Pawson. September 1977, iii +
15 p., 3 figs. For sale by the Superintendent of Documents, U.S. Gov-

415. A basis for classifying western Atlantic Sciaenidae (Teleostei:
Perciformes). By Labbish Ning Chao. September 1978, v -i- 64 p., 41
figs.. 1 table.

416. Ocean variability: Effects on U.S. marine fishery resources -
1975. (20 papers.) By Julien R. Goulet. Jr. and Elizabeth D. Haynes,
Editors. December 1978, iii -I- 350 p.

417. Guide to the identification of genera of the fish Order
Ophidiiformes with a tentative classification of the order. By Daniel M.
Cohen and -Jyfrgen G. Nielsen. December 1978, vii + 72 p.. 103 figs., 2

NOAA Technical Report NMFS Circular 434

^'^m

^"'i^Emof^^

Osteology, Phylogeny, and
Higher Classification of
the Fishes of the
Order Plectognathi
(Tetraodontiformes)

James C. Tyler

OCTOBER 1980

Marine Biological Laboratory
LIBRARY

OCT 14 1992

Woods Hole, Mess.

U.S. DEPARTMENT OF COMMERCE

Philip M. Klutznick, Secretary

National Oceanic and Atmospheric Administration

Richard A. Frank, Administrator

National Marine Fisheries Service

Terry L. Leitzell, Assistant Administrator for Fisheries

The National Marine Fisheries Service (NMFS) does not approve, rec-
ommend or endorse any proprietary product or proprietary material
mentioned in this publication. No reference shall be made to NMFS, or
to this publication furnished by NMFS, in any advertising or sales pro-
motion which would indicate or imply that NMFS approves, recommends
or endorses any proprietary product or proprietary material mentioned
herein, or which has as its purpose an intent to cause directly or indirectly
the advertised product to be used or purchased because of this NMFS
publication.

CONTENTS

INTRODUCTION 1

Preface 1

Methods 2

PROLOGUE 4

Synopsis of the phylogeny of the Plectognathi 4

Synopsis ofthe higher classification of the Plectognathi 7

Historical review of the classification of the Plectognathi 7

Historical review of the nonclassificatory anatomical work on the Plectognathi 28

Relationship of the Plectognathi to the Perciformes 40

SYSTEMATIC SECTION 41

Definition and synopsis of the osteology of the Order Plectognathi (Tetraodontiformes) 41

Suborder Balistoidei (Sclerodermi) 44

Comparative diagnosis (contrast with that ofthe Tetraodontoidei) 44

Infraorder Triacanthoideo [and Superfamily Triacanthoidea] 44

Comparative diagnosis (contrast with that of the Balistoideo) 44

Family Triacanthodidae 46

Comparative diagnosis (contrast with that of the Triacanthidae) 46

Detailed description of Para/io//ardia lineata 47

Comparative diagnoses of subfamilies (Spinacanthinae, Eoplectinae, Hollardiinae, Triacan-

thodinae) 56

Anatomical diversity 56

Generic relationships 69

Subfamilial relationships, and relationships to the Triacanthidae and Gymnodontes 73

Family Triacanthidae 78

Comparative diagnosis (contrast with that of the Triacanthodidae) 78

Detailed desciiption of Pseudotriacanthus strigilifer 79

Comparative diagnoses of subfamilies (Protacanthodinae, Cryptobalistinae, Triacan-

thinae) 88

Anatomical diversity 92

Generic relationships 98

Infraorder Balistoideo 99

Comparative diagnosis (contrast with that of the Triacanthoideo) 99

Superfamily Balistoidea 99

Comparative diagnosis (contrast with that ofthe Ostracioidea) 99

Family Balistidae 101

Comparative diagnosis (contrast with that of the Monacanthidae) 101

Detailed description of Ba/ts£apusundu/atus 102

Anatomical diversity 110

Generic relationships 120

Relationship to the Monacanthidae 123

Relationship to the Triacanthidae 131

Family Monacanthidae 135

Comparative diagnosis (contrast with that ofthe Balistidae) 135

Detailed description of Monacant/iusci7iatus 136

Anatomical diversity 1^

Generic relationships 1"^^

Superfamily Ostracioidea 1^3

Comparative diagnosis (contrast with that of the Balistoidea) 183

Family Aracanidae ^85

Comparative diagnosis (contrast with that ofthe Ostraciidae) 185

Detailed description of Kentrocapros acu/eaf us 185

Anatomical diversity 1^5

Generic relationships 204

Relationship to the Ostraciidae 206

Relationship to the Balistidae 207

Family Ostraciidae 211

Comparativediagnosis(contrast with that ofthe Aracanidae) 211

Detailed description of A canthostracionquadncomis 211

Anatomical diversity 220

Generic relationships and comparative diagnoses ofsubfamilies(Ostraciinae and Lactophrysinae) . . . 238

Suborder Tetraodontoidei (Gymnodontes) 243

Comparative diagnosis (contrast with that of the Balistoidei) 243

Infraorder Triodontoideo [and Superfamily Triodontoidea, Family Triodontidae] 243

Comparative diagnosis (contrast with that of the Tetraodontoideo) 243

Detailed description of Triodon macropterus 244

Anatomical diversity 254

Relationships to the Balistoidei and to the other Tetraodontoidei 260

Infraorder Tetraodontoideo 263

Comparative diagnosis (contrast with that of the Triodontoideo) 263

Superfamily Tetraodontoidea 264

Comparative diagnosis (contrast with that of the Moloidea) 264

Family Tetraodontidae 265

Comparative diagnosis (contrast with that of the Diodontidae) 265

Comparativediagnosesof subfamilies (Tetraodontinae and Canthigasterinae) 265

Detailed description of Lagocep/ia/us/aeui^atus 266

Detailed description of Canthigaster rostrata 274

Anatomical diversity and generic relationships 286

Summary of generic relationships and intrafamilial classification 341

Relationships to the Diodontidae and to the other Tetraodontoidei 342

Family Diodontidae 343

Comparative diagnosis (contrast with that of the Tetraodontidae) 343

Detailed description of D(odon/io/ocan(/ius 343

Anatomical diversity 351

Generic relationships 365

Relationship to the other Tetraodontoidei 368

Superfamily Moloidea [and Family Molidae] 368

Comparative diagnosis (contrast with that of the Tetraodontoidea) 368

Detailed description oiMola mola 368

Anatomical diversity 377

Generic relationships 391

Relationship to the other Tetraodontoidei 392

Material examined 393

Recent species 393

Fossil species 398

EPILOGUE, Remaining problems 400

ACKNOWLEDGMENTS 401

LITERATURE CITED 403

INDEX OF DISCUSSION OF GENERA 421

Figxires

1. Flowing-line illustrated version of hypothesized familial and subfamilial groupings of the Plectognathi 5

2. Angular-line versionof hypothesized familial and subfamilial groupings of the Plectognathi 6

3. Flowing-line illustrated version of hypothesized familial, subfamilial, and generic relationships of the
Superfamily Triacanthoidea, the basal Plectognathi 45

4. Range of diversity in body form in the Recent Triacanthodidae: Parahollardia lineata and Halimo-
chirurgus alcocki 47

5. Parahollardia lineata: external features 48

Parahollardia lineata, composite based on several specimens, 45.7-86.1 mm SL, Gulf of Mexico to

South Carolina:

6. lateral view of entire specimen 57

7. lateral view of head 58

8. dorsal and ventral views of skull 58

9. posterior view of orbit; posterior view of skull; posterior and lateral views of first abdominal verte-

bra 59

10. dorsal view of branchial arches; lateral view of hyoid arch and urohyal 60

11. lateral view of pectoral girdle 60

12. lateral view of caudal fin supporting structures 61

13. Parahollardia lineata, 62.2 mm SL, Florida: dorsal, lateral, and ventral views of pelvis and pelvic fin 61

14. Parahollardia lineata, 82.0 mm SL, Florida: lateral and anterior views of first two dorsal spines 61

15. Protobalistum imperiale, lateral view of holotype, 522 mm SL, Eocene of Monte Bolca, Italy 62

16. SpiVwcanf/juscunei/ormis, lateral view of holotype, 104 mm SL, Eocene of Monte Bolca, Italy 63

17. £op/ecfUA' 6/o(i, lateral view of holotype, 65.2 mm SL, Eocene of Monte Bolca, Italy 64

18. Zjgnoic/if/iysofe/ongus, lateral view of holotype, ca. 161 mm SL, Eocene of Monte Bolca, Italy 65

Lateral views of entire specimens of other representative triacanthodid genera of the Subfamily Tria-
canthodinae:

19. Triacanf/iodesanoma/us, 71.5 mm SL, Japan, example of a relatively generalized member 67

20. Tydemanfa/iat^igaforjs, 64.4 mm SL, Bay of Bengal, example of a moderately specialized member .... 68

21. Macror/jomp/iosodesuradoi, 69.9 mm SL, Japan, example of a highly specialized member 69

22. Hollardia hollardi: lateral views of moderate-sized specimen, 62.7 mm SL, and extremely large
specimen, 174 mm SL, both from Caribbean 70

23. Para/3o//ardiasc/im;d(i, lateral view of head, 62.0 mm SL, Nicaragua 71

24. Tn'acanf/iodesanoma/us, lateral view of head, 71.5 mm SL, Japan 71

25. Parafn'acant/iodesrefrospmis, lateral view of head, 85.3 mm SL, Mozambique 71

26. Johnsonina eriomma, lateral view of head, 67.0 mm SL, Puerto Rico 72

27. Atrophacanthus japonicus, lateral view of head, 43.1 mm SL, Celebes 72

28. Tydemanianai'igatons, lateral view of head, 64.4 mm SL, Bay of Bengal 72

29. Macrorhamphosodes uradoi, lateral view of head, 69.9 mm SL, Japan 73

30. //a/imoc/z(>urgusceAj(nscoides, lateral view of head, 99.2 mm SL, Bay of Bengal 73

31. Typical body form in the Recent Triacanthidae iPseudotnacant/iusstrJ^i/i/er 78

Pseudotriacanthus strigilifer:

32. external features 79

33. lateral view of entire specimen, 79.0 mm SL, India 89

34. lateral view of head, 79.0 mm SL, India 90

35. dorsal, ventral, and posterior views of skull, 79.0 mm SL, India 90

36. posterior view of orbit, 130 mm SL, Thailand 90

37. Tripodichthys angustifrons: dorsal view of branchial arches; lateral view of hyoid arch and urohyal;

137 mm SL, Australia 91

38. Triacanthus biaculeatus: lateral and anterior views of first two dorsal spines, 113 mm SL, Borneo 91

39. A, Parahollardia lineata, 45.7 mm SL, Florida, and B, Triacanthus biaculeatus, 113 mm SL, Borneo;
lateral views of pelvis and pelvic fin; anterior views of the base of the pelvic spine 91

40. Trixiphichthys weberi: lateral view of entire specimen, 101 mm SL, Bay of Bengal 92

41. Trixiphichthys weberi: lateral view of head, 101 mm SL, Bay of Bengal 93

42. Protacanthodesombonii: lateral views of holotype, 112 mm SL, Eocene of Monte Bolca, Italy 94

43. Protacanthodes ombonii: lateral view of abdominal region, holotype, 112 mm SL, Eocene of Monte
Bolca, Italy 95

44. Acanthopleurus serratus: lateral view of entire specimen, syntype, ca. 91 mm SL, Oligocene of Canton
Glarus, Switzerland 96

45. Acanthopleurus serratus: lateral views of entire specimens, 153 mm SL and 127 mm SL, and A. col-

/e«ei, ca. 120 mm SL and 96.2 mm SL, paratypes, Oligocene of Canton Glarus, Switzerland 96

46. Charts showing two proportional measurements as percent of standard length that differentiate Acan-
thopleurus serratus and A. collettei 97

47. Marosichthys huismani: lateral view of anterior part of body of holotype, ca. 36 mm head length, Mio-
cene of the Celebes 98

48. Crs'ptobalistes brevis: lateral view of holotype, ca. 88 mm SL, Oligocene of Canton Glarus, Switzer-
land 99

49. Rangeof diversity in body form in the Recent Balistidae: Balistapus undulatus and Odonus niger 101

50. Balistapus undulatus: external features 102

Balistapus undulatus, composite based on several specimens, ca. 120-124 mm SL., western Pacific:

51. lateral view of entire specimen Ill

52. lateral view of head 112

53. dorsal and ventral views of skull 112

54. posterior views ofskuU and orbit; posterior and lateral views of first abdominal vertebra 113

55. dorsal view of branchial arches; lateral view of hyoid arch and urohyal 114

56. lateral view of pectoral girdle 114

57. medial view of upper jaw; lateral view of pelvis and encasing scales at end of pelvis; posterior view of

end of pelvis 115

58. lateral view of spiny dorsal fin and its pterygial supports 116

59. dorsal view of pterygial supports of spiny dorsal fin; posterior view of first dorsal spine and anterior

view of second dorsal spine 116

60. lateral view of caudal fin supporting structures 116

External features of other representative balistid genera:

61. Abalistes stellaris 117

62. Canthidermis maculatus 117

63. Xanthichthys ringens 117

64. Odonusniger 118

65. Fossil balistids: A, Balistomorphus ovalis, lateral view of holotype, 121 mm SL; B, B. ovalis, lateral
view of entire specimen, 119 mm SL; C, B. spinosus, lateral view of holotype, ca. 90 mm SL; D, B. orbi-
culatus, lateral view of holotype, 65.6 mm SL — specimens A through D from Oligocene of Canton Glarus,
Switzerland; E, Oligobalistes robustus, lateral view of syntype, ca. 60 mm SL, Maikop deposits of the
Oligocene of the Caucasus 119

66. Hypothesized phylogenetic relationships ofthe genera of Balistidae 121

67. Dentition of representative balistids: A, Batistes polylepis, 56.1 mm SL, Galapagos; B, Xanthichthys
lineopunctatus, 181 mm SL, Hawaii; C, Odonus niger, 173 mm SL, New Guinea; D, Melichthys niger,

147 mm SL, locality unknown 121

68. Scalesjustabovetipof pectoral fin of fia/tstoidesuindescens, 114 mm SL, Indonesia 122

69. Batistes polylepis, lateral view of entire specimen, 56.1 mm SL, Galapagos 124

70. Su/Z/amen/renatus, lateral view of entire specimen, 74.3 mm SL, Somalia 125

71. fl/imecanf/iusrecfangu/us, lateral view of entire specimen, 37.4 mm SL, Phoenix Islands 126

72. Cant/iidermismacu/atus, lateral view of entire specimen, 80.1 mm SL, Alabama 127

73. Xant/iic/it/iys/meopunctatus, lateral view of entire specimen, 181 mm SL, Hawaii 128

74. Lateral views of relatively normal balistid skulls: Batistes capriscus, ca. 360 mm SL, Gulf of Mexico; B.
polylepis, ca. 390 mm SL, Mexico 129

75. Lateral view of most specialized balistid skull and suspensorium, Odonus niger, 173 mm SL, New
Guinea 130

76. Batistes polylepis: ventral and dorsal views of skull, ca. 390 mm SL, Mexico 131

77. Batistes capriscus: ventral view of skull, ca. 360 mm SL, Gulf of Mexico 132

78. Range of diversity in body form in the Monacanthidae: A, Monacanthus citiatus; B, Amanses scopas;

C, Brachatuteres trossulus; D, Psitocephalus barbatus 135

Monacanthus citiatus:

79. external features 136

80. lateral view of entire specimen, 81.3 mm SL, Florida 145

81. lateral view of head, 81.3 mm SL, Florida 145

82. dorsal and ventral views of skull, 51.3 mm SL, Colombia 146

83. posterior views of skull and orbit, 81.3 mm SL, Florida 147

84. dorsal view of branchial arches; lateral view of hyoid arch and urohyal; 75.7 mm SL, Florida 147

85. lateral view of pelvis and encasing scales at end of pelvis; detail of the end of pelvis; 81.3 mm SL,

Florida 148

External features of other representative monacanthid genera:

86. Monacanthus chinensis 148

87. Stephanotepis cirrhifer 149

88. Chaetoderma spinosissimus 149

89. Paramonacanthus curtorhynchus 149

90. Pervagormetanoceplmlus 150

91. Rudarius ercodes 150

92. Navodon hypocrepis 150

93. Eubalichthys spilomelanurus 151

94. Cantherhines pultus 151

95. Amanses scopas 151

96. Alutera monoceros 151

97. Psitocephalus barbatus 152

98. Oxymonacanthus longirostris 152

99. Pseudatuteres nasicornis 152

100. Paratuteres prionurus 152

101. Brachatuteres trossulus 153

102. Step/iano/epw/iispidus, lateral view of entire specimen, 50.4 mm SL, Florida 155

103. Paramonacant/iuscrypfodon, lateral view of entire specimen, 68.5 mm SL, Thailand 156

104. C/iaetoderma spinosissimus, lateral view ofentire specimen, 33.0 mm SL, Malaya 157

105. Peryagorspi/osomus, lateral view of entire specimen, 80.0 mm SL, Hawaii 158

106. i?udanusmmuf us, lateral view of entire specimen, 17.9 mm SL, Borneo 159

107. Amansesscopas, lateral view of entire specimen, 167 mm SL, Japan 160

108. /lcanf/ra/uferesspi7ome/anurus, lateral view of entire specimen, 63.7 mm SL, Australia 161

109. A/uiera/ieude/otii, lateral view of entire specimen, 107 mm SL, Florida 162

110. PsiVocep/ja/us 6ar6atus, lateral view of entire specimen, 137 mm SL, Singapore 162

111. Brac/ia/uferestrossu/us, lateral view of entire specimen, 55.5 mm SL, Australia 163

112. Para/uterespnonurus, lateral view of entire specimen, 46.4 mm SL, Seychelles 164

113. Oivmomicani/ius/ongirostns, lateral view of entire specimen, 34.2 mm SL, Seychelles 165

114. Pseuda/uteresmjstcornjs, lateral view of entire specimen, 108 mm SL, Philippines 166

115. Sfep/iano/epis/iispidus, lateral view of head, 50.4 mm SL, Florida 168

116. ParamonacoMt/iuscrypiodon, lateral view of head, 68.5 mm SL, Thailand 168

117. Pervagor spilosomus, lateral view of head, 80.0 mm SL, Hawaii 169

118. Amanses scopas, lateral view of head, 167 mm SL, Saipan 169

119. A/ufera /leude/ofii, lateral view of head, 107 mm SL, Florida 170

120. Psilocephalus barbatus, lateral view of head, 137 mm SL, Singapore 170

121 Brachaluteres trossulus, lateral view of head, 55.5 mm SL, Australia 171

122. OAfymo/iacant/jus fofigirosfm, lateral view of head, 26.5 mm SL, Seychelles 171

123. Pseudaluteres nasicornis, lateral view of head, 108 mm SL, Philippines 172

124. Dorsal views of skulls of: A, Pseudaluteres nasicornis. 108 mm SL, Philippines, highly specialized; B,
Paraluteres prionurus, 46.4 mm SL, Seychelles, moderately specialized; C, Chaetoderma spinosissimus,

33.0 mm SL, Malaya, relatively generalized 174

125. A, Paraluteres prionrus, ventral view of skull, 46.4 mm SL, Seychelles; B, Pervagor spilosomus, ventral
view of skull, ca. 65 mm SL, Hawaii; C, Psilocephalus barbatus, dorsal and ventral views of skull, 137

mm SL, Singapore 175

126. Dorsal view of branchial arches and lateral view of hyoid arch and urohyal of: Alutera heudelotii, 114

mm SL, Florida, and Psilocephalus barbatus, 139 mm SL, Singapore 176

127. Medial views of upper and lower jaws of: A, Alutera heudelotii, 114 mm SL, Florida; B, Monacanthus
ciliatus, 51.3 mm SL, Colombia; C, Psilocephalus barbatus, 137 mm SL, Singapore; D, Amanses scopas,

167 mm SL, Saipan;E, Acanthaluteres spilomelanurus, 63.1 mmSL, Australia 177

128. Hypothesized phylogenetic relationships of the genera of Monacanthidae 178

129. Typical body form in the Recent Aracanidae: Aracana aurita 185

Range of diversity in body form and external features of representative aracanid genera:

130. Strophiurichthys inermis and S. robustus 186

131. Aracana aurita 187

132. Anoplocapros amygdaloides 187

133. Capropygia unistriata 187

134. Caprichthys gymnura 187

Kentrocapros aculeatus:

135. lateral view of entire specimen, 90.0 mm SL, Japan 196

136. lateral view of head, 90.0 mm SL, Japan 197

137. dorsal and ventral views of skull, 90.7 mm SL, Japan 197

138. posterior views of skull and orbit, 90.7 mm SL, Japan 198

139. dorsal view of branchial arches, lateral view of hyoid arch and urohyal, 90.7 mm SL, Japan 199

140. Strop/iiunc/jf/j.vs ro6usf us, lateral view of entire specimen, 150 mm SL, Australia 200

141. Aracana aunta, lateral view of entire specimen, 87.3 mm SL, Tasmania 201

142. Aracana omafa, lateral view of entire specimen, 70.5 mm SL, Australia 202

143. Lateral views of heads of typical aracanids: Strophiurichthys robustus, 150 mm SL, Australia, and
Aracana aurita, 87.3 mm SL, Tasmania 203

144. Ventral views of the two types of configuration of the ventral surface of the vertically expanded anterior
portion of the parasphenoid: A, Caprichthys gymnura, 74.1 mm SL, Australia; B, AracarM aurita, 87.3

mm SL, Tasmania 203

145. Proaracana dubia: lateral view of entire specimen; composite based on three specimens, including the
holotype, 31.4-54.5 mm SL, all specimens from the Eocene of Monte Bolca, Italy 204

146. Hypothesized phylogenetic relationships ofthe genera of Aracanidae 205

147. Typical body form in the Recent Ostraciidae: Ostracion lentiginosum 210

148. Acant/iostracionQuadricomis: external features 211

Acanthostracion quadricornis, composite based on several specimens, 58.2-350 mm SL, Gulf of Mexico

and Caribbean:

149. lateral view of entire specimen 221

150. lateral view of head 222

151. dorsal and ventral views of skull 223

152. posterior views of skull and orbit 224

153. lateral and medial views of palatopteroquadrate region 224

154. Acanthostracion quadricornis, 77.7 mm SL, Florida: lateral view of occipital region of skull and the fused

firstfiveabdominal vertebrae; posterior view of fused first five abdominal vertebrae 224

Acanthostracion quadricornis, composite based on several specimens, 58.2-350 mm SL, Gulf of Mexico

and Caribbean:

155. medial and lateral views of pectoral girdle 225

156. ventral views of vertebral column from the fused first five abdominal vertebrae to the fourth caudal

vertebra and of the basal pterygiophores of the anal fin 225

157. dorsal view of branchial arches; lateral view of hyoid arch; lateral and dorsal views of urohyal 226

158. lateral view of caudal fin supporting structures 226

External features of other representative ostraciid genera:

159. A, Rhinesomus bicaudalis, B and C, R. triqueter 227

160. Lactophrys trigonus 227

161. Lactoria fornasinii 227

162. Tetrosomus gibbosus ■ 227

163. Ostracion lentiginosum 227

164. Rhynchostracion nasus and R. rhinorhynchus 227

Number of teeth in upper and lower jaws in relation to standard length and the relatively negligible in-
crease in number of teeth with increasing standard length at sizes greater than about 40 mm SL:

165. Lactophrys trigonus, Rhinesomus triqueter, R. bicaudalis 228

166. Ostracion tuberculatus, O. lentiginosum, Acanthostracion quadricornis, A. polygonius 229

167. /?/imesomustn(7ue(er, lateral view of entire specimen, 134 mm SL, Colombia 230

168. Lactona cornuta, lateral view of entire specimen, 104 mm SL, Philippines 231

169. Tetrosomus concatenatus, lateral view of entire specimen, 43.4 mm SL, Thailand 232

170. Os(rac(on(Li6ercu/a(us, lateral view of entire specimen, 54.5 mm SL, Aldabra Atoll 233

171. Rhinesomus triqueter, lateral view of head, 134 mm SL, Colombia 234

172. Ostracion fu6ercu/atus, lateral view of head, 54.5 mm SL, Aldabra Atoll 234

173. Tetrosomus concaferiatus, lateral view of head, 43.4 mm SL, Thailand 235

174. Lactoria cornuta, ventral view of skull, 88.2 mm SL, Philippines 235

175. Lactoria fornasinii, posterior view of orbit, 65.2 mm SL, Hawaii 236

176. Rhynchostracion rhinorhynchus: ventral and dorsal views of skull, 88.2 mm SL, Java 237

177. Lactoria cornuta, dorsal view of branchial arches; lateral view of hyoid arch and urohyal; 88.2 mm SL,
Philippines 238

178. Hypothesized phylogenetic relationships ofthe genera of Ostraciidae 239

179. Eolactoria sorbinii: lateral view of holotype, 15.5 mm SL, Eocene of Monte Bolca, Italy 242

180. Body form in the only known Recent species of Triodontidae: Triodon macropterus 243

Triodon macropterus:

181. external features 244

182. lateral view of entire specimen, 391 mm SL, Philippines 255

183. lateral view of head, 391 mm SL, Philippines 256

184. dorsalandlateral views of skull, 391 mm SL, Philippines 256

185. posterior views ofskull and orbit, 391 mm SL, Philippines 257

186. dorsal view of branchial arches; lateral view of hyoid arch; dorsal and lateral views of urohyal; 391 mm

SL, Philippines 257

187. lateral view of pectoral girdle, 391 mm SL, Philippines 258

188. lateral view of scapula and first actinost in representatives of all superfamilies 258

189. lateral views of alternate pleural ribs and their epipleurals; dorsal view of lower jaw; 391 mm SL, Phil-

ippines 259

190. lateral and dorsal views of rudimentary spiny dorsal fin and its basal pterygiophores, 463 mm SL,

Japan 259

191. ventral, dorsal, and lateral views of pelvis, 391 mm SL, Philippines 259

192. lateralviewofcaudalfinsupportingstructures, 391 mm SL, Philippines 260

193. Range of diversity in body form in the Recent Tetraodontidae: A, Lagocephalus spadiceus, B, Xenop-
terusnaritus;C, Amblyrhynchotes piosae; andD, Canthigaster rostrata 264

194. Lagocephalus laeuigatus, with L. lunaris and L. sc/erafus for comparison: external features 266

Lagocephalus laevigatas, composite based on several specimens, 61.4-166 mm SL, Gulf of Mexico:

195. lateral view of entire specimen 275

196. lateral view of head 276

197. dorsal and ventral views of skull 276

198. posterior view of skull; lateral and posterior views of first abdominal vertebra; posterior view of

orbit 277

199. dorsal view of branchial arches and lateral and medial views of hyoid arch 278

200. lateral view of pectoral girdle 278

201. lateral view of caudal fin supporting structures 279

202. Canthigasterrostrata, with C. ualentini and two renditions of C. amboinensis: external features 279

Canthigaster rostrata, composite based on several specimens, 55.2-96.5 mm SL, Texas:

203. lateral view of entire specimen 286

204. lateral view of head 287

205. dorsal and ventral views of skull 288

206. posterior view of skull; lateral and posterior views of first abdominal vertebra; posterior view of or-

bit 288

207. dorsal view of branchial arches; lateral and medial views of hyoid arch 289

208. lateral view of pectoral girdle 290

209. lateral view of caudal fin supporting structures 290

210. External features of other representative tetraodontid genera: A, Sphoeroides nephelus; B, S. testudi-

neus; C, S. dorsalis; D, S. lobatus; E, S. pachygaster 291

211. Nasal organs and olfactory lamellae of representative species of Sphoeroides and Lagocephalus: A,
Sphoeroides trichocephalus; B, S. spengleri; C, S. maculatus; D, Lagocephalus spadiceus; E, L. lago-
cephalus; F, L. inermis 292

External features of other representative tetraodontid genera:

212. Guentheridia formosa 292

213. Colomesus psittacus 292

214. A, Amblyrhynchotes piosae; B, A. honckenii; C, A. richei 293

215. Torquigener pleurostictus and T. hamiltoni 293

216. Fugu vermicularis 293

217. Arothron armilla and A. immaculatus 294

218. A, Tetraodon miurus; B, T. lineatus; C, T. kretamensis; and D, T. schoutedeni 294

219. Ephippionguttifer of decreasing sizes (325, 232, and 101 mm SL) 295

220. Photographic detail of interdigitated scale plates from lower side of body of Ephippion guttifer, 325

mm SL 295

External features of other representative tetraodontid genera:

221. Chelonodon patoca and C. fluviatilis 295

222. Monotreta palembangensis and M. leiurus 296

223. Chonerhinos modestus 296

224. Xenopterus naritus 296

225. Carinotetraodon lorteti 297

226. Sphoeroides macu/atus, lateral viewof entire specimen, 97.5 mm SL, Virginia 299

227. Sp/ioeroides pachygaster, lateral view of entire specimen, 80.6 mm SL, Nigeria 300

228. Guent/iendia/ormosa, lateral view of entire specimen, 175 mm SL, Panama (Pacific) 300

229. Co/omesuspsjUacus, lateral view of entire specimen, 178 mm SL, Surinam 301

230. Amftiyr/iync/ioteshoncfeemi, lateral view of entire specimen, 70.7 mm SL, India 301

231. Amb/yr/iynchoiesric/iei, lateral view of entire specimen, 59.4 mm SL, New Zealand 302

232. Amb/yr/iynchotes piosae, lateral viewof entire specimen, 33.8 mm SL, Australia 302

233. Fu^u c/irysops, lateral view of entire specimen, 98.6 mm SL, Japan 303

234. Fugu o6/orj^us, lateral view of entire specimen, 46.2 mm SL, India 303

235. i4ro(/iroMs«e//atus, lateral view of entire specimen, 292 mm SL, India 304

236. Arot/iron armi/te, lateral view of entire specimen, 61.3 mm SL, Australia 305

237. Tetraodofimbu, lateral view of entire specimen, 47.7 mm SL, Congo 306

238. Ephippion guttifer, lateral view of entire specimen, 108 mm SL, Guinea 306

239. Che/onodon/ZuDJah/is, lateral view of entire specimen, 84.7 mm SL, Thailand 307

240. Monotreta leiurus, lateral view of entire specimen, 61.5 mm SL, Thailand 308

241. Choner/iinos modestus, lateral view of entire specimen, 31.0 mm SL, Borneo 308

242. Xenopterus mzn'tus, lateral view of entire specimen, 143 mm SL, Bay of Bengal 309

243. Carmotetraodon/orteti, lateral view of entire specimen, 35.9 mm SL, locality unknown 310

244. Cant/iigaster am boinerasi's, lateral view of entire specimen, 80.3 mm SL, Hawaii 311

245. Sphoeroides maculatus, lateral view of head, 97.5 mm SL, Virginia 313

246. Co/omesuspsiffacus, lateral view of head, 178 mm SL, Surinam 314

247. /lm6/yr/ivnc/T0fesric/iej, lateral view of head, 59.4 mm SL, New Zealand 314

248. Fugu chrysops, lateral view of head, 98.6 mm SL, Japan 315

249. Fugu oblongus, lateral view of head, 46.2 mm SL, India 315

250. Arothron stellatus, lateral view of head, 292 mm SL, India 316

251. Tetraodon mbu, lateral view of head, 47.7 mm SL, Congo 316

252. C/ie/onodon/Zuuiafi'/is, lateral view of head, 84.7 mm SL, Thailand 317

253. Monotreta leiurus, lateral view of head, 61.5 mm SL, Thailand 317

254. C/ioner/imosmodestus, lateral view of head, 31.0 mm SL, Borneo 318

2.55. Xenopferus aarif us, lateral view of head, 143 mm SL, Bay of Bengal 318

256. Carmofetraodofi /ortett, lateral view of head, 35.9 mm SL, locality unknown 319

Dorsal views of skulls:

257. Sphoeroides maculatus, 12.4-201 mm SL, New Jersey and Virginia 320

258. A, Sphoeroides dorsalis, 155 mm SL, Florida; B, S. nephelus, 128 mm SL, Bahamas; C, S. lobatus,

67.5 mm SL, Panama (Pacific) 321

259. A, Sphoeroides angusticeps, 172 mm SL, Galapagos; B, S. spengleri, 92.8 mm SL, Nicaragua; C, S.

maculatus, 179 mm SL, Virginia 322

260. A, Sphoeroides greeleyi, 60.8 mm SL, locality unknown; B, S. trichocephalus, 57.1 mm SL, Panama

(Pacific); C, S. pachygaster, 117 mm SL, Mozambique 323

261. Sphoeroides testudineus, 68.5 mm SL, Venezuela, and S. annulatus, 174 mm SL, locality unknown . . . 324

262. Guentheridia formosa, 175 mm SL, Panama (Pacific), and Colomesus psittacus, 179 mm SL, Suri-

nam 325

263. A, Lagocephalus inermis, 52.2 mm SL, Bay of Bengal; B, L. scleratus, 80.9 mm SL, Philippines; C,

L. lunaris, 62.3 mm SL, Bay of Bengal 326

264. Lagocephalus spadiceus, 98.2 mm SL, Mozambique, and L. lagocephalus, 214 mm SL, Malpelo Is-

lands 327

265. /lm6/yr/iync/iotes/ioncfeem7, 97.4 mm SL, Mozambique, and A. n'c/iei, 59.4 mm SL, New Zealand .... 328

266. Torquigener pleurogramma, 113 mm SL, Australia, and Amblyrhynchotes piosae, 33.8 mm SL, Aus-

tralia 329

267. Torqujgenerp/eurosfjctui, 87.1 mm SL, Australia, and Fugu c/irysops. 98.6 mm SL, Japan 330

268. Fugu oblongus, 46.2 mm SL, India, and F. rubripes, 151 mm SL, Japan 331

269. Aro</irofjste//a(us, ca. 420 mm SL, Seychelles, and A. armi7/a, 61 .3 mm SL, Australia 332

270. Tetraodon lineatus, 222 mm SL, French Equatorial Africa, and T. mbu, 47.7 mm SL, Congo 333

271. Ephippionguttifer, 101 mm SL, Guinea 334

272. Chelonodon patoca, 70.6 mm SL, New Guinea, and C. fluviatilis, 84.7 mm SL, Thailand 335

273. Monotreta gularis, 46.8 mm SL, Burma, and M. leiurus, 61.5 mm SL, Thailand 336

274. C/ioner/imosmodestus, 54.5 mm SL, Borneo, and Xenopf ems narjtus, 143 mm SL, Bay of Bengal .... 337

275. Carinotetraodon lorteti, 33.1 mm SL, locality unknown 338

276. Ventral views of skulls of: Sphoeroides maculatus, 179 mm SL, Virginia, and Arothron stellatus, ca.

420 mm SL, Seychelles 339

277. Dorsal views of branchial arches and lateral views of hyoid arches of: Xenopterus naritus, 108 mm SL,
Bayof Bengal, and Arot^onmgropunctatus, 56.8 mm SL, Solomon Islands 339

278. Hypothesized phylogenetic relationships ofthe genera of Tetraodontidae 341

279. Range ofdiversity in body form in the Diodontidae: Diodonholocanthus and Chilomycterusschoepfi . . 343

280. Diodon holocanthus, with Chilomycterus schoepfi for comparison: external features 344

Diodon holocanthus, composite based on several specimens, 12.3-124 mm SL, Gulf of Mexico and Carib-
bean:

281. lateral view of entire specimen 352

282. lateral view of head 353

283. dorsal and ventral views of skull 354

284. posterior view of skull; lateral and posterior views of first abdominal vertebra; posterior view of orbit;

dorsal view of lower jaw 355

285. dorsal view of branchial arches; lateral and medial views of hyoid arch 356

286. lateral view of pectoral girdle 357

287. lateral view of caudal fin supporting structures 357

288. Lateral view of entire specimen of another representative diodontid genus: Chilomycterus schoepfi,

60.1 mm SL, Louisiana 359

Dorsal views of skulls:

289. Chilomycterus schoepfi, 60.1 mm SL, Louisiana, and C. affinis, 310 mm SL, Galapagos 360

290. Diodon jacuUferus, 53A mmSh, AuatialiA, and Chilomycterus orbicularis, 12.9 mmSL, Somalia .... 360

291. Dicotylichthys punctulatus, 226 mm SL, Australia 361

292. Chilomycterus schoepfi: dorsal view of branchial arches and lateral view of hyoid arch, 168 mm SL,
Florida 361

293. Nasal apparatuses of representative diodontids: A, Chilomycterus mauretanicus; B, C. orbicularis; C,

C. schoepfi; D, C. antennatus; E, C. reticulatus; F, C. atinga; G, Diodon hystrix; H, D. holocanthus; I,

D. jaculiferus; J, Dicotylichthys punctulatus; K, D. nicthemerus 361

294. Ventral views of upper jaws to show trituration tooth plates: A, Chilomycterus orbicularis, 77.0 mm SL,
Somalia, and 158 mm SL, Philippines; B, C. schoepfi, 62.5 mm SL, Texas, and 168 mm SL, Florida .... 362

295. Chilomycterus schoepfi: lateral view of upper jaw; cross section through middle of one of the individual
flattened disklike trituration teeth; outline of about one-half of a trituration tooth; 168 mm SL, Flor-
ida 362

Cross sections of upper and lower jaws just to one side of the midline:

296. Diodon hystrix, 88.3 mm SL, 228 mm SL, 505 mm SL 363

297. Diodon holocanthus, 63.4 mm SL, 122 mm SL, 183 mm SL, 375 mm SL 364

298. Chilomycterus schoepfi, 97.6 mm SL, 166 mm SL 365

299. Chilomycterus antillarum, 116 mm SL 365

300. Chilomycterus atinga, 173 mm SL 365

301. Chilomycterus mauretanicus, 92.2 mm SL 366

302. Chart showing the increased number of trituration plates in the upper jaw in Diodon holocanthus, D.
hystrix, Chilomycterus schoepfi, C. mauretanicus, C. atinga, C. antillarum, and C. reticulatus 366

303. Hypothesized phylogenetic relationships of the genera of Diodontidae 367

304. Rangeofdiversity in body fnrmintheMolidae: Mola mola and Ramania laevis 368

305. Mola mola: external features 369

Mola mola, composite based on two specimens, 306 and 310 mm SL, California:

306. lateral view of entire speciinen 378

307. lateral view of head 379

308. dorsal and ventral views of skull 380

309. posterior view of skull; lateral and posterior views of first abdominal vertebra; posterior view of orbit;

dorsal viewof lower jaw 381

310. dorsal viewof branchial arches; lateral viewof hyoid arch 382

311. lateral view of pectoral girdle 382

312. lateral view of pseudocaudal fin supporting structures 383

External features of other representative molid genera:

313. Masturus lanceolatus 383

314. Romania laevis 383

315. Masf urns /anceo/af us, lateral view of entire specimen, 127 mm SL, Florida 385

316. /Tanzania /aei'is, lateral viewof entire specimen, 65.1 mm SL, Hawaii 386

317. Masturus /anceo/af us, ventral and dorsal views of skull, 127 mm SL, Florida 387

318. Kanzania /aei;js, ventral and dorsal views of skall, 65.1 mm SL, Hawaii 388

319. Masturus lanceolatus, lateral viewof head, 127 mm SL, Florida 388

320. Ramania laevis, lateral viewof head, 65.1 mm SL, Hawaii 388

321. Dorsal views of branchial arches and lateral views of hyoid arches: Ramania laeuis, 66.9 mm SL, Ha-
waii, and Masturus lanceolatus, 127 mm SL, Florida 389

322. Masturus lanceolatus: dorsal viewof lower jaw, 127 mm SL, Florida 389

323. Ramania laevis: dorsal view of lower jaw and ventral view of upper jaw, 493 mm SL, Hawaii; dorsal view

of lower jaw, 65.1 mm SL, Hawaii 389

324. Ramania laevis: diagrammatic representation of the relationships between the basal pterygiophores of
the dorsal, anal, and pseudocaudal fins and the neural and haemal spines of the supporting vertebrae;

65.1 mm SL, Hawaii 390

325. Ramania laevis: scales from upper middle region of body; 493 mm SL, Hawaii 390

326. Hypothesized phylogenetic relationships of the genera of Molidae 392

Tables

1. Characteristics ofthe hyoid and branchial arches in families of Tetraodontiformes 413

2. Total number of vertebrae and vertebral formulas of Recent plectognath fishes examined as radiographed

and as cleared and stained specimens 414

3. Numbers of teeth in the upper and lower jaws of ostracioids 420

Osteology, Phylogeny, and Higher Classification

of the Fishes of the

Order Pleetognathi (Tetraodontiformes)^

JAMES C. TYLER'

ABSTRACT

The osteologj- of over 160 species of fossil and Recent plectognath or tetraodontiform fishes is
described and illustrated in relation to the supposed phylogeny and proposed higher classiflcation
(subfamilial to ordinal levels) of this group of approximately 320 Recent species of primarily tropical
and temperate forms of the Atlantic. Pacific, and Indian oceans. The history of the classification and
of the previous work on the osteology of the order is reviewed, while one new species (^ Acanthopleurus
collettei, Oligocene of Canton Glarus. Switzerland) and one new genus {lEotetraodon, Eocene of
Monte Bolca. Italy) are described. Comparative inclusive and exclusive definitions are given for all
higher categories based on both external and internal anatomical features. The Order Pleetognathi
(Tetraodontiformes) is divided into two suborders, the Sclerodermi or Balistoidei and the Gym-
nodontes or Tetraodontoidei, with a variety of other infraordinal and superfamilial categories, and 10
families, with subfamilial groupings in 4 of the latter.

INTRODUCTION

Preface

The Pleetognathi are widely, but not unanimously,
recognized as one of the major orders or phyletic lines of
teleostean fishes derived of perciform ancestors. The
primary reason that there is any doubt concerning the
naturalness of the Pleetognathi, regardless at what tax-
onomic level it is recognized, is that it is a highly diver-
sified group for which adequate definitions have not yet
been given either for it as a whole or for most of its sub-
divisions. Until there appears and is accepted a con-
sistent and systematic general osteological survey and
comparison of the basic types of organization found
among the fossil and Recent species of plectognaths,
both generalized and highly modified, it will be impos-
sible to state without reasonable refutation that they are
all closely related, and more similar to one another than
to any other group of fishes.

The aims of the present monograph are: first, to
anatomically define, especially with osteologically based
characters, the higher categories in the classification of
the plectognaths proposed here, these categories being
from the subfamilial to ordinal levels; and second, to in-

Contribution No. 77-25M from the Southeast Fisheries Center Miami
Laboratory, Miami, Fla.

'Southeast Fisheries Center Miami Laboratory, National Marine Fish-
eries Service, NOAA, Miami, FL 33149, and Research Associate, Depart-
ment of Ichthyology, American Museum of Natural History, New York,
NY 10024; present address; Office of Marine Mammals and Endangered ,
Species, National Marine Fisheries Service, NOAA, 3300 Whitehaven
Street, N.W., Washington, DC 20235.

terpret the phylogenetic relationships between these
divisions and most genera of the fossil and Recent Plee-
tognathi, dating back to the upper portion of the lower
Eocene, approximately 60 million years ago.

This is done by pointing out, as much by illustrations
as by written descriptions, the many common os-
teological features of numerous representatives of each of
the families of plectognaths, including both relatively
generalized and specialized species when available. The
totality of these characteristics of the fossil and Recent
species demonstrates beyond question the distinc-
tiveness and phylogenetic naturalness of the plectognath
fishes, while at the same time providing the bases for the
detailed comparative diagnoses given here for the tax-
onomic subdivisions of the Pleetognathi.

As classified here, the Order Pleetognathi (or
Tetraodontiformes), comprising today about 320 species
of mostly shallow-water, cireumtropical, and sub-
tropical marine forms, is divided into two suborders
(Sclerodermi or Balistoidei and Gymnodontes or
Tetraodontoidei), a variety of categories from infra-
orders to superfamilies, and into 10 families, in 4 of
which subfamilies are recognized. The Pleetognathi are
much more diversified than the great majority of fish
groups of a comparable number of species. Relative to its
comparatively few species, mostly marine, the Pleetog-
nathi are as highly diversified as that prime example of
diversity among fishes, the speciose freshwater characids
of South America and Africa.

Just as interesting as the plectognath's diversity itself.

is an analysis of the evolutionary mechanism leading to
it. This presents a strikingly clear case of intraordinal
evolution primarily through the processes of reduction,
simplification, and loss of elements, especially of the
bony parts. Reductive tendencies in the evolution of
vertebrates, including many groups of fishes, are not un-
common (Myers 1958), and the reductions that have
taken place in plectognaths are of especial interest only
because of the extreme exaggeration of the tendency
in this order. The reductive trend in the evolution of
plectognaths obviously is not absolute, and certain
morphological units in circumscribed groups show an
opposite tendency, e.g., the stomach and complex mus-
culature of tetraodontoids, the pectoral girdle of ostra-
coids, the scales in nearly all groups except a few tetra-
odontids, etc.

The main point, however, is that the generalized mem-
bers of the order, the triacanthoids, possess only slightly
less than the percoid complement of bones, but many
bones have become reduced or simplified, fused with
others, or entirely lost in what appear to be clear
evolutionary lines derived from these basal plectog-
naths. This is especially evident in the: branchial ap-
paratus; caudal, dorsal, and anal fin supporting struc-
tures; dorsal fin spines; pelvic fin and girdle; jaws and
teeth; and myodome.

The plectognaths are also of biological interest because
of their supposed position at one of the major end lines of
modern teleost radiation and because of their great
diversification in structure, size, behavior, way of life,
and habitat (Tyler 1965c). They range from 22 mm and
30 g to 2 m and 1,000 kg in adult length and weight, from
relatively normal shapes to strangely specialized forms
with long tubular snouts or aborted caudal regions, from
scaleless to heavily armored, from nearly toothless to

equipped with massive crushing beaks, from drably
colored to gaudy, from palatable to poisonous flesh, etc.,
and with comparable contrariety in behavior, habits, and
habitats. They include the spikefishes, triplespines, trig-
gerfishes, filefishes, boxfishes, trunkfishes, pursefishes,
pufferfishes, porcupinefishes, and giant ocean sun-
fishes.

The plectognaths are of some commercial interest
directly, for the dried skins or encasements of the weirder
forms are sold as curios, while a few species are sold for hu-
man consumption, such as tetraodontids, called sea squabs
on the east coast of the United States and fugu in Japan, but
of greater worth indirectly — the young of many plectog-
nath species are important forage for such popular large
oceanic fishes as dolphins, tunas, and billfishes.

The Plectognathi are thought to have been derived
from a percoid ancestry related to the same line which
gave rise to the acanthuroids (the surgeonfishes and their
allies) in the late Cretaceous, but the evidence of this
hypothesized relationship is not yet conclusive (Tyler
1968, 1970c). The osteology of the fossil and Recent
acanthuroids is not well known, but with the osteologj' of
the plectognaths being made better known here, it will be
easier to search among the acanthuroids for a group
which shows a preplectognath type of organization and
which may be related to the speculated ancestral stock
common to the two groups.

Because the plectognaths are often of exotic form and
occur in European waters, even though their center of
diversification is the Indo-Pacific, a few species were
described by early naturalists from Aristotle and Pliny to
Linnaeus. But some species still remain to be described,
many others are inadequately known, and much con-
fusion remains about the systematics of the group, even
at the familial level.

Methods

The majority of the specimens of the Recent species
studied for this monograph were prepared by potassium
hydroxide clearing, alizarin staining, and glycerin preser-
vation, prior to the advent of techniques using trypsin. A
few specimens were prepared by maceration as dry or
alcohol wet whole skeletons, or parts thereof, while
limited dissections were made on alcohol - preserved
whole materials. For fossil species, only superficial sur-
face preparation with chipping and weak acid washes
was employed.

A list of the species examined, along with the num-
bers of specimens and their sizes, general localities, and
catalogue (or other identifying) numbers of their reposi-
tories, appears toward the end of this work.

Length of specimens is always standard length (SL),
unless otherwise noted, taken from the anteriormost
middle point of the upper jaw (often from the tip of the
exposed upper teeth) to the middle of the line of flexure
at the caudal fin base, and measured and recorded with a
needlepoint dial caliper to the nearest 10th of a millimeter
(mm), but with lengths given here of over 100 mm

rounded off to the nearest mm (0.5 and above to the next
highest integer).

The survey of the osteology of the order was hindered
by the poor showing of internal features in several fossil
forms, by the relative unavailability of specimens for
clearing and staining of about one-fifth of the described
species currently considered valid, by a lack of time and
purpose to describe and illustrate each available species
(especially in genera with large numbers of osteologically
similar forms), and by the demands of space conser-
vation and publication costs. It is felt, however, that the
species selected for osteological inclusion (167 studied,
and 115 illustrated, for one skeletal region or another, of
the approximately 320 Recent species, and nearly all of
the fossil forms) adequately cover the anatomical diver-
sity of the group, and almost always cover the entire
range of external and osteological differences of the more
highly modified as well as more normal represen-
tatives of each family. I find it difficult to believe that
there is a missing link or living relic among those Recent
and fossil plectognaths not examined for this work which

would materially change the concepts of the phylogeny
and classification presented here.

For each of the 10 families treated here, one relatively
generalized species is described in detail and extensively
illustrated, with other species as appropriate described
and illustrated less fully, except in the Tetraodontidae,
with two species representative of the two subfamilies
that have often been considered as separate families
being given the fuller descriptive and illustrative treat-
ment to aid in comparisons of their osteological distinc-
tiveness. Line drawings show the most diagnostically
important or anatomically interesting external features
of representatives of each group.

Lateral view illustrations of the entire skeleton are
usually supplemented by other views of various parts of
the skeleton showing features of particular phylogenetic
or morphological interest.

Statements concerning the numbers of such serial
elements as basal pterygiophores, epipleurals, neural
and haemal spines, epurals, and hypurals are based
mostly on cleared and stained specimens and not on
radiographs, because all of the members of such series of
these elements usually are not clearly defined in radio-
graphs due to the often thickened nature of the scales.
Vertebral counts are based on both cleared and stained
specimens and radiographs, except for those of diodon-
tids in which radiographs of the vertebral column are
usually undeciferable because of the massive roots of the
skin spines.

The lateral view illustrations of entire skeletons and of
most of the skeletal parts were prepared by photo-
graphing the skeleton or its parts submerged in a gly-
cerin solution. The 35 mm negatives were projected and
outline drawings showing as much detail as possible
made, of about 40 to 50 cm length in the case of entire
skeletons. Details were filled in by examination through
a dissecting microscope of the photographed skeleton or
part. The completed pencil drawings were transferred by
ink tracing on paper, with contours and depth approxi-
mately shown by varying intensities of stippling. While
the pencil drawings and ink tracings are all by the
author, the stippling is by a combination of the author
and his illustrators listed in the Acknowledgments.
Photographed entire specimens and/or one or more ad-
ditional specimens of the same species were partially
disarticulated in order to prepare, if necessary, drawings
of selected parts or views of regions of diagnostic or mor-
phological interest, as well as generally to examine the
skeleton and its parts in greater detail. A few of the sim-
pler supplemental illustrations were made with the aid of
proportional dividers rather than photography. Some of
the drawings are composite, based on several specimens,
and are so indicated.

Fin rays usually are shown only partially and diagram-
matically as an outline of the positions of the bases of the
rays, with, for the caudal fin, the unbranched rays in-
dicated by solid bases and the branched rays by open
bases. Only the pelvic fin rays and uppermost pectoral
fin ray routinely were fully drawn. Epipleurals, especial-
ly if slender, usually are shown in solid black rather than

stipple. The finer details of the surface sculpturing of the
bones usually is not shown, for, even though it is beauti-
ful, it is exceedingly time consuming to accurately por-
tray and normally is of little value to an understanding of
the relationships of the species in question.

The otoliths of the species treated here are not illus-
trated, for they only rarely were intact and uncorroded
enough in the cleared and stained study material to be
useful.

The osteological descriptions of the species treated in
greatest detail were written from microscopic reexamina-
tion of the skeletons or parts thereof, and corrections
were made in the drawings whenever necessary. Correc-
tions were frequent, for nothing brings to light the errors
in illustrations more quickly than having to verbally
describe the shapes of bones and, particularly, their ar-
ticulations.

The osteological terminology used here is conserva-
tive, and hopefully that most readily understandable by
the majority of ichthyologists. For the most part, it is
that of Starks (1901), although such terms as ptero-
sphenoid, ectopterygoid, first pharyngobranchial,
cleithrum, and scapula are substituted, respectively, for
Starks alisphenoid, pterygoid, suspensory pharyngeal,
clavicle, and hypocoracoid. Certain names, such as
prefrontal rather than lateral ethmoid or dermethmoid,
are used in order to be deliberately topological and to
avoid implications of the dermal versus endochondral
origin of the bone.

Subsequent to the beginning of this monograph, and of
the pervasive decisions on the names of bones to be
employed in the text and illustrations, several
researchers have provided more precise and homologous-
ly correct terminologies for certain skeletal regions than
used here. It has not been possible for reasons of prac-
ticality to make the numerous desirable changes, es-
pecially in labeling of the illustrations, that would be
necessitated by incorporating these changing ter-
minologies into the present work. Examples of the more
modem and accurate names for certain skeletal regions,
such as the branchial arches and lower jaw, are found in
the exemplary publications of Weitzman (e.g., 1967 et
seq.) and Nelson (e.g., 1969 et seq.).

In describing the types of articulations between bones,
purely descriptive phrases often are used, for the tech-
nical terminology is not standardized and it is sometimes
more precise than my observations. Whenever an
amount of cartilage observable under the dissecting
microscope (ca. 30x ) is seen to intervene between the ar-
ticular faces of two bones, these bones are said to ar-
ticulate through cartilage. When bones have relatively
smooth articular faces held to one another without the
intervention of cartilage, they are said to articulate by
fibrous tissue. A large number of bones £ire said to ar-
ticulate with one another by interdigitation, and this
simply means that the closely apposed surfaces of two
bones have their articulation strengthened by delicate to
coarse emarginations of one bone fitting into similar in-
dentations of the other bone. At its fullest development,
this interdigitation can be distinguished from fusion only

by attempting to disarticulate the bones after long
maceration. When a bone is said to be cartilage filled
along a particular edge, this refers to the presence of car-
tilaginous material extending into the otherwise ossified
substance of the bone so that an upper and a lower layer
of bone, separated by cartilage, are distinguishable at the
edge of the bone. Such a condition, of course, is only pos-
sible in endochondral bones. As a general rule, the
amount of interdigitation between bones increases with
increasing specimen size, while at the same time the
amount of cartilage at the edges of endochondral bones
decreases.

The suffix oideo is adopted for the names of infraor-
dinal categories.

Throughout the text, vernacular versions of the
classically based and more formal superfamilial,

familial, and subfamilial names, ending respectively in
oidea, idae, and inae, are employed, with the last two
vowels dropped for adjectival usage and with these
replaced by an "s" for nominative usage, so that super-
families end in -oid or -oids (triacanthoid, balistoid, os-
tracioid, triodontoid, tetraodontoid, moloid), families in
-id or -ids (triacanthodid, triacanthid, balistid, mona-
canthid, aracanid, ostraciid, triodontid, tetraodontid,
diodontid, molid), and subfamilies in -in or -ins (spina-
canthin, eoplectin, hollardiin, triacanthodin, protacan-
thodin, cryptobalistin, triacanthin, ostraciin, lactophry-
sin, tetraodontin, canthigasterin). The names of two of
the familial categories, the Triacanthodidae and Triacan-
thidae, unfortunately are rather similar, and the distinc-
tion between such terms as triacanthoid, triacanthodid,
and triacanthid is a necessary mental excercise.

PROLOGUE

Synopsis of the Phylogeny of the Plectognathi

The anatomical data on the fossil and Recent Plectog-
nathi presented in this monograph indicate that the
family Triacanthodidae contains the most generalized
fossil and Recent species of the order, and that these
basal plectognaths gave rise on the one hand to the line
leading to the triacanthids, their nearest relatives, and
through them to the balistoids and ostracioids, while, on
the other hand, and with even greater modification away
from the ancestral level of organization, the triacan-
thodids gave rise to the triodontids, and through them to
the tetraodontoids and molids. The triacanthodids and
their anatomically and phylogenetically closest deriva-
tive groups, the triacanthids, balistoids, and ostracioids,
are here considered to represent the suborder Scleroder-
mi or Balistoidei, while the other major line of plectog-
nath radiation, that has diverged even further from the
ancestral triacanthodid type, comprising the triodon-
tids, tetraodontoids and molids, is here considered to
represent the suborder Gymnodontes or Tetraodon-
toidei.

Two fossil forms of the basal triacanthoids from the
Eocene are thought to be especially pertinent to the phy-
logeny of the plectognaths, for Protacanthodes forms a
strong link between the triacanthodids and the triacan-
thids and thus to the other derived scleroderms
(balistoids and ostracioids), while Eoplectus is nearly a
perfect intermediary between the triacanthodids and the
triodontids, the most generalized of the gymnodonts, and
thus to the derived tetraodontoids and molids.

More particularly, of the four subfamilies of Triacan-
thodidae recognized here, it is obvious that the Triacan-
thidae are derived from the HoUardiinae rather than
from the Triacanthodinae or the fossil Spinacanthinae
and Eoplectinae because of similarities in the shapes and
positions of the bones in the rear of the skull and of the
shape of the pelvis and because there is every reason to
believe that the Spinacanthinae were evolutionary dead

ends without issue surviving today and that the Eoplec-
tinae were ancestral to the triodontids and the other
gymnodonts.

The evolution of the Triacanthidae from the Triacan-
thodidae was probably mediated through a form like
Protacanthodes, the Eocene representative of the most
generalized subfamily of triacanthids. Of the other two
subfamilies of triacanthids, the triacanthins include all
of the Recent species and obviously evolved from such
Oligocene forms as Acanthopleurus, while the aberrant
cryptobalistins of the Oligocene were, like the
spinacanthin triacanthodids, failed evolutionary ex-
periments that did not give rise to forms living today.
The balistoids and ostracioids are reasoned to share a
common ancestral line to the triacanthids in the Eocene,
and it is unquestioned that the balistids gave rise to the
monacanthids and that the aracanids are ancestral to the
ostraciids. Two subfamilial lines of evolution are evident
in the ostraciids.

The recent discovery of the eoplectin triacanthodids,
with a well-developed spiny dorsal fin, the best
developed pelvic fin among plectognaths, and a generally
triacanthodid appearance, but with a fully developed
gymnodontlike crushing beak, tends to confirm that the
triacanthodids gave rise to a line of triodontidlike forms
which in turn were ancestral to the two main radiations
of the gymnodonts, the molids on the one hand and the
tetraodontids and diodontids on the other hand. The
tetraodontids show two subfamilial phyletic lines, the
tetraodontins for the vast majority of diversified forms,
related through Carinotetraodon to the canthigasterins.

In summary, it seems clear, especially in the light of
our present knowledge of Protacanthodes and Eoplec-
tus, as discussed by Tyler (1973b), that the basic diver-
sification of the plectognaths was a relatively rapid event
of the lower Eocene. In the upper portion of the lower
Eocene of Monte Bolca, Italy, alone, there are represen-

"lis

Hill

£ ^ •§ a ■« _^ £ - « §
II I S I ^ J ll S

7. c 's =■ 'i £ c 'E « -5,

tatives of most of the families of plectognaths (and nearly
modern acanthurids as well). The triacanthodids are
represented there by the Spinacanthinae, which
probably became extinct, and the Eoplectinae, which are
ancestral to the gymnodont families. The triacanthids,
derived from the triacanthodids, are represented by the
Protacanthodinae, in many ways intermediate between
the triacanthodids and triacanthids. The aracanids are
represented by Proaracana, not much different from Re-
cent genera, while the ostraciids, derived from the
aracanids, are also found in the Eocene beds of Monte
Bolca in the form of Eolactoria, again not much dif-
ferent from Recent genera. Among these scleroderms.

only the balistids and the derivative monacanthids are
not found in the Eocene. The balistids, of triacanthid
derivation, are first known from the Oligocene, with
genera rather similar to those alive today, while the
monacanthids as yet have no known fossil record. The
fossil record of the gymnodonts is not as impressive for
overall antiquity, quantity, or quality as that of the
scleroderms, with only the diodontids and, probably, the
tetraodontids known from the Eocene, again from Monte

ll;-s-s

■3 i ■§ « S 2 - £
.1 « X fi E- -S 5;

Bolca. Triodontids have been described from the Eocene
of Africa and Europe (not including Monte Bolca), but
only on the basis of jaws alone, with the dentaries fused
into a single piece and the premaxillaries separate. The
teeth in these jaws are small rounded units such as are
found in Triodon and diodontids, but this is scarcely suf-
ficient evidence of a fish of truly 7>iodon-like general
configuration in the Eocene. Molids are first known from
the Miocene.

Synopsis of the Higher Classification of the Plectognathi

Order Plectognathi
Suborder Balistoidei (Sclerodermi)
Infraorder Triacanthoideo
Superfamily Triacanthoidea
Family Triacanthodidae: spikefishes
Subfamily Spinacanthinae (Eocene)
Subfamily Eoplectinae (Eocene)
Subfamily HoUardiinae (Recent)
Subfamily Triacanthodinae (Recent)
Family Triacanthidae: triplespines
Subfamily Protacanthodinae (Eocene)
Subfamily Cryptobalistinae (Oligocene)
Subfamily Triacanthinae (Oligocene to Recent)
Infraorder Balistoideo
Superfamily Balistoidea

Family Balistidae (Oligocene to Recent): trigger-
fishes
Family Monacanthidae (Recent): filefishes
Superfamily Ostracioidea
Family Aracanidae (Eocene to Recent): boxfishes
Family Ostraciidae (Eocene to Recent): trunk-
fishes
Subfamily Ostraciinae
Subfamily Lactophrysinae
Suborder Tetraodontoidei (Gymnodontes)
Infraorder Triodontoideo
Superfamily Triodontoidea
Family Triodontidae (Eocene to Recent): pursefishes
Infraorder Tetraodontoideo
Superfamily Tetraodontoidea
Family Tetraodontidae (Eocene to Recent): puffer-
fishes

Subfamily Tetraodontinae
Subfamily Canthigasterinae

Family Diodontidae (Eocene to Recent): porcupine-
fishes

Superfamily Moloidea

Family Molidae (Miocene to Recent): giant ocean
sunfishes.

Alternative arrangements of subordinal and lower
groupings are obviously possible and perhaps equally
appealing to some. One of these would be to recognize the
triacanthodids as subordinally distinct from both of the
two major lineages to which they gave rise, a suborder
Sclerodermi (Balistoidei), in this case comprising the
triacanthids, balistoids, and ostracioids but minus the
triacanthodids, and a suborder Gymnodontes
(Tetraodontoidei), with the triodontids, tetraodontoids,
and molids. Another would be to recognize both the
triacanthodids and triacanthids as subordinally distinct
from both the remainder of the suborder Sclerodermi (in
this case only the balistoids and ostracioids) and the sub-
order Gymnodontes (as above). Or, as Winterbottom
(1974) has so ably suggested, a recognition of a suborder
for the triacanthodids and triacanthids as distinct from a
suborder for all of the other plectognaths, with the latter
divided into two superfamilies, one for the balistoids and
ostracioids (the remnants of the Sclerodermi) and one for
the gymnodonts (as above) along with the Eocene
Eoplectus and its relatives.

It is felt here that the entire constellation of overall
similarities and differences in both generalized and
specialized features between the various species and
higher categories support the reasonableness of con-
tinuing to divide the Plectognathi into two sub-
orders: the Sclerodermi for the basal triacanthodids
and their closely related triacanthids as well as the
balistoids and ostracioids derived from the latter, and
the Gymnodontes for the more modified tetraodontoids
and molids linked together through the triodontids to the
Eocene eoplectin triacanthodids.

Historical Review of the Classification of the Plectognathi

Since representatives of most of the major subgroups
of plectognaths occur in the Mediterranean and off the
Atlantic coast of Europe, they were familiar to the early
Greeks and Romans. A rather long article could be
devoted to just the pre-Linnaean occidental knowledge of
the plectognaths, for almost every naturalist who dealt
with fishes, from Aristotle to Marcgrave, made promi-
nent mention of various balistoids, ostracioids,
tetraodontoids, and molids (the similar oriental,
primarily Chinese, history of the plectognaths cannot be
dealt with here). Passing over those nearly 2,000 years
that intervened between Aristotle and the person who is
most often considered the founder of modem ich-
thyology, Peter Artedi, only one example of the type of
classificatory scheme that preceded Artedi's work need
be cited.

In his "Historia Piscium," WiUughby (1686:22-25)
divided "fishes" (including elasmobranchs and
cetaceans) into 11 major categories, one of which was the
Pinnis uentalibus carentes, containing the plectognaths,
syngnathids, and swordfishes.

Within any one of the 11 categories were described a
variable number of species, most of them based on the
literature of such previous naturalists as Clusius, Ron-
delet, Gesner, Marcgrave, Belon, Salviani, and Al-
drovandi, but with a few supposedly new species also
described. The types of fishes that occurred (Willughby
1686:143-164) in the "de Piscibus corpore contractiore vel
saltern non admodum lubrico, qui pinnis ventralibus
carent," or "Pinnis ventralibus carentes," were for the
most part plectognaths. This section consisted of 14
chapters, each of which described either a single species

or a group of supposedly related species. For example,
Chapter I (p. 143-148) was devoted to "De Orbe Pisce"
and described 11 species in which the first word of the
binomial or polynomial was, in all but two cases, Orbis.
The two exceptions had the species name beginning with
Orbibus or Histrici and are diodontids. One might ex-
pect that all of the Orbis species would be tetraodontids
or diodontids and in fact they are, with a single excep-
tion: the Orbis Ranae rictu (p. 145, fig. 2 of pi. 19, after
Clusius) is Cyclopterus lumpus. Another species, Orbis
oblongus Testudinis capite (p. 147-148, fig. 3 of pi. 19,
after Clusius) is probably a tetraodontoid, but the figure
shows a creature whose front end reminds one of a fetal
seal but whose hind end is fishlike. Chapters II (p. 148)
and III (p. 148-149) described, respectively, Ostracion
Nili and Ostracion prior, both of which are ostraciids, as
is Pisces triangularis of Chapter IV (p. 149-150). Chap-
ter V (p. 150-151) described Monoceros pisces, which is
probably a monacanthid, while Chapter VI (p. 151-152)
was devoted to the Mola of Salviani or Orthagoriscus
of Rondelet. Chapter VU (p. 152-156) described the
balistid Capriscus pesce Balestra, but an appendix to
this chapter described three species of Orbes: a
tetraodontid, a diodontid, and an ostraciid. The remain-
ing seven chapters described nonplectognaths, these be-
ing, to use only the "generic" term, Stromateus (a but-
terfish), Hippocampus (a seahorse), Acus (a pipefish),
Acui (a syngnathid), Scolopax (Macrorhamphosus), and
Xiphias (a swordfish). The latter seven types have little,
if anything, in common with the plectognaths, making
for an odd mixture of forms within the "Pinnis ven-
tralibus carentes," but it is more realistic to dwell on the
fact that by the 17th century the plectognaths were al-
ready placed together in the same subgroup of fishes.

In an appendix, written by Lister, to Willughby's
(1686) "Historia Piscium," an illustration of a shallow-
water triacanthid was given, based on the Hoornvisch of
the Dutch East Indies first described by Nieuhof (1682)
in his work on the fishes of Batavia. An account of
this species of triacanthid was not to reappear until
Bleeker (1852b) formally described this first-mentioned
triacanthid as Triacanthus Nieuhofii, while in the in-
terim the first triacanthid to be binomially described was
Batistes Biaculeatus Bloch 1786. By contrast, the deep-
water triacanthodids were not described until Schlegel
(1850).

With the advent of Artedi, the classification of fishes
in general as well as that of the plectognaths was placed
on a much more refined basis, for the groupings of sup-
posedly related forms were far superior to any that had
previously appeared. In his "Genera Piscium," Artedi
(1738) recognized five basic divisions within his Pisces.
One of these, the Plagiuri, had the caudal fin horizontal
and is the Cetacea, while the other four divisions (Mala-
copterygii, Acanthopterygii, Branchiostegi, and
Chondropterygii) had the caudal fin perpendicular and
are true fishes. The Branchiostegi contained the fol-
lowing genera: Batistes, Ostracion, Cyctopterus, and
Lophius.

The generic categories of Artedi are usually recog-

nized today at the familial or ordinal levels. The name
Branchiostegi {branchio, gills, and stegein, to be
covered) implies that Batistes, Ostracion, Cyctopterus,
and Loph'us should have the superficial similarity of the
gills being relatively well hidden from external view by
the constricted aperture of the branchial cavity. We now
know, of course, that the last two genera mentioned
above are neither closely related to one another nor to
Batistes and Ostracion. Ostracion contained 22 species
which represent not only the ostraciids but also the
tetraodontids, diodontids, and molids. It should be noted
that the Ostracion category was composed of species of
extreme diversity in external appearance, which are
nevertheless closely related to one another. Moreover,
the group was not as heterogeneous as it might super-
ficially seem to be, for within it the species were de-
scribed successively in a manner which shows that Artedi
had an insight into the relative closeness of relationship
of the various forms. Thus, the first 10 species are os-
traciids, and the next 3 species are tetraodontids. All of
the next seven species (nos. 14 to 20) are diodontids, with
the exception of species no. 17, which is the Orbis Ranae
rictu of Willughby and of Clusius, a cyclopterid. The
next to the last species (no. 21) is probably a tetraodon-
tid, but since it was based on the poorly figured Orbis ob-
tongus Testudinis capite of Willughby and of Clusius,
one could scarcely have expected Artedi to have placed it
with his other tetraodontids. The last species (no. 22) is
the commonest (as an adult, at least) of the molids, Mola
mola. The Batistes group contains five balistoids and a
macrorhamphosid (the only nonplectognath included).
In short, Artedi did justice to those plectognaths which
he had actually seen or which were reasonably described
in the literature, and the several errors were only the
result of occasionally being forced to rely solely on poor
descriptions by earlier naturalists.

The plectognaths appeared in a somewhat different
assemblage in Klein's (1742) "Historiae Piscium
Naturalis." Like previous workers since the time of
Aristotle, Klein included in his Pisces both cetaceans
and fishes, separating them from one another on the
basis of possession of lungs or gills. Whereas Artedi's first
dichotomous division of the true fishes was on the basis
of bone as opposed to cartilage in the skeleton, Klein's
was on whether the gills were hidden or easily observed.
Thus, the Branchiis Apertis contained a large array of
relatively normal fishes, such as Siturus, Acipenser,
Xiphias, Mastacembetus, Sotea, Tetragonopterus, etc.
By contrast, the Branchiis Occultis contained a less
plausible association of 15 genera, including elas-
mobranchs and cyclostomes.

The plectognaths were all included in Crayracion
(Klein 1742:17) and Capriscus (p. 23), which, along with
Batrachus and Conger, supposedly had the following
combination of characters: paired fins present;
branchial cavity with a single aperture; gills placed
laterally and hidden from view. Klein's Batrachus is the
same as Artedi's Lophius, while Artedi's Cyctopterus ap-
peared as a part of Klein's Crayracion. The addition of a
number of eels, as Conger, to a place of proximity with

the plectognaths and Lophius and the placement of all of
these with the elasmobranchs and cyclostomes were ma-
jor differences in the treatment of the plectognaths by
Artedi and Klein. Klein's Crayracion and Capriscus are
more or less synonymous with, respectively, Artedi's
Ostracion and Batistes. Both of Artedi's groups con-
tained only plectognaths (except for Macrorham-
phosus). but the same cannot be said of Klein's. The lat-
ter described 32 species of Crayracion, of which most are
ostraciids, tetraodontids, diodontids, or molids, but
species 7 (p. 19) and 11 (p. 19) are Cyclopterus and
species 32 (p. 23) is a syngnathid. On the other hand, all
11 species of Klein's Capriscus are balistoids.

In his "Museum Ichthyologicum," Gronovius (1754)
followed precisely the classification adopted by Artedi,
with Pisces containing the Plagiuri, Malacopterygii,
Acanthopterygii, Branchiostegi, and Chondropterygii.
The Branchiostegi again contained four types: Balistes,
Ostracion, Cyclopterus, and Lophius. Gronovius made
no innovations in the Balistes category; on the contrary,
one of his species of Balistes appears to be Marcgrave's
antenneriid frogfish from Brazil. However, Gronovius did
go a step beyond Artedi in his handling of the Ostracion
group. Gronovius divided his Ostracion into three sub-
groups on the basis of body shape, thus: Corpore quad-
rangulo, with three ostraciids; Corpore triangutato, with
three more ostraciids; and Corpore cathetoplateo, uel
rotundo. with two tetraodontids and a molid. Although
Gronovius' (1763) subsequent "Zoophylacium" is post-
Linnaean chronologically, it is pre-Linnaean in its ter-
minology. His Pisces were now divided into Plagiuri,
Chondropterygii, Branchiostegi, and Branchiales, the
latter group being roughly equivalent to his previously
used Malacopterygii and Acanthopterygii. Within these
groups the species were segregated according to fin struc-
ture. Thus, his Branchiostegi were arranged as follows:

Pinnis Ventralibus nullis: Muraena, Gymnotus, Syn-

gnathus, Ostracion.
Pinnis Ventralibus spuriis: Balistes, Cyclopterus,

Cyclogaster.
Pinnis Ventralibus veris praesentibus: Gonorynchus,

Cobitis, Uranoscopus, Lophius.

New groups were added to the Artedian Branchio-
stegi, and, beyond that, Gronovius (1763) added a fur-
ther subdivision to the Ostracion category. The Balistes
category remained a simple descriptive list of a number
of balistoids and a macrorhamphosid. The four Ostracion
subgroups that Gronovius recognized were: Corpore
quadrangulo, with four ostraciids; Corpore triangulato,
with three more ostraciids; Sphaerico, uel oblongo-
rotundo corpore, with two diodontids and one tetraodon-
tid; Corpore cathetoplateo, with two tetraodontids and
two molids. When comparison is made between the
classification of the nonbalistoid plectognaths of
Gronovius and that of Linnaeus, it is obvious that
Gronovius simply incorporated into his Ostracion sub-
groups a few of Linnaeus' ideas without an excess of
■recision.

With the 10th edition of Linnaeus' (1758) "Systema
Naturae," a basic division of the plectognaths into four
genera (Balistes, Ostracion, Tetraodon, Diodon) was
presented for the first time. Five groups of Pisces were
recognized by Linnaeus: Apodes, Jugulares, Thoracici,
Abdominales, and Branchiostegi. The latter contained
the following genera: Mormyrus, Balistes, Ostracion,
Tetraodon, Diodon, Centriscus, Syngnathus, and
Pegasus.

But to the above Pisces must be added the fishes which
Linnaeus placed in the Amphibia Nantes, containing
Petromyzon, Raja, Squalus, Chimaera. Lophius, and
Acipenser.

Such a classification shows an eclectic approach to the
systems of previous workers. The Branchiostegi of Artedi
is a much smaller group than that of Linnaeus, and the
only species that are common to the Branchiostegi of
both are the plectognaths. Thus, Cyclopterus and
Lophius appeared in Artedi's Branchiostegi, but Lin-
naeus placed Cyclopterus in the Thoracici and Lophius
in the Amphibia Nantes. Mormyrus, Centriscus, Syn-
gnathus, and Pegasus of the Linnaean Branchiostegi are
found among the Acanthopterygii and Malacopterygii of
Artedi. In short, as far as the plectognaths are con-
cerned, Linnaeus (1758) had followed Artedi's views,
only adding generic names to the four plectognath sub-
groups that Artedi had at least implicitly recognized.
The Linnaean Balistes contained seven balistoids and a
macrorhamphosid; Ostracion, nine ostraciids; Tetraodon
(spelled Tetrodon on p. 243 in the list of genera of
Branchiostegi but Tetraodon on p. 332 where defined),
five tetraodontids and one molid; Diodon, seven diodon-
tids. Linnaeus thus recognized 29 plectognath species,
only two more than Artedi had recognized 20 years
previously. Many of Klein's 43 species of Crayracion and
Capriscus were synonymized by Linnaeus, or dis-
regarded as unrecognizable.

In the 12th edition of Linnaeus' (1766) "Systema
Naturae" a major change took place in the classification
of the plectognaths — Branchiostegi was eliminated. All
of the genera of the former Branchiostegi were placed in
the Amphibia Nantes, except for Mormyrus, which was
placed in the Abdominales. The Amphibia Nantes then
consisted of an odd assortment of types: Petromyzon,
Raja, Squalus, Chimaera, Lophius, Balistes, Ostracion,
Tetrodon, Diodon, Cyclopterus, Centriscus, Syng-
nathus, Acipenser, and Pegasus. The motive behind the
placement of the plectognaths in the Amphibia Nantes
was the erroneous report sent to Linnaeus by Dr. Gar-
den, a physician in Charleston, S.C, who informed Lin-
naeus that Diodon possessed a lung. A combination of
three anatomical peculiarities in Diodon probably led to
the confusion, for Diodon not only has a distensible
diverticulum of the oesophagus that can be filled with
either air or water (at the time of Linnaeus it was thought
to undergo inflation only by the intake of air), but the
swim bladder is bilobed anteriorly and the well-
developed kidneys are placed far forward. Linnaeus
wrote (1766:348) that "Garden in America habitant!,
petere, vellet dissecare Diodontis respirationis organa &

inquirere numne Pulmones haberent." It is possible that
Garden mistook either the swim bladder, kidneys, or dis-
tensible diverticulum of the oesophagus for lungs. Brous-
sonet (1780:680) believed that it was the distensible
diverticulum that had been misinterpreted, but Cuvier
(1817:146) thought that it was the kidney, and the par-
ticular source of Garden's error is impossible to identify.
Severe as was the elimination of the Branchiostegi, the
only significant change in the assemblage of species of
plectognaths described was the synonymizing of several
of his previously numerous species of Diodon, so that
only two species of that genus were recognized.

Among the 29 Linnaean species were representatives of
the balistids, monacanthids, both subfamilies of os-
traciids, the tetraodontin tetraodontids, diodontids, and
molids. As discussed in the preceding section on
Willughby, the triacanthids were known before 1758,
even though Linnaeus did not recognized them, and they
were not binomially described until Bloch (1786), while
triacanthodids were not described until Schlegel (1850).
As discussed in the succeeding section on Cuvier, the
aracanids and triodontids were not formally described
until well after Linnaeus, respectively by Shaw (1798)
and Cuvier (1829), but see Tyler (1967) for a history of
the impressionistic pre-Linnaean illustrations of TVt'o-
don. Several names for canthigasterin tetraodontids
were inadvertently listed by Linnaeus as synonyms for
several species of tetraodontin tetraodontids (e.g., see
Shipp 1974 for Lagocephalus) , but the first binomial
description of a canthigasterin was in a now mostly
forgotten publication by Paterson (1786), as Tetrodon
electricus, and the currently used specific name of one of
the more common Indo-Pacific species of Canthigaster
may have to be changed.

Gmelin's 13th edition (1788) of the "Systema Naturae"
reestablished the Branchiostegi in the Pisces and entirely
eliminated the Amphibia Nantes, while at the same time
the "Systema Naturae" introduced the Chrondropte-
rygii. Fishes were thus all included in six categroies of
Pisces: the Apodes, Jugulares, Thoracici, Abdominales,
Branchiostegi, and Chondropterygii.

This general classification was a distinct improve-
ment over any of those in previous editions of the
"Systema Naturae" and was largely the synthesis of
Gmelin himself. Within the Branchiostegi the genera ap-
peared in the following order: Mormyrus, Ostracion
with 10 species of ostraciids, Tetrodon with 12 species of
tetraodontids and 2 species of molids, Diodon with 6
species of diodontids and 1 species of molid, Syng-
nathus, Pegasus, Centriscus, Batistes with 18 species of
balistoids, Macrorhamphosus, Cyclopterus, and
Lophius. Thus, the treatment of the individual genera of
plectognaths was inferior to that of previous editions, for
molids were placed in both Tetrodon and Diodon, while
Balistes was spatially separated from the other plectog-
naths by the intervening Syngnathus, Pegasus, and
Centriscus. Even if this separation of Balistes was
deliberate, its significance, if any, is unknown.

The classification adopted by Goiian (1770:94) in his
"Historia Piscium" divided the Pisces into three main

groups, the Acanthopterygii, Malacopterygii, and
Branchiostegi, each with the same four sub-
divisions: the Apodes, Jugulares, Thoracici, and Ab-
dominales.

Such subdivisions of the three major categories have a
deductive simplicity that is truly alluring, but, unfor-
tunately, fishes are not constructed to fit into such
logical pigeonholes. The Acanthopterygii and Malacop-
terygii contained the expected genera, and the Branchi-
ostegi sorted out to:

Apodes: Syngnathus, Balistes (with a footnote

saying that certain Balistes have ven-
tral fins and thus must be placed in
the Abdominales), Ostracion, Tetrao-
don, Diodon.

Jugulares: Lophius.

Thoracici: Cyclopterus.

Abdominales: Centriscus, Pegasus (and, according to
the footnote, a few Balistes, which are
not specifically named, but may imply
Macrorhamphosus).

Only genera were described, but the species of Balistes
or of other plectognaths that would have to have been
placed in the Branchiostegi Abdominales are the triacan-
thoids, which had not yet been described binomially.
The above treatment of the Branchiostegi is basically
that of Gmelin's 13th edition of the "Systema Naturae,"
except that Mormyrus was placed by Goiian in the
Malacopterygii Abdominales.

In Bonnaterre's (1788) "Tableau Encyclopedique," the
plectognaths reverted to the cartilaginous fishes, for five
classes were recognized: the Cartilagineux, Apodes,
Jugulares, Thoracici, and Abdominales. These five
groups are those of the 12th edition of Linnaeus' (1766)
"Systema Naturae," the only difference being the sub-
stitution of the term Cartilagneux for Amphibia Nantes.
The Cartilagineux contained exactly the same genera as
the Amphibia Nantes, differing only in that Petromyzon
was called Lampetra. The value of Bonnaterre's work to
the study of plectognaths rests on the fact that of the 41
species described, the great majority were figured with
some accuracy, and outline cross sections of the body
were given for many.

If fishes were ever to fit into a deductively logical
scheme of classification, they were given a chance to do
so in Lacepede's (1798, 1800, 1802a, b, 1803) cat-
egories, as outlined in his "Histoire Naturelle," even
more so than in Goiian's work. The "Classe des Pois-
sons" was divided into the Sous-classe Poissons Car-
tilagineux and the Sous-classe Poissons Osseux. Each of
these subclasses was split into the same four divisions,
and each of these divisions was split into the same four
orders. Thus, there was a dichotomy followed by two suc-
cessive sets of quadrichotomies, to give a total of 32
orders in eight sets of Apodes, Jugulaires, Thoracins, and
Abdominaux (following Gouan). The balistoids were in
the third order (Thoracins) of the second division of the

Poissons Cartilagineux, while the other plectognaths
were in the first order (Apodes) of the fourth division,
well separated from the balistoids.

When one considers the "Poissons Cartilagineux" as a
whole, the plectognaths were associated with much the
same genera as they were in Gmelin's 13th edition (1788)
of the "Systema Naturae." Again, just as in Gmelin,
Batistes was separated from the other plectognaths,
although in a more definite manner. One innovation was
Lacepede's handling of the plectognath genera. His con-
temporaries had accepted four genera {Balistes, Ostra-
cion. Tetrodon, and Diodon) but Lacepede admitted six
(Balistes, Ostracion, Tetrodon, Les Ovoides, Diodon,
and Les Spheroides). It happened that his two new
genera were based on artifacts or misinformation and
thus lost their value, except for nomenclatural purposes.
Les Ovoides (Lacepede 1798:520) was based on a damag-
ed specimen described by Commerson which lacked dor-
sal, caudal, and anal fins. Lacepede had never seen the
single specimen used by Commerson but thought that it
probably represented a new genus, and so named it. Les
Spheroides (Lacepede 1800:22) was likewise said to lack
dorsal, anal, and caudal fins. It was based on an unpub-
lished figure by Plumier of the anterior view of the body
of what we know now as Sphoeroides spengleri. Lacepede
did somewhat better with his other genera, for each of
these was divided into unnamed subgenera.

Balistes (Lacepede 1798:332) contained four sub-
genera; the first for those forms with more than one spine
in both the pelvic and first dorsal fins; the second for
those with more than one spine in the pelvic fin but only
one spine in the first dorsal fin; the third for those with
only one spine in the pelvic fin but more than one spine
in the first dorsal fin; the last for those with but a single
spine in both the pelvic fin and first dorsal fin.

The first subgenus should only contain triacanthoids,
which by common definition are those plectognaths with
well-developed pelvic spines. Lacepede mistook the en-
larged scales that occur in the midventral line of the ab-
dominal region of many balistoids between the end of the
pelvis and the anus for pelvic fin spines. These scales
strengthen the "abdominal fan" of those balistoids
whose pelvis is especially movable around its anterior ar-
ticulation with the pectoral girdle. The distal ends of the
enlarged scales are tapered into narrow shafts which
often project out through the skin on either side of the
midventral line (see Monod 1959a), and it is not sur-
prising that they should have been considered to be
some kind of pelvic fin spines. Lacepede also observed
the thickened scales that encase the posterior end of the
pelvis and mostly obscure from view the modified fin-ray
element of most balistoids, and understandably thought
them also to be a pelvic fin spine, as had others before
him and afterward (e.g., Garman 1891). For these
reasons Lacepede's first subgenus contained not only the
single binomially described triacanthid then known but
also three balistids. His second subgenus contained a
single monacanthid, whose modified scales in the ab-
dominal fan were again mistaken for pelvic fin spines.
The third subgenus was by far the largest of his sub-

genera, containing numerous balistids and a few
monacanthids. The fourth subgenus contained two
monacanthids.

Intervening between the descriptions of Balistes and
the next group of plectognaths {Ostracion) were the
descriptions of Chimaera, Polyodon, and Acipenser.
Ostracion (Lacepede 1798:441) was divided into four sub-
genera on the basis of the four logical combination pos-
sibilities of the presence, or absence, of cuirass spines in
the ocular region and of the presence, or absence, of
cuirass spines in the caudal region. These categories are
superficial and have not withstood the test of time.
Tetrodon (Lacepede 1798:474) had three subgenera, of
which the first two were described as not having the body
particularly compressed and consisted of tetraodontids,
while the third subgenus was said to have a compressed
body and contained a single molid. Diodon (Lacepede
1800:1) had no subgenera and contained four diodontids
as well as two molids, one of which was the same species
that also occurred in the third subgenus of Tetrodon.

Lacepede's only subgeneric groups of lasting value
were those of Balistes, which have since, at least in part,
been elevated to familial rank. His contribution to the
plectognaths was primarily the recognition of the fact
that the Linnaean genera could, and should, be sub-
divided. His subgenera tended to be arbitrarily drawn
and factual errors pervaded his descriptions, but he
began the trend of subdivision of the Linnaean categories
of plectognaths.

Like that of Lacepede, the classification adopted by
Bloch and Schneider (1801) in their "Systema Ich-
thyologiae" was deductively logical and equally ar-
tificial. Eleven classes of fishes were recognized on the
basis of the number of fins, i.e., Hendecapterygii (11 fins)
to Monopterygii (1 fin). Each class was divided, on the
basis of pelvic fin position, into such orders as Apodes,
Jugulares, Thoracici, Abdominales, and Achiri ( =
without hands, but with reference to pelvic rather than
pectoral fins, and thus would seem to be equivalent to
Apodes). The plectognath genera fell within the fol-
lowing classes and orders (Bloch and Schneider
1801:Lm-LVni):

Classis VI: Hexapterygii [i.e., two dorsals, an anal, a

caudal and two pectorals]
Ordo Apodes: Balistes, Rynchobdella [a mastacembe-

lid]
Ordo Pinna Anali Carentes: Trachypterus, Gymne-

trus [= Regalecus]
Classis VII: Pentapterygii [i.e., a dorsal, anal, caudal,

and two pectorals]
Ordo Apodes: Ophidium, Pomatias, Gnathobolus,

Muraena, Stromateus, Ammodytes, Sternoptyx,

Anarrhichas, Channa, Sternarchus, Ostracion,

Tetrodon, Orthagoriscus, Diodon, Syngnathus.

This was the manner of listing in the index, but in the
text the Pentapterygii were placed in two orders (Bloch
and Schneider 1801:484-516), Apodes for the genera from

Ophidium to Anarrhichas, Achiri for the genera Channa
to Syngnathus. Thus, Batistes, along with the
mastacembelids, was one of the Apodes Hexapterygii,
while the other plectognaths, along with Channa and the
syngnathids, were Achiri Pentapterygii. Such a system
could scarcely be more artificial, but several new species
were described and the use of Orthagoriscus for molids
set a generic usage to be followed for the next hundred
years.

One of the widest separations of the balistoids from the
other plectognaths was that made by Rafinesque (1810)
in his "Indice d'lttiologia Siciliana." He recognized two
subclasses of fishes, the Pomniodi (branchial apparatus
with both an operculum and branchiostegal membrane)
and the Atelini (branchial apparatus incomplete). The
Sotto-Classe Pomniodi contained four divisions based on
pelvic fin position; Guigulari, Toracici, Abdominali, and
Apodi. Each of these four "Divisione" was subdivided
into a variable number of "Sezione," which in turn con-
tained the orders. Tetrodon, Diodon, and Orthagoriscus
were placed in the Ordine Gli Odontini (Rafinesque
1810:40; all but the pelvic fins present; jaws in the form
of a bony beak), while Ostracion was placed in the Or-
dine Ostracidi (p. 39; body covered by a cuirass; all but
pelvic fins present; jaws with teeth not formed as a bony
peak). Both of these orders were in the Sezione
Branchiosomi (body short; spherical or elliptical) of the
Divisione Apodi of the Pomniodi. Batistes, however, was
elevated as the Ordine Balistini of the Divisione Gli Om-
nanchidi (branchial apparatus with a branchiostegal
membrane but without an operculum) of the Atelini.
Rafinesque's treatment of the plectognaths is best
remembered as a nomenclatural curiosity.

The name of Cuvier looms large, of course, in the study
of plectognaths, and not only because it was he who es-
tablished them as a natural group at the ordinal level. In
his "Legons d'Anatomie Comparee," Cuvier (1805) fol-
lowed a classification (folding sheet at end of volume 1)
that borrowed heavily from previous systems. Fishes
were divided into those with "a squelette cartilagineux"
and those with "a squelette osseux," with the bony fishes
subdivided into Apodes, Jugulares,Thoraciques,and Ab-
dominaux. The cartilaginous fishes were subdivided into
two groups of Chondropterygiens and six groups of
Branchiosteges, of which two were the "bouche au bout
du museau; des dents" for Batistes and Ostracion, and
the "os des machoires tenant lieu de dents" for Tet-
rodon, Ouoides, Mota, and Diodon.

The plectognaths were thus associated with much the
same genera as in the 12th edition of Linnaeus' (1766)
"Systema Naturae." An important improvement was
made, however, by the fact that Batistes and Ostracion
were placed together as a subgroup distinct from the
other plectognaths. This was the first time that such a
distinction had been made so clearly. Cuvier further con-
tributed to the study of plectognaths in this work, for
throughout all five volumes numerous anatomical notes
were given for this group.

Cuvier (1817) put this anatomical information to good
use in his "Le Regne Animal," in which the Order Plec-

tognathi was established, as one of the six orders of Pois-
sons Osseux (in contrast to Chondropterygiens), as fol-
lows;

Ordre Plectognathes.
La premiere famille, ou les Gymnodontes.
La deuxieme famille, ou les Sclerodermes.

Cuvier placed the Plectognathi as the first order of
Poissons Osseux, just after the Chondropterygiens,
because (1817:144) "il se rapproche un peu par I'imper-
fection des machoires, et par le durcissement tardif du
squelette; cependant ce squelette est fibreux, et en
general toute sa structure est celle des poissons or-
dinaries." He continued with the diagnosis of the order,
of which "Le principal caractere distinctif tient a ce que
I'os maxillaire est soude ou attache fixement sur le cote
de I'intermaxillaire qui forme seul la machoire, et a ce
que I'arcade palatine s'engrene par suture avec le crane,
et n'a par consequent aucune mobilite." He added that
the operculum and branchiostegal rays were hidden un-
der the thick skin and that the branchial aperture was
restricted. He further stated that there were only ves-
tiges of ribs, no pelvic fins, a large intestine without
caeca and usually a large swim bladder. The Gymno-
dontes (p. 145) were defined as those Plectognathi which
"A, au lieu de dents apparentes, les machoires gamies
d'une substance d'ivoire, divisee interieurement en
lames," while the operculum was small and there were
five branchiostegal rays. Three genera were de-
scribed; Diodon, Tetrodon, and Orthagoriscus. The
Sclerodermes (p. 149-150) were defined as being dis-
tinguished by "le museau conique ou pyramidal prolonge
depuis les yeux, termine par une petite bouche armee de
dents distinctes en petit nombre a chaque machoire.
Leur peau est generalement apre ou revetue d'ecailles
dures; leur vessie natatoire ovale, grande et robuste."
Five generic groups were recognized, three of them for
the first time: Batistes, Les Monacanthes, Les
Aluteres, Les Triacanthes, and Ostracion.

This, then, was the basic anatomical classification of
the Plectognathi upon which all subsequent work on the
group rests. It is perhaps unfortunate that the order was
not more carefully diagnosed, for subsequent workers
found it easy to erode the structure that Cuvier had
erected. For instance, the principal character used by
Cuvier, the suturing (or at least immovable attachment)
of the maxillary to the premaxillary is true for all plec-
tognaths, except for the triacanthoids. But since Cuvier
had examined Triacanthus biaculeatus he might have
been expected to have noticed that his principal diag-
nostic character of the Plectognathi did not apply to one
of the genera. The definition of the order can also be
criticized because some nonplectognaths also have the
two bones of the upper jaw intimately connected. These
criticisms, of course, are valid, but by far the more im-
portant point is that it was Cuvier who clearly saw in the
early 19th century the naturalness of the plectognaths as
a group, elevated them to ordinal rank, and defined the

basic subgroups within the two families. His families are
now of subordinal rank, and his genera have become the
bases for most of the families now recognized.

In only one group did Cuvier miss an important major
distinction, i.e., the ostracioids can be divided easily into
ostraciids and aracanids, but Cuvier could not have been
very familiar with the aracanids, for these mostly Japan-
ese and Australian deepwater forms were rarely collected
at that time and known primarily from a single species
(aurita, Shaw 1798 illustration and 1804 description). Of
the 10 families of plectognaths presently recognized, only
the triacanthodids and triodontids were unmentioned by
Cuvier, and they were not to be discovered for another
decade or more. Cuvier's description of Reinwardt's
specimen of the unique Triodon bursarius (= mac-
ropterus) appeared in the second edition of Cuvier's "Le
Regne Animal" (1829:370) as the fourth genus of the
family Gymnodontes, while the first triacanthodid was
described by Schlegel (1850). In summary, most of the
basic groupings of plectognaths were made known by
Cuvier, with the notable exception of the two sub-
divisions of the ostracioids. At about the same time that
his (1817) first edition of "Le Regne Animal" appeared,
Cuvier (1818) published a short paper on Diodon, de-
scribing some new species, but, more importantly, cor-
recting a number of anatomically erroneous statements
about diodontids made by such workers as Broussonet,
Plumier, and Bloch.

The appearance of the classifications of Latreille
(1825) and Risso (1826) shortly after Cuvier's (1817)
"Le Regne Animal" set the precedent for the thereafter
nearly unanimous ordinal recognition of the Plectog-
nathi. Risso followed Cuvier rather closely, and as far as
the plectognaths are concerned, added nothing to their
classification. In his (1825) "Families Naturelles,"
Latreille used Cuvier's definition of the Ordre Plectog-
nathes, but placed them in a different relationship with
other orders (with the sturgeons and lophobranchs).

At about this time Geoffroy Saint-Hilaire (1827) pub-
lished his "Poissons du Nil." His description (p. 176-214,
pis. 1-2) of Tetrodon physa (= lineatus Linnaeus), the
common pufferfish of the Nile, was the finest descrip-
tion of a plectognath that had appeared up to that time,
describing the general anatomy, habits, distribution, and
nomenclature of the fish and gently setting aside the er-
rors made by Bloch and by Lacepede on the inflation
mechanism, showing the sac to be a diverticulum of the
oesophagus and describing its general structure and mus-
culature.

The year 1833 saw the begiiming of the publication of
Agassiz's "Recherches sur les Poissons Fossiles" [vols. 1
(1844a) and 2 (1833, 1842, 1844b)], which was to make a
major (or at least temporarily so) change in the
classification of nearly all groups of fishes. Based on the
configuration and composition of the scales, Agassiz
hoped to establish a more natural grouping of orders than
had previously been available. Agassiz's system as a
whole did not stand the test of time, and his association
of the plectognaths with the ganoids was a major error.
His Ordre des Ganoides (1844a: 169) contained six

groups: Acipenserides, Siluroides, Lophobranches,
Gymnodontes, Sclerodermes, and Sauroides.

To Agassiz the plectognaths were ganoides for various
reasons. The scales of Ostracion were described (Agassiz
1844a:75) as consisting of a homy substance deposited in
strata and covered with a thick layer of dentine charac-
terized by ramifying calcareous tubes like those of the
teeth. The major references to plectognaths, however,
were contained in a section (1844b:248-267) entitled "De
la famille des Gymnodontes" where brief osteological de-
scriptions of a Tetraodon and a Diodon were given, along
with a comparison of the structure of the teeth and spines
of these two gymnodonts. Agassiz found the gymnodont
spines to be similar to the scales of scleroderms, in that
both possessed a dentine layer with calcareous tubes,
while the teeth of gymnodonts were said to be similar to
the teeth of sharks. Agassiz (1857) also published a brief
note on the structure of Mola.

Agassiz (1833:1-2) listed the six families of his Ordre
Ganoides as: Famille Lepidoides (with Acanthodes,
Palaeoniscus, Osteolepis, Lepidotus, etc.); Famille
Sauroides (with Leptolepis, Sauropsis, etc.); Famille
Pycnodontes (with Placodus, Pycnodus, Microdon, etc.);
Famille Sclerodermes; Famille Gymnodontes; and
Famille Lophobranches (with Syngnathus, etc.). It is
strange to see the plectognaths associated in the same
order with such forms as the placoderm Acanthodes, the
crossopterygian Osteolepis, and the isospondyl Lep-
tolepis. Even stranger, however, was the assemblage in-
cluded in the Sclerodermes (1844b: 248- 267). Not only
were Balistes, Ostracion, the Balistes-like fossil Acan-
thoderma, and the Triacanthus-Vike fossil Acan-
thopleurus placed there, but also included in the
Sclerodermes were such obviously nonplectognath fos-
sils as Blochius (which Woodward, 1901:591, referred to
the Blenniiformes), Dercetis (which Woodward, p. 171,
referred to the Isospondyli), and Rhinellus (which Wood-
ward, p. 266, referred to the Isospondyli). Agassiz rec-
ognized in later years that his classification was highly
artificial (see Agassiz 1860).

With the appearance of Miiller's (1844) "Ueber den
Bau und die Grenzen der Ganoiden," the Ganoides of
Agassiz were modified to a considerable extent. Miiller
purged the Ganoides of the plectognaths, lophobranchs,
and most of the other living species placed there by Agas-
siz. Muller pointed out that the true ganoids had, among
other characteristics, more than two valves in the conus
arteriosus, while the teleosts had only two. Among the
supposed ganoids of Agassiz that Muller had examined
for this characteristic were Balistes, Ostracion, and Tet-
raodon. The classification at which Muller arrived
(1844:85-88) recognized six subclasses of Pisces, with the
Subclass Teleostei containing the Order Plectognathi
and its three families: Balistini, Ostraciones, and Gym-
nodontes.

Muller pointed out (1844:6) that whereas the true
ganoids are physostomes with abdominal pelvics, the
Plectognathi are physoclists and that when obvious pel-
vic fins are present, as in the triacanthoids, they are not
abdominal. He also disagreed (p. 7) with Agassiz's con-

tention that the scales of plectognaths have much in
common with those of ganoids. Just as Muller had purg-
ed the Ganoides, so he also rid (p. 26) the Scleroderms of
Agassiz's Blochius, Dercetis, and Rhinellus. Elsewhere
(p. 78-79) Muller gave systematic notes on the tetraodon-
toids, based on the structure of the nasal aperture and
tube. The character of the nasal organ was thereafter to
play an important (and exaggerated) role in the classi-
fication of the tetraodontoids. Muller made a clear state-
ment (p. 84) of the fact that the maxillary and premaxil-
lary are not fused or firmly attached to one another in all
of the plectognaths, as Cuvier had said that they were,
for he noted that in the triacanthoids these two bones
have a more normal relationship to one another.

Muller's contribution to the classification of fishes was
so outstanding that the outline given above in relation to
the plectognaths cannot do it justice, for his was the
finest classification of fishes to appear up until that time,
and in the magnitude of the changes it wrought in ich-
thyology it was at least equal to the work of Artedi and
Cuvier.

With Agassiz saying one thing about the highly
modified scales and teeth of plectognaths and with
Muller saying something else, a controversy broke out in
the literature which has not yet been fully resolved (see
references under subsequent discussion of Owen).

Some of Muller's classificatory conclusions came in for
immediate criticism. Vogt (1845) took exception to
Muller's recognition of the Orders Pharyngognathi and
Plectognathi. Vogt thought these orders to have been
founded on insufficient criteria, i.e., the fusion of the
lower pharyngeals in the Pharyngognathi and the fusion
of the upper jaw bones in the Plectognathi.

Muller's classification is discussed here slightly out of
chronological order so that it can be more easily com-
pared with that of Agassiz. Intervening between these
two monumental works were the classifications of Swain-
son, Nardo, and Bonaparte.

In his (1832) "Saggio d'una Distribuzione Metodica"
Bonaparte adopted a classification that was basically
that of Cuvier's (1817) "Le Regne Animal," but with sig-
nificant changes in the rank of the groups recognized.
Bonaparte introduced here for the first time, in a con-
sistent manner, the "idae" ending for familial and the
"ini" for subfamilial names for his Classe Pisces. Bona-
parte recognized two monotypic orders in his Sezione 3,
or Plectognathi: the Gymnodontes (for the family
Tetraodontidae) and the Sclerodermi (for the family
Balistidae).

The Tetraodontidae contained the usual four genera:
Diodon, Tetraodon, Orthagoriscus, and Triodon. The
Balistidae contained three genera: Ostracion, Triacan-
thus, and Balistes (with the latter divided into four sub-
genera: Batistes, Balistapus, Monacanthus, and Aluterus) .

In his (1841a) magnum opus, the "Iconografia della
Fauna Italica," Bonaparte treated the plectognaths as
one of the three sezione (along with the Micrognathi and
Teleostomi) of the Sottoclasse Pomatobranchii (in con-
trast to the Elasmobranchii, Lophobranchii, and Mar-
sipobranchii) as follows:

Sezione Plectognathi
Ordine Sclerodermi
Famiglia Balistidi [sic]
[Subfamily] Balistini
[Subfamily] Ostraciontini
Ordine Gymnodontes
Famiglia Tetraodontidae
[Subfamily] Tetraodontini
[Subfamily] Diodontini
Famiglia Orthagoriscidae
[Subfamily] Orthagoriscini
[Subfamily] Molini.

In the same year, Bonaparte (1841b) published his "A
New Systematic Arrangement of Vertebrae Animals,"
and the classification adopted was that outhned above,
with only minor changes. In the plectognaths, for ex-
ample, the one difference was that in the family
Orthagoriscidae only one subfamily was recognized, the
Orthagoriscini, which was thus equal to his previous
two subfamilies, Orthagoriscini and Molini.

The classification used by Swainson (1838, 1839) in his
"Natural History of Fishes, Amphibians, and Reptiles"
was, like the rest of his work, a careless hodge-podge. The
pseudometaphysical "systems" by which Oken (1816)
compared orders of fishes with various classes of inver-
tebrates and vertebrates obviously made an impression
on Swainson, for the latter carried on where the former
had left off. To Swainson the plectognaths were equated
with such groups as amphibians and turtles, whereas
Oken had equated them with mammals. Swainson
(1838:189) included all of the plectognaths (or Cheloni-
form Order) in a family Balistidae, in which were
recognized five subfamilies: Balistinae, Ostracinae,
Cephalinae, Diodoninae, and Tetraodinae, each of which
included the expected forms.

Nardo (1842) presented a brief outline of his classi-
fication of the Sclerodermi, in which he closely followed
Bonaparte (1832), but attributed all of the familial
names to himself.

The first person after Cuvier to extensively analyze the
classification of the plectognaths as a whole was Dareste.
His first paper ( 1849) dealt with the osteology of Triodon
macropterus, the monotypic representative of its family.
Dareste saw that Triodon had many features in common
with the gymnodonts, but that it also possessed certain
characteristics of the scleroderms. One might expect that
this would strengthen the union of these two groups in an
Order Plectognathi. Such was not the interpretation of
Dareste, however, for he concurred with Vogt's (1845:67)
belief that the Order Plectognathi was ill-defined, un-
natural, and not long for this world. Although Dareste
made this point in his work on Triodon, he did not sub-
stantiate it with a discussion until 1850, in his
"Recherches sur la Classification de i'Ordre des Plectog-
nathes." In the latter work Dareste took apart Cuvier's
definition of the Order Plectognathi sentence by sen-
tence. To review very briefly Dareate's critique, he stated

that: Mola is the only plectognath with a skeleton less
ossified than that of the majority of fishes (but Cuvier
had stated only that the ossification was "tardif," while
otherwise like that of other fishes); the premaxillaries
and maxillaries are fixed firmly to one another in a num-
ber of fishes other than plectognaths; the palatine is
movable in certain scleroderms (Cuvier implied that it
was sutured to the cranium in all plectognaths); the
operculum and branchiostegal rays are hidden beneath
the skin in a number of fishes other than plectognaths;
Triodon possesses ribs; some scleroderms possess pelvic
fins (Cuvier said that they are always absent in plectog-
naths); Mola does not have a swim bladder (Cuvier said
that it is well developed in all plectognaths).

Dareste (1850) went too far when he criticized Cuvier
for not knowing that Triodon possessed ribs. In the first
edition of the "Le Regne Animal," Cuvier (1817) stated
that the plectognaths had only very small vestiges of ribs.
Triodon was not to be discovered until after the first edi-
tion of the "Le Regne Animal," and while its original de-
scription appeared in the second edition (1829), it was
still not known that Triodon possessed well-developed
ribs until Dareste published his 1849 paper. Cuvier could
scarcely have been expected to have known the osteology
of an as yet undiscovered fish, or to have personally in-
vestigated the ribs of every species offish, plectognath or
otherwise, described in the "Le Regne Animal."

The only features that Dareste (1850) believed the
plectognaths to have in common were a reduced oper-
culum, a rodlike interoperculum, and a low number of
vertebrae. To Dareste these characteristics were not im
portant enough to merit the recognition of the plectog
naths as an order, or even as a natural group.

One must admit that Cuvier's definition was im
precise, but Dareste temporarily neglected (see Dareste
1872c:1019) the fact that ordinal definitions are besi
founded on a combination of characters possessed by the
group in question and not by other fishes. Even ac
cepting Cuvier's far-from-perfect diagnosis, it is doubt
ful that any fishes other than plectognaths would fit into
it. Unfortunately, a significant number of plectognaths
also would be excluded from it, but this could be cured
by a little rewording.

Dareste proceeded to split the plectognaths into
(1872c: 117) "cinq petites families bien distinctes les unes
des autres" which were numbered as follows:

Premiere Famille - Diodon and Tetraodon

Deuxieme Famille - Triodon

Troisieme Famille - Orthagoriscus

Quatrieme Famille - Batistes (which included "les
petits generes Batistes, Alutere,
Monacanthe et Triacanthe")

Cinquieme Famille - Ostracion.

His definitions of these families were anatomically, and
especially osteologically, founded, and included not only
a good summation of the contemporary knowledge of the
plectognaths but also some original observations.
In his later publications Dareste (1872a, b, c) con-

fined himself to the placement of the scleroderms
"among the Acanthopterygians, in the vicinity of the
Acanthuri and other fishes belonging to the small family
of the Teuthyes" (1872b:68). However, he despaired of
showing the relationships of the gymnodonts, for he said
(1872c: 1088): "Ce type est assez difficile a definir, par
suit de la diversite des formes sous lesquelles il se
presente, et qui en font une famille par chaine, plutot
qu'une famille en groupe" and (1872c: 1089): "Le
Triodon differe notablement des autres Gymnodontes, et
forme par consequent un type a part, quoique voisin."
Dareste (1872b:69-70) pointed out the many charac-
teristics held in common by the "Acanttiuri and Balis-
tidae, especially the true Balistes, which are more nearly
allied to the Acanthuri than the Triacanthi, Monacanthi
and Aluterae." The particular features he noted in
Balistes and acanthurids were: firm and immovable
union of the premaxillary and maxillary; general shape
of skull; supraoccipital with a long forward extension and
an elevated crest; ethmoid elongate; parasphenoid as a
vertical plate anterior to the orbit; vomer small and
toothless; palatine small, toothless, and movable; in-
teroperculum at least partially rodlike and hidden
beneath the preoperculum; pelvis elongate, its two
halves more or less fused together; few vertebrae. In com-
parison to these similarities, Dareste said (1872b:70) that
"The differences between the skeletons of the Acanthuri
and Balistes are but few and of slight importance," these
differences being: the presence of suborbitals in acan-
thurids and their absence in balistids; the spiny dorsal
and soft dorsal fins of acanthurids being continuous, but
separated in balistids; the branchial aperture of acan-
thurids being larger than in balistids, due to the shape of
the preoperculum; the acanthurids possessing true ribs,
but not balistids.

Great credit is due to Dareste for having so definitely
pointed out these similarities. They are for the most part
correct, although just as Dareste had torn apart Cuvier's
definition so also could a person familiar with both acan-
thurids and balistids find many osteological details at
variance with the above information. But the only
reasonable ana useful criticism that can be made of
Dareste's scheme is that the comparison was made
between acanthurids and balistids rather than between
acanthurids and triacanthoids. The triacanthoids are ob-
viously the most generalized of the plectognaths and the
balistids show every indication of having arisen from
triacanthoid ancestors, perhaps not long after the
triacanthoids had themselves arisen from an ancestral
stock which also gave rise to the acanthurids. Dareste's
arguments are still convincing, but they would have been
even more so had the comparison been between the acan-
thurids and triacanthoids.

At the same time that Dareste was attempting to dis-
mantle the Order Plectognathi, another Frenchman took
a diametrically opposed view. Hollard stands second to
no one in the study of the plectognaths. While Dareste's
primary contribution was to relate scleroderms to acan-
thurids, Hollard's was to describe the anatomy and rela-
tionships of the families within the Plectognathi.

HoUard's (1853, 1854a, b, 1855) first contribution
was the "Monographie de la Famille des Balistides,"
which described the differences and similarities of Ba-
tistes, Monacanthus, and Triacanthus, the three genera
he recognized. Whereas Alutera had previously been
recognized as a distinct genus by many workers, Hollard
relegated it to subgeneric rank, with the genus Mona-
canthus containing two subgenera, Monacanthus and
Aluteres. He believed these two types to be essentially
the same, except that the posterior end of the pelvis
protruded and was covered by modified scales (his
"pointe pelvienne") in Monacanthus, while the pelvis
remained covered by skin in Aluteres. One might notice
that in the excellent review of the group by Berry and
Vogele (1961) this concept still stands. Not only did
Hollard described the general anatomy, including os-
teology, viscera, integument, muscles, etc., but he also
gave systematic descriptions of all of the then known
species of balistoids and triacanthoids and described
many new species.

Several years later, Hollard's (1857a) "Monographie
de la Famille des Ostraciontides" treated trunkfishes in
the same fine style as he had done with the other sclero-
derms. Hollard pointed out that, in the progression from
Triacanthus to Batistes to Monacanthus and Aluteres,
the pelvic fin and spiny dorsal fin elements decrease in
number and that in Ostracion there is no trace of either.
Thus, to Hollard (1857a: 125) the acanthopterygians and
malacopterygians "ne s'en separent pas d'une maniere
absolue, et que le caractere tire de la presence des rayons
epineux n'a qu'une valeur relative et conditionelle, bien
inferieure a celle que lui attribuaient Artedi et G.
Cuvier." Regarding the actual classification of the trunk-
fishes, it must be remembered that at this time a great
number of new species of plectognaths were being made
known by Bleeker and a host of others, and that Kaup
(1855) had just reviewed the ostracioids. Kaup synony-
mized many of Bleeker's species and defined a number of
new generic groups, as follows: Cibotion Kaup,
Laetophrys (sic) Swainson, Ostracion Linnaeus, Acerana
(sic) Gray (with four subgenera — Acerana Gray,
Capropygia Kaup, Kentrocapros Kaup, Anoplocapros
Kaup), Centaurus Kaup (for the larval Mola described
by Richardson [1845:52] as Ostracion boops). Hollard
did not wish to recognize as many generic and sub-
generic categories as Kaup had given. Whereas Kaup
had placed great importance on the number and place-
ment of cuirass spines, Hollard thought them to be of a
superficial or secondary nature and of no real value to an
understanding of the phylogeny of the trunkfishes. Thus,
Hollard recognized only two genera, Ostracion and
Aracana. Since Hollard had never seen a specimen of
Ostracion boops, he was correctly skeptical of it and
mentioned it only in passing — quite in contrast to
Kaup's establishing a new genus for it.

The next in Hollard's (1857b) series of papers was the
"Etude sur les Gymnodontes," of the same high quality
as the preceding works. The osteological illustrations
that Hollard presented here were to be used by all sub-
sequent workers on plectognaths, and they were even

reproduced by Gill (1892a). Hollard had at his dis-
position the unpublished manuscript left upon the death
of Bibron, who had been reviewing the pufferfishes and
had established in his manuscript numerous new genera,
some of which were in name only, for the diagnoses had
not yet been written. This manuscript was published al-
most in its original condition, with only a few notes add-
ed by Dumeril (1855), its editor. Bibron had completed
the diodontids, but was still working on the tetraodon-
tids at the time of his death. Suffice it to say that, on the
whole, Bibron's generic groupings were highly artificial
and are unacceptable in a modem phylogenetic classi-
fication. Hollard, for example, synonymized seven of
Bibron's genera {Dilobomycter, Aphanacanthus,
Amblyrhynchotus, Stenometopus, Geneion,
Epipedorhynchus, Promecocephalus) into one genus, to
which, unfortunately, he gave yet another new name,
Apsicephalus.

The last, summarizing, article in Hollard's (1860)
series was the "Memoire sur le Squelette des Poissons
Plectognathes," in which he condensed all his
anatomical observations and arrived at the following
classification (p. 46; genera were not listed, but are here
inserted below as he recognized them in his previous
three articles):

Plectognathes ou Echinoides
Sclerodermes
Balistides
Triacanthiens
TYiacanthus
Balistines
Batistes
Monacanthiens
Monacanthus

subgenus Monacanthus
subgenus Aluteres
Ostracionides
Aracaniens
Aracana
Ostraciens
Ostracion
Gymnodontes
Loganiosomes ou
Triodoniens
Triodon
Spherosomes (Orbes epin.)
Tetrodoniens
Rhynchotus
Xenopterus
Batrachops
Apsicephalus
Brachycephalus
Monotreta
Diodoniens
Diodon

mes ou
Orthagorisciens

Ort- agoriscus.

The above classification was rarely accepted in its en-
tirety by subsequent researchers on the plectognaths,
and it was not a radical departure from previous ar-
rangements of the order. By present-day standards,
Hollard's osteological observations were often incorrect,
with two inaccuracies being of particular prominence.
The bone now called the parasphenoid was thought by
Hollard to be composed of two entities, a large anterior
piece or "sphenoide anterieur" and a small posterior
piece or "sphenoide posterieur." What is presently term-
ed the frontal was recognized as such by Hollard, except
that he believed a small portion at its posteromedial end
to be separated from it, and called this small piece the
"parietal." But Hollard had to rely, for the most part, on
dried skeletons, and if he occasionally saw a few too
many bones in the cranium, it is readily under-
standable. With the present techniques of clearing and
staining, the sutural regions between bones are much
easier to define, and surface sculpturing of individual
bones is less likely to be confusing. Hollard's con-
tribution, then, was that he systematically made known
for the first time the general morphological structure of
all of the primary types of plectognath fishes.

Coming after Dareste's (1850) critique of Cuvier's or-
dinal recognition of the Plectognathi but before the
general plectognath classification of Hollard (1860) was
the "Ichthyologie Analytique" by Andre Dumeril (1856),
father of the August Dumeril who had edited Bibron's
manuscript. In probable deference to Dareste, the elder
Dumeril did not recognize an Order Plectognathi, but, on
the other hand, his families Sclerodermes and "Gym-
nognathes" followed one another in his subclass Chon-
drostichthes, as outlined in the large folding sheet
between pages 92 and 93:

Famille Lophobranches (syngnathids)

Famille Podopteres (Cydopterus, Lophius, anten-

nariids, ogcocephalids, etc.)
Famille Sclerodermes (scleroderms)
Famille Gymnognathes (gymnodonts)
Famille Hypostomates (climaerids, Polyodon, Aci-

penser, Pegasus).

Whereas the two plectognath groups were listed one
after the other in the folding sheet, in the text they were
described discontinuously. Dumeril did not seem to have
any confidence in the validity of the Plectognathi. In the
section on "Les Chondrostes Gymnognathes," Dumeril
(1855:159-160) simply and conservatively listed the four
genera he recognized: Diodon, Triodon, Tetraodon,
and "Cephale" (= Mola). "Les Chondrostes Sclero-
dermes" (p. 173-182) were treated in only a slightly
more detailed manner. Two groups were recognized, the
Ostracides and the Balistides. There were four genera of
trunkfishes {Ostracion, Aracana, Cibotion, and
Doryophrys) and the standard four genera (Monacanthe,
Alutere, Triacanthe, and Baliste) in his Balistides. One
will notice that the term "Gymnognathes" (of Bleeker)
replaced Cuvier's "Gymnodontes." But the term Gym-
nodontes was not simply stricken from the record, for

Dumeril used it as the name of one of the families of his
Ordre Hemisopodes. This "Famille Gymnodontes" in-
cluded such items as Gerres, sparids, Upeneus, mullids,
etc.

In Dumeril's (1806) much earlier "Zoologie Analyti-
que," the fish classification of Lacepede was followed
with only the addition of a large number of familial and
ordinal names.

Perhaps the fish classification that appears in ths vari-
ous editions of the Encyclopaedia Britannica can be
taken as an example of the standard or average condition
of our knowledge of that subject. Richardson's (1856)
article on "Ichthyology" (8th edition) recognized the
Order Plectognathi (p. 312-314) within the teleosts. For
the arrangement of the families of plectognaths Richard-
son followed Kaup (1855). Kaup, however, had not fin-
ished his major groupings of plectognaths in that paper,
and Richardson filled in the missing categories with
what he thought to be Kaup's logic, as follows:

Family Balistidae
Sub-Family Balistini: Pyrodon, Melichthys,

Xanthichthys, Canthidermis, Batistes, Balistapus.
Sub-Family Monacanthini: Monacanthus, Aluterius,
Triacanthus.
Family Ostraciontidae

Cibotion, Doryophrys, Ostracion, Aracana (with
four subgenera: Aracana, Capropygia, Kentrocapros,
Anoplocapros) , Centaurus.
Family Diodontidae

Sub-Family Diodontini: Diodon, Dicotylichthys,

Cyclichthys, Cyanichthys, Chilomycterus.
Sub-Family Tetraodontini: Tetraodon (with four
subgenera, after Miiller: Physogaster, Chelono-
don, Cheilichthys, Arothron).
Sub-Family Orthagoriscini: Orthagoriscus.

There are two obvious errors in the above scheme, one
of omission and the other of commission. The unique
Triodon is not even mentioned, and all of the triacan-
thids (Triacanthus) were placed in the Monacanthini.
The explanation is that when Kaup treated the
Balistidae he included all the balistid species he de-
scribed under the Sub-Family Balistini, but did not
mention £uiy other subfamily since he was not describ-
ing any triacanthoids or monacanthids. Richardson as-
sumed that if in the Balistidae there was a Sub-Family
Balistini, then there should be a counterbalancing sub-
family, so he set up the Sub-Family Monacanthini to in-
clude all the other nonostracioid scleroderms. This treat-
ment of the Plectognathi was later adopted by Fitzinger
(1873), but after that it fortunately disappeared from
use.

After the turn of the 18th century there was an almost
unanimous, with the few exceptions to be discussed later,
belief that the plectognaths of Cuvier were indeed a
natural group, although there would continue to be much
discussion about the rank at which they were to be recog-
nized. Attention was then turned to better descriptions,
diagnoses, and eirrangements of the species within the

order. This task would have been long delayed had it not
been for Bleeker, that great describer of the Indo-Aus-
tralian ichthyofauna. As with so many other orders of
fishes, a knowledge of the Indo-Pacific representatives of
the plectognaths is indispensable to an understanding of
their evolution. Here is to be found a far greater number
and diversity of species of plectognaths than in all other
areas of the world combined. If the Indo-Pacific was not
the cradle of the evolution and distribution of the plec-
tognaths, then at least it is their present center of
greatest speciation and morphological diversity.

It is impossible to give here a full review of Bleeker's
contributions to the study of plectognaths, or to mention
more than a few of the numerous papers in which he de-
scribed new species of that order. Of particular descrip-
tive, but not classificatory, interest is Bleeker's
(1852a) "Bijdrage tot de Kennis der Blookakige (Gym-
nognathen)" in which numerous new species were de-
scribed and his (1852b) "Bijdrage tot de Kennis der
Balistini en Ostraciones." Of most importance to this
discussion, however, is Bleeker's complete classification
of the plectognaths, which appeared in two places: in
volume 5 (1865) of his monumental "Atlas Ich-
thyologique" and in his 1866 "Systema Balistidorum,
Ostracionidorum, Gymnodontidorumque Revisum." The
classification was the same in both works, and that
which is here condensed is that of the "Systema" only
because the definitions of the categories were more com-
plete there and all of the ordinal names were latinized,
which they were not in the "Atlas." Bleeker never used
the term Order Plectognathi, since he gave each of his
three major groups of plectognaths ordinal rank, but he
obviously thought of them as a natural assemblage, as
examination of the titles of his articles will show. His
classification was as follows:

Ordo Balistidi
Familia I. Triacanthoidei
Subfamilia I. Triacanthiformes
TriacanthusA Acanthopleurus
Subfamilia II. Paratriacanthiformes
Triacanthodes, Hollardia
Familia 11. Balisteoidei

Subfamilia I. Balistidiformes
Leiurus, Erythrodon, Melichthys, Balistes (with
five subgenera: Parabalistes, Pseudobalistes, Ba-
listapus, Balistes, Canthidermis)
Subfamilia II. Monacanthiformes
Phalanx 1. Monacanthini
Monacanthus, Chaetodermis, Paramonacan-
thus, Amanses, Pseudomonacanthus, Liomona-
canthus, Oxy monacanthus
Phalanx 2. Aluterini
Brachaluteres, Acanthaluteres, Ceratacanthus,
Paraluteres, Pseudaluteres, Aluteres
Phalanx 3. Psilocephalini
Psilocephalus

Ordo Ostracionidi
Familia Ostracionoidei

Ostracion (with four subgenera: Ostracion, Laeto-
phrys, Tetrosomus, Acanthostracion), Aracana (with
four subgenera: Aracana, Capropygia, Kentrocapros,
Anoplocapros) , Centaurus
Ordo Gymnodontidi
Familia I. Orthagoriscoidei

Orthagoriscus
Familia 11. Tetraodontoidei
Subfamilia I. Diodontiformes
Phalanx 1. Trirhizacanthini

Chilomycterus, Diodon
Phalanx 2. Dirhizacanthini
Atopomycterus, Paradiodon, Trichodiodon
Subfamilia 11. Tetraodontiformes
Phalanx 1. Tetraodontini

Tetraodon, Crayracion, Leiodon, Chonerhinus,
Ephippion
Phalanx 2. Canthogasterini
Canthogaster
Familia HI. Triodontoidei
Triodon.

The precise separation of the triacanthids from the
triacanthodids and of the tetraodontins from the canthi-
gasterins appeared here for the first time. Although his
groupings of the balistoids owed much to the work of Hol-
lard, they were well defined and concisely executed. If
one bears in mind that what Bleeker called Diodon we
now refer to as Chilomycterus, then one sees that he also
realized the natural division of the diodontids into those
with erectile two-rooted spines (Phalanx Dirhiza-
canthini) and those with permantly fixed three-rooted
spines (Phalanx Trirhizacanthini). In short, Bleeker
made good use of what had come before him and added
much new information based on both observation and
synthesis in his own right. His is a reasonable classifica-
tion even today, and at the time of its appearance it was
superb.

Owen (1840 & 1845) in his "Odontography" gave what
are now the classic descriptions of the teeth of plectog-
naths. The fish classifications he used in his various
publications were not his own, but they are cited here to
show the variable handling of the plectognaths that one
finds up until the turn of the 19th century. In his "Odon-
tography," which appeared several years after Agassiz's
(1833) "Poissons Fossiles," Owen placed the plectog-
naths in two groups (scleroderms and gymnodonts)
within the ganoid fishes, and an Order Plectognathi was
not recognized. But in his "Lectures on Comparative
Anatomy," which appeared after Miiller's (1844) "Ueber
den Bau," Owen (1846) recognized the Order Plectog-
nathi, with three families: Balistinae, Ostraciones,
and Gymnodontes. The same plectognath groups were
recognized by Owen in his (1853) "Descriptive Cata-
logue of the Osteological Series," but a new term for a

higher plectognath category appeared in his (1866) "On
the Anatomy of Vertebrates." In the latter work, the
Order Plectognathi was composed of two sub-
orders: Sclerodermi, with the family Balistini; and
Apleuri, with the families Ostraciontidae and Gym-
nodontidae.

This is not the place to discuss the histological struc-
ture and composition of the plectognath teeth and scales,
nor to attempt to summarize the highly conflicting infor-
mation available. But since some recent workers are not
aware of all the older literature dealing with this sub-
ject, it is useful to list the more important references to
the histology of the scales and teeth of plectog-
naths: Bom 1827 (teeth of gymnodonts). Owen 1839
(teeth of Tetraodon and Diodon); 1840 & 1845 (teeth of
Balistes, Tetraodon, and Diodon). Williamson 1851
(scales of Balistes and Ostracion). Hollard 1857a
(scales of Ostracion). Cleland 1862 (bony tubercles in
skin of Mola). Turner 1862 (bony tubercles in skin of
Mola). Hertwig 1881 (scales of numerous plectog-
naths, mostly gross morphology, but some his-
tology). Hilgendorf 1886 (mostly gross morphology of
teeth of Tetraodon and Diodon); 1893 (teeth of Mola,
Tetraodon, and Diodon). Wortman 1886 (teeth of
Balistes and Diodon). Green 1901 (subdermal tissue of
Mola). Green and Tower 1902 (albuminoids in scales of
Mola and Sphoeroides) . Ghigi 1905a, 1905b (teeth of
Balistes); 1921 (teeth of Ostracion and
Tetraodon). Rosen 1913 (scales of numerous plectog-
naths); 1916b (scales of plectognaths in comparison
with those of other fishes). Kaschkaroff 1914a (scales of
numerous plectognaths). Rauther 1919, 1927a (scales of
Balistes); 1927b (teeth of numerous plectognaths).
Tretjakoff 1924a, b, 1926a, b (scales of numerous
plectognaths); 1925 (pectoral fin rays of Tetraodon com-
pared with body scales); 1926c (teeth of numerous plec-
tognaths). Pflugfelder 1930 (teeth of Tetraodon and
Diodon in particular, but also of Ostracion and
Balistes). Grieb 1935 (pharyngeal teeth of
Sphoeroides). Arsuffi 1939 (teeth of Tetraodon). Bar-
tolini 1941 (general discussion of plectognath
teeth). Isokawa 1955 (teeth of Monacanthus and
Cantherhines). Soule 1969a, b, 1970 (tooth attach-
ment in balistids). More recently Roberto Andreucci
and his colleagues in Brazil have begun a systematic
reexamination of the histology of plectognath teeth (so
far confined to gymnodonts): Andreucci 1966a, b,
1967a, b, 1968a, b, 1969, 1970; Andreucci and
Britski 1968a, b, 1969a, b, 1970, 1971; Britski and
Andreucci 1971; Andreucci and Blumen 1971.

A landmark in the study of almost all groups of fishes
was the publication of Gunther's "Catalogue of the Fishes
in the British Museum." Here was a consistent, stan-
darized description of nearly every order, family, genus,
and species of known fishes. The Plectognathi appeared
in volume 8, published in 1870. In the subclass Teleostei,
six orders were recognized: Acanthopterygii,
Acanthopterygii Pharyngognathi, Anacanthini,
Physostomi, Lophobranchii, and Plectognathi. The plec-
tognaths were divided as follows:

Order Plectognathi
Family Sclerodermi

First Group: Triacanthina

Triacanthodes, Hollardia, Triacanthus

Second Group: Balistina

Balistes (with six subgenera: Liurus, Balistes,
Canthidermis, Parabalistes, Melanichthys, Ery-
throdon), Monacanthus (with two subgenera:
Monacanthus and Aluteres), Anacanthus

Third Group: Ostraciontina

Ostracion (with two subgenera: Ostracion and
Aracana)
Family Gymnodontes

First Group: Triodontina
TYiodon

Second Group: Tetrodontina
Xenopterus, Tetrodon (with 10 subgenera: Tetro-
don, Hemiconiatus, Gastrophysus, Cheilichthys,
Liosaccus, Crayracion, Chelonodon, Monotretus,
Arothron, Anosmius), Diodon, Chilomycterus,
Dicotylichthys, Atopomycterus, Trichodiodon,
Trichocyclus

Third Group: Molina

Orthagoriscus (with two subgenera: Orthagoriscus
and Ranzania).

The classification of the Plectognathi followed by
Gunther (1880) in his "An Introduction to the Study of
Fishes" was essentially the same as that given above.

There was nothing special about Gunther's (1870) treat-
ment of the Plectognathi, and it could be held to be dis-
tinctly inferior to that given 4 years earlier by Bleeker
(1866). Gunther's most notable failure was his handling of
the admittedly difficult "Tetrodontina." The tetraodon-
tids are a far larger and more diversified group than the
diodontids, yet Gunther recognized only two genera of the
former (albeit with 10 subgenera, but these were based
primarily on the form of the nostril, an unreliable charac-
ter when used alone rather than in conjunction with a
number of other anatomical systems) but six genera of
the latter (defined by the form of the nasal tentacle and
the movability of the spines).

In his (1871a) "Observations on the Systematic
Relations of the Fishes," Cope, while giving credit to
Muller as (p. 580) "the father of modern ichthyology,"
nevertheless thought that within the teleosts the Plectog-
nathi and Lophobranchii were not of equal rank with
the Physostomi and Physoclisti. Thus, within the Sub-
class Actinopteri (= Muller's Teleostei and Ganoidei),
Cope recognized three tribes: the Chondrostei ( =
Mailer's Order Chondrostei of the Subclass Ganoidei),
Physostomi, and Physoclisti. The Physoclisti contained
10 orders, including the Plectognathi. Cope stated
(1871a:582) that phylogenetic lines radiate out from the
Percomorphi and that "one leads from the Chaetodon-
tidae, through the Acroneuridae [= Acanthuridae], to
the Plectognathi, by the similarity in the arrangement of

the posttemporal and forms of the pharyngeal ap-
paratus." In the same year Cope (1871b) again ex-
pressed the idea that the Plectognathi and Lopho-
branchii were natural groups, but that Miiller had given
them too high a rank in comparison with the other
teleosts. Cope's opinion is almost unanimously accepted
today.

Winther (1877), in part three of his "Fiskenes Ansigt:
en Comparativ-Anatomisk Unders«igelse," briefly de-
scribed the general anatomy, especially of the head
region, of such representative plectognaths as
Triacanthus, Batistes, Monacanthus. Ostracion,
Tetraodon, Diodon, and Mola. He was the last person
with a knowledge of the plectognath fishes as a whole
who did not believe that they were a natural group. Like
others before him, Winther criticized Cuvier's character
of the intimate fixation of the upper jaw bones to one
another as being neither unique to, nor characteristic of,
all of the plectognaths. He believed the plectognaths to
represent three families: Triacanthini, Sclerodermi (in-
cluding balistoids and ostracioids), and Gymnodontes.
He was unsure of the relationship of the triacanthoids to
his Sclerodermi, but he concurred with Dareste that the
balistoids were related to the acanthurids. For some
unaccountable reason, Winther thought that the Gym-
nodontes were related to the Discoboli (Cyclopterus).

A good summary of the then prevailing feeling about
plectognaths by knowledgeable ichthyologists is found
in Gill's (1882) "Arrangement of the Families of
Fishes . . .," published shortly after Bleeker's "Atlas,"
Giinther's "Catalogue," and Cope's "Systematic Relation-
ships." While Gill's arrangement was drawn up to serve as
the fish filling system to be used at the Smithsonian In-
stitution, it was more than a mere list of families, for Gill
gave numerous comments on the phylogeny of the higher
categories. After reviewing the characteristics of the plec-
tognaths. Gill expressed the contemporary view that
(1872:XLI) "the many common characters justify their
association together, and the characters that are peculiar
to them sanction their isolation as a group," while "Their
differences sink into comparative insignificance." Gill
agreed that the scleroderms "have been deemed more
related to ordinary Acanthopterygian types than to the
other admitted Plectognaths. And it is quite true that
they (and especially the Triacanthids) are much more
similar to the ordinary fishes than are the typical Plectog-
naths. This, however, is quite explicable by the sup-
position that they are the most generalized, and repre-
sent the immediate line of descent . . . ." It was thus
agreed that the work of Dareste and others related the
scleroderms to the acanthurids and that the gym-
nodonts were obviously highly specialized offshoots of
the scleroderms. This was then, as it is now, the more or
less accepted view. In the actual arrangement used by
Gill, the Plectognathi were divided into three major sub-
groups, as had been done by Muller and Bleeker, in-
stead of into two major subgroups, as had Cuvier,
HoUard, and Giinther. The families that Gill recognized
in each of the three subgroups were those as recognized
by Bleeker.

Several years later Gill (1885) published his "Synop-
sis of the Plectognath Fishes." This was a skillful diag-
nosis, based almost entirely on the literature, of the
characteristics of the various plectognath subgroups, and
of the nomenclatural labyrinth that surrounds the order.
Gill relied heavily for the osteological information in his
diagnoses on the various publications of HoUard, whose
papers he thought had been unwisely neglected. Gill's
classification of the plectognaths was much more
elaborate than that of his "Arrangement of the Families
of Fishes," and although it has much in common with
Bleeker's system, it is sufficiently different to merit
citation:

Order Plectognathi
Suborder Sclerodermi
Family Triacanthidae
Subfamily Triacanthodinae
Subfamily Triacanthinae
Family Balistidae
Subfamily Balistinae
Subfamily Monacanthinae
Subfamily Psilocephalinae
Suborder Ostracodermi

Family Ostraciontidae
Suborder Gymnodontes
Superfamily Triodontoidea

Family Triodontidae
Superfamily Tetrodontoidea
Family Tetrodontidae
Subfamily Tetrodontinae
Subfamily Colomesinae
Family Psilonotidae
Family Chonerhinidae
Superfamily Diodontoidea

Family Diodontidae
Superfamily Moloidea
Family Molidae
Family Molacanthidae.

The above scheme differs only slightly from Bleeker's
handling of the scleroderms, the difference being that the
elongate monacanthid Psilocephalus was elevated by
Gill to familial rank. In regard to the gymnodonts. Gill
recognized a great many more higher categories than had
Bleeker, but few of Gill's extra divisions have continued
to be recognized. To Gill's credit, however, was the fact
that all of the subgroups were diagnosed, and his sys-
tematic handling of the maze of names that have been
applied to the various plectognath groups was a boon to
all subsequent researchers.

A number of years later Gill (1892a) published his
"Notes on the Tetraodontoidea," in which he expanded
on his nomenclatural treatment of the tetraodontids,
without any basic change in the classification he had
previously used. In a series of notes, Gill (1888a, b,
1889, 1892b, 1897) dealt effectively with a number of
other nomenclatural problems which need not be dis-
cussed here. In the same vein, Eigenmann's (1885) only

paper on plectognaths might be noted, since he worked
out the 36 specific names that had been applied to the six
species of diodontids that occur off the American coasts,
while that of Goode (1880) did the same for ostraciids.

No mention has yet been made of works on fossil plec-
tognaths, other than Agassiz's species descriptions, for
the simple reason that until part IV of Woodward's
"Catalogue of Fossil Fishes" (1901) appeared, there was
no systematic account of them. The order Plectognathi
was not recognized by Woodward; rather, the plectog-
naths occurred in the families Balistidae and Gym-
nodontidae, which, along with the Chaetodontidae and
Acronuridae (= acanthurids), composed the Division
Chaetodontiformes of the Suborder Acanthopterygii. Ad-
mittedly the higher categories adopted by Woodward for
the purpose of his pioneering catalogue were somewhat
artificial, but it is significant that the one person who
had reviewed the fossil plectognaths felt that they were
so closely related to the acanthurids and chaetodontids
that all of these groups should be placed in the same
higher category.

The plectognaths were the first of the long series of
orders that Regan (1903a) revised; his "On the Classi-
fication of the Fishes of the Suborder Plectognathi" was
in sharp contrast to Gill's plectognath work. Whereas
Gill concentrated on examining the literature, and par-
ticularly Hollard's figures, Regan made direct use of the
skeletal material of the British Museum and nearly com-
pletely shunned nomenclatural matters. Gill having al-
ready straightened out most of the latter. Not only were
the research styles much different but the classifications
arrived at were equally dissimilar.

Regan recognized the Plectognathi as one of the sub-
orders of the Acanthopterygii in close relationship with
the acanthurids. He (1903a:285) felt that the "feature of
most importance in diagnosing the suborder Plectog-
nathi is the absence of ribs, although in some well-os-
sifed epipleurals are present which have been mistaken
for ribs," but he pointed out that, like everyone else since
Dareste and HoUard, he had not had the opportunity to
examine Triodon and only supposed that it also lacked
ribs. On the other hand, Regan believed that "the
coalescence of the teeth in the jaws is a feature of little
importance, and has, as probably as not, originated in-
dependently . . ." in the gymnodonts and scleroderms.
One could not do justice to Regan's classification without
quoting in full his brief but excellent osteologically
grounded diagnoses for the various subdivisions, but
space forbids and in outline his system was as follows:

Suborder Plectognathi
Division Sclerodermi

Family Triacanthidae: Triacanthus, Triacanthodes,

Halimochirurgus
Family Triodontidae: Triodon
Family Balistidae: Balistes, Monacanthus, Para-

luteres, Pseudaluteres, Pseudomonacanthus,

Alutera, Psilocephalus
Family Ostraciontidae: Aracana, Ostracion, Lac-

tophrys

Division Gymnodontes
Family Tetrodontidae: Tetrodon, Ephippion,

Tropidichthys, Chonerhinus, Xenopterus
Family Diodontidae: Diodon, Lyosphaera
Family Molidae: Mola, Ramania.

The most obvious feature of Regan's classification is
that it was much more conservative than that of such
predecessors as Bleeker and Gill, but the differences were
more basic than a simple and inevitable "lumper-split-
ter" disagreement. The unique Triodon had always been
placed in the Gymnodontes because of the fusion of its
teeth, following the lead of Cuvier (1829) in the second
edition of the "Le Regne Animal." Regan (1903a:285),
however, thought that "the structure of the pectoral arch
and vertebral column, as well as the presence of a pelvis
and well-ossifed epipleurals" related it to the triacan-
thoids and balistids, and thus Triodon for the first time
appeared in the Sclerodermi. Contrary to Miiller and
Bleeker, who recognized three subdivisions of the Plec-
tognathi, Regan followed Hollard and Giinther in recog-
nizing only the original two Cuvierian subdivisions, for
the "Ostraciontidae do not seem to me to differ suf-
ficiently from the Sclerodermi to rank as another
division— Ostracodermi" (Regan 1903a:285). Whereas
Gill recognized three families of tetraodontids, Regan
preferred a single family. Regan pointed out that Gill
based his diagnoses of the tetraodontid families on the
figures given by Hollard of the top of the skull of various
puffers, but that "the figures of Hollard have met with
too ready an acceptance, that author having mistaken
ridges on and fissures in the frontal bones for sutures" (p.
293). Regan's objection to Hollard's figures and Gill's use
of them was entirely correct. In short, Regan gave
excellent diagnoses for the limited number of plectognath
subgroups that he recognized, and corrected numerous
osteological errors that had prevailed up to that time. Un-
fortunately, as was typical of most of Regan's revisions,
only a few figures were given to accompany the great
amount of osteological information that was discussed in
the text. It is impossible, for example, to tell what Regan
meant by the presence of an "ossified praeorbital" in the
Balistidae. But these are relatively minor criticisms of
the work'that became the standard treatment of the ma-
jor plectognath subgroups.

Regan's only other effort devoted solely to the plectog-
naths was a brief, but useful at its time, revision of the
genus Triacanthus (1903b). In the only complete fish
classification that Regan ever presented, his "Fishes"
(1929) in the 14th edition of the Encyclopaedia Britan-
nica, the plectognaths were discussed and diagnosed as
in his 1903 revision of the order. But now the Plectog-
nathi were an order of the Neopterygii rather than a
suborder of the Acanthopterygii. Thus, in the years that
intervened between his revision of the plectognaths and
his general fish classification, the plectognaths would
seem to have grown in stature in Regan's view.

After Regan's (1903a) plectognath classification, the
diagnoses that were there given were used to a greater or

lesser extent by all subsequent ichthyologists. For exam-
ple, Boulenger's "A Synopsis of Suborders and Families
of Teleostean Fishes" (1904a) and "Fishes, Systematic
Account of Teleostei" (1904b) followed Regan without
deviation, and Boulenger's (1907) only other interest in
plectognaths was his lateral view illustration of the
skeleton of the Nile puffer, Tetraodon fahaka (=
lineatus Linn.).

On the other hand, Jordan (1905) accepted Regan's
work more critically, for while sparingly using Regan's
diagnoses, Jordan had his own ideas about the plectog-
nath subdivisions. For one thing, Jordan continued to
recognize three suborders of the Plectognathi, with the
Ostracodermi being one of them, rather than being
relegated, in Regan's fashion, to familial rank within the
Sclerodermi. Like Regan, Jordan (1905, 2:411) wished to
stress the close association of the plectognaths and
acanthurids, and did so by recognizing a "Series Plec-
tognathi" which is "derived directly from the Acan-
thuridae, from which they differ by progressive steps
of degeneration." But since the plectognaths "differ
from one another more widely than the highest or most
generalized forms differ from the Acanthuridae, we do
not regard it as a distinct order. The forms included in it
differ from the Acanthuridae much as the swordfishes
differ from ordinary mackerel" (Jordan 1905, 2:411). It is
a matter of opinion whether the magnitude of dif-
ference, external and internal, between such plectog-
naths as Mola and Triacanthus is of a higher degree
than that between the swordfish and mackerel, but in
any case Jordan was restating Dareste's acanthurid-plec-
tognath relationship. In Jordan's somewhat vague ter-
minology, the Chaetodontidae, Zanclidae,
Acanthuridae, and related forms comprised the
Squamipinnes, which gave rise to the Series Plectog-
nathi. But in 1898 Jordan and Evermann, in their "The
Fishes of North and Middle America" (part 2) made the
same point, the Plectognathi being called either an
"Order" (p. XVU) or a "Group" (p. 1696); and in 1923,
Jordan, in his "A Classification of Fishes," recognized an
Order Plectognathi with the following conservative sub-
groupings:

Order Plectognathi
Suborder Sclerodermi

fFamily Spinacanthidae [for all fossil triacanthoids]

Family Triacanthidae

Family Balistidae

Family Monacanthidae

Family Psilocephalidae
Suborder Ostracodermi

Family Ostraciidae
Suborder Gymnodontes

Family Triodontidae

Family Tetraodontidae

Family Canthigasteridae

Family Diodontidae

Family Molidae.

Of more importance than the above classification were

Jordan's collaborative reviews of various families of plec-
tognaths, making known many new species and rede-
scribing poorly known ones: Jordan and Edwards (1887);
Jordan and Snyder (1901); Jordan and Fowler (1903). Be-
fore continuing on to other researchers, it should be noted
that Jordan's belief in the close relationship of the plec-
tognaths and acanthurids was reinforced by the osteologi-
cal observations of his colleague Starks (1907:217) on the
Siganidae and Acanthuridae.

An example of the combined effect that Woodward's
(1901) "Catalogue" and Jordan's numerous pub-
lications had on the subsequent handling of the plectog-
naths can be seen in Goodrich's (1909) "Cyclostomes
and Fishes." Woodward had used the Division
Chaetodontiformes for plectognaths, acanthurids, and
chaetodontids, while Jordan had stressed again and
again the close relationship of his "Series Plectognathi"
with his Squamipinnes. These messages were taken to
heart by Goodrich, for, even though he made free use of
Regan's osteological diagnostic information, he adopted
a modification of Woodward's and Jordan's views about
the placement of the Plectognathi. The eclectic result
was as follows:

Subtribe Chaetodontiformes
(of the Tribe Perciformes)

Division A, Squammipennes
Chaetodontidae
Drepanidae
Division B, Plectognathi
Subdivision A
Teuthididae
Siganidae
Acanthuridae
Subdivision B
Branch 1, Sclerodermi
Series 1

Triacanthidae
Balistidae
Monacanthidae
Series 2

Ostraciontidae
Branch 2, Triodontes

Triodontidae
Branch 3, Gymnodontes
A

Tetrodontidae
Diodontidae
B

Molidae.

Although the above "Plectognathi" was the logical ex-
tension of Jordan's championing of Dareste's ideas, it is
nevertheless the first time that the term Plectognathi
had included within its folds the acanthuridlike fishes.
More important, however, was the fact that the plectog-
naths proper (Subdivision B) were divided into three
major groups, one of which had never been recognized
previously at that level. The Triodontidae had formerly

22

been placed in either the Sclerodermi or Gymnodontes,
and thus had always had a subordinate position until
Goodrich gave it equal ranking with the two Cuvierian
divisions. However, Goodrich's (1909:439) remarks on
why Triodon had been so elevated were only that this is
"a family intermediate between the first and third sub-
groups, whose exact position it is difficult to deter-
mine." Thus, the new third category was only a matter of
convenience. In his (1930) "Studies of the Structure and
Development of Vertebrates," Goodrich followed exactly
the same classification.

The handling of the acanthurid-plectognath line in
Gregory's (1933) "Fish Skulls" was modified from that of
Goodrich. Gregory recognized a Balistoidei as one of the
23 subdivisions of the Acanthopterygii. The Balistoidei
contained five large assemblages: acanthurids, zan-
clids, siganids, teuthids, and plectognaths (Gregory
hesitated to give any of them a precise rank). His
Balistoidei was therefore equal to the expanded "Plec-
tognathi" (or Division B of the Subtribe Chaetodonti-
formes) of (Goodrich. Goodrich's "Squammipennes"
(Division A of the Subtribe Chaetodontiformes) became
the Chaetodontoidei of Gregory, a group equal in rank
with the Balistoidei. Gregory felt that the chaetodon-
tids, acanthurids, and plectognaths formed a natural
series, but he split apart Goodrich's Chaetodontiformes
because it was his opinion (1933:281) "that the cleft
between the typical chaetodonts and acanthurids is
much greater than that between the latter and the plec-
tognath stem as represented by the balistids." Just as
with Dareste, one wonders why the comparison was not
made between acanthurids and triacanthoids, which are
the most generalized of the plectognaths, rather than
between acanthurids and balistids, which are obviously of
triacanthoid derivation. Gregory even illustrated the
skull of a triacanthid (1933:282, fig. 160), but no men-
tion at all was made of it in the text. Highly useful as
Gregory's plectognath figures (1933:figs. 160-172) are for
obtaining a general idea of cranial construction, they
lack detail.

The tentative classification as seen in Gregory's (1933)
"Fish Skulls" was a great change from that Gregory
(1907) used in his much earlier "Orders of Teleostomous
Fishes." In the earlier work Gregory recognized the Plec-
tognathi as one of the 10 orders of the Superorder
Acanthopteroidei. The Plectognathi were subdivided in
the same way as Regan had done, with the exception that
the monacanthids were recognized as two separate
families (with the Alutera-like species as a distinct fami-
ly) rather than being placed in the Balistidae. One other
paper should be mentioned in connection with Gregory.
Gregory and Raven (1934) discussed the anatomy and
relationships of Mola and supported previous workers'
placement of the molids in close association with the
diodontids.

During the period 1912 to 1916 a number of papers of
great importance to the study of plectognaths were pub-
lished by Rosen in Sweden and Kaschkaroff in Russia.
Rosen produced five successive articles in his "Studies on
the Plectognaths" which dealt, respectively, with the

anatomy of the blood-vascular system (1912), the air-
sac and intestine (1913a), the integument (1913b), the
body muscles (1913c), and the skeleton (1916a). In the
introduction to the series, Rosen (1912:1) explained his
purpose to be the study of a number of anatomical sys-
tems, since "to base the building up of a phylogenetical
tree solely on one anatomical system is a great error
which far too many anatomists commit." He then
promised to "end this series of studies on the Plectog-
naths with a section in which all the phylogenetical
results attained independently from the study of each or-
gan will be discussed together" (1912a:2). Unfortunately,
such a last summary article on phylogeny never ap-
peared, the last of the series being the 1916 paper on the
skeleton, in which almost no phylogenetic conclusions
were drawn. One can go to his earlier papers for his ten-
tative views on the phylogeny of plectognaths, but he cer-
tainly would not have drawn his final conclusions until
after the osteological paper had appeared. The ten-
tativeness of the conclusions that will be quoted from his
earlier work should be stressed, for the series is a good
anatomical survey of the plectognaths, but his phylo-
genetic remarks based on the blood-vascular, intestinal,
and integumentary systems do not seem to me to be
substantiated by the osteology of the order. Rosen's ideas
were most clearly expressed in his paper on the air-sac
and intestine (1913a:19) in which he stated that, "The
Molids and the Ostraciontids, which possess no such sac,
must have developed independently of the other
families. The inquiry given above on the air-sac has
clearly shown that we have every reason to believe that
the families Triacanthidae, Balistidae (including the
genus Monacanthus), Diodontidae, and Tetrodontidae
form a continuous series of development. This is an evi-
dent proof against the present classification of the Plec-
tognaths into two subgroups, Sclerodermi and Gym-
nodontes, as proposed by Regan." Rosen's work on the
integument confirmed his opinions and he (1913c:25-26)
added that "The Molids and Ostraciontids are primitive
Plectognaths, both groups separately specialized." The
fourth paper in the series, like the fifth, contained no
phylogenetic remarks.

Regardless of how the above statements would have
been modified in a summarizing article, Rosen's series
was highly useful as a pioneering survey of the soft, as
well as the osteological, anatomy of the plectognaths. His
first four papers are presently the only publications that
even attempt to treat as a whole all of the main non-
osteological systems of the plectognaths, although Win-
terbottom's (1974) superb myological analysis is a major
step in that direction. Rosen added much new infor-
mation and reviewed the then existing knowledge.
Without going into details, a few of the leading struc-
tural peculiarities summarized by Rosen in his first
paper of the series should be mentioned as examples of
the type of information with which he dealt. He was par-
ticularly impressed by the fact, known since the time of
Cuvier, that Mola has a larger, and supposedly rather
primitive, number of valves in its conus arteriosus and
atrioventricular region than do the other plectognaths,

which are normal in this respect. This fact, along with his
observations of the primitive condition of the "circulus
cephalicus" vessels in Lactophrys and Balistes (the only
scleroderms in which he was able to examine the vas-
cular system) and of its highly specialized condition in
the gymnodonts, led Rosen (1912:19) to comment on
"the occurrence of a great number of both primitive and
highly specialized characters not only within the group
but also in the same form." Thus, he felt that "Judging
from the number of primitive characters that we have
found in some Plectognaths we have all reason to believe,
that this fish group has branched off rather early from
the Teleostean main stock" (1912:20).

On matters of osteology, and considering that he had
examined only four species of Plectognathi, Rosen's ob-
servations were generally quite accurate. He was able to
substantially improve our knowledge of the plectog-
naths and to correct one of Regan's major errors. Regan
had said in his diagnoses of the Sclerodermi and Gym-
nodontes that the former had all of the neural arches
forming single spines, while in the latter the anterior
vertebrae had bifid neural spines. Rosen correctly
pointed out that in balistoids the first vertebra has a bifid
neural spine and no bony roof over the neural canal. One
would probably disagree, however, with Rosen's assess-
ment of Regan, whose contribution Rosen (1916a:l) said
"does not seem to me to have increased our knowledge
either of the skeleton or of the phylogeny of the group."
Rosen's description of the developmental and adult
cranial osteology of Sphoeroides was well done, for he de-
scribed it with more precision than had been given to a
plectognath up until that time. But like Regan before
him, Rosen included very few figures of the structures he
discussed so well. However, since Rosen described these

was not until 3 years after he had finished his manuscript
that he had seen Rosen's first three articles. Kaschkaroff
(1914a:365) simply said that he found them "sehr in-
teressante." Whereas only a small, but highly sig-
nificant, part of Rosen's output was devoted to osteology,
the majority of Kaschkaroff s work concerned the os-
teology of the plectognaths and the histology of their
bones, and only minor excursions were taken into soft
anatomy. In this respect Kaschkaroff s treatment of the
order was similar to that of HoUard. Whereas HoUard
produced the first work that could be called a monograph
on the osteology of the plectognaths, Kaschkaroff
produced what until now is the last such work.

Kaschkaroff began his study at one of the end points of
plectognath radiation, for he (1914a:265) admitted that
"Meines Erachtens ist Orthagoriscus mola die interes-
san teste Form genannter Gruppe." That view has many
advocates, as attested to by the practically infinite num-
ber of distributional and natural history notes that have
been devoted to that species. Kaschkaroff proceeded to
describe, with many more illustrations than had Rosen,
the general osteology of members of all the major groups
of plectognaths, with the exception of Triodon, of which
he had no specimens. While on the whole Kaschkaroff s
work is highly valuable, it contains a number of serious
observational errors that will be mentioned in the os-
teological section of this monograph. Regardless of os-
teological errors, Kaschkaroff had intelligently examined
a large number of plectognaths and the conclusions at
which he arrived are important to note. These con-
clusions were best summarized in the diagram he
presented (1914a:359), while cautioning the reader that
this was only a chart of anatomical similarity and not
necessarily a phylogenetic tree.

Tetrodon Diodon Triodon Orthagoriscus Balistes Monacanthus Ostracion

Triacanthus /
Der Vorfahr

structures in some detail, it is not overly difficult to
reconstruct what he saw. The same cannot be said of
Regan's telegraphic summaries of his osteological obser-
vations.

In the middle of the 4-year period during which
Rosen's series was published, there appeared Kasch-
karoff s (1914a) "Vergleichendes Studium der Or-
ganisation von Plectognathi." This was evidently to be
the first of a series of anatomical monographs concerned
with "Materialen zur Vergleichenden Morphologie der
Fische," but, just as with Rosen, something in those
troubled times was to cut short the series. Kaschkaroff
was not aware of Rosen's work until after completing his
own studies, for he said at the close of his paper that it

The only outstanding peculiarity of the above was the
close association of Orthagoriscus (= Mola) with Triodon,
but there was no explanation in the text as to the reasoning
that lead to this association. Kaschkaroff went on to say,
however, that HoUard's classification was acceptable,
and the latter was presented by Kaschkaroff (1914a:360)
with the statement that "die alte Klassifikation von
Hollard erscheint mir vollkommen begrundet."

Kaschkaroff discussed what "Der Vorfahr" of the Plec-
tognathi should be, listing the characteristics of this
hypothetical ancestor. The list (1914a:359) added up, in
essence, to a triacanthoid with a large amount of carti-
lage in its skeleton. He did not believe that the plectog-
naths arose from an acanthurid stock, for the latter did

not possess a large amount of cartilage in the skeleton, a
characteristic he considered to be of great importance.
Rather, he felt that the plectognaths might possibly best
be placed in association with the Lophobranchii and
Lophius. If one wishes to place such an emphasis on the
amount of cartilage present in the skeleton of some plec-
tognaths, then one could just as well relate the order to
Cycloptenis as well as to Lophius, and return to the
Artedian Branchiostegi.

In his "Studien iiber die Flossenmuskulatur der
Teleostier," Grenholm (1923:245) supported Rosen's
view that "Die alte Zweiteilung der Plectognathen in
Gymnodontes und Sclerodermi ist nach meiner Ansicht
. . . schwer aufrecht zu erhalten," while at the same time
admitting that the Plectognathi were a distinct group
related through the Acanthuridae to the other
acanthopterygians. It is of interest that Grenholm's
research on the fin muscles and Rosen's on the soft
anatomy lead to the same conclusion, a conclusion not
supported by the great majority of osteologically based
works, nor by Winterbottom's (1974) general myology of
the plectognaths.

Whenever one wishes to identify most plectognath
genera, there is only one place to which one usually
turns, and that is to the long series of revisionary papers
by Eraser- Brunner (1935a, b, 1940a, b, c, 1941a, b, c,
1943, 1950, 1951). For species identification the
story is somewhat different. Fraser-Brunner
simply listed under the keyed-out genus the species
which he found to be recognizable. Only for the molids
and several genera of monacanthids were keys to the
species provided. However, since Fraser-Brunner
typically drew his generic lines rather finely, there are
usually only a few species within any one genus and the
practical matter of species identification is not as dif-
ficult as it might first appear. The exemplary service that
Fraser-Brunner performed will be appreciated by anyone
who attempts to identify plectognaths, especially those
from the Indo-Pacific.

Throughout the series Fraser-Brunner promised even-
tually to publish a comprehensive monograph on the
order, but, since that has not yet appeared, there is
presented below his tentative classification as brought
together from his various publications. No genera will be
listed by name, for with Fraser-Brunner the number
usually accepted was greatly increased:

Order Plectognathi
Suborder Balistoidea
Division Triacanthiformes

Family Triacanthodidae (10 genera)
Family Triacanthidae (2 genera)
Division Balistiformes

Family Balistidae (13 genera)
Family Aluteridae
Subfamily Aluterinae (21 genera)
Subfamily Anacanthinae (1 genus)
Suborder Ostraciontoidea

Family Aracanidae (6 genera)

Family Ostraciontidae

Subfamily Ostraciontinae (3 genera)
Subfamily Lactophrysinae (3 genera)
Suborder Tetraodontoidea
Division Moliformes
Family Molidae
Subfamily Ranzaniinae (1 genus)
Subfamily Molinae (2 genera)
Division Tetraodontiformes
Subdivision Tetraodontines
Family Canthigasteridae (1 genus)
Family Lagocephalidae
Subfamily Lagocephalinae (1 genus)
Subfamily Sphaeroidinae (3 genera)
Family Colomesidae (1 genus)
Family Tetraodontidae
Subfamily Tetraodontinae (2 genera)
Subfamily Arothroninae (1 genus)
Family Chonerhinidae (2 genera)
Subdivision Diodontines
Family Diodontidae (3 genera).

The above is obviously not complete, for Fraser-Brun-
ner has not yet assigned a place to Triodon, although he
(1943:4) did say that the "Tetraodontoidea are almost
certainly derived from the Triacanthiformes, a some-
what fragile connection between the two being provided
by the existing Triodon."

Fraser-Brunner's classification is basically similar to
that of Bleeker's, with, of course, the refinements made
possible by Fraser-Brunner's own skill and by the results
of the researchers who intervened between these two. It is
nevertheless a compliment to Bleeker that Fraser-
Brunner should, nearly a hundred years later, arrive at
essentially the same classificatory scheme — although
Fraser-Brunner's diagnoses are, naturally, incomparably
better. Fraser-Brunner relied heavily on Regan's os-
teological observations and pointed out (1943:1), e.g.,
that Regan's diagnosis of the Gymnodontes "here re-
garded as the Suborder Tetraodontoidea, cannot be
surpassed, and its salient features are given below, prac-
tically unaltered." This is not to imply that Fraser-Brun-
ner relied only on previous diagnoses of the subgroups,
for he contributed many original and significant obser-
vations on the osteology of various families, notably in
the tetraodontids, triacanthodids, and ostracioids. His
strong point was his combination of his own and
previously published osteological observations with his
knowledge of the relative importance of external charac-
teristics in arriving at his familial categories. This is par-
ticularly evident in his diagnoses of the tetraodontid sub-
groups, in which osteology (albeit only the top of the
cranium in most cases) is uniquely combined with the
structure of the nostril and lateral line, two systems
notoriously misleading in this group when used by them-
selves.

In one of his more recent papers, Fraser-Brunner
(1950) speculated about the basic phylogeny within the
Plectognathi. There is little doubt today about the

naturalness of the triacanthoid-balistid-monacanthid
line, but from what stock the tetraodontoids and ostra-
cioids arose is an open question. After examination of the
larvae of a triacanthodid, Fraser-Brunner suggested that
both the tetraodontoid and ostracioid lines may have
been derived neotenously from the triacanthodids.

Whereas Miiller and Bleeker had started the trend
toward the recognition of three major subgroups of plec-
tognaths, rather than the original two of Cuvier, Berg's
(1940) "Classification of Fishes" went one step further
and recognized four. Berg came to the following classi-
fication on the basis of his analysis of the works of Regan,
Rosen, Kaschkaroff, Gregory, and Fraser-Brunner:

Order Tetraodontiformes (Plectognathi)
Suborder Balistoidei (Sclerodermi)
tFamily Spinacanthidae
Family Triacanthidae
Subfamily Triacanthini
Subfamily Halimochirurgini
Family Triodontidae
Family Balistidae
Subfamily Balistini
Subfamily Monacanthini
Subfamily Psilocephalini
Suborder Ostracioidei (Ostracodermi)

Family Ostraciidae
Suborder Tetrodontoidei (Gymnodontes)
Family Tetrodontidae
Family Diodontidae
Suborder Moloidei
Family Molidae.

Other than the fact that the molids were given subor-
dinal rank, this was a very conservative classification,
based entirely on works already mentioned.

Breder and Clark's (1947) "A Contribution to the Vis-
ceral Anatomy, Development and Relationships of the
Plectognathi" is a useful, and interesting, comparison of
the anatomy of the acanthurid-plectognath line. It is in-
dicative of the then current state of our knowledge of
plectognath anatomy (Breder and Clark 1947:312-313,
table 2) that these two authors, on the basis of the litera-
ture and their own research, were unable to state whether
the palatine was movable in either Triodon or the trunk-
fishes. One also sees in table 2 that even greater gaps
exist in our knowledge of the characteristics associated
with reproduction. It has been mentioned several times
in this historical review that the rarity of museum
specimens of Triodon had hampered the phylogenetic in-
vestigations of all workers after the time of Dareste and
Hollard. Breder and Clark were able to make a partial
dissection of one of the Stanford University specimens of
Triodon obtained by Herre in the Philippines in 1931.
Their report on the intestine was the first addition that
had been made in 1(K) years to our knowledge of Triodon.
This same specimen has been utilized for the present
monograph.

The classification adopted by Breder and Clark was
modified from that of Fraser-Brunner:

Order Plectognathi
Suborder Sclerodermi
tFamily Spinacanthidae
Family Triacanthidae

Subfamily Triacanthodinae
Subfamily Triacanthinae
Family Balistidae
Family Monacanthidae
Subfamily Monacanthinae
Subfamily Aluterinae
Family Triodontidae
Suborder Ostracodermi
Family Ostraciidae
Subfamily Aracaninae
Subfamily Ostraciinae
Suborder Gymnodontes
Family Tetraodontidae

Subfamily Canthigasterinae
Subfamily Lagocephalinae
Subfamily Colomesinae
Subfamily Tetraodontinae
Subfamily Chonerhininae
Family Diodontidae
Suborder Moloidei
Family Molidae.

On the basis of their research, Breder and Clark (1947)
felt that the trunkfishes deserved subordinal rank rather
than inclusion in the Sclerodermi: "In view of the
marked differences in the ontogeny from that of Mona-
canthus and its resemblance to that of Spheroides, we
cannot feel justified in holding these fishes within the
Sclerodermi almost solely on the basis of cranial charac-
ters" (p. 311). Their justification for following Berg's
handling of the molids was that "Although these fishes
represent an offshore modification of the gymnodontid
type, they are so much further modified as, in our judge-
ment, to warrant such a separation. The fact they have
lost their tail and its associated musculature ... in-
dicates a fundamental change in the whole plan of or-
ganization" (p. 311). Breder and Clark's views were sum-
marized in a phylogenetic tree (p. 315, fig. 8) that is one
of the most sensible and understandable that has been
presented for the plectognaths, and the reader is referred
to it.

The extensive work by Y. Le Danois (1954, 1955, 1956,
1959, 1961a, b) on the osteology, myology and other
soft anatomy, taxonomy, and phylogeny of the plectog-
nath fishes, including monographs on the tetraodon-
toids and ostracioids, has not had a favorable reception.
The osteology and systematics were criticized by Tyler
(1963b) and the myology by Winterbottom (1974). To
more fully comment on the controversial aspects of Le
Danois' work would require hundreds of pages, and it
must suffice to summarize Le Danois' conclusions: the
triacanthoids and balistoids are of acanthurid origin; the
other plectognaths (Orbiculati) are not even of percoid

derivation, being related to the isospondylous fishes; and
Canthigaster is related to the ostracioids rather than the
tetraodontids. Le Danois' classification of the "Orbicu-
lati" is as follows:

Sous-ordre Orbiculati
Division Orbispiniformes ou Tetraodontoidea

Families: Xenopteridae, Colomesidae, Diodontidae,
Tetraodontidae, Lagocephalidae.
Division Ostracioniformes ou Ostraciontoides
Families: Canthigasteridae, Aracanidae, Ostra-
cionidae.
Division Moliformes ou Moloidea
Families: Ranzanidae, Molidae.
Division Triodontiformes ou Triodontoidea
Famille: Triodontidae.

To what order the "Orbiculati" belong is not stated.
Le Danois' work can be dismissed, except nomen-
claturally.

Although not directly concerned with the general phy-
logeny and higher classification of the plectognath fishes,
several outstanding recent works on the systematics of
limited groups of plectognaths have made certain species
complexes far better known than in the past, and repre-
sent the basic building blocks upon which such revisions
as this monograph are based, de Beaufort (1962) sys-
tematically treated all of the more common species of the
Indo-Pacific plectognaths (104 species) for the first time
since Bleeker, immensely aiding the often difficult task
of specifically identifying the plectognaths of that region.
Berry and Vogele (1961) carefully analyzed the sys-
tematics of the Atlantic species of monacanthids, and
Berry and Baldwin (1966) did the same for the Pacific
balistids. Shipp (1970, 1972a, b, 1974) and Shipp
and Yerger (1969a, b) have brought order to the nu-
merous Atlantic species of Sphoeroides, discussing ex-
ternal morphological features, delimiting distributional
patterns, and describing new species. With such exem-
plary work as that of Berry and Shipp and colleagues, the
American balistoids and tetraodontoids have become the
best known of their respective families. In Asia the ex-
ceptionally speciose and mainly freshwater pufferfishes
of the genus Tetraodon received their first systematic
review by Dekkers (1975), who placed them in five
species groups and unraveled most of the systematic
snarls that previously surrounded them. Similarly, the
marine species (primarily of Arot/iroM and Canthigaster)
of tetraodontids from Taiwan were redescribed by Shen
and Lim (1974a) and Shen et al. (1975) in that general
area for the first time since the series of articles by Abe
(1942, 1944, 1949a, b, c, 1950-51, 1952, 1954, 1960), while
Shen and Lim also reviewed the ostracioids (1973) and
balistids (1974b) of Taiwan. Allen and Randall (1977) pro-
vided an excellent and superbly photographed review of
the Indo-Pacific pufferfishes of the subfamily Canthigas-
terinae, recognizing 22 species, of which 7 were described
as new, in this group of notoriously difficult tetraodontids,
while Leis (1978) has performed a similarly fine service for

the five species of Diodon he recognizes worldwide. The fos-
sil and Recent triacanthoids were redefined and revised by
Tyler (1968), which monograph was reviewed by Myers (1970).

The provisional reclassification of fishes by Green-
wood et al. (1966), which is wisely accepted as the con-
temporary standard, does not attempt to include new in-
formation on the plectognaths. While I agree with their
basic subgroupings of the Recent Plectognathi, I consider
their families as superfamilies, a matter of opinion that I
hope is supported by the numerous differences given here
between the 6 superfamilies and their included 10
families as recognized in this monograph.

In Gosline's (1971) combination of functional mor-
phology and fish classification, the plectognaths are
recognized in 7 superfamilies with 12 families (the
Aluteridae recognized as distinct from the Mona-
canthidae, and the Canthigasteridae from the
Tetraodontidae) in the usual two suborders, while Gos-
line (1968:16) had previously related the acanthuroids
and plectognaths.

The tentative classification of the plectognaths
proposed by Tyler (1974), with 7 superfamilies and 11
families, is modified here by the combination of the
Canthigasteridae with the Tetraodontidae and of the
Diodontoidea with the Tetraodontoidea.

It is pleasant to be able to conclude this historical
review of the higher classification of the plectognaths
with praise for the latest significant publication, that by
Winterbottom (1974) on "The Familial Phylogeny of the
Tetraodontiformes (Acanthopterygii: Pisces) as Evidenc-
ed by Their Comparative Myology." This is a beautifully
illustrated and thoroughly analyzed survey of the
muscles of numerous representative plectognaths and of
its implications to the phylogeny and classification of the
order. It is unquestionably the finest comparative
myological study of any major group of fishes. A phy-
logenetic interpretation is given for the condition of every
main muscle (75 in toto) in representatives of every plec-
tognath family and of various reinterpretations of these
data. It is a forcefully reasoned analysis based on an ex-
ceptional amount of myological data.

A hallmark of Winterbottom 's work is a rigorous
application of the principal that only synapomorphies
(shared specialized characters) can be used to link
together related groups, while symplesiomorphies
(shared generalized or primitive characters) are not valid
as phylogenetic indicators. The excellence of Winterbot-
tom's contribution can best be seen by perusal of his
brilliant monograph, and only the highlights of his phy-
logenetic conclusions and an outline of his proposed
major reclassification can be given here.

Winterbottom (1974:ii) states in his abstract that
"... using only those characters of a shared specialized
(synapomorph) nature . . . suggests that the Triacantho-
didae and Triacanthidae are sister groups, and separated
off first from the ancestral stock. The remaining families
are divided into two lines. On the one hand, the
Balistidae form the sister group of the Monacanthidae,
and together form the sister group of the Aracanidae and
Ostraciidae. In the other linage, the Triodontidae

separated off first, followed by the Molidae. The Diodon-
tidae form the sister group of the Tetraodontidae plus
Canthigasteridae." The classification based on this phy-
logeny divides the plectognaths into two suborders in a
rather different way than ever proposed before, revised as
follows (Winterbottom 1974:99):

Order Tetraodontiformes
Suborder Triacanthoidei
Family Triacanthodidae

Subfamily Triacanthodinae
tSubfamily Cryptobalistinae
Subfamily HoUardiinae
Family Triacanthidae

Subfamily Triacanthinae
tSubfamily Protacanthinae
Suborder Tetraodontoidei
Superfamily Balistoidea
tFamily Spinacanthidae
Family Balistidae

Subfamily Balistinae
Subfamily Monacanthinae
Family Ostraciidae

Subfamily Aracaninae
Subfamily Ostraciinae
Superfamily Tetraodontoidea
tFamily Eoplectidae
tFamily Zignoichthyidae
Family Triodontidae
Family Tetraodontidae

Subfamily Tetraodontinae
Subfamily Canthigasterinae
Family Diodontidae
Family Molidae.

The phylogenetic conclusions reached by Winter-
bottom on the basis of the myology of the plectognaths
and those presented here on the basis of their osteology
are in general agreement, although the hypotheses of this
monograph emphasize the role on the one hand of a
hollardiinlike Eocene stock of triacanthodids as an-
cestral to triacanthids through a protacanthodinlike line,
with the basal triacanthids giving rise to the balistoids
and ostracioids, and on the other hand of an eoplectin-
like Eocene stock of triacanthodids as ancestral to the
triodontids and hence to the tetraodontoids and molids.
Even though the phylogenies based on Winterbottom's
myological analysis and that of the osteology given here
are not far apart, and differ mainly in details of the rela-
tionships of fossil forms for which there is no myological
evidence, the classifications adopted are widely diver-
gent.

While finding Winterbottom's philosophy of classifica-
tion highly worthwhile, eminently logical and entirely
defensible, the preference followed here is to weigh and
balance the value of both the generalized and specialized
features of the fossil and Recent species, rather than to
rely on the specialized alone, and the classification sug-
gested here reflects that bias.

Winterbottom and I plan to prepare a joint paper after
the publication of this monograph discussing the merits
of alternatives to our respective modes of phylogenetic
interpretation and classification of various plectognath
linages and subgroups. Until then, the reader is left to
his own inclination.

Historical Review of the Nonclassificatory Anatomical Work on the Plectognathi,
with notes on certain anatomical systems of special historical interest.

Comments are given in the preceding section on the
contributions of such anatomically oriented workers as
Cuvier, Dareste, Hollard, Winther, Regan, Kaschkaroff,
Rosen, Gregory, Fraser-Brunner, and Le Danois, all of
whom were using an osteological framework for their
classifications of the plectognath fishes. A number of
other workers have described the general osteology of one
or several species of Plectognathi from a more or less
purely anatomical point of view, without concerning
themselves with questions of phylogeny and classifica-
tion. Specific details of these works are mentioned in the
subsequent analysis, but some of these shorter contribu-
tions are of sufficient importance to merit brief general
comment.

Wellenbergh (1840) was the first person to treat an in-
dividual plectognath species in a manner similar to that
of Geoffroy Saint-Hilaire's (1827) anatomy of the Nile
puffer. Wellenbergh's description of the bones of the
skull of Mola leaves much to be desired by con-
temporary standards, but the branchial and hyoid arches
were described rather accurately. After Wellenbergh's
work there came a veritable flood of literature on the
anatomy of molids. Goodsir (1841) described with some
detail the histology of the shagreenlike skin and the
gelatinous subdermal tissue of Mola, as well as its
general myology. His work is of importance here because
of his description of the abortive nature of the posterior
end of the molid vertebral column. Cleland (1862) like-

wise gave a general account of the soft anatomy of Mola,
with an emphasis on the structure of the caudal region,
its bony supports, cartilaginous pterygials, and unusual
fin rays. Two of Cleland's most often quoted statements
were that Mola possessed no otoliths and had only two
semicircular canals. Cleland was not aware that Cuvier
(1805, 2:456-457) had described the otoliths of Mola as
being gelatinous and mucoid, rather than calcareous,
and that he had even figured (pi. 18, fig. 1) the three
semicircular canals. Cleland's error has been corrected a
number of times: vide, Harting (1865), Thompson
(1888), and Meek (1904).

Harting's (1865) general anatomy of Mola contained
very few references to osteology, but he did give a brief
description of the long, sharp-pointed teeth on the upper
pharyngeals. Walgren's (1867) work on Mola was most-
ly confined to external characteristics, but brief notes
were given on the skeleton, and the jaws, with their
trituration plates, were well described and figured. The
first worker to publish on the general anatomy of the dis-
tinctive molids of the genus Ranzania was Trois ( 1883-
1884), whose osteological interest, however, was mainly
confined to the comparison of the histology of the bones
of Ranzania with that of Masturus and Mola. He pointed
out, for example, that Ranzania has a much more com-
pletely ossified skeleton than that of the other molids.
Trois' histologically oriented work was soon followed by
Beauregard's (1893) descriptive osteology of Ranzania.
Beauregard gave us our basic knowledge of the general
osteology of Romania, and the information that he
presented was not supplemented until Raven (1939b)
published his brief article on its anatomy. One of the
more accurate of the early general osteologies of Mola
was that of Steenstrup and Liitken (1898); the various
figures (unnumbered; occurring on p. 91-94) of the skull
and branchial apparatus are, with the exception of the
orbital region, notably accurate, and the lateral view of
the entire skeleton (pi. 2) is only questionable in the
region of the pseudocaudal fin. Two other papers should
be mentioned here in relation to molid anatomy, that of
Gregory and Raven (1934) on Mola and that of Raven
(1939a) on Masturus. Both of these papers were primari-
ly concerned with the myology, but information was also
given on the osteology of the supporting structures of the
pseudocaudal fin.

Of a much more ambitious nature than any of the
above-mentioned papers on molid anatomy was Briihl's
(1856) "Osteologisches aus dem Pariser Pflanzengar-
ten," in which there appeared (p. 58-70) brief os-
teological descriptions of Balistes, Ostracioh, Alutera,
Tetraodon, and Diodon. Briihl confined himself almost
entirely to the cranium, and the few figures which he
gave of various regions of the plectognath skeleton show
very little detail, but if one takes the time to translate
the sirchaic terminology of the bones, one finds that
Bruhl's osteological observations were substantially cor-
rect. It is hard to explain, however, how he mistook part
of the sphenotic in Diodon for a parietal; possibly it was
because Wellenbergh had found a "parietal" in Mola,
and thus one might expect to also find one in Diodon, if

one believed very strongly in the close relationship of
those two types. In his "Zootomie," Bruhl (1880)
presented excellent illustrations (pi. 24, with two un-
numbered pages of explanation; also see Briihl 1891) of
Balistes, and the only serious error in his various figures
of the skeleton is the presence of an extra suture between
the parasphenoid and basioccipital.

A work very similar to Briihl's (1856) "Osteologisches"
in its conception, but broader in its scope of coverage,
was Klein's (1884, 1885, 1886) "Bildung des Schadels der
Knochenfische." In this general work on the teleost skull,
Klein discussed the cranial osteology of representatives
of most of the major plectognath subgroups. His descrip-
tions were, on the whole, accurate, and since his illus-
trations were so well executed, it is unfortunate that
there were so few of them. Klein corrected a great many
of Hollard's osteological errors, and stressed (1886:230-
234) the presence of a myodome in the nonostracioid
scleroderms and its absence in the gymnodonts. In an
earlier paper Klein (1872) had described the cranial os-
teology of four species of Balistes, and while his descrip-
tions were relatively accurate, his illustrations were so
reduced in size for publication that little detail can be
seen. Klein (1881) published one other paper on plectog-
nath osteology, devoted primarily to the description of
the spiny dorsal fin locking mechanism in triacanthoids
and balistoids.

Rosenthal (1839) produced rather accurate lateral
views of the entire skeletons and other detailed views of
Balistes, Ostracion, Tetraodon, and Diodon.

Several other papers of special importance to the os-
teology of particular species should be mentioned here.

The osteology of the skull of Balistes and Mola was
described and reasonably well illustrated by Supino
(1905), as was the branchial apparatus of each species,
the latter being an important anatomical system which
had all too often been neglected by previous workers.
Much of the information that Supino published in his
1905 paper appeared again in his more general work
(1907) on "II Crano dei Pesci."

Awati and Bal ( 1933) initiated a series of papers on the
anatomy of Indian puffers with a description of the
skeleton of Tetraodon (= Fugu) ohlongus, but their os-
teological descriptions suffer from numerous serious
defects, as will be detailed later. The second paper in this
series (Awati and Bal 1934) dealt with the blood vas-
cular system, and the third paper (Bal 1937) with the
nervous system. A far more detailed and accurate survey
of the functional anatomy of the head of that species was
given by Sarkar (1960) in a published doctoral disser-
tation.

A work of importance to the generic classification of
the tetraodontids is Kuronuma's (1943) survey of the
configuration of the top of the cranium of 14 species of
Japanese puffers. Besides giving a photograph of the top
of the cranium of each of these species, diagrammatic il-
lustrations were presented of lateral views of the skulls of
species representative of the three genera to which the
species belong. Kuronuma's osteology is correct and he
has made available much information of diagnostic value

for the genera and species studied. Kiausewitz (1964) ac-
curately described and illustrated the skull of an
Arothron, and Rishi (1969) that of a Chelonodon.

The otic region of the skull of plectognaths has been
the subject of much misinterpretation in the literature.
One source of the error has evidently been the feeling
that the plectognaths, like most other fishes, should
possess a pair of parietals. But the plectognaths have lost
all trace of the parietals and what have been called the
parietals in plectognaths are portions of the epiotics,
sphenotics, or frontals. Wellenbergh (1840) referred to
a portion of the epiotic in Mola as a parietal. Bruhl (1856)
described the parietals as being absent in all plectog-
naths except Diodon and Mola, calling the posterior
part of the sphenotic in Diodon a parietal and accepting
Wellenbergh's statement that a parietal existed in Mola.
In HoUard's (1853, 1854a, b, 1855, 1857a, b, 1860) various
publications the posterior portion of the frontal is called
the parietal, although a definite suture between the
bones is not usually shown in his figures. Goeldi (1884)
pointed out that many of the sutures shown by Hollard
were artifacts and criticized Hollard for the excessive
number of bones that had been described as separate ele-
ments. What Siebenrock (1901) described as a parietal in
Balistes, even though he could not find one in Tria-
canthus, or Monacanthus, is probably a part of the
epiotic. Supino (1905) thought that in Balistes the
parietals were fused to the frontals, but in Mola he
described the medial half of the pterotic as being a
parietal. Kaschkaroff (1914a) described a parietal as be-
ing present only in Balistes, his parietal evidently being
part of the epiotic. Kaschkaroff believed that the frontal
of Mola might represent the product of the fusion of the
frontal with the parietal. Rosen (1916a:13) said that the
parietals were absent in plectognaths, and expressed the
prevailing view that "they have perhaps fused with the
frontals." The above statements are exclusive of those in-
stances in which there is a purely nomenclatural dif-
ference, e.g., Regan's (1903a) use of the term "parietal"
as being synonymous with "epiotic." Since Rosen's time,
no knowledgeable worker has described a parietal in a
plectognath fish, and if that bone has fused with the
frontal it has done so indistinguishably. Not even a trace
of a separate ossification which could conceivably be
referred to as a parietal has been seen in any of the
developmental stages of various plectognaths examined
for this work. Under these circumstances it would seem
best to assume that the parietal simply fails to develop in
plectognaths.

The other major source of error in the interpretation of
the otic region has been the description of the upper and
lower surfaces of the epiotic as two separate and distinct
bones, because of a certain anatomical peculiarity of the
development of the endochondral bones, particularly in
the otic region. The otic bones begin to develop in the
normal manner as small ossifications in the appropriate
region of the chondrocranium. From this center of os-
sification in a plate of cartilage, the gradual peripheral
movement of the ossification does not proceed evenly.
Rather, the ossification spreads through the upper sur-

face and through the lower surface of the cartilage, so
that toward its edges the bone is composed of an upper
and a lower layer separated by cartilage. The two layers
are continuous with one another only at their original
center of ossification. As the specimen becomes larger,
there is a slow and gradual ossification that fills in the
otherwise cartilaginous area between the two peripheral
layers of the bone. It is only in extremely large specimens
that all of the cartilage between the double-layered
peripheral parts of the bone is replaced by ossified tis-
sue. Thus, in the descriptions given here of the otic and
other regions, there is reference to these endochondral
bones having the edges filled with cartilage, simply
because in all but the largest specimens of any par-
ticular species the replacement of the cartilage between
the upper and lower layers in the peripheral region of the
bone is incomplete. This double-layered condition is par-
ticularly evident in the epiotic and sphenotic, and is
progressively less evident in the pterotic, prootic, exoc-
cipital, supraoccipital, and the other endochondral bones
of the skull. In adult specimens the cartilage-filled edges
of these endochondral bones are usually not evident on
external examination, and they can be seen only when
the bones are disarticulated.

The reason that this condition is not particularly ap-
parent is that the edges of the upper layers of two ad-
jacent bones, e.g., epiotic and sphenotic, may inter-
digitate with one another, hiding the fact that just below
this interdigitated surface there is a thin layer of car-
tilage separating the upper and lower layers of the bones.
It is usually true, except in large adults, that when the
upper layers of two bones articulate by interdigitation,
the lower layers of these bones are separated from each
other by a small amount of cartilage, this cartilage ex-
tending for a short distance into the substance of each of
the bones. It must be emphasized that even though a
bone such as the epiotic has an upper and a lower layer
separated by cartilage peripherally, these two layers
converge toward one another centrally to be connected
by a core of bone representing the original center of ossifi-
cation. This central core of bone is delicate enough in
some cases to be easily broken when skeletal material is
prepared by drying, because of the contorted shrinkage
of the cartilage that lies between the double-layered
peripheral regions of the bone. If the drying of the skele-
ton does not break the central core, then the process of
disarticulation of the skull often will.

It is thus not surprising to find that Klein (1872, 1881,
1884, 1885, 1886) stated that in plectognaths there is an
epiotic which is covered over by a parietal. Klein's
"parietal" was simply the upper layer of the epiotic,
which had become separated from its lower layer during
the drying and disarticulation of the skull by breaking of
the central ossified core which normally makes the two
layers continuous. Awati and Bal (1933) have done essen-
tially the same thing in their study oiFugu oblongus, for
they refer to the ventral surface of the epiotic as the
"epiotic," but to its dorsal surface as the "parietal."
Furthermore, they refer to the ventral surface of the
sphenotic as the "sphenotic," but to its dorsal surface as

the "squamosal." Like Klein, they were using dried
skeletal material.

Rosen (1916a) had some insight into this problem of
the "double-layered" condition, which is especially
prominent in the otic region of tetraodontids. In his brief
description of Sphoeroides testudineus he contradicted
Klein's statement that the epiotics are covered by the
parietals, for Rosen stated (1916a: 19) that "I have not
been able to find distinguishable such elements either in
adult specimens or in young ones (18 mm. in length)."
But, quite unaccountably, Rosen went on a few lines
later to say, without any elaboration, that there were
"Parietals present." Unfortunately, there were no figures
of plectognath skulls in that paper, and one can only
guess that Rosen was referring to a portion of the pos-
terior end of the frontal in Sphoeroides as the parietal.
Rosen's most interesting statements, however, were
about the sphenotic and pterotic in Sphoeroides, for he
described them as follows: "Sphenoticum. Sections of
specimens 18 mm in length show that this bone orig-
inated as two elements: a dermosphenoticum and an
autosphenoticum in the same mode as, e.g., squamosum.
Squamosum is formed by the fusion of a dermo-
squamosum and an autosquamosum." No other com-
ments were made about this condition of the otic region,
nor were any figures presented of it. Rosen was obviously
aware of the double-layered condition in the otic region,
but why he did not notice it in the epiotic is a mystery.
Rosen was probably incorrect in assuming that the upper
layers of the sphenotic and pterotic were of dermal
origin, for in the many young tetraodontids examined for
the present work, this upper layer is obviously ossified in
the upper region of the cartilage of the chondrocranium.
This is not to say, however, that there is not any dermal
ossification eventually incorporated into the external
surface of the upper endochondral layer of the otic bones,
but only that the great bulk of these bones is endochon-
dral in origin. In ParahoUardia and other triacan-
thodids, for example, most of the substance of the part of
the pterotic that overlies the sphenotic and epiotic ap-
pears to be of dermal origin, but this is only a very small,
and irregular, portion of the pterotic.

A number of workers have erroneously referred to cer-
tain portions of the pterotic as the "opisthotic." Awati
and Bal (1937) did so in Fugu oblongus, and Gamaud
(1956) did likewise in Batistes. Kaschkaroff (1914a:351)
seems to be the one who began the mistake, for he said
that an opisthotic was present in scleroderms but not in
gymnodonts. Kaschkaroff s "opisthotic" is only a region
of the pterotic.

A well-developed myodome has long been known to be
present in triacanthoids and balistoids, and more recent-
ly in triodontids (Tyler 1962a), but it was thought to be
absent in all other plectognaths. Klein (1884, 1885, 1886)
was the first person to point out the diagnostic value of
this structure (see part 1:150-152 and part 3:230-234) in
plectognaths, but Regan (1903a) rather unaccountably
stated that the Sclerodermi have the "basis cranii more
or less distinctly double" (p. 280) and that the Gym-
nodontes have the "basis cranii simple" (p. 291), which

implies that the ostracioids have a myodome. Kasch-
karoff (1914a:351) described the longitudinal concavity
on the ventral surface of the basioccipital in Triacan
thus, Balistes, and Monacanthus, but he did not men
tion that this channel leads into the myodome. Holm
gren and Stensio (1936:488) contended that the
myodome of plectognaths and a few other fishes cor-
responds to the dorsal part of the myodome of Salmo
because it encloses only the M. recti externi. Be that as it
may, the important fact that Triodon possesses a well
developed myodome was overlooked by Dareste (1849)
and Hollard (1857b), and it is significant to know that at
least a rudiment of the dorsal roofing of the myodome is
present in some tetraodontids (Tyler 1963b).

Frost (1930:621) gave cursory descriptions of the
otoliths of a few plectognaths and said that they "are
curiously aberrant in form, showing little affinity with
those of the other orders."

In the orbital region there has been much controversy
in the literature concerning the bones with which the
frontal articulates posteriorly. The difficulty stems from
Hollard's (1857b) inaccurate figures of the top of various
gymnodont skulls. Gill's (1885, 1892a) reproduction and
use of these figures could only lead to erroneous diag-
noses, with the tetraodontids being said to have the fron-
tals in contact with the supraoccipital, except for
Canthigaster and Chonerhinus, which were supposed to
have these bones separated from one another by the in-
tervention of the sphenotics. This erroneous difference
between the tetraodontins and canthigasterins was used
by Jordan and his associates (Jordan and Edwards 1887;
Jordan and Evermann 1898; Jordan and Snyder 1901)
and thus it gained popularity. However, Regan (1903a)
showed that the differences were purely in Hollard's
figures and presented figures of his own which showed
the correct relationships of the bones of the top of the
skull in a tetraodontin and a canthigasterin. Fraser-
Brunner (1943) used the fact that the sphenotic separates
the frontal and epiotic in diodontids as one of the diag-
nostic characteristics of that group. Sexual dimorphism
in the size of the frontals has been pointed out by Ebina
(1932) in filefishes, and supposedly by Le Danois (1954)
in diodontids. In relation to this discussion of the frontal,
it is worthwhile to mention that none of the dermal bones
of the head develop in conjunction with the sensory canal
system, a fact first stated by Rosen (1916a:20). The der-
mal bones show no trace of canals, bridges, or troughs
that could in any way be associated with the cephalic
sensory canal system, which is confined entirely to the
skin in plectognaths. Le Danois' (1956) statements to the
contrary are unproven.

A basisphenoid is present only in two groups of plectog-
naths— Triodon and molids. Neither Dareste (1849) nor
Hollard (1857b) observed the basisphenoid in Triodon,
but that of molids has been described often, although un-
der a variety of names. A few workers have described the
anterior region of the prootic in various scleroderms as a
"basisphenoid" (Klein 1884; Supino 1905; Rosen 1916a),
but, strangely, there has been agreement (as early as
Stannius 1854:61) that gymnodonts, except for molids.

do not possess a basisphenoid. It is indeed food for
thought that a basisphenoid occurs only in Triodon,
which by all odds is the most primitive of the Recent
gymnodonts, and in molids, which are one of the end
lines of gymnodont radiation. Starks (1905:755), in dis-
cussing the myodome of fishes, noted that "the dichost
( = basisphenoid of Huxley) is always absent when the
myodome is. I know of no case where it is at all ossifed
when the myodome is absent." Molids are thus an ex-
ception to the rule. The basisphenoid of Triodon is
relatively normal in appearance, but that of molids is ex-
panded anterodorsally until it makes contact with the
pterosphenoids (in Mola and Masturus) or frontals (in
Ranzania).

In the ethmoid region the size and shape of the eth-
moid have been prominently used in the classification of
the tetraodontids by Fraser-Brunner (1943). While the-
ethmoid of tetraodontids is sometimes rather small, by
far the most reduced state of the bone is found in diodon-
tids, where there is a thin plate of bone that represents
the only remains of the ossifications of the entire eth-
moid region of the skull in a position analogous to that of
both the vomer and the ethmoid. For that reason, this
bone is called here the ethmoid-vomer, supposing that it
represents the fused rudiments of both elements. Starks
(1926:286) recognized the dual anatomical position of
this bone in Diodon, but preferred to call it the ethmoid.
It should be mentioned briefly that in HoUard's various
publications the medial portion of the plectognath eth-
moid was called the "ethmoid," while its lateral por-
tions were referred to as the "nasals." Hollard did not,
however, show any sutures between these two regions, so
his terminology was undoubtedly based on theoretical
considerations. Awati and Bal (1933) divided the eth-
moid of their tetraodontid in the Hollardian manner, but
referred to the medial portion of the ethmoid as the
"lateral ethmoid" and to the lateral portions of the eth-
moid as the "nasals." Swinnerton (1902) found the eth-
moid and palatine regions of balistids and acanthurids to
be similar.

The vomer of tetraodontids and ostracioids is some-
times so closely interdigitated, or even fused, with the
ethmoid and parasphenoid that even a relatively knowl-
edgeable worker such as Kaschkaroff (1914a:350) mis-
takenly said that the vomer was absent in those groups.

In the mandibular region the presence of a symplec-
tic has sometimes been overlooked. Cope (1871:591)
stated that in the Phyaoclisti "the symplectic is present,
except in Ostracion, where it is not ossified," but Regan
(1903a:290) corrected Cope's error. In view of Regan's
statements, it is surprising to find that Kaschkaroff
(1914a:352) extended Cope's misstatement to include
Triacanthus as well as Ostracion, both of which Kasch-
karoff said lacked a symplectic. Raven (1939b:4) said that
the symplectic was absent in Ranzania, but figures 316
and 320 given here show it to be long and prominent.

In the palato-pterygoid region the size, shape, and
articulations of the palatine have been used as impor-
tant diagnostic features for a number of plectognath
subgroups, sometimes erroneously. A summary of the

condition of the palatine in plectognaths, with brief com-
ments on the literature, follows. In triacanthoids the
palatine is basically a flattened rectangular or squarish
bone with a small posterodorsal lobe about midway along
its upper edge for firm anchoring through fibrous tissue
with a region of the ethmoid-prefrontal-vomer, while
ventrally the palatine is firmly anchored by fibrous tis-
sue with the ectopterygoid, except in the long-snouted
species in which it forms a portion of the tubelike snout;
an anterior process of the palatine movably articulates
with the recessed dorsolateral surface of the maxillary so
that the latter can rotate around this articulation and
allow for a slight protraction of the upper jaw, except in
the long-snouted species in which the protractibility of
the upper jaw is not so directly associated with a maxil-
lary-palatine articulation. In balistids what is the squar-
ish ventral portion of the palatine in most triacanthoids
becomes shaftlike and the whole palatine T-shaped,
with the ventral shaft firmly anchored by fibrous
tissue with the ectopterygoid, the posterodorsal end
similarly held to the ethmoid and vomer, and the antero-
dorsal end movably articulated with a slight concavity
on the lateral surface of the maxillary, around which
point of articulation the upper jaw rotates, without being
protracted. In monacanthids the palatine is rodlike, but
sometimes with a bulge on its ventral surface represent-
ing the long foot of the T as found in balistids, con-
nected ventrally by a long ligament to the anterior edge
of the ectopterygoid, while posteriorly the rod is
anchored by fibrous tissue to the ethmoid and anteriorly
is movably articulated with a slight concavity on the
lateral surface of the premaxillary and maxillary.

The differences in the shape of the palatine in the tria-
canthoids, balistids, and monacanthids have long been
known, but with Regan's (1903a) use of these features as
diagnostic characteristics an error crept into the
literature. Regan said that the triacanthoids had the
palatine immovably articulated with the other bones of
the skull, but that the balistoids had it either movably
articulated with the ectopterygoid (balistids) or entirely
free from it (monacanthids), and Fraser-Brunner (1941b)
has said much the same thing. The differences in the
shape of the bone are real enough, but the palatine im-
movably articulates with the ectopterygoid and with the
ethmoid-vomerine region in all of these groups. It is the
firm anchoring of the palatine to these two regions which
provides the upper jaw with the immovable prop around
which it can rotate. The ventral edge of the palatine and
the dorsal edge of the ectopterygoid in triacanthoids and
balistids are often not in close contact with one another,
but the sheet of tough fibrous tissue that binds the
palatine to the ectopterygoid and to the ethmoid-
vomerine area makes the palatine immovable for all
practical purposes.

In ostracioids the palatine is a relatively small and
elongate block of bone, representing an enlarged foot
of the T-shaped balistid palatine, which is usually
extensively interdigitated with the dorsal end of the
ectopterygoid and mesopterygoid. The rounded dorso-
medial edge of the palatine is firmly held by fibrous

tissue primarily to the vomer, usually to a slight con-
cavity on the surface of the latter. The palatine is so
small and closely apposed to the ectopterygoid that most
authors have been noticeably vague about its articula-
tion or even about is presence, although Hollard (1860:
fig. 6 of pi. 3) presented an accurate figure of it.
The palatine in the gymnodonts is a much larger bone
than it is in the scleroderms. The ventral edge of the
palatine in gymnodonts always immovely articulates,
usually by interdigitation, with the ectopterygoid.
Posteroventrally the palatine articulates with the meso-
pterygoid, and in some cases with the metapterygoid as
well. The anterodorsal edge of the palatine supports the
upper jaw, primarily through a flexible fibrous tissue ar-
ticulation with the concave facet on the maxillary. There
is some variability, however, in the immovable ar-
ticulation of the palatine with the ethmoid region of the
skull. In Triodon the medial surface of the palatine ar-
ticulates with the ethmoid, vomer, parasphenoid, and
prefrontal. Rather unexpectedly, the shape and ar-
ticulation of the palatine in molids is strikingly similar to
that in Triodon. The characteristic feature of the
palatine in tetraodontids is that its posterodorsal region
possesses an anteriorly directed cleft which fits closely
around some type of flange on the lateral surface of the
vomer. With the almost total reduction of the ethmoid
region in diodontids, the palatine becomes extensively
interdigitated with the frontals and makes only a sec-
ondary and functionally unimportant contact medially
with the ethmoid-vomer.

Gregory (1933:287) presented highly diagrammatic
figures of the palato-pterygoid region of the skull in
Triacanthus, Balistes, Alutera, and Psilocephalus to il-
lustrate the progressive elongation of the snout in that
series. Fraser-Brunner (1941a:423) showed, with more
detailed illustrations, the changes that take place in the
palato-pterygoid region when the same type of elonga-
tion of the snout takes place in the triacanthodid genera
Johnsonina, Tydemania, and Macrorhamphosodes,
while Tyler (1968) did so for all the genera of that family.
The figures and descriptions given by Thilo (1920) of the
jaws and palato-pterygoid region of Triacanthus,
Balistes, and Tetraodon are too crude to be of any value,
but those of Winther (1877) are much more useful.
Others have presented accurate descriptions and figures
of the palato-pterygoid region in particular species, e.g.,
Steenstrup and Liitken (1898) for Mala, and Supino
(1905, 1907) for Mola and Balistes.

In the opercular region the length and posterior
articulation of the interoperculum have been used as
diagnostic familial characters. A summary of the inter-
operculum in plectognaths with brief comments on the
literature follows. The interoperculum is best developed
in the triacanthoids, in which it is relatively broad and
rounded posteriorly and only gradually tapers to a point
anteriorly. Posteriorly the interoperculum articulates
closely with the suboperculum. In balistoids the inter-
operculum is a short rod which never extends posteriorly
past the level of the epihyal. The posterior end of the
interoperculum in balistoids has its main articulation

with the epihyal, while more posteriorly it connects by a
long band of fibrous tissue with the operculum, or with
the region of articulation between the operculum and
the suboperculum. In ostracioids the interoperculum
is even slightly shorter than it is in most balistoids, and
similarly it has its main posterior articulation with
the epihyal. In contrast to the balistoids, however,
the long band of fibrous tissue running posteriorly from
the ostracioid interoperculum attaches exclusively to
the suboperculum, rather than to the operculum or to
both the operculum and suboperculum. Among the
gymnodonts the interoperculum is relatively long and
well developed in all groups, except in the molids. In
the latter group the interoperculum is reduced to a thin
but sometimes long splint of bone entirely embedded in
the long ligament which runs between the angular and
suboperculum. In tetraodontids and diodontids the rod-
like interoperculum always has a ventral flange in about
the middle of its length wWch articulates by fibrous
tissue with the epihyal. The bifurcate posterior end of
the interoperculum in Triodon is unique among the
plectognaths, with it being hypothesized that the ventral
process of the bifurcate portion corresponds to the
ventral flange of the interoperculum in tetraodontids and
diodontids. The interoperculum articulates posteriorly
with the anteriorly directed process of the suboperculum
in Triodon, diodontids, and molids, but in tetra-
odontids it articulates with the operculum.

While many early workers described the unusual shape
of the interoperculum in this order, and Dareste (1850)
said that it was one of the few characters that the plec-
tognath fishes had in common, it was not until Regan
(1903a) that the structure of the interoperculum was
used diagnostically for various plectognath subgroups.
Regan incorrectly said that in the Sclerodermi the inter-
operculum was attached to the suboperculum, while in
the Gymnodontes it was attached to the suboperculum in
all groups except the tetraodontids. Fraser-Brimner
(1943) similarly used this supposed difference in the pos-
terior articulation of the interoperculum as a diagnostic
character.

In his various publications on the respiratory ap-
paratus of plectognaths and related forms, Willem (1941,
1942, 1944, 1945, 1947, 1949) pointed out the correlation
between the presence of a distensible diverticulum of the
oesophagus and the size of the suboperculum (i.e., the
suboperculum is relatively well developed in the tetra-
odontids and diodontids) and discussed the role of the in-
teroperculum in coordinating the opening of the lower
jaw with the expansion of the opercular apparatus at the
beginning of the inspiration phase of the respiratory cy-
cle.

The structure of the jaw8 of plectognaths has played
such an important role in the diagnoses of various
families since the founding of the order that it must be
summarized here. Cuvier's (1817) term Plectognathi
(from the Greek: plektos, plaited or twisted together;
gnathos, jaw) refers to the intimate and inflexible union
of the premaxillary and maxillary in all families except
the triacanthoids.

In all plectognaths except the triacanthoids, the upper
jaw is nonprotractile and the premaxillary immovably
articulates, usually by extensive interdigitation, with the
maxillary. In triacanthoids the premaxillary and maxil-
lary movably articulate with one another and the upper
jaw is slightly protractile. The premaxillary of triacan-
thoids is more or less L-shaped, with the long arm of the
L forming the posteriorly directed ascending process
which slides over the dorsal surface of the ethmoid in the
process of protracting and retracting the upper jaw. In
balistoids and ostracioids the posteriorly directed process
of the premaxillary is essentially absent, and the postero-
medial end of the premaxillary articulates with the an-
terior edges of the ethmoid and vomer, particularly with
the former. Along its anterior edge the premaxillary of
the scleroderms bears a variable, but usually small,
number of discrete teeth. The premaxillary of gym-
nodonts is modified, in conjunction with the teeth, into a
crushing plate which is often and aptly referred to as a
parrotlike beak. The posteromedial end of the premaxil-
lary articulates with the ethmoid, and to a lesser extent
with the vomer, in Triodon, but in most tetraodontids
this articulation is primarily with the vomer. In Triodon
and in the tetraodontids the medial edges of the two pre-
maxillaries are closely apposed and articulate with one
another by fibrous tissue. Emarginations from the
medial surfaces of each of the premaxillaries in tetra-
odontids alternate with one another and strengthen the
articulation between the two elements. In diodontids and
molids the two premaxillaries are indistinguishably fus-
ed together into a large U-shaped bone.

Hollard (1857b:308) did not believe that fusion of the
premaxillaries, or dentaries, was a character of much sys-
tematic importance, possibly because it was known that
the premaxillaries and dentaries of extremely large
specimens of tetraodontids occasionally fuse together in
the same way as do those of diodontids. More interesting,
however, is Reuven's (1894:130) observation of a
phenomenon in Mola of just the opposite nature to that
occasionally observed in tetraodontoids. In his descrip-
tion of a young Mola, Reuven said that he "found that
the maxilla superior is splitted up in the middle, the in-
ferior is not cleft," and a figure was presented (pi. 5) to
show this "tetraodont" condition of the upper jaw.
Hollard's view that too much importance has been at-
tributed to the fusion of the premaxillaries, or dentaries,
as a major indicator of phylogenetic affinity seems well
taken. It might be mentioned here that the premaxillary
and maxillary are often said to be fused or coalesced,
when in actuality they, at least in the great majority of
instances, only extensively interdigitate with one
another. Only very rarely is an extremely large specimen
of a tetraodontid or diodontid found in which the pre-
maxillary and maxillary are for all practical purposes
fused. One of the more extreme statements about the fu-
sion of the various jaw bones in plectognaths is that of
Wagner (1845:190) who said that the upper jaw bone
"coalesces with the vomerine, palatal and nasal bones to
form a single bone, which unites, however, with that of
the other side by suture, as in Orthagoriscus." Such a

condition would certainly be the death of any plectog-
nath and whether Wagner had ever actually examined a
molid is problematical, although he had (1841) once
presented a figure of the skeleton of a balistid.

Only in the triacanthoids is the maxillary movably ar-
ticulated and not closely apposed to the premaxillary, for
in all other plectognaths the maxillary firmly articu-
lates, usually by extensive interdigitation, with the pre-
maxillary. In triacanthoids the lateral surface of the
rounded dorsal end of the maxillary movably articulates
with the anterior process of the palatine. In balistids the
palatine articulates with a slight concavity on the sur-
face of the maxillary and premaxillary, but in monacan-
thids the maxillary scarcely makes contact with the
palatine or is even entirely excluded from articulation
with it. With reduction of the palatine in ostracioids the
maxillary articulates with only the anterior edges of the
vomer and ethmoid. In gymnodonts the maxillary al-
ways articulates over a broad, and usually slightly con-
cave, area with the anterodorsal end of the palatine, and
it is this articulation which is the major source of sup-
port for the upper jaw in gymnodonts. In triacanthoids
the support of the upper jaw is shared about equally
between the articulation of the maxillary with the
palatine and the articulation of the premaxillary with
the ethmoid and vomer, but in balistoids the pala-
tine-maxillary articulation becomes progressively less
important and in ostracioids it is completely absent. The
posterior edge of the maxillary of most balistoids and ostra-
cioids bears an indentation which is characteristic of
these two groups, but not of the other plectognaths. The
lateral surface of the maxillary is relatively flat in all
plectognaths except the diodontids, in which this sur-
face is upraised into a stout flange for muscle attach-
ment. The lower end of the maxillary forms the lower
part of the anterior edge of the upper jaw in all plec-
tognaths. Among Shufeldt's (1917) nearly totally
erroneous observations on the osteology of Diodon, there
occurs the statement that an "admaxillary" is present on
the dorsal edge of each maxillary, but since no descrip-
tion or illustration was given of this structure, that
author's statement is best forgotten, as is his other work
(1926) on plectognaths.

In scleroderms the teeth are usually few in number but
they are always individually recognizable units. In the
gymnodonts, however, the teeth are so highly modified
that they are scarcely recognizable as such. The number
and shape of the teeth in the jaws are discussed in the
diagnoses of the familial groupings recognized here and
are summarized below.

The number and shape of the teeth vary most in the
triacanthodids, and Gunther (1870), Myers (1934),
Fraser-Brunner (1941a), and Tyler (1968) have all made
prominent use of dental formulae in their revisions of the
family. The teeth of Recent triacanthodids are usually
conical and usually occur in a single series. This outer
series contains the majority of teeth, which usually vary
in number from about 10 to 50, but with reduced num-
bers (sometimes absent) in the long-snouted genera.
When teeth in an inner series are present, in the more

generalized genera, they vary from 1 to about 10 in num-
ber. The teeth of the upper and lower jaws are basically
similar, but there are usually a few more teeth in the
lower jaw than in the upper. In a few moderately to
highly specialized genera the teeth are truncate rather
than conical, while the fossil genera sometimes have en-
larged teeth.

In triacanthids there is an outer series of about 8 to 10
heavy incisors in each jaw, internal to which are several
more or less molariform teeth, usually 4 (2 in one genus)
in the upper jaw and 2 in the lower. The structure and
number of the teeth in balistids are strikingly similar to
those found in triacanthids, for in balistids the heavy in-
cisors invariably (with rare individual exceptions) occur
in an outer series of eight teeth and an inner series of six
teeth in the upper jaw, while in the lower jaw there is
only a single series (corresponding to the outer series of
the upper jaw) of eight teeth. The number of teeth
becomes further reduced in monacanthids to six in an
outer series in the upper jaw and six (sometimes only
four) in the outer series of the lower jaw, with an inner
series of four teeth in the upper jaw and no inner series
teeth in the lower jaw.

The dental formulae of the various balistoids have
been common features of the diagnoses of that group ever
since the time of Bleeker ( 1866:8- 10) , who was the first to
give prominent attention to these differences. The
development of balistoid teeth from deep sockets in con-
tinuity with the pulp cavity was accurately described for
the first time by Owen (1840 & 1845:82-85). Goodrich
( 1909:438, fig. 448) gave a figure of the replacement teeth
developing in Balistes, and Isokawa (1955) presented a
description of the general development and histology of
the teeth in monacanthids. The teeth of ostracioids are
elongate and rodlike, with between 6 and 17 teeth in a
single series in each jaw.

The teeth of gymnodonts are so highly modified that
they have attracted a great deal of attention in the
literature. There is, however, a phyletically important
distinction to be made about the teeth of the various
gymnodont subgroups which usually is not clearly
stated. In Triodon and diodontids the teeth always take
the form of relatively small £uid discrete units which
become densely packed with one another and incor-
porated with the bony matrix of the premaxillary and
dentary. In tetraodontids the teeth are relatively more
discrete long rods which lie parallel to the biting edge of
the jaws. There are no discrete teeth present in the jaws
of molids, nor can any individual tooth primordia be
found in the pulp cavity. The intimate incorporation of
the teeth with the bony matrix of the premaxillary and
dentary has reached its zenith in the molids, and one can
only speculate that the highly irregular surface of the
floor of the pulp cavity, with its numerous individual
"tooth" or bone forming cavities, indicates that the
molid condition has evolved from the Triodon and
diodontid type rather than from the tetraodontid type.

The structure, and even the mere presence or absence,
of a trituration plate is also of diagnostic interest. In
Triodon a trituration plate composed of right and left

halves is present in the upper jaw, while the trituration
plate of the lower jaw is undivided. In both jaws about
five individual teeth can be seen at the posterior edge of
the trituration plate to each side of the midline. In
diodontids and molids there is a trituration plate in each
jaw, but the number of dental units that go into its make
up differs. In diodontids there is a single dental unit
along the posterior edge of each half of the trituration
plate in each jaw, except for one species with several den-
tal units. In molids there is a single dental unit along the
posterior edge (and sometimes anteriorly) of each half of
the trituration plate in the upper jaw, but in the lower
jaw several such teeth are present.

Steenstrup and Lutken (1898) described the dif-
ferences between the trituration plates in two genera
{Mola and Rumania) of molids, as did Hilgendorf (1893).
Wahlgren (1867) gave one of the earliest and best de-
scriptions of the trituration plate in Mola, and his figure
of that structure was reproduced by Steenstrup and Lut-
ken. Trituration plates are never developed in canthigas-
terins, but, contrary to Fraser-Brunner (1943), they are
sometimes present in tetraodontins. In the upper jaw of
many tetraodontids there is a single longitudinal series of
a small number of teeth, to each side of the midline. In
the lower jaw, however, these trituration teeth are of less
frequent occurrence. They have been reported in the
lower jaws of several tetraodontids by Kaschkaroff
(1914a:317-318), in the lower jaw of the tetraodontid
Colomesus by Pinto (1959), and in many other tetra-
odontids described here. There are a number of papers
describing the general structure of the teeth in plectog-
naths to which the reader should be referred before
passing on to other topics. Cuvier (1805, vol. 3) gave de-
scriptions of the gross morphology of the teeth in plectog-
naths, but it was not until Owen (1840 & 1845) that their
detailed structure was made known. Owen took excep-
tion to the view advocated by Cuvier (1805, 3:125) and by
Bom (1827) that the dental lamellae in gymnodonts
develop by the "apposition or the transudation of layers
of calcareous matter from the pulp's surface" (Owen
1840:77). On the contrary, Owen showed that these
lamellae developed by "intus-susception or the deposi-
tion of calcareous tubes in the pulp's substance." Owen
pointed out that whereas the teeth of the biting part of
the jaws in diodontids and tetraodontoids were dif-
ferent, there was a marked similarity between the dental
lamellae of the trituration plate of diodontids and the
elongate dental lamellae of the biting portion of the jaws
in tetraodontids. A more modem treatment of the
general anatomy of the teeth in gymnodonts is that
presented by Pflugfelder (1930), Andreucci (1966a, b,
1967a, b, 1968a, b, 1969, 1970), Andreucci and Britski
(1968a, b, 1969a, b, 1970, 1971), and Andreucci and
Blumen (1971). The rodlike structure of the dental units
of tetraodontids and the platelike trituration lamellae
have been described by a number of workers other than
those mentioned above, notably by Goodrich (1909),
Ghigi (1921), and Tretjakoff (1926c). Jenyns (1842:150)
was probably the first to call attention to the discrete
dental units that are visible on the external surface of the

crushing beak of diodontids in his description of a species
of Diodon: "the true teeth appear on the surface of the
jaws like minute scales, as in several species of the genus
Scarus."

It will be recalled that Regan (1903a) placed the genus
Triodon in the Sclerodermi, and, as a consequence of
this, it was necessary for him to state (p. 285) that "the
coalescense of the teeth in the jaws is a feature of little
importance, and has, as probably as not, originated in-
dependently in . . . [the Sclerodermi] . . . and in the
Gymnodontes." That view is not followed here, as the
placement of Triodon in the Gymnodontes indicates.

For a discussion of the surface structure of plectog-
nath teeth in relation to sound production by stridula-
tion the reader is referred to Burkenroad (1931:22-24),
Fish et al. (1952:189-190), Fish (1954:62-77), Moulton
(1958:359-362), and Vincent (1963).

The two dentaries are indistinguishably fused with one
another in Triodon, diodontids, and molids. The fibrous
tissue articulation of the two dentaries in tetraodontids is
strengthened by emarginations from the medial edges of
the two bones, just as is the case with the premaxil-
laries.

In triacanthodids and molids the articular and den-
tary are firmly held to one another mainly by fibrous tis-
sue, but in other plectognaths there is normally an ex-
tensive interdigitation between the surfaces of these two
bones. It is not true, however, that the dentary and ar-
ticular are fused into a single piece, as was maintained
by Gill (1872:XL; 1885:412) for plectognaths in general,
and as Smith (1935:359) gave as a characteristic of the
balistoids. Regan (1903a:286) correctly stated that the
dentary and articular are not fused in the gymnodonts,
but, rather unaccountably, he agreed with Gill that these
two bones are fused in the scleroderms. In the descrip-
tions of the representative plectognaths given here no at-
tempt is made to differentiate between a dermal and an
endochondral portion of the articular, for this requires
histological study of the developing bone for consistant
accuracy. Only two papers in the literature treat this
subject in plectognaths. Rosen (1916a: 19) simply said,
without any elaboration, that in Sphoeroides the "ar-
ticulare develops as an autarticulare and a dermar-
ticulare." Haines (1937:14-21) gave a detailed descrip-
tion of the growth and fate of Meckel's cartilage and the
surrounding ossifications in Tetraodon, and concluded
that the endochondral portion of the articular was either
greatly reduced or absent. The sesamoid articular is pres-
ent in all plectognaths, except some tetraodontids and
molids. It has been described in a number of different
plectognath genera by Starks (1916:32-33), who said that
in Sphoeroides annulatus "the sesamoid articular more
evidently originates within the tendon than in any other
examples I have encountered."

With the exception of Rumania, which has lost the
sixth branchiostegal ray, all gymnodonts have six such
rays, and statements in the literature giving a lesser
number probably reflect the loss of a ray during dissec-
tion, while Hubbs (1919:69) pointed out that "in Tetrao-
don the uppermost ray basally is an unossified liga-

ment," which is true also of some other tetraodontids
where the sixth ray tends to be an especially slender
shaft. McAllister (1968) accurately surveyed the num-
ber of branchiostegal rays and the structure of the hyoid
arch in a wide selection of plectognaths, far more so than
ever previously attempted, and found them to be related
to the acanthopterygian type.

Wellenbergh (1840) gave an accurate description and
figure of the six branchiostegal rays in Mala, but Kasch-
karoff (1914a:282) and Van Dobben (1935) stated that
there are only five branchiostegal rays present in Mola.
Roon and Pelkwijk (1939) and Roon (1942) have again
called attention to the fact that there are six branchio-
stegal rays in Mola. Thilo (1899a, 1914) first pointed out
that species such as Stephanolepis setifer and
Brachaluteres trossulus possess only five branchiostegal
rays. In contrast to Ranzania, however, it is not the sixth
branchiostegal that is lost; rather, it is the second
branchiostegal. The branchiostegal counts given by
Willen (1941, 1942, 1944, 1945, 1947, 1949) are not
reliable. He said that there were six branchiostegals in
Triacanthus, Monacanthus, Ostracion and Mola, but that
there were seven in Batistes and five in Tetraodon and
Diodon. The supposed "seventh ray" of balistids was
neither figured nor well described, and it is only possible
to guess that Willem may have mistaken the rodlike in-
teroperculum for a seventh branchiostegal.

The branchial region is one of the most neglected
anatomical systems of fishes, for cleared and stained
specimens and especially delicate dissections are re-
quired, although Nelson (1969 et seq.) has made great
strides in rectifying this situation. In plectognaths the
extensive variation in branchial structure reported here
(Table 1) forms an important part of the diagnoses of the
higher categories recognized and of the phylogenetic inter-
pretations offered, and yet the only previous survey of
the plectognath branchial apparatus is the cursory one
by Kaschkaroff( 1914a). There are, however, a few papers
containing descriptions of the branchial or hyoid arches
of particular genera, such as Awati and Bal (1933) for
Tetraodon, Borcea (1907) for Batistes, Steenstrup and
Liitken (1898) for Mola, Supino (1905) for Balistes and
Mola, and Wellenbergh (1840) for Mola.

The condition of the plectognath branchial ap-
paratus, with comments on the literature, is sum-
marized below. A basihyal is always present but the dor-
sal hypohyal is absent in many tetraodontids and a few
diodontids. The hypohyals are enlarged in ostracioids,
and it was probably the size of the dorsal hypohyal that
lead Horshelmann (1866:23) to say that Ostracion
possessed a tongue while the other plectognaths did not.

The ceratohyal and epihyal are always present, while
the interhyal is absent in diodontids and most tetra-
odontids. The urohyal is well developed only in triacan-
thoids and balistoids, although it is present in a much-
reduced state in ostracioids and in Triodon. The
branchiostegal rays usually are present in the 2-1-4
arrangement typical of percoid fishes (Hubbs 1919), al-
though the second and sixth branchiostegal rays are ab-
sent in a few species of monacanthids and aracanids, and

the sixth absent in one of the molids. There are a num-
ber of diagnostically important differences given here in
the manner by which the branchiostegal rays are articu-
lated to the ceratohyal and in the form of the modified
first branchiostegal ray of many gymnodonts.

Thilo (1899b, 1914) drew attention to the obvious cor-
relation between the presence of an inflation mechanism
in many gymnodonts and the enlargement of the first
branchiostegal ray, and described the pumping action of
that platelike element. Thilo (1914) contended that when
the distensible diverticulum of the oesophagus is not
well developed (as in Fugu rubripes and Lagocephalus
scleratus, according to Thilo) the first branchiostegal ray
is as small as it is in most balistoids. Thilo is guilty of
exaggeration, for the first branchiostegal ray in those two
species is much larger than it is in any balistoid, even
though it is somewhat reduced in size from that seen in
more typical gymnodonts.

There are always three basibranchials (except in the
monacanthid Psilocephalus with two), three pairs of
hypobranchials, five pairs of ceratobranchials, and four
pairs of epibranchials (except in Psilocephalus with
three), but the number of pharyngobranchials is variable
and of much diagnostic interest. There are always at
least two pharyngobranchials, those of the second and
third arches, except that in diodontids they are of the
first and second arches. The pharyngobranchial of the
first arch is absent in monacanthids, molids, and some
ostraciids, while that of the fourth arch is absent in
balistids, monacanthids, ostraciids, tetraodontids, and
diodontids and that of the third arch absent only in
diodontids. The dentition of the pharyngobranchials is
also of great diagnostic use, for teeth are present on the
pharyngobranchial of the first arch in diodontids and
some tetraodontids, on that of the second arch in all but
a few aracanids, on that of the third arch in all but some
ostraciids and diodontids, and on that of the fourth in
triacanthodids, triacanthids, aracanids, triodontids, and
molids. The fifth ceratobranchial is toothed in triacan-
thodids, triacanthids, balistids, triodontids, and, with
minute teeth, in a few diodontids.

In Canthigaster the first three ceratobranchials are
more compressed and deeper than in other tetraodon-
tids, while in molids the ceratobranchials are so deep
that in Mola they have been called (Barnard 1935:657)
"knife-like bones." These support the enormous gills of
molids, which were well illustrated as long ago as Ales-
sandrini (1839). The first pharyngobranchial in triacan-
thoids, balistoids, and triodontids is usually a toothless,
often rodlike, suspensory element.

Grieb (1935) has said that teeth are present on the
lower pharyngeals of "Spheroides sp.," but that has not
been found to be the case in any tetraodontids examined
here. The reader is referred to Al-Hussaini (1947) for an
interesting discussion of the correlation between diet,
length of intestine, and development of pharyngeal teeth
in a number of plectognaths. Iwai (1964) has carefully
surveyed the taste buds on the gill rakers and gill arches
of Fugu and Rudarius.

In the pectoral girdle the position of the supracleith-

rum varies from more or less vertical in scleroderms to being
obliquely placed in relation to the skull in triodontids
and tetraodontids, while in diodontids and molids it is
almost horizontal to the skull. Regan (1903a) used the
position of the supracleithrum as one of the distinguish-
ing features between the Sclerodermi and Gym-
nodontes.

It was probably Gill (1872, 1885) who made best known
that there is some rudiment of a posttemporal present
between the supracleithrum and pterotic in at least some
plectognaths, and much the same was said by Regan
(1903a:285), Rosen (1916a:21), and Berg (1940:495).
Even though relatively small and closely sutured and in-
terdigitated to the skull, careful examination reveals its
presence in all scleroderms except one monacanthid, and
its absence in all gymnodonts.

Siebenrock's (1901) descriptions of the pectoral girdle
in a number of plectognaths are so erroneous that they
are useless, and Regan (1903a:291) called attention to the
incorrectness of Seibenrock's contention that the supra-
cleithrum of Mola was really a posttemporal. Sjirensen's
(1883) mainly myological description of the pectoral
region of Tetraodon is highly detailed.

Both the dorsal and the ventral postcleithra are pres-
ent in most plectognaths as distinct elements, hut in
monacanthids, diodontids, and molids the two elements
tend to indistinguishably fuse. Fraser-Brunner (1941c:307)
used size of the postcleithra as one characteristic to distin-
guish aracanids from ostraciids. The suture between the dor-
sal and ventral postcleithra is difficult to trace in many
ostracioids, and it is possible that the two elements oc-
casionally fuse together. The anteriorly directed process,
which lies just lateral to the actinosts, of the fused post-
cleithra of molids is a diagnostic feature of that group.
Kuronuma (1943) has used the shape of the ventral post-
cleithrum in tetraodontids as a systematic character and
Parr (1927) gave an interesting discussion of the func-
tion of the postcleithra in gymnodonts, while Winterbot-
tom (1971) has thoroughly surveyed the matter. Thilo
(1899b) was surely in error in saying that the post-
cleithra could be moved through an arc of 90° . The role
played by the postcleithra and cleithra in sound produc-
tion has been discussed by Mobius (1889), Cunningham
(1910:117), and Fish (1954:69). For a general survey of
the endochondral bones of the plectognath pectoral gir-
dle the reader is referred to Starks' (1930:206-210) excel-
lent work on that subject. Sound production by pectoral
fin drumming is reviewed by Salmon et al. (1968), and by
this and other methods by Schneider (1961).

The scapula completely encloses the scapular foramen
in most scleroderms and in Triodon, but in tetraodontids
and diodontids the foramen is incomplete, being enclosed
by both the scapula and the cleithrum. In molids the
scapular foramen is essentially absent or highly incom-
plete, being represented only by a small hole in the sheet
of fibrous tissue holding the bottom half of the scapula to
the cleithrum. In scleroderms and in Triodon the dorsal
edge of the scapula always bears an emargination to
which is attached the first pectoral fin ray, but in the
other plectognaths no such emargination is present. Four

actinosts are present in all plectognaths, except in
molids, which have three.

The close connection of the scapula with the first ac-
tinost is such that Regan (1903a:291) drew attention to
the fact that "On a superficial examination there ap-
pears to be no scapula, and the pectoral fin to be sup-
ported by a series of four enlarged pterygials. In fact, the
united upper pterygial and scapula together resemble
one of the enlarged pterygials." Even after this state-
ment, others (e.g., Kaschkaroff 1914a:355; Awati and Bal
1933:91) continued to misidentify the actinosts and
scapula. On the other hand, Regan's (1903a) statement
that the scleroderms have the "pterygials (pectoral
basalia) not enlarged, movably attached by ligament to
the scapula and coracoid, three to the former and one to
the latter" (p. 286) while the gymnodonts have the
"lower three pterygials enlarged and immovable united
to the coraco-scapula" (p. 291) is only partially true (see
above). In molids what was the first actinost of other
plectognaths is either lost or fused with the scapula, so
that only three actinosts are present. Somewhat under-
standably, a number of workers have mistaken the
scapula of Mola for an actinost and have concluded that
Mola possessed four actinosts and no scapula. However,
Gregory and Raven (1934:148) described Mola as having
four actinosts (i.e., three actinosts and a scapula), as well
as a "vestigial scapula" represented by a "bone which is
crowded between the expanded first pterygial and the
posterior recurved border of the cleithrum." It is impos-
sible to state what it was that Gregory and Raven saw,
for their description is brief and there is no illustration of
the vestigial scapula.

The structure of the pelvis and pelvic fin in plectog-
naths has been the cause of much confusion in the
literature, for the relatively normal pelvic apparatus of
triacanthoids is replaced by a highly modified and
specialized structure in balistoids that has been often
misinterpreted. Tyler (1962b) has surveyed in detail the
reduction and eventual loss of the pelvic apparatus in
plectognaths, and these features are used prominently in
the diagnoses given here, while Winterbottom (1970) sur-
veyed the muscles of the triacanthoid pelvic fin. The
locking mechanism of the triacanthoid pelvic spines
(continuous series of positions of erection in triacan-
thodids and two positions in triacanthids) is also treated
by Tyler (1962b). HoUard (1853:107) and Thilo (1896a,
1896b:327, 1898, 1899a, 1900) incorrectly described only a
single position of erection in Triacanthus, but the two-
position mechanism has been correctly described by
Sorensen (1884:69, 1897) and Monod (1959c).

The true nature of the rudimentary pelvic fin element
in balistids was first made known by Monod (1959a), in
excellent style for Balistes forcipatus.

As detailed in the accounts of each family given here, the
number of vertebrae in plectognaths varies from 16 to 30
(Table 2) but is usually less than 24 and normally in the
range of 17 to 21, while the structure of the vertebral
column is highly modified in some groups, especially the
ostracioids (Tyler 1963a) and molids. In aracanids the
first two, and in ostraciids the first four or five, ab-

dominal vertebrae are at least partially fused to the
skull, and the whole vertebral column is relatively in-
flexible except in the caudal peduncular region ex-
tending outside of the carapace. Another unusual feature
of the ostraciid vertebral column, first pointed out by
Fraser-Brunner (1941c:307), is that the course of the
haemal canal is displaced to one side or the other from
the midline, while Tyler (1963a) called attention to the
uniquely divergent positions away from the midline of
the anal fin basal pterygiophores. Abe (1942) presented
an extensive study of the variation in the form and num-
ber of elements of the tetraodontid vertebral column.
Ford (1937) compared the balistid and zeoid vertebral
columns.

The fact that true or pleural ribs are found in the
monacanthid Pseudaluteres and in the gymnodont
Triodon negates Regan's (1903a:285) statement that "the
feature of most importance in diagnosing the suborder
Plectognathi is the absence of ribs."

The sequence of vertebrae to which the epipleural
bones are attached is a character of systematic impor-
tance, as first shown by Fraser-Brunner (1941b: 176) in
his differentiation of the triacanthodids from the triacan-
thids and of the balistids from the monacanthids. The
epipleurals in some monacanthids tend to become swol-
len (hyperostotic) and closely applied to the transverse
processes of the vertebrae, and it is not surprising that
they have been misinterpreted (e.g., Smith 1935). The
development of the epipleurals has been studied in
several monacanthids by Goeppert (1895) and Goette
(1879).

The structure and reduction in number of elements in
the caudal fin supports is documented by Tyler (1970b),
while comments are given by Pope (1945), Okada (1950),
Randall (1964), and Tyler (1970d) on abnormalities of
vertebral structures. The work of Abe (1949b) should be
consulted for a detailed survey of hypural fusion in tetra-
odontids. The description given by Whitehouse (1910)
was one of the earlier accurate accounts of a balistid
caudal skeleton. Monod (1968) accurately surveyed and
illustrated the ural structures of a wide variety of plec-
tognaths, and agreed with Dareste (1850) and Le Danois
(1955) that the triacanthoids and balistoids are not close-
ly related to the ostracioids and gymnodonts (the Or-
biculates).

Gosline (1961:269) pointed out that "there seems to
have been a general trend toward fusion of parts in the
caudal skeleton of teleostean fishes that has occurred
repeatedly and independently in various lineages." The
Plectognathi are certainly a perfect example of this, since
there are several uroneurals, one epural, and six hy-
purals in the primitive triacanthodids, but the number of
elements becomes reduced in the triacanthid-balistid-
monacanthid series and becomes even more reduced in
the ostracioids. The same reduction takes place in-
dependently in the gymnodonts. Gosline said that the
caudal skeletons of Zanclus, Acanthurus, and the plec-
tognath fishes form a series, but Monod (1959b:728) has
said that the caudal skeleton of Acanthurus and Balistes
are "toutefois profondement diff6rente" and thus the

acanthurids and balistids "ne sauraient etre tenues pour
tres eloignees." What Monod did not realize was that if
the caudal skeleton of a triacanthodid is compared with
that of an acanthurid, one finds the two to be rather
similar.

Just as the number of epurals, uroneurals, and hy-
purals becomes reduced in the more specialized plectog-
naths, so also the number of caudal fin rays becomes
reduced. In triacanthoids and balistoids there are 12
caudal fin rays; the uppermost ray and the lowermost ray
unbranched, the others branched (except Psilocephalus) .
In aracanids there are 11 rays, but in ostraciids the num-
ber is further reduced to 10. As with the other sclero-
derms, the uppermost ray and the lowermost ray are un-
branched, but the others are branched. Triodon has 12
principal caudal fin rays, but in addition to these it also
possesses a series of procurrent rays. Triodon is thus the
only Recent plectognath with more than 12 fin rays in the
caudal fin, but its number of principal caudal fin rays is
still the same as in triacanthoids and balistoids. In tetra-
odontids there are usually 11 caudal fin rays; the upper-
most ray and the lowermost two rays unbranched, the
others branched. As a part of his long and useful series of
papers (1942, 1944, 1949a, 1950-51, 1952, 1954, 1960),
Abe extensively and accurately surveyed the variability
of the caudal fin rays (1949b) in tetraodontids, show-
ing the i, 8, ii arrangement to be the normal condition
for all of the numerous species he studied. In diodontids
the number of caudal fin rays is either 9 or 10, with the
uppermost ray and the lowermost ray unbranched,
and the other rays branched.

The complicated subject of the pseudocaudal fin of
molids can only be summarized here. A number of early
workers (Wellenbergh 1840; Goodsir 1841; Cleland 1862;
Wahlgren 1867; Putnam 1871) gave brief descriptions of
the highly modified structures of the caudal fin of
molids, but it was not until Ryder's (1886) work that the
origin of this fin was adequately discussed. Ryder
described some of the developmental stages of the molid
caudal fin and came to the conclusion that that struc-
ture was a gephyrocercal tail; the true caudal fin having
been lost and replaced by posteriorly migrated dorsal
and anal fin rays. Ryder said that molids and carapids
were the only two groups with gephyrocercal tails.
Ryder's opinion did not meet with unanimous accept-
ance, and Kaschkaroff (1914a), Grenholm (1923), and
Gregory and Raven (1934) continued to describe it as a
true caudal fin, while Whitehouse (1910) and Regan
(1910) both agreed that the structure was a gephyrocer-
cal tail. With typical thoroughness, Gudger (1935; 1936;
1937a, b, 1939) and Gudger and MacDonald (1935)
reviewed the literature on the subject and also per-
sonally examined a few developmental stages of molids.
Gudger came to the conclusion that both Mala and
Masturus have a gephyrocercal tail, and Raven (1939a)
agreed with that analysis. Raven (1939b) described the
tail of Ramania as also being gephyrocercal. Fraser-
Brunner (1951) was in basic agreement with Gudger and
Raven on this subject, but Fraser-Brunner stated that
the few fin rays that are found in the "nipple" part of the

tail of Masturus are the remnants of the true caudal fin
rays, whereas the rest of the caudal structure is
gephyrocercal. In short, there is general agreement that
the entire caudal structure in Mola and in Ramania is
gephyrocercal, and that in Masturus at least the great
majority of the caudal structure is gephyrocercal. It is
also agreed that the bony supporting elements of the
pseudocaudal fin in molids are posteromedially migrated
dorsal and anal fin basal pterygiophores. Owen's
(1846:64) description of these pterygiophores as
"rudimental vertebrae . . . blended together at right
angles to the rest of the column" does not seem plausible.
The variability in the number of fin rays said to be pres-
ent in Mola (summarized by Beauregard 1893) attests to
the fact that it is difficult to delimit the pseudocaudal fin
from the dorsal and anal fins, as well as to the fact that a
great many of the descriptions of Mola in the literature
are actually based on Masturus (summarized by Gudger
1937a). To complicate matters even more, Fraser-Brun-
ner (1951) attempted to distinguish two species of Mola,
partially on the basis of the number of fin rays borne on
modified basal pterygiophores. Barnard (1935) has
pointed out the diagnostic value of the degree of branch-
ing of the fin rays of the pseudocaudal fin in molids.

The dorsal and anal fins, and especially the dorsal fin
spines, being highly visible and variable between groups
of plectognaths, have been featured prominently in near-
ly all diagnoses, including here, as summarized below.
In triacanthodids there are six dorsal fin spines sup-
ported by five basal pterygiophores, with the first two
spines always prominent and visible externally, the third
to fifth either normally developed or rudiments buried
beneath the skin or barely protruding to the surface, and
the sixth spine either a short protruding element or a
rudiment buried beneath the skin or barely protruding to
the surface. There are 12 to 18 dorsal fin rays and 11 to 16
anal fin rays in Recent species.

In triacanthids there are nearly always six dorsal fin
spines, except that the fifth and sixth spines are
sometimes absent in one Recent species and perhaps in
several fossil forms, supported by four, rarely only three,
basal pterygiophores, with the first four spines always
visible externally, the fifth spine usually very short but
nearly always protruding at least a short distance
through the skin, and the sixth spine nearly always pres-
ent as a buried rudiment. There are 19 to 26 dorsal fin
rays and 13 to 22 anal fin rays.

In balistids there are three dorsal fin spines, the second
more than one-half the length of the first, supported by
two basal pterygiophores and a supraneural strut. In
monacanthids there are usually two dorsal fin spines, but
sometimes only one, the second spine, when present, not
more than one-third the length of the first, with both
spines supported by a single basal pterygiophore,
without a supraneural strut. In balistoids there are 23 to
52 dorsal fin rays and 20 to 66 anal fin rays. The locking
mechanism of the spines of triacanthoids is described by
Tyler (1968), and that of balistoids by the references
given in the description of Balistapus undulatus.

In ostracioids there is no spiny dorsal fin, and the dor-

sal and anal fins are both short-based, with 9 to 13 rays
each.

Among the gymnodonts only Triodon has even a trace
of a spiny dorsal fin, as a minute structure composed
usually of two spines, occasionally only one but with pos-
sible rudiments of a third spine, borne on two basal
pterygiophores, present in most specimens of one of the
populations (Indonesia to Japan) of the single species,
with the second basal pterygiophore succeeded by two
supraneural elements.

It is possible that the supraneural elements in gym-
nodonts are rudiments of the basal pterygiophores of the
now absent spiny dorsal fin. The structure and position
of the single supraneural element in tetraodontids give
no indication of its possible derivation from a spiny dor-
sal fin pterygiophore, but the structure and number of
the supraneurals in Triodon are reminiscent of the spiny
dorsal fin pterygiophores of triacanthids. An at least
analogous situation is present in ostracioids, for the
aracanids have a very long supraneural element which
extends throughout almost the entire distance between
the supraoccipital and soft dorsal fin, but in ostraciids
this supraneural is short and confined to a position just
in front of the soft dorsal fin.

One should refer to Bridge (1896) and Rosen (1916a)
for descriptions of the cartilaginous distal pterygio-
phores in the dorsal and anal fins of several plectog-
naths. Hora's (1924, 1925) descriptions of Kanduka as a
genus of tetraodontid without a dorsal fin and with the
anal fin at least rudimentary, if not absent, are to be
looked upon skeptically. His two small (14 and 54 mm
SL) specimens were highly inflated, and the fins were
undoubtedly hidden beneath the skin. The fact that
Hora's radiographs did not show these fins probably
only means that the pictures were overexposed and that
the specimens were still weakly ossified.

Much has been said in the literature about the general
structure of the bones of plectognaths. Some of the ear-
ly workers obviously thought the plectognaths to be

similar to the "cartilaginous" fishes, probably because of
the relatively late ossification of the bones that is seen in
some plectognaths and because of misinterpretation of
the spongy nature of the ossifications in Mala and of the
large amount of cartilage that is present even in the adult
Mola. The consensus, however, seems to be that while
many plectognaths simply have a relatively late ossifica-
tion of the endochondral bones of the skull, the com-
position of the ossification is normal. This is more or less
what Cuvier (1817:144) said in his definition of the Order
Plectognathi. The principal references to the histology of
the bones of plectognaths are the following: Quekett
(1850-1855, bones of a number of plectognaths), Leydig
(1857, bones of Mola), Kolliker (1859, bones of
numerous plectognaths; 1860, bones of Mola), Harting
(1865, bones of Mola), Goette (1879, epipleurals of
Monacanthus), Trois (1883-1884, bones of Ranzania
and Mola), Goeldi (1884, general bone structure in
Balistes), Goeppert (1895, development of epipleurals
of Monacanthus), Stephan (1900, bones of Mo/a, Tetra-
odon, and Balistes), Supino (1904, bones of Mola),
Nowikoff (1910, bones of Mola), Kaschkaroff (1914a
and 1925, bones of numerous plectognaths; 1914b and
1916, bones oi Mola), Studnicka (1916, bones of Mola),
Rauther (1927b, epipleurals and vertebrae oi Monacan-
thus), Haines (1937, Meckel's cartilage in Tetrao-
don).

The literature on the muscles of plectognaths has been
thoroughly reviewed by Winterbottom (1974) and dup-
lication of that laudable effort is avoided here.

The smatterings of information available on other sys-
tems (such as the blood-vascular, digestive, renal, neu-
ral, chromosomal, cellular, physiological, developmental,
etc.) are too incomplete for plectognaths to be of im-
mediate value to the present studies, and this literature
will not be routinely reviewed here. It will be analyzed in
a forthcoming synopsis of the biological data on plectog-
naths, and specific papers will only be cited here if they
have direct bearing on other matters discussed.

Relationship of the Plectognathi to the Perciformes

As indicated in the historical review of the classifica-
tion of the Plectognathi, it is commonly thought that the
plectognaths are a derivative of the perciform fishes, and
perhaps most closely related to the acanthurid surgeon-
fishes. The features of the osteology of the Recent acan-
thurids most pertinent to an interpretation of their
classification and interrelationships have been reviewed
by Tyler (1970c), aiming especially at ascertaining what
is generalized versus specialized in that family. The com-
plementary analysis of the fossil forms of acanthurids is
not yet completed, and the necessary comptirisons
between them and the fossil plectognaths studied here,
especially those of both groups from the Eocene, must
await a subsequent publication under preparation by
J. Blot (Mus. Nat. Hist. Nat., Paris) and the author.

However, as indicated by Tyler (1968), it is obvious
that the skeletal structure of acanthurids is not far

removed from that of triacanthoids, the basal plectog-
naths, differing mainly in the presence of a number of
bones absent in triacanthoids and in a greater number of
certain meristic elements. In general appearance the
acanthurid skeleton is closer to that of triacanthids
among the plectognaths rather than to that of the more
generalized triacanthodids. This might suggest that if
the triacanthoids and thus the other plectognaths share a
common Upper Cretaceous ancestral stock (as sug-
gested by Patterson 1964:400) with the acanthurids, that
the Recent acanthurids are not as generalized a group as
the triacanthodids are relative to the triacanthids.

Acanthurids differ most significantly from plectog-
naths by having: two or three anal fin spines (none in
plectognaths); a relatively better developed pelvic fin
varying from 1,5 to 1,3, with the rays cross-striated and
branched (one spine followed by, at the most, two much

40

smaller unbranched rays without cross-striations in
Recent triacanthodids, but with the Eocene Eoplectus
having a spine and four long branched and presumedly
cross-striated rays); 16 principal caudal fin rays (12 in
triacanthodids, the maximum for the order); the presence
of a folding spine or fixed plates on the caudal peduncle
(folding spine lacking in all plectognaths but with fixed
plates in some scleroderms); a complete series of subor-
bital bones (none in plectognaths); nasal or other bones
lateral to the ethmoid and anterior to the prefrontal
(none in plectognaths); parietals (none in plectognaths);
true or pleural ribs (lacking in all plectognaths except in
one monacanthid and in the primitive gymnodont Trio-
don); a basisphenoid (lacking in all plectognaths except
Triodon and molids); a relatively normal percoid type
pelvis without a long extension posterior to the level of
the pelvic fin origin (a long extension of the pelvis pres-
ent only in triacanthoids posterior to the level of the pel-
vic fin origin, this portion being at least half, and usually
more, of the total length of the pelvis); very short
premaxillary pedicels and a nonprotractile upper jaw
(premaxillary pedicels well developed and upper jaw
protractile only in triacanthoids); and a dorsal fin con-
sisting of an anterior portion of 4 to 9 (rarely 3 or 10)
spines not decreasing much, if at all, in length pos-

teriorly in the series (except in the young of some species)
and followed without spatial interruption or indentation
of the membrane by about 20 to 33 rays (as many as six
dorsal spines present only in triacanthoids, decreasing
greatly in length posteriorly in the series, except in the
Eocene Spinacanthus, and followed either without
spatial interruption or with only a slight spatial inter-
ruption, but always with an indentation of the mem-
brane to the base of the fin, by 12 to 26 rays in Recent
species of triacanthoids and as few as about eight in
Spinacanthus; no more than three dorsal spines in other
plectognaths).

A number of other features of acanthurids (e.g., the
ridged surfaces, deep indentations, and hooks found on
the distal articular areas of the first basal pterygio-
phores of the spiny dorsal and anal fins and the laterally
expanded distal ends of many of the others; the usually
highly modified dentition; less restricted gill opening;
longer and more anteriorly placed ethmoid; two or three
more vertebrae; fewer and less well-developed branchios-
tegal rays anteriorly in the series, etc.) distinguish them
from triacanthoids and other plectognaths, but these will
be dealt with in detail along with their similarities to
triacanthoids in a subsequent work comparing the fossil
acanthurids and triacanthoids.

SYSTEMATIC SECTION

Definition and Synopsis of the Osteology of the Order Plectognathi (Tetraodontiformes)

Until such time as most other orders of fishes have
been extensively studied anatomically, and especially os-
teologically, it is impossible to give a comparative in-
clusive diagnosis of the Plectognathi that would clearly
and succinctly distinguish them at once from all other
orders of Acanthopterygii, and especially from all sub-
groups of the Order Perciformes, of which order the plec-
tognaths are most likely a specialized derivative. The
perciform suborder Acanthuroidei has often been
thought to be the closest extraordinal relative of the plec-
tognaths, the implication being that the acanthurids and
plectognaths share a common ancestral line in the
Cretaceous, with Patterson (1964:400) having suggested
that this was the Pharmacichthyidae. Tyler (1968:42-43)
briefly pointed out that the Recent acanthurids differ
from the most generalized plectognaths, the triacan-
thodids, mainly in the retention of a number of bones not
now found in triacanthodids and in a greater number of
certain meristic elements. I subsequently studied the os-
teology of numerous additional Recent species (the major
features summarized in Tyler 1970a, c) and of a large
majority of the specimens of fossil acanthurids available
(unpubl. data), mainly from the Eocene of Monte Bolca,
Italy. These subsequent studies of acanthurids tend to
support the view that acanthurids are probably relatively
generalized (except in dentition, caudal peduncle ar-
mature, and dorsal and anal spine locking mechanisms)

representatives of the same ancestral line which gave rise
to the plectognaths, with at least the majority of plectog-
naths becoming far more specialized than the acan-
thurids in most respects.

These matters will be discussed in a subsequent paper
on the anatomy of fossil and Recent acanthurids by
J. Blot and the author, and then on their differences
from and similarities to the plectognaths.

For the moment, the Order Plectognathi is defined
below by listings of the bones that are either constantly
present throughout the order or are of variable oc-
currence.

Certain bones apparently are always present in plectog-
naths, although in many cases these bones are highly
variable in size and shape from one family or genus to the
next. Those elements which are constantly present are:
basioccipital; exoccipital; supraoccipital; pterotic;
sphenotic; epiotic; prootic; frontal; parasphenoid; eth-
moid; vomer (with the possible exception of diodontids);
hyomandibular; quadrate; metapterygoid; symplectic;
palatine; ectopterygoid; operculum; suboperculum; pre-
operculum; premaxillary; maxillary; dentary; articular;
angular; ventral hypohyal; ceratohyal; epihyal; four
branchiostegal rays (at least one in the anterior group
and three in the posterior group); two basibranchials (of
the second and third arches); three hypobranchials (of
the first to third arches); five ceratobranchials (support-

ing at least three gills, of the first to third arches); three
epibranchials (of the second to fourth arches); two
pharyngobranchials (at least of the second and third
arches; except in diodontids, of at least the first and sec-
ond arches); supracleithrum; cleithrum; postcleithrum
as at least a single piece; coracoid; scapula; three acti-
nosts (second to fourth sequentially from anterodorsal to
posteroventral); between 9 and 25 pectoral fin rays (fin
placed in about middle of side of body); between 7 and 52
dorsal fin rays and a similar or only slightly lesser num-
ber of basal pterygiophores; between 7 and 66 anal fin
rays and a similar or only slightly lesser number of basal
pterygiophores; between 16 and 30 vertebrae (usually
less than 24); gill rakers on the anterior and posterior
edges of the first to third arches.

Those elements which are present in some plec-
tognaths, but not in others, are: basisphenoid (con-
sistently present only in triodontids and molids, but rare-
ly present in at least one species of triacanthodid);
sesamoid articular (absent only in several species of
tetraodontids and molids); basihyal (present only in
triacanthodids); dorsal hypohyal (absent only in a few
diodontids and many tetraodontids); interhyal (absent
only in diodontids and most tetraodontids); second and
sixth branchiostegal rays (one or both absent only in a
few species of monacanthids and aracanids; the sixth
absent in one species of molid), urohyal (absent in all
Tetraodontoidei except triodontids); first and fourth
pharyngobranchials (pharyngobranchial of first arch
absent in monacanthids, molids, and some ostraciids;
that of the fourth arch absent in balistids, monacan-
thids, ostraciids, tetraodontids, and diodontids) and
third pharyngobranchial (absent only in a few diodon-
tids); teeth on the pharyngobranchials (that of the
first arch with teeth in diodontids and some tetra-
odontids; that of the second with teeth in all but a few
aracanids; that of the third with teeth in all but some os-
traciids and diodontids; that of the fourth with teeth in
triacanthodids, triacanthids, aracanids, triodontids, and
molids); teeth on fifth ceratobranchial (toothed in
triacanthodids, triacanthids, balistids, and triodontids
and with minute teeth in a few diodontids);
basibranchial and epibranchial of the first arch (absent
only in one species of monacanthid); first actinost (ab-
sent only in molids); pterosphenoid (absent only in one
species of tetraodontid); prefrontal (absent only in two
species of tetraodontids and several species of diodon-
tids); posttemporal (absent in one species of monacan-
thid and in all Tetraodontoidei); mesopterygoid (ab-
sent in most species of triacanthids and in one species of
tetraodontid); interoperculum (absent only in one
species of molid); pelvis (absent in aracanids, ostraciids,
tetraodontids, diodontids, molids); well-developed pel-
vic fin spine (present only in triacanthodids, with the
possible exception of several Eocene species, and triacan-
thids; thoracic in position, placed under pectoral fin
base); well-developed pelvic fin rays (present only in one
Eocene species of triacanthodid, with four rays); poorly
developed pelvic fin rays immediately following the spine
(one or two present in many species of triacanthodids;

one even less well-developed ray present in some species
of triacanthids); a rudimentary but complex composite
pelvic fin element at the posterior end of the pelvis most-
ly hidden from view by enlarged encasing scales (present
only in balistids and many monacanthids); pleural ribs
(present only in triodontids, one species of monacan-
thid, and probably in one of the Eocene species of both
the triacanthodids and the tetraodontids); epipleurals
(present in all triacanthodids, with the possible exception
of one of the Eocene species, and in all triacanthids,
balistids, monacanthids, and triodontids); a separate
epural (absent in aracanids, ostraciids, and molids and
in at least most diodontids); uroneurals (present as one
or two pairs in triacanthodids, triacanthids, and triodon-
tids, and, rarely, in at least one species of balistid); a
separate parhypural (autogenous in triacanthodids,
balistids, monacanthids, triodontids, and tetraodon-
tids); one or more separate hypurals (usually five,
sometimes only three or four, separate hypurals in
triacanthodids, four in triodontids, one in triacanthids,
balistids, and tetraodontids, and one in at least most
monacanthids); a separate haemal spine of the penul-
timate vertebra (autogenous in triacanthodids, balistids,
monacanthids, aracanids, triodontids, tetraodontids,
and a few species of ostraciids); a separate haemal spine
of the antipenultimate vertebra (autogenous only in trio-
dontids); procurrent caudal fin rays (present in triodon-
tids and in at least one species of Eocene triacanthodid);
principal caudal fin rays (apparently absent in at least
two of the three species of molids, but perhaps represent-
ed in the other species by the rays in the central nipple of
the pseudocaudal fin otherwise formed of posteriorly
migrated dorsal and anal fin rays; 12 principal rays in
triacanthodids, triacanthids, balistids, monacanthids,
and triodontids, 11 in aracanids (10 in one species) and
tetraodontids, 10 in ostraciids, and 9 or 10 in diodontids;
in all cases the uppermost ray and the lowermost ray be-
ing unbranched and the intervening rays branched, ex-
cept in tetraodontids in which the uppermost ray and the
two lowermost rays are unbranched, and in one species of
monacanthid in which several rays both above and below
are unbranched); dorsal fin spines and their basal pte-
rygiophores (six spines, the last four of which may be
rudimentary and mostly buried beneath the skin, the
spines borne on five basal pterygiophores in triacantho-
dids; usually six spines, rarely only four or five, the sixth
spine nearly always a buried rudiment, the spines borne
on four, rarely three or five, basal pterygiophores in
triacanthids; three spines borne on two basal pteryg-
iophores in balistids; two spines, rarely only one, borne
on one basal pterygiophore in monacanthids; two, pos-
sibly three, rudimentary spines borne on two basal pte-
rygiophores in triodontids; no spines in aracanids, os-
traciids, tetraodontids, diodontids, and molids);
supraneural elements (one, as a strut supporting the sec-
ond basal pterygiophore, in balistids; one in aracanids
and ostraciids; one in several molids and in most tetrao-
dontids; absent in all monacanthids and diodontids); os-
sified Baudelot's ligament (present only in aracanids and
ostraciids); trituration teeth or teeth internal to the ma-

jor outer series (present in both the upper and lower jaws
of a few triacanthodids and in all triacanthids, diodon-
tids, and molids; present only in the upper jaw of
balistids and monacanthids; present in the upper jaw of
many tetraodontids and, to a lesser extent, in the lower
jaw; absent in all aracanids and ostraciids); gill rakers
along anterior edge of first gill slit (present only in
molids); gill rakers along anterior edge of fourth arch (ab-
sent only in a few diodontids); gill rakers along posterior
edge of fourth arch (absent in tetraodontids and diodon-
tids); gill rakers along anterior edge of fifth arch (present
in monacanthids, aracanids, ostraciids, and molids);
scales (present on at least a part of the body in all species
except a few tetraodontids).

The few major features in the soft anatomy of plec-
tognaths treated here are about as variable between or
within various families as the bony parts just discussed.
Of most importance here, neglecting for the moment the
musculature that has been so superbly studied by Win-
terbottom (1974), is the presence of four gills and a gill
slit between the fourth and fifth arches in all groups ex-
cept tetraodontids and diodontids, which have lost the
gill of the fourth arch and the slit between it and the
fifth. A pseudobranch is present in all species, but the
degree of development and number of lamellae are highly
variable, the number of lamellae ranging from only 4 or 5
to about 70, with balistids and a few tetraodontids tend-
ing to have the pseudobranch the least well developed.
The olfactory epithelium is in the form of a rosette in
triacanthodids, triacanthids, and triodontids, but it is
variously modified in all other families as folds (often ex-
tensive but never as a rosette), ridges, pits and
plications, or relatively smooth. There are two nostrils in
all families, except in tetraodontids and diodontids in
which there are either one or two nostrils. There is
basically a single lateral line on the body in all families,
except in tetraodontids in which there are one, two, or
three lateral lines on the body. A well-developed air
bladder is present in all families, except that it is absent
in molids, at least as adults. An inflatable diverticulum
of the gut is well developed only in tetraodontids and
diodontids, while an expansible dewlap of skin between
the end of the rotatable pelvis and the anus is moderately
developed in most balistids, well developed in many
monacanthids, and greatly developed in triodontids.
Whether a slighly inflatable diverticulum of the gut is
present in a few monacanthids (especially
Brachaluteres) is still debateable (see Thilo 1899b, 1914;
Rosen 1912; Breder and Clark 1947; Clark and Gohar
1953; Arbocco 1957), but, if so, it is far less well de-
veloped than in tetraodontids and diodontids and has
undoubtedly been independently evolved. Molids may
be unique among plectognaths by having a single ovary
(according to Cleland 1862 and Wahlgren 1867 for Mola,
and Pellegrin 1912 for Ranzania). With the possible ex-
ception of molids, all plectognaths have hard otoliths of
normal size, but these are usually so corroded or other-
wise dissolved in the formaldehyde-preserved, cleared,
and stained study material that they cannot be ade-
quately described or figured, and are not further dis-

cussed here. Cleland (1862) said that otoliths were ab-
sent in Mola, while Cuvier (1805) and Thompson (1888;
repeated by Kaschkaroff 1914a) found that the otoliths in
Mola were represented by small calcareous granules
grouped together. It is not known whether this is also the
case in Masturus and Ranzania, but there is no evidence
of otoliths in the present study specimens.

The large number of bones not mentioned above in
either of the two lists (constant versus variable occur-
rence) that are often found in perciform fishes are ab-
sent in plectognaths and will help form a large segment
of the defining characteristics of the Order Plectognathi
when the other orders and suborders of acanthopteryg-
ian fishes are better known anatomically. Some of these
structures typical of perciforms that have been lost by
plectognaths are: anal fin spines; suborbital bones;
parietals; at least one and usually several of the other-
wise five pelvic fin rays, and in all but the Eocene
triacanthodid Eoplectus of the branched and cross-
striated structure of the fin rays that are retained;
nasals; tabulars; scale bones; intermuscular bones other
than 5 to 11 epipleurals; intercalars or "opisthotics";
supramaxillae; any sheath or surface concavity for the
upper jaw; canals in the skull bones for the lateral line
system; vomerine and palatine teeth, and of any other
teeth except those of the dentary, premaxillary, pharyn-
gobranchials, and fifth ceratobranchials; reduced or-
namentation of the skull bones and lack of spiny
processes; simplification of the posttemporal to a
relatively flat rod sutured along all of its medial surface
to the skull (mostly to the pterotic); development of a
small and only slightly, if at all, protractile mouth, al-
though often with massive jaws, and the relatively great
restriction of the gill opening (never ending ventrally
very far below the pectoral fin base); the complete cover-
ing by scaly to scaleless skin of the branchiostegal region,
and the usual development of relatively spinulose or
heavy scales probably derived from ctenoid scales.

The two suborders of plectognaths (Balistoidei and
Tetraodontoidei) are difficult to comparatively diag-
nose, primarily for two reasons. One reason is that both
groups contain an anatomically highly diverse assem-
blage of families. For example, the diagnosis of the Bal-
istoidei must encompass such widely anatomically dif-
ferentiated families as the Triacanthodidae and Ostra-
ciidae, and that of the Tetraodontoidei has to include
such similarly diverse families as the Triodontidae and
Molidae or Diodontidae. Moreover, the diagnoses must
also take into account the great diversity found within
certain families, such as the differences between a gener-
alized monacanthid such as Stephanolepis and a highly
specialized one like Psilocephalus, or the differences be-
tween a generalized tetraodontid such as Sphoeroides
and a highly specialized one like Xenopterus. However,
the most important reason that the two suborders of plec-
tognaths are difficult to comparatively diagnose is that
we are fortunate in having alive today a very generalized
or primitive Recent member of the Tetraodontoidei,
Triodon, and that we also have available the speciaHzed
Eocene fossil subfamily of triacanthodid Balistoidei, the

Eoplectinae, that was ancestral to Triodon and the other
Tetraodontoidei. From the very fact of their close ances-
tor-derivative relationship, many of the characteristics of
the eoplectins and triodontids bridge what are otherwise
the morphological gaps between the known fossil and Re-
cent Balistoidei and Tetraodontoidei.

SUBORDER BALISTOIDEI
(SCLERODERMI)

Comparative diagnosis (contrast with that of the
Tetraodontoidei). — Teeth relatively large (small in one
highly specialized genus of triacanthodids) and discrete
separate units protruding out from sockets in the jaws,
except in the Eocene eoplectin triacanthodids in which
the teeth are small, nonprotruding, and fully incor-
porated into the matrix of the jaw bones just as in many
of the Tetraodontoidei, to which they are ancestral; den-
taries and premaxillaries never fused to their opposite
members, except in at least some eoplectins; the medial
articulation of the premaxillaries to one another not
strengthened by alternating emarginations and inden-
tations; posttemporal present, except in one highly
specialized species of monacanthid; urohyal present; pel-
vis present, except in the ostracioids, the most specializ-
ed superfamily; pelvic fin present, at least as a rudiment
at the posterior end of the pelvis, except completely ab-
sent in many of the more specialized monacanthids and
in all ostracioids; palatine relatively small and not firmly
sutured to both the ethmoid-vomerine region and the
pterygoid arch, except probably so sutured in the eoplec-
tins; myodome with a complete dorsal roof present, ex-
cept small to absent in ostracioids; prootic shelf under
the orbit present, except in triacanthoids, the most
generalized superfamily, and in several highly specializ-
ed monacanthids; supracleithrum placed vertically or
only slightly obliquely to the axis of the skull; scapular
foramen complete, except in two highly specialized
species of monacanthids; scapula with a knob or crest for
articulation with the uppermost pectoral fin ray; at least
some of the distal pterygiophores of the soft dorsal and
anal fins ossified, except in ostracioids; spiny dorsal fin
varying from six well -developed spines to a single slen-
der spine, or, in ostracioids, absent altogether, but when
present having its origin above or close behind the skull;
first branchiostegal ray relatively unmodified, without
an inturned dorsomedial edge.

Infraorder Triacanthoideo

Comparative diagnosis (contrast with that of the
Balistoideo), which is also that of its only contained
Superfamily, the Triacanthoidea.— Premaxillary with
a well-developed pedicel or ascending process, the
pedicel usually a sturdy rodlike shaft (but variously

modified in the two long-snouted genera of triacantho-
dids) sliding along the dorsal surface of the ethmoid
and, sometimes, vomer as well; premaxillary movably
articulated with the maxillary, allowing for a slight
protraction of the upper jaw; maxillary deeply indented
dorsally for articulation with the anterior end of the
palatine, except in the long-snouted triacanthodids;
palatine usually with a squarish or oblong major portion
(as seen laterally) from which arises an anterior prong for
articulation with the maxillary, although much modified
in the long-snouted triacanthodids, but never as a T-
shaped bone or a simple rod, or as a bony column sutured
to the palato-pterygoid arch; ethmoid with a more or less
evenly convex upper surface, without a laterally ex-
panded dorsal or dorsolateral region and always nar-
rower dorsally than ventrally; no prootic shelf developed
under the orbit in front of and above the major articula-
tion of the posterior region of the parasphenoid with the
prootics; dorsal end of the hyomandibular articulated
with the prootic, pterotic, and sphenotic; interoper-
culum long and, at least posteriorly, relatively deep and
flattened, never as a stout rod throughout its length, and
extending posteriorly well behind the level of the epihyal
and interhyal to approach closely and connect by a short
ligament to the anterior edge of the suboperculum; dor-
sal fin spines usually six, rarely only five, supported by
five or four, and rarely only three, basal pterygiophores;
the first dorsal spine not lockable in an erected position
through the agency of the second spine, but with an in-
dependent locking mechanism between the base of the
first spine and its basal pterygiophore; first basal pteryg-
iophore of spiny dorsal fin with a high dorsomedial
flange articulating through an anteroposterior canal in
the basal region of the first dorsal spine; pelvic fin con-
sisting of at least a large erectile and lockable spine,
sometimes followed by one to four rays of greatly varying
degrees of development, at about the middle of the
length of the pelvis, except in several Eocene
triacanthodids in which the spine may have been reduc-
ed in size and possibly absent; pelvis either shaftlike or
basinlike posterior to the pelvic spines and never with a
dorsal lobe posteriorly, but always with a laterally ex-
panded portion anterior to the spines; pelvis not par-
ticularly rotatable in life around its anterior articulation
with the cleithra and no expansible dewlap of skin pres-
ent between the posterior end of the pelvis and the anus;
one or two sets of uroneurals usually present; vertebrae
nearly always 8 -I- 12 = 20 (9 -(- 11 = 20 in one Eocene
species); three pharyngobranchials with prominent large
protruding teeth; posteromedial edges of epiotics slightly
to decidedly inturned to form, in association with the
exoccipitals and neural spine of the first vertebra, a
pocket in which the shaftlike base of the first basal
pterygiophore of the spiny dorsal fin is held, with the
probable exception of the Oligocene triacanthid Crypto-
balistes: mesopterygoid in neither direct nor indirect
contact with the quadrate and ectopterygoid, except
sometimes very slightly so with the quadrate by the
agency of the symplectic making contact with both the
quadrate and mesopterygoid.

Family Triacanthodidae

Comparative dia^osis (contrast with that of the
Triacanthidae) (modified from Tyler 1968:58-62).— A
not especially strongly sutured skull and a variety of
body forms none of which are built for particularly ac-
tive, strong swimming; many areas of cartilage visible on
the external surface of the skull between the limited areas
of suturing between the bones in the otic and occipital
regions; teeth in the jaws various, conical or truncate,
in a single series or with a numerous outer series fol-
lowed by a few teeth in an inner series, or absent al-
together (in adults of one of the long-snouted genera,
Macrorhamphosodes) in the upper jaw, or of numerous
rounded dental units incorporated into the matrix of the
premaxillary and dentary in beaklike jaws in the Eocene
Eoplectus, but never as heavy incisor teeth (except per-
haps somewhat so in the Eocene Protobalistum); a nor-
mal ethmoid-frontal complex for support of jaws without
massive dentition (except in the Eocene Eoplectus, with
a modified ethmoid-vomer-palatine complex to support
the massive beaklike jaws); the prefrontal without a long
anterior extension sutured to the ethmoid and vomer,
and the vomer without posterolateral extensions toward
the prefrontal; premaxillary pedicels when retracted
reaching most of the way along the dorsal surface of the
ethmoid, almost to the tip of the frontals, except in both
of the long-snouted genera (Halimochirurgus and Mac-
rorhamphosodes), in which the premaxillary is far
removed from both the ethmoid and frontal, and in the
Eocene Eoplectus, which has only very short premaxil-
lary pedicels and a nonprotractile jaw; supraoccipital
either domelike with a convex posterior surface, or most-
ly flattened with a small dome or laterally compressed
crest; epiotics meeting one another in the midline on the
dorsal surface of the skull and separated from the fron-
tals by the sphenotics in one group (those with a flat-
tened supraoccipital) but separated from one another on
the dorsal surface of the skull in the other group by the
supraoccipital and meeting medially only on the pos-
terior surface of the skull, and articulating anterolateral -
ly with the frontals (for illustrations see Tyler 1968:fig.
4a-d); pterosphenoids not meeting in the midline of the
posterior wall of the orbit; neural processes of the first
vertebra not meeting in the midline above the neural
canal, forming in conjunction with the exoccipitals an
only partially enclosed bony region (without a bony bot-
tom) in which the shaftlike end of the first basal pte-
rygiophore of the spiny dorsal fin is immovably held;
parasphenoid in region of orbit relatively straight or with
a ventral arch and with a poorly developed ventral flange
below the orbit, the flange not deeper than the depth of
the shaftlike portion of the bone, except in the Eocene
Eoplectus, with a much deeper flange; hyomandibular
without a prominent groove and crest along its lateral
surface; pterotic without a ventral process overlying the
hyomandibular; supracleithrum usually placed obli-
quely to the horizontal axis of the skull, only the approx-
imate lower half of its length overlying the cleithrum;
mesopterygoid always present as a separate element.

sometimes sutured along its ventral edge with the
metapterygoid; olfactory cavity between the ethmoid
and prefrontal not well defined by bony outlines; basi-
hyal present; lower two branchiostegal rays more or less
like the others, not enlarged; operculum more or less
triangular; air bladder usually thin walled, somewhat
rounded, extending posteriorly to no more than the level
of about two-thirds the length of the abdominal cavity;
pelvis either a sturdy shaft more or less triangular or
broadly heart-shaped in cross section (the apex or round-
ed surface ventrally) just behind the level of the pelvic
spine or a flat basin with upturned edges; the two halves
of the pelvis only lightly sutured to one another; the side
of the pelvis at the level of the flange of the pelvic spine
either smooth and allowing for only a single position of
erection of the spine or with numerous small grooves
allowing for numerous, continuous positions of erection
of the spine, but never with a single relatively large ob-
lique crest allowing for only two positions of erection; the
ventrolateral surface of the pelvis at the base of the spine
with a complete foramen through which the two sides of
the base of the spine meet medially; the haemal arch and
spine of the penultimate vertebra usually autogenous
(occasionally extensively sutured to the centrum), the
other haemal arches and spines fused to their centra; the
epural, parhypural, and five hypurals usually separate
elements articulated by connective tissue to each other
and to the centrum, but occasionally a few of the middle
hypurals may be partially to fully fused to one another;
usually two pairs of uroneurals present; slender
epipleurals present from the first or second abdominal
vertebra to the seventh or eighth abdominal vertebra,
and, sometimes, on the first or second, rarely third,
caudal vertebra, except in the Eocene Eoplectus, in
which epipleurals, if present, probably did not occur on
at least the more posterior abdominal vertebrae; fifth
basal pterygiophore of spiny dorsal fin present, normally
developed; first basal pterygiophore of spiny dorsal fin
with anterior and posterior medial flanges well
developed, many times the width of the relatively slen-
der shaftlike portion of the bone, except in the Eocene
Eoplectus in which the flanges are only moderately
developed and in Halimochirurgus in which they are
poorly developed (in the Eocene Protobalistum and
Spinacanthus the more anteriorly placed spiny dorsal fin
probably was supported high on the rear of the skull by
basal pterygiophores with short shafts and poorly
developed anterior and posterior medial flanges); first
basal pterygiophore of spiny dorsal fin with a medial
flange dorsally completely enclosing a foramen through
which the two sides of the first dorsal spine meet medial-
ly; second to fifth basal pterygiophores of spiny dorsal fin
well developed, their shafts reaching ventrally to or
between the distal regions of the neural spines of the
fourth to eighth abdominal vertebrae; at least the first
two basal pterygiophores of spiny dorsal fin sutured
together distally; spiny dorsal fin base (including rudi-
ments) much longer than soft dorsal fin base, except in
the Eocene Eoplectus, with these two bases of equal
length; neural spines of all eight abdominal vertebrae

placed anterior to first basal pterygiophore of soft dorsal
fin (the Eocene Eoplectus with nine abdominal verte-
brae, all but the last with their neural spines anterior to
the first soft dorsal fin basal pterygiophore); first anal fin
basal pterygiophore either with or without a medial
flange anterior to its lateral flanges, the pterygiophore
either T- or + -shaped in cross section; well-developed
lateral flanges for muscle attachment present only on the
first anal fin basal pterygiophore, those present, if any,
on the other pterygiophores very poorly developed; no
lateral flange present horizontally along the last cen-
trum or on the hypurals; soft dorsal fin basal pterygio-
phores 11 to 16, anal fin basal pterygiophores 11 to 13,
these pterygiophores not sutured to one another distally
(probably fewer basal pterygiophores in both fins in the
Eocene Spinacanthus); distal pterygiophores of soft dor-
sal and anal fins always ossified in adults, usually as two
separate halves; sixth dorsal spine either a short
(probably longer in the Eocene Spinacanthus)
protruding element or a rudiment buried beneath the
skin or barely protruding to the surface; third to fifth
spines either normally developed or rudiments buried
beneath the skin or barely protruding to the surface; first
and second spines always prominent and visible exter-
nally; dorsal fin rays 12 to 18; anal fin rays 11 to 16 (fewer
rays in both fins in the Eocene Spinacanthus); pelvic fin
with a large (smaller in the Eocene Protobalistum and
Spinacanthus) spine followed in the Eocene Eoplectus by
four well-developed branched rays but in all Recent
species by no more than one or two small unbranched
rays, both of which, but especially the second, become in
some species buried rudiments in adults; adults of some
species with one or two protruding pelvic rays; dorsal and
pelvic spines with deep lengthwise grooves, obscured by
the overlying scale plates except at the naked distal end
(one-half to one-tenth or less of the length); uppermost
pectoral fin ray short but not reduced to a nubbin, the
two halves of the ray of about equal length and the basal
region of the medial half not immensely larger than that
of the lateral half; the slightly overlapping basal plates of
the scales of the body bearing a vertical row of upright
spinules arising from individual bases, large specimens of
some species acquiring supplemental spinules in front of
and behind the principal row (the Eocene Eoplectus with
a single upright spinule per scale plate and the body not
completely covered with scales, some of the plates being
isolated and nonoverlapping, and the Eocene Protobalis-

Figure 4.— Range of diversity in body form in the

Recent Triacanthodidae: Parahollardia Uneata (left)

and Halimochirurgus alcocki (right).

turn with huge more or less hexagonal plates bearing
numerous rounded tubercules, as in ostracioids);
peritoneum light tan to jet black, except in Hollardia
meadi in which it is pale or silvery; coloration basically
reddish, often with darker red, blue, yellow, or green
markings; lateral line inconspicuous; scaly skin not form-
ing a definite low sheath along the bases of the soft dor-
sal and anal fins; olfactory lamellae 9 to 20, relatively
plump; anterior nostril with a tube, lowest in front,
highest in back; posterior nostril more or less flush with
the surface, or with a slightly upraised rim anteriorly; gill
rakers laterally on first arch relatively long, as long as or
longer than the width of the fleshy arch; two to seven,
rarely only one, rakers laterally on the upper limb of the
first arch above the angle; caudal peduncle relatively
short, 10 to 19'^c SL (24'~c SL in the Eocene Eoplectus),
not distinctly tapered and without a narrow transversely
indented region above and below just in front of the
caudal fin base; least depth of the caudal peduncle 5 to
12'^c SL (21% SL in the Eocene Eoplectus); caudal
peduncle deeper than wide; caudal fin rounded to al-
most truncate.

Detailed description of Parahollardia Uneata.

Material examined. — Seven cleared and stained
specimens, 4.5.7-86.1 mm. The large amounts of cartilage
visible on the surface of the skull, particularly in the otic
and occipital regions, is the norm for the family, in con-
trast to the closely related and derivative Triacanthidae.
Even as large specimens, all species of Triacanthodidae
retain cartilaginous regions between some of the cranial
bones, although the amount slightly decreases with in-
creasing specimen size. The largest specimen of a Recent
triacanthodid presently recorded (Tyler 1968:64) is a 174
mm female Hollardia hollardi (ANSP 97654) which has
subsequently been cleared and stained. As can be seen in
Figure 22, areas of cartilage are still present on the sur-
face of the skull in the occipital and otic regions, al-
though the various bones are more extensively inter-
digitated to one another than in the 62.7 mm specimen
illustrated by Tyler (1968:fig. 198) and relatively less car-
tilage is present.

SKULL.

Occipital Region.

Basioccipital. — A short column, dorsolateral^ ex-
panded; cartilage filled along its anterior and anterodor-
sal edges; articulates through cartilage posterolaterally

Figure H.—Parahollardia lineata: with
dots representing the course of the inconspicuous
lateral line; lower left, scales from upper middle
region of body, including two lateral line canal
bearing scales; lower right, nasal region as seen
externally (above) and the olfactory lamellae as
seen with the top of the nasal sac removed.

with the exoccipitals, anterolaterally with the prootics,
and, with sHght interdigitation, anteroventrally with the
overlying posterior end of the parasphenoid. The rim of
the round concave posterior end of the basioccipital ar-
ticulates by fibrous tissue with the rim of the concave an-
terior face of the centrum of the first vertebra. A deeply
concave medial channel is present on the anterior half of
the ventral surface of the basioccipital, but the channel
is mostly hidden from view by the overlying para-
sphenoid. Posteriorly this channel is open to the exterior
at the base of the posterior bifurcation of the para-
sphenoid, while anteriorly it opens into the myodome
where the anterior end of the basioccipital forms the
posterodorsal and posterolateral walls of the myodome.

Exoccipital. — Cartilage filled at its dorsal, lateral,
and ventral edges; articulates through cartilage dorsally
with the epiotic, anteroventrally with the prootic, ventro-
medially with the basioccipital; articulates laterally
through cartilage and slight interdigitation with the
overlying pterotic. Medially the exoccipital forms the
lateral and ventral walls of the foramen magnum, the
foramen being closed dorsally by the cartilage between
the dorsomedial edges of the two exoccipitals. Postero-
medially the exoccipitals are firmly attached by fibrous
tissue and slight interdigitation to the anterior surface of
the elongate, laterally expanded, bifid neural spine of the
first vertebra. The exoccipital condyle is a short pos-
terior prolongation of the posterior edge of the bone just

above its articulation with the basioccipital, forming a
plate which overlies the lower anterolateral surface of the
neural arch of the first vertebra just above the region of
the centrum.

Supraoccipital. — Dome-shaped, its stout rounded
anterodorsal edge forming the apex of the cranium and
its posterior surface convex; cartilage filled along all of
its ventral edges; articulates posteroventrally through
cartilage with the epiotics and anteroventrally through
cartilage and slight interdigitation with the overlying
frontals.

Otic Region.

Pterotic. — Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates
posterodorsally through cartilage and extensive inter-
digitation with the epiotic, posteroventrally through car-
tilage and slight interdigitation with the exoccipital,
anterodorsally through cartilage with the sphenotic, and
anteroventrally through cartilage with the prootic. Along
the middle of the anterior half of its ventral surface the
pterotic articulates with the hyomandibular through
fibrous tissue, the articulation being somewhat flexible.
Posterolaterally the pterotic is broadly overlain by the
posttemporal to which it is firmly interdigitated.

Sphenotic. —Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates
through cartilage posterodorsally with the epiotic,
posteroventrally with the slightly overlying pterotic, ven-
tromedially with the prootic and dorsomedially with the
pterosphenoid. Dorsally the sphenotic is broadly over-
lain by and slightly interdigitated with the frontal. On its
ventral surface the medial edge of the sphenotic ar-
ticulates by fibrous tissue with the hyomandibular.

Epiotic. — Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates
through cartilage dorsally with the supraoccipital,
anterodorsally with the occasionally slightly overlying
frontal, anteriorly with the sphenotic, and posteroven-
trally with the exoccipital. Ventrolaterally the epiotic ar-
ticulates anteriorly through cartilage with the pterotic,
but more posteriorly the pterotic broadly overlies and in-
terdigitates with the epiotic. Along the lower portion of
its posteromedial edge the epiotic bends anteriorly so
that a shallow, vertical depression or groove is formed
where the two epiotics articulate through cartilage with
one another medially. The ventral end of the stout shaft
of the first basal pterygiophore of the spiny dorsal fin fits
into this depressed area and is held tightly to it by
fibrous tissue. This depression in the back of the skull is
continued ventrally by a similar intuming of the upper
third of the medial edges of the exoccipitals.

Prootic. — Cartilage filled along all of its edges of
articulation with the other cranial bones, except an-
teriorly; articulates through cartilage anterodorsally with

the pterosphenoid, anterolaterally with the sphenotic,
posterolaterally with the pterotic, posteriorly with the
exoccipital, and posteromedially with the basioccipital.
Anteromedially the prootic articulates for most of its
length through cartilage with the parasphenoid, but at
its extreme anteromedial end the articulation of the
prootic with the short dorsal wing of the parasphenoid is
by slight interdigitation. It is at this place of interdigita-
tion between the prootic and parasphenoid that the first
pharyngobranchial makes a firm ligamentous connec-
tion with the cranium. Along its ventral surface the an-
terior half of the lateral edge of the prootic helps support
the dorsal head of the hyomandibular. The lateral walls
of the myodome are formed by the ventromedial sur-
faces of the prootics, while the dorsal walls of the myo-
dome are formed by medially directed horizontal shelves,
which are attached to one another medially by fibrous
tissue, from the ventromedial surfaces of the two
prootics. The prootics also form most of the anterior wall
of the myodome.

edges of the prootics and thus form the lower part of the
anterior edge of the myodome. Behind this restricted
area of interdigitation with the anteroventral edge of the
prootic, the lateral edges of the parasphenoid articulate
through cartilage with the medial edges of the prootics
and thus form the floor of the myodome. Anteriorly the
parasphenoid broadly overlies and interdigitates with
the dorsal surface of the posterior shaftlike portion of the
vomer, and, to a much lesser extent, with the ventral sur-
face of the latter. The anterior end of the parasphenoid
thus essentially possesses a deep concavity into which
the shaft of the vomer fits. Anterodorsally the para-
sphenoid is tightly held to the ethmoid cartilage, and
hence to the ethmoid and prefrontals.

Pterosphenoid. —Cartilage filled along all of its
edges, except medially; articulates through cartilage
ventrally with the prootic and ventrolaterally with the
sphenotic. Dorsolaterally the pterosphenoid is held by
fibrous tissue to the overlying frontal.

Orbital Region.

Ethmoid Region.

Frontal. — Wide posteriorly but tapering to a point
anteriorly; articulates posterodorsally and posteroven-
trally by overlying and slightly interdigitating with,
respectively, the supraoccipital and sphenotic. Oc-
casionally the most posterior portion of the frontal slight-
ly overlies the anterodorsal portion of the epiotic.
Posteromedially on its ventral surface the frontal some-
what overlies and articulates by fibrous tissue with the
pterosphenoid. Anteriorly the frontal overlies the eth-
moid cartilage and articulates by fibrous tissue with the
ethmoid and prefrontal. Ventrally the medial edges of
the frontals are widely separated by a large cartilaginous
mass which is continuous anteriorly with the ethmoid
cartilage.

Prefrontal. — In the form of a basally expanded
column; cartilage filled along its dorsomedial and ventro-
medial edges; articulates through cartilage along all of its
medial surface with the ethmoid and ethmoid cartilage;
articulates by fibrous tissue dorsally with the frontal and
anteroventrally with the posterior end of the palatine.
The prefrontal articulates ventrally with the dorsal sur-
faces of the vomer and parasphenoid, by way of the fact
that the ethmoid cartilage is firmly attached to the dor-
sal surfaces of the posterior region of the vomer and the
anterior region of the parasphenoid.

Parasphenoid. — An elongate shaft with a slight
keel along its ventral surface in the region of the orbit. A
deep, anteriorly directed cleft is present in the wide pos-
terior portion of the parasphenoid, giving rise to the fork-
ed region which broadly overlies and interdigitates with
the basioccipital and at the same time covers over the
medial groove on the ventral surface of the basioccipital
leading into the rear of the myodome. About two-thirds
the way back, the parasphenoid possesses paired dorso-
lateral wings which interdigitate with the anteroventral

Ethmoid. — The ethmoid remains largely
cartilaginous, the ossification being restricted to the sur-
face regions. It articulates through cartilage dorsally with
the slightly overlying frontals, posterolaterally with the
prefrontals, and ventrally with the vomer and para-
sphenoid.

Vomer. — Large and rounded anteriorly, but
tapering to a stout shaft posteriorly which interdigitates
with the concavity at the anterior end of the para-
sphenoid. Dorsally the vomer is attached by fibrous tis-
sue to the base of the ethmoid cartilage and thus second-
arily with the bases of the ethmoid and prefrontals. At
the lateral edges of its rounded anterior portion the
vomer articulates by tough fibrous tissue with the medial
surfaces of the palatines.

Mandibular Region.

Hyomandibular. — Somewhat expanded dorsally,
tapering to a stout shaft anteroventrally; cartilage filled

at its dorsal and anteroventral edges; articulates by fi-
brous tissue at its dorsal head with the pterotic pos-
teriorly, the sphenotic anterolaterally, and the prootic
anteromedially. Along the lower three-fourths of its pos-
terior edge it articulates by fibrous tissue with the pre-
operculum. Just above the dorsal end of the preoper-
culum, the hyomandibular articulates by fibrous tissue
at a small groove on its posterior edge with the knob at
the dorsal end of the operculum. Anteriorly the hyoman-
dibular ends at the cartilaginous plate that lies between
it and the symplectic and metapterygoid.

Quadrate. — Wide posteriorly, tapering to a knob
anteriorly for articulation through fibrous tissue with the
articular in the lower jaw; cartilage filled along its pos-
terior edge; with a long posteriorly directed process from

its posteroventral region; articulates by fibrous tissue
dorsally with the ectopterygoid, ventrally with the
preoperculum, and anteriorly with the articular.
Posteriorly the quadrate articulates through cartilage
with the metapterygoid and symplectic.

Metapterygoid. — Broad anteriorly, narrowing pos-
teriorly; cartilage filled along its anterior and ventral
edges; articulates through cartilage anteriorly with the
quadrate, ventrally with the symplectic, and posteriorly
with the cartilaginous area in front of the anterior end of
the hyomandibular. Dorsally the metapterygoid ar-
ticulates by fibrous tissue with the mesopterygoid.

Symplectic. —Long and rod-shaped; cartilage filled
at its anterior and posterior ends; articulates through
cartilage anteriorly with the quadrate, dorsally with the
metapterygoid, posteriorly with the cartilaginous area in
front of the anterior end of the hyomandibular. Ven-
trally the symplectic articulates by fibrous tissue with
the preoperculum.

Palato-Pterygoid Region.

Palatine. — Cartilage filled at its anterior and pos-
terior ends; articulates by fibrous tissue anteriorly with
the dorsolateral surface of the maxillary, medially with
the lateral surface of the expanded portion of the vomer,
posteriorly with the mesopterygoid. A posterodorsal pro-
jection from its dorsal edge articulates by fibrous tissue
with the base of the ventrolateral surface of the pre-
frontal, and also connects to a band of fibrous tissue that
runs over and above the premaxillary pedicels to join
with the posterodorsal wing of the palatine on the other
side. The palatine is firmly held in place by these fibrous
tissue articulations even though it is not sutured to any
of the surrounding bones,

Ectopterygoid. — Somewhat triangular in shape; ar-
ticulates closely by fibrous tissue along its ventral edge
with the quadrate, while along its posterodorsal edge it
articulates through a sheet of fibrous tissue with the
palatine and mesopterygoid.

Mesopterygoid. — Relatively well developed, only
slightly smaller than the metapterygoid; articulates by
fibrous tissue anterodorsally with the palatine, antero-
ventrally with the ectopterygoid, and ventrally with the
metapterygoid.

Opercular Region.

Operculum. — Thin and broad ventrally, nar-
rowing to a stout shaft dorsally; articulates by fibrous tis-
sue dorsally with a groove on the upper posterior edge of
the hyomandibular. Ventrally the operculum slightly
overlies and is held by fibrous tissue to the suboper-
culum.

Suboperculu

-Rounded anteriorly, tapering to

a point posteriorly; with a small dorsal projection located
in about the middle of its dorsal edge; articulates by fi-
brous tissue dorsally with the operculum, and anteriorly
with the interoperculum.

Interoperculum. — Very elongate; wide and rounded
posteriorly but tapering to a point anteriorly; articulates
by fibrous tissue posteriorly with the suboperculum,
while anteriorly it makes a ligamentous attachment
to the angular in the lower jaw. In the posterior half
of its length the dorsal edge of the interoperculum con-
nects by a band of fibrous tissue to the epihyal in the
region where the epihyal articulates with the interhyal.

Preoperculum. — Slightly expanded posteroven-
trally; the dorsal edge along the anterior half of the bone
slightly expanded laterally for articulation with the
quadrate; articulates by fibrous tissue along the pos-
terior region of its dorsal edge with the hyomandibular,
along the middle region of its dorsal edge with the car-
tilaginous plate between the symplectic, metaptery-
goid, hyomandibular, and quadrate, along the anterior
region of its dorsal edge with the quadrate.

Upper Jaw.

Premaxillary. — L-shaped, with the long posterior
arm (pedicel) movably articulated by fibrous tissue
along the dorsal surface of the vomer and ethmoid, allow-
ing for protraction of the upper jaw. When fully retracted
the pedicel reaches most of the way along the dorsal sur-
face of the ethmoid, almost to the tip of the frontals. The
shorter ventral arm of the premaxillary forms the an-
terior edge of the upper jaw, except for a short distance
ventrally where the maxillary forms the border. The pre-
maxillary articulates by fibrous tissue along the middle
of the lateral surface of its posterior arm with the medial
surface of the upper exptmded portion of the maxillary,
while along the medial surface of its posterior arm the
premaxillary articulates with its opposite member. The
ventral arm of the premaxillary articulates by fibrous tis-
sue along its posterior edge with the anterior edge of the
maxillary. Six to seven bluntly conical teeth, decreasing
in size laterally, are borne in deep sockets along the an-
terior edge of the premaxillary. Behind and internal to
this outer row of teeth there is usually a single conical
tooth, about half the size of the largest tooth in the outer
series, but several such inner teeth may be present on
each premaxillary. The teeth are replaced by new ones
developing in new sockets just above and external to the
sockets of the outer row of teeth. After a new tooth erupts
through the surface of the premaxillary, it gradually
migrates slightly forward and downward into the area of
the socket of the older tooth which it is replacing.

Maxillary. — Rounded dorsally, constricted in the
middle, and only slightly expanded ventrally; ar-
ticulates by fibrous tissue dorsally along the lateral sur-
face of its rounded portion with the anterior end of the
palatine, while along the medial surface of its rounded

portion it attaches to the lateral and ventral surfaces of
the premaxillary. Ventrally the medial surface of the
maxillary articulates by tough fibrous tissue with the
lateral surface of the posterodorsal part of the dentary,
while along its anteroventral edge the maxillary ar-
ticulates similarly with the premaxillary.

Lower Jaw.

Dentary. — The posterior end concave to accom-
modate the anterior end of the articular, which it broad-
ly overlies laterally but only slightly overlies medially;
articulates by fibrous tissue anteromedially with its op-
posite member, posteroventrally with the angular, and
posteromedially with the articular. Along the postero-
dorsal region of its lateral surface the dentary ar-
ticulates by tough fibrous tissue with the medial surface
of the ventral portion of the maxillary. Seven to eight
bluntly conical teeth, decreasing in size laterally, are
borne in deep sockets along the anterior edge of the den-
tary. Behind and internal to this outer row of teeth there
is usually a single conical tooth, about half the size of the
largest tooth in the outer series, but several such inner
teeth may be present on each dentary. The teeth of the
outer series are replaced by new ones developing in new
sockets just below and external to the sockets of the outer
row of teeth, in the same manner as described for the
premaxillary.

Articular. — More or less triangular in shape;
cartilage filled at its anterior edge, where it is con-
tinuous with the remains of Meckel's cartilage; ar-
ticulates by fibrous tissue dorsally, ventrally, and
laterally with the broadly overlying dentary, postero-
ventrally with the angular, and posteriorly at the groove
on its posterior edge with the knob at the anterior end of
the quadrate. The sesamoid articular is a small ossifica-
tion held by fibrous tissue to the region of juncture of
Meckel's cartilage and the anteromedial surface of the
articular.

Angular. — Small; articulates by tough fibrous tis-
sue with the dentary and articular. Posteriorly the
angular connects by ligament to the anterior end of the
interoperculum.

BRANCHIAL APPARATUS.

and with the ceratohyal, while they articulate by fibrous
tissue anteromedially with their opposite members, and
in the case of the dorsal hypohyal with the region of ar-
ticulation between the basihyal and first basibranchial.
The ventral edge of the ventral hypohyal articulates by
fibrous tissue with the urohyal.

Ceratohyal. — Large, somewhat expanded pos-
teriorly; cartilage filled at its anterior and posterior
edges; articulates through cartilage anteriorly with both
of the hypohyals and posteriorly with the epihyal. The
six branchiostegal rays articulate by fibrous tissue with
the ceratohyal; the first two rays to about the middle of
the ventral edge and the last four rays to the lateral sur-
face of its posteroventral region.

EpihyaL — Large; cartilage filled at its anterior and
ventral edges; articulates through cartilage anteriorly
with the ceratohyal and posterodorsally by fibrous tis-
sue with the interhyal.

Interhyal. — Short, columnar; cartilage filled at its
dorsal and ventral edges; articulates by fibrous tissue
ventrally with the epihyal and dorsally with the car-
tilaginous plate between the symplectic and hyoman-
dibular.

Branchiostegal rays. — Six in number, increasing in
length posteriorly in the series; the first ray somewhat
flattened or laterally compressed, the succeeding rays
less compressed so that the sixth ray is rodlike; ar-
ticulate by fibrous tissue with the ceratohyal as de-
scribed above.

Urohyal. — Thick at its dorsal and anterior edges,
otherwise very thin; articulates by fibrous tissue with the
ventromedial surfaces of the ventral hypohyals.

Branchial Arches.— All the elements are cartilage
filled at their edges of articulation with the other ele-
ments of the series, and the articulations are usually
through cartilage and fibrous tissue. The branchial
arches are composed of three basibranchials, three pairs
of hypobranchials, five pairs of ceratobranchials, four
pairs of epibranchials, and four pairs of pharyngobran-
chials. Four gills are present, with a slit between the
fourth arch and the lower pharyngeal.

Hyoid Arch, Branchiostegal Rays, and Urohyal.

Basihyal. — Elongate; slightly expanded laterally
at posterior end; cartilage filled at posterior end; ar-
ticulates by fibrous tissue posteriorly with the first
basibranchial and laterally with the dorsal hypohyals.

Hypohyals. —Both hypohyal elements well de-
veloped; dorsal hypohyal cartilage filled at its ventral
and posterior edges, the ventral hypohyal cartilage filled
at its dorsal and posterior edges. The dorsal and ventral
hypohyals articulate through cartilage with one another

First Arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. First basibranchial
short, laterally compressed in the middle of its length;
displaced forward so that it articulates posteriorly main-
ly with the second basibranchial and only secondarily
posterolaterally with the first hypobranchials, while an-
teriorly it articulates with the basihyal. First hypo-
branchial the largest of the hypobranchial elements,
which decrease in size posteriorly in the series; ar-
ticulates mainly with the anterior half of the lateral sur-
face of the second basibranchial, and to a lesser extent
along its extreme anteroventral edge with the first basi-

branchial. First ceratobranchial the shortest of the
ceratobranchial elements, which, except for the fifth, in-
crease slightly in size posteriorly in the series; possesses a
ventrally directed flange along most of its ventral sur-
face, this flange on the succeeding ceratobranchials
becoming progressively less developed until it is non-
existent on the fifth. First epibranchial with the dorsal
one-third of its length bifurcate into two projections;
the anterior projection articulating with the base of the
first pharyngobranchial, and the posterior projection ar-
ticulating with the second pharyngobranchial. First
pharyngobranchial (suspensory pharyngeal) a narrow
rod articulating ventrally with the anterior projection of
the first epibranchial and dorsally by fibrous tissue with
the area of articulation of the dorsolateral wing of the
parasphenoid with the anteroventral edge of the prootic.

Second arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Second basi-
branchial the largest of the three basibranchial elements;
articulates anteriorly with the first basibranchial,
anterolaterally with the first hypobranchials, posteriorly
with the third basibranchial, and posterolaterally with
the second hypobranchials. Second hypobranchial very
wide ventrally; articulates ventrally with the posterior
end of the second basibranchial and anterior end of the
third basibranchial, articulates dorsally with the second
ceratobranchial, which in turn articulates dorsally with
the somewhat laterally expanded second epibranchial.
The second pharyngobranchial is the first of the tooth
bearing pharyngobranchials, having seven or eight teeth,
set in sockets, in a single row on its ventral surface. The
teeth are of the same type as those of the jaws, although
slightly smaller and with sharper points. They are
replaced by new teeth developing in new sockets just an-
terior to the sockets of the old teeth, in the same manner
as in the replacement of the teeth in the jaws.

Third arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Third
basibranchial articulates anteriorly with the second
basibranchial, anterolaterally with the second hypo-
branchials, posterolaterally with the third hypo-
branchials, and posteriorly with the fourth cerato-
branchials. Third hypobranchial with an anteroventral
process which articulates by fibrous tissue with the
posterodorsal edge of the urohyal; at its posterodorsal
end it articulates medially with the posterior end of the
third basibranchial and laterally with the ventral end of
the third ceratobranchial. Third ceratobranchial ar-
ticulated dorsally with the third epibranchial, which in
turn articulates with the third pharyngobranchial, the
largest of the three tooth-bearing pharyngobranchial
elements. It possesses six to seven teeth in a row along its
anterior edge and about three more teeth irregularly
placed behind the anterior row. The teeth are similar to
those described for the second pharyngobranchial. The
third pharyngobranchial closely articulates by fibrous
tissue with the other two toothed pharyngobranchials
(second and fourth).

Fourth arch. — Cerato-, epi-, and pharyngobranchial
elements. With the disappearance of the fourth basi-
branchial and fourth hypobranchial, the fourth cerato-
branchial articulates ventrally with the posterior end of
the third basibranchial. Fourth epibranchial an elongate
rod which articulates ventrally with the fourth cerato-
branchial and dorsally with the small fourth pharyngo-
branchial. The latter is the smallest of the toothed
pharyngobranchial elements, and closely articulates by
fibrous tissue with the third pharyngobranchial. It bears
five or six teeth in an anterior row and three or four more
teeth, irregularly placed, behind this anterior row.

Fifth arch. — Ceratobranchial (lower pharyngeal)
element only. Fifth ceratobranchial wide anteroven-
trally but tapering to a short stout column dorsally.
Teeth more or less placed in three rows, with those of the
posterior row the largest; 7 to 10 teeth in the anterior
row, 7 to 9 teeth in the middle row, and 7 to 11 teeth in
the posterior row. The teeth are of the same type as those
of the pharyngobranchials. Ventrally the fifth cerato-
branchial articulates with the base of the fourth cerato-
branchial.

PAIRED FIN GIRDLES.

Pectoral Fin.

Posttemporal. — A straight shaft of bone without
evidence of a forked condition, broadly overlying and
strongly interdigitated to the pterotic in the lower two-
thirds of its length and to the epiotic above. The rounded
ventral head of the posttemporal articulates by fibrous
tissue with the concave dorsal end of the supra-
clei thrum.

Supracleithrum. — Located slightly obliquely
anterodorsally to posteroventrally in relation to the axis
of the body; articulates by fibrous tissue dorsally with
the ventral head of the posttemporal, and ventrally with
the cleithrum, which it broadly overlies.

Cleithrum. —Laterally expanded along the ventral
two-thirds of its length; articulates by fibrous tissue dor-
solaterally with the overlying supracleithrum, while dor-
somedially it overlies the dorsal postcleithrum. Along the
middle of its posterior surface the cleithrum articulates
by fibrous tissue with the scapula and the rounded
anterodorsal portion of the coracoid. Ventromedially the
cleithrum articulates by tough fibrous tissue with its op-
posite member, while just above this region the anterior
end of the pelvis is firmly held between the cleithra.

Postcleithra. — The postcleithra form a long
posteroventrally directed strut from the dorsomedial
region of the cleithrum along the abdominal wall
musculature to the region just above the middle of the
posterior half of the pelvis. The dorsal postcleithrum is
expanded into a thin plate dorsally and articulates by

fibrous tissue anterolaterally with the somewhat overly-
ing cleithrum and posteroventrally with the ventral
postcleithrum. The ventral postcleithrum is a slightly
compressed shaft which articulates by fibrous tissue
anterodorsally with the dorsal postcleithrum in the
region behind the pectoral fin rays.

Coracoid. — Rounded dorsally, tapering ventrally to
a narrow shaft; produced posterodorsally into a prong
below the lowermost actinosts; cartilage filled along the
dorsal and anterior edges of its rounded dorsal region and
at the end of its ventral shaft; articulates by fibrous
tissue anterodorsally and anteroventrally with the
posterior surface of the cleithrum, dorsally through car-
tilage with the scapula and through fibrous tissue with
the bases of the last two actinosts.

Scapula. — Completely encloses the scapular
foramen; cartilage filled at its anterior and ventral edges;
articulates anteriorly by fibrous tissue with the
cleithrum and ventrally through cartilage with the cor-
acoid. Posteriorly the scapula articulates by fibrous tis-
sue with the following elements, in order from dorsal to
ventral; the first pectoral fin ray, the small first actinost,
the second actinost, and the dorsal part of the base of the
third actinost. The first fin ray is supported on a stubby
projection, and the first actinost on a flattened surface
just posteroventral to the articulation of the first fin ray.

the lateral edge of the pelvis, as well as ventrally with the
pelvis by an anteroventral flange from the base of the
spine, and dorsally with the pelvis by a smaller dor-
somedial flange (the anterior direction of the ventral
flange and the medial direction of the dorsal flange are
for the unerected spine). The spine can be envisioned to
lock in an erect position by a combination of three move-
ments. The whole spine is rotated approximately 70°
forward and outward around its base of articulation with
the pelvis. By this movement the ventral flange, that was
originally directed anteriorly, is now directed medially,
and the dorsal flange, that was originally directed
medially, is now pointed posteriorly and is no longer in
contact with the dorsal surface of the pelvis, but rather is
lying just lateral to the edge of the pelvis. The spine is
then slightly rotated so that the dorsal flange very slight-
ly overlies the ventrolateral edge of the pelvis. When the
spine is then rotated slightly backward and inward, the
dorsal flange hits flatly against the ventrolateral surface
of the pelvis and any further backward and inward move-
ment of the spine is stopped. The spine is unlocked by
the simple rotation of the spine slightly forward and out-
ward, and then slightly backward and upward, until the
dorsal flange is out of contact with the ventrolateral sur-
face of the pelvis and is lying just lateral to it. The spine
can then be drawn back to its unerected position, with
the dorsal flange sliding over the dorsal surface of the
pelvis to its original medially directed position.

Actinosts. — Four elements; all cartilage filled at
both ends; the first two actinosts articulating with the
scapula, the third actinost articulating with the dorsal
edge of the region of articulation between the scapula
and coracoid, the fourth actinost articulating with the
coracoid; distally the actinosts support all of the pectoral
fin rays, except for the first. The actinosts increase in size
from the first to the fourth.

Fin rays. — Thirteen fin rays in most specimens;
the first ray normal, of two equal halves, but only one-
fifth to one-fourth the length of the second ray and ar-
ticulating directly with the scapula rather than with the
actinosts, as do the other fin rays; the first two rays and
the last or lowermost ray unbranched, the intervening
rays branched. First ray without cross-striations; all
other rays with cross-striations.

Pelvic fin.

Pelvis.— Formed of two distinct lateral halves
tightly bound together medially by fibrous tissue; tapers
to a sharp point posteriorly, but terminates bluntly
anteriorly where it is tightly held by fibrous tissue
between the medial edges of the cleithra; expand-
ed ventrolaterally in the area anterior to the articulation
of the pelvic spines so that a large basin for muscle at-
tachment is formed on the ventral surface of the pelvis.

Pelvic spine. — Large and strong; articulates
broadly by fibrous tissue at its concave medial edge with

Fin rays. — Two rays present, only the first of which
can be seen externally. The first ray short, about one-
third the length of the spine, unbranched and without
cross-striations, the two halves of the ray distinct from
one another throughout their lengths and held together
by fibrous tissue. Basally the two halves of the first ray
are widely divergent, the base of the medial half of the
ray articulating broadly across the ventral surface of the
pelvis, while the base of the lateral half of the ray articu-
lates along the lateral surface of the pelvis. The second
fin ray is small and easily overlooked, for it lies flat
against the ventral surface of the pelvis and is completely
covered by the thick spinulose skin of the body. Like the
first ray, the second ray is composed of two distinct
halves that are held together by fibrous tissue. The
medial half of the second ray is wider than the lateral
half and both halves are unbranched and without cross-
striations. No pterygial elements are present. See Tyler
(1962b) for additional details on the pelvis and pelvic fin,
and Winterbottom (1970) for their muscular evolution.

VERTEBRAL COLUMN. —All vertebrae with bi-
concave centra, except the last, which ends posterodor-
sally in the urostyle.

Abdominal Vertebrae.

First vertebra. —Neural spine enlarged, laterally
expanded dorsally, bifid to the centrum and hence
without a bony roof over the neural canal; articulates
over all of the anterior surface of its neural spine by

fibrous tissue with the epiotics and by slight interdigita-
tion with the exoccipitals. Over the anterior half of the
neural arch area just above its centrum, the first vertebra
articulates by fibrous tissue with the overlying posterior-
ly directed exoccipital condyles. The rim of the concave
anterior end of the centrum articulates with the rim of
the concave posterior end of the basioccipital. Posteriorly
the first vertebra articulates with the second vertebra by
apposition of the rims of their centra and by the short,
but wide, neural postzygapophysis of the first vertebra
slightly overlying the neural prezygapophyseal area of
the second vertebra. No haemal zygapophyses are pres-
ent.

Other abdominal vertebrae. — In 17 specimens the
abdominal vertebrae numbered eight. Except for the
first vertebra, all of the abdominal vertebrae, as well as
the caudal vertebrae, have a bony roof over the neural
canal and a single, undivided, neural spine. The basal
regions of the neural spines and dorsal regions of the
neural arches become progressively more antero-
posteriorly expanded from the second to the eighth ab-
dominal vertebra. This expansion involves both the
neural pre- and postzygapophyses, which become
broadened into large articular surfaces, with the pre-
zygapophyses of the seventh and eighth (sometimes sixth
to eighth) abdominal vertebrae slightly overlying the
postzygapophyseal area of the vertebrae anterior to
them. No haemal pre- or postzygapophyses are present.
Each neural arch has a neural foramen along the middle
of its lateral surface. The first three abdominal vertebrae
have no transverse or haemal processes, but the fourth
and fifth vertebrae have transverse processes from the
anterolateral edges of their centra, the process of the fifth
being somewhat longer and more ventrally curved than
that of the fourth. The sixth vertebra has a slightly
longer transverse process, which differs from that of the
fourth and fifth vertebrae in that the two processes of the
sixth vertebra possess medial projections that meet and
are continuous in the midline beneath the centrum,
enclosing the haemal canal and thus representing a
haemal arch without a haemal spine. The seventh
vertebra has the haemal arch more strongly developed
and possesses ventrally directed processes from each side
of the haemal arch representing a bifid haemal spine.
The haemal apparatus of the eighth vertebra is like that
of the seventh, except larger. Six epipleurals or inter-
muscular bones are present in the myocommata between
the epaxial and hypaxial musculature, and borne
on the second to seventh abdominal vertebrae. The first
and second epipleurals articulate with the neural arches
of, respectively, the second and third abdominal
vertebrae. The third epipleural articulates with the dor-
sal surface of the short transverse process of the fourth
abdominal vertebra; the fourth epipleural with the dor-
sal surface of the transverse process of the fifth ab-
dominal vertebra. The fifth and sixth epipleurals articu-
late with the ventrolateral surfaces of the haemal arches
of, respectively, the sixth and seventh abdominal
vertebrae.

Caudal Vertebrae. — In 17 specimens the caudal
vertebrae numbered 12. As is true with the last two or
three abdominal vertebrae, the first four to six caudal
vertebrae have the neural spine area around and above
the neural zygapophyses expanded into large articular
surfaces so that the neural prezygapophyses of the first
four or five caudal vertebrae overlie the neural postzyga-
pophyseal area of the vertebrae anterior to them. With
the exception of the last two vertebrae, each neural arch
has a completely enclosed neural foramen. All of the
abdominal vertebrae have a bony roof over the neural
arch and a single, undivided, neural spine. The haemal
arches and spines are well-developed and normal, except
for those of the first caudal vertebra, which has the
haemal spine single and undivided for the upper one-
third of its length, but bifid for the ventral two-thirds
of its length. The bifid portion of this spine overlies the
upper anterior surface of the first basal pterygiophore of
the anal fin. The haemal arch of the penultimate verte-
bra is not fused to its centrum, but, rather, is tightly
held to it by fibrous tissue. The haemal arch of the last
vertebra is described below.

Caudal Skeleton. — The caudal skeleton consists of
the last vertebral centrum, one epural, two pairs of
uroneurals, five hypurals, and the parhypural, with the
whole complex receiving support from the prolonged
neural and haemal spines of the penultimate vertebra.
The last vertebral centrum is prolonged posterodorsally
into a urostyle and supports along its posterior edge the
five hypurals and the parhypural. From the dorso-
lateral surface on each side, the centrum bears antero-
dorsal projections which do not meet in the midline dor-
sally. These projections represent the lateral walls of the
neural arch of the last vertebra and its neural pre-
zygapophyses, since they form the lateral borders of the
neural canal and at their dorsal ends articulate by
fibrous tissue with the neural postzygapophyses of the
penultimate vertebra. The epural is a long rod articulat-
ing by fibrous tissue along its posterior edge with the
uroneurals and at its anteroventral edge with the area
between the dorsal ends of the two anterodorsal projec-
tions of the last vertebral centrum. The epural thus forms
the dorsal roof of the neural canal of the last vertebra and
it is, evidently, the very little modified neural spine of
the last vertebra. The uroneurals in this and the other
species of triacanthodids are highly variable in shape,
size, number and condition of pairing, as described in
detail by Tyler (1970b; see fig. 1 for illustrations of the
uroneurals in six specimens of P. lineata). There
is usually clear evidence of two pairs of uroneurals,
with the first pair usually smaller and more variable
than the second, more posterior, pair. In the caudal
skeleton of P. lineata illustrated here, the first pair
of uroneurals is represented by a single small rounded
plate above the anterior end of the second uroneural,
it being obvious that either one half of the first uroneural
has failed to develop or that the two halves have fused
into one piece. One other study specimen also has a
single first uroneural ossification, but three others have a

pair of pieces and another has three small pieces, as dis-
cussed by Tyler (1970b:5). The second pair of uroneurals
is represented by two long separate pieces in three of the
specimens, while in the other three specimens the two
halves are fully fused anteriorly.

The hypurals decrease in length from the first (lower-
most) to the fifth. Fifth hypural directed posterodorsally
from just behind and slightly below the uppermost point
of the urostyle; more or less rod-shaped; supports, along
with the uppermost edge of the fourth hypural, the up-
permost caudal fin ray. Fourth hypural the deepest of the
hypural elements; articulates anteriorly with the pos-
terior edge of the urostyle and posteriorly supports the
uppermost five caudal fin rays, although it shares the
support of the uppermost and lowermost of these five
rays with, respectively, the fifth hypural and third
hypural. Third hypural broad posteriorly but becoming
narrow anteriorly where it articulates with the posterior
edge of the urostyle; posteriorly it supports the upper of
the middle two caudal fin rays and partially supports the
lowermost of the upper five caudal fin rays. Second
hypural similar to third hypural, but slightly larger; ar-
ticulates anteriorly with the posterior edge of the
urostyle, while posteriorly it supports the lower of the
middle two caudal fin rays. First hypural articulated
anteriorly with the posterior edge of the urostyle, while
posteriorly it supports three fin rays. The parhypural the
largest of the hypural series and enclosing the end of the
haemal canal with its deeply concave anterior end; the
lateral edges of its concave anterior end articulate by
fibrous tissue with the posterolateral surfaces of the cen-
trum. A slight indentation on the anterodorsal edge of
the parhypural marks the end of the haemal canal. The
parhypural is thus the not greatly modified haemal arch
and spine of the last vertebra.

Caudal fin rays. — Twelve in number; the upper-
most ray and the lowermost ray unbranched, the others
becoming increasingly branched toward the two middle
fin rays, which are branched in triple dichotomies. The
fin rays articulate at their bifid bases by fibrous tissue
with the hypurals, as described above.

DORSAL AND ANAL FINS.

Dorsal Fin.

Spines and pterygiophores. —Six spines borne on
five basal pterygiophores, without the intervention of
distal pterygial elements. The first spine long, the others
decreasing in length posteriorly in the series. First spine
with its generally concave base somewhat expanded ven-
trolaterally. The ventrolateral flanges of the first spine
are in close contact with one another medially, although
they are not fused together. Just above the basal area of
close apposition of the flanges, a hole is present antero-
posteriorly through the middle of the spine. The general-
ly concave basal surface of the spine fits over the general-
ly convex upper surface of the first pterygiophore, while

at the same time a high medial flange from the upper
surface of the pterygiophore fits into the hole through the
base of the first spine. The medial flange of the pterygio-
phore possesses a large hole through which the two
medial surfaces of the base of the first spine come into
close fibrous tissue contact. The first spine is thus im-
possible to disarticulate from its pterygiophore without
cutting the fibrous tissue or breaking either the base of
the spine or the medial flange of its pterygiophore. The
first spine can be held in an erect position by a very sim-
ple mechanism. The ventral articular surface of the first
spine and the dorsal articular surface of the first
pterygiophore, on either side of its medial flange, are
rough and irregular. When the spine is erected at a right
angle to the dorsal surface of the pterygiophore and then
pulled slightly downward, a firm contact is established
between the two roughened articular surfaces and the
spine is effectively locked in position. The spine is un-
locked by relaxation of the strong, vertically directed
pressure of the spine against the pterygiophore, so that
the articular surface of the spine can slide forward over
the articular surface of the pterygiophore without undue
frictional resistance. The second spine articulates at its
concave ventral end with the generally convex surface of
the first pterygiophore directly behind the articular area
of the first spine, but the second spine has no function in
the locking mechanism of the first spine. There is a small
medial flange of the pterygiophore that fits loosely into
the slight concavity in the middle of the base of the sec-
ond spine. The third to the sixth spines articulate to
their individual basal pterygial elements and have, from
the third to sixth, progressively less concave articular
surfaces fitting against the progressively less convex ar-
ticular surfaces of their pterygiophores, so that the ar-
ticulation of the sixth spine to its pterygiophore is at the
relatively flat surfaces of both elements.

The first basal pterygiophore has a strong antero-
ventrally directed columnar process which fits into a
groove formed by the intumed posteromedial surfaces of
the ventral half of the epiotics and the dorsal one-third of
the exoccipitals, as well as by the medial surfaces of the
bifid neural spine of the first vertebra, which overlies
these regions of the epiotics and exoccipitals. The colum-
nar portion of the first pterygiophore is held in this
groove by tough fibrous tissue. The second to fifth
pterygiophores bear, respectively, the third to sixth dor-
sal fin spines. The ventral regions of the second to fifth
pterygiophores articulate, respectively, between the
neural spines of the third and fourth vertebrae, the fifth
and sixth vertebrae, the sixth and seventh vertebrae, and
the seventh and eighth vertebrae. The pterygial elements
are cartilage filled at their extreme ventral edges.

Fin rays and pterygiophores. — Sixteen fin rays
usually present; the first two rays and the last ray un-
branched, the others branched in single or double
dichotomies. Each fin ray has a small pair of distal
pterygiophores as two distinct halves between the bifur-
cate base of the ray, except for the last one or two rays, in
which these elements, if present, are unossified. Basally

the fin rays are supported by about 14 basal pterygia-
phores, which decrease in size posteriorly in the series.
The more anterior of the pterygiophores have weakly
developed lateral flanges along their lengths, but these
lateral projections for muscle attachment become reduc-
ed posteriorly in the series and are entirely absent from
the more posterior pterygiophores. The basal pterygio-
phores articulate by fibrous tissue ventrally with the
neural spines of the vertebrae. The first and second basal
pterygiophores of the soft dorsal fin are usually located
between the neural spines of the eighth abdominal and
first caudal vertebrae while the last basal pterygiophore
is located between those of the sixth and seventh caudal
vertebrae. The basal pterygial elements are cartilage fill-
ed at their dorsal and ventral ends.

Anal Fin.

Fin rays and pterygiophores. — Fourteen fin rays
usually present; the first ray and the last ray un-
branched, the others branched in single or double
dichotomies. Each fin ray has a small pair of distal
pterygiophores as two distinct halves between the bifur-
cate base of the ray, except for the last one or two rays, in
which these elements, if present, are unossified. Basally
the fin rays are supported by about 13 basal pterygio-
phores, which decrease in size posteriorly in the series.
They possess weakly developed lateral flanges for muscle
attachment along their lengths, as in the case of the dor-
sal fin basal pterygiophores, and they are cartilage filled
at their dorsal and ventral ends. The first basal pterygio-
phore is by far the largest of the series and has well-
developed lateral flanges for muscle attachment along
each side of its anterior surface. It articulates by fibrous
tissue to the posterior surface of the bifid portion of the
haemal spine of the first caudal vertebra and to the
anterior surface of the haemal spine of the second caudal
vertebra. The 2d to 13th basal pterygiophores articulate
by fibrous tissue to the haemal spines of the 2d to 7th
caudal vertebrae.

pelvic fin, if it is present at all. Two species: Protobalis-
tum imperiale (Massalongo 1857) and Spinacanthus
cuneiformis (Blainville 1818).

The Eocene Eoplectinae differ from both the
Spinacanthinae and the Recent species in having:
gymnodontlike jaws and upper jaw suspension, with
numerous small rounded dental units incorporated in the
matrix of the premaxillary and dentary, with the
premaxillary and maxillary immovably articulated, and
the jaws probably nonprotrusible; a well-developed
pelvic fin of one spine and four long branched rays; the
ventral shaft of the second basal pterygiophore of the
spiny dorsal fin directed anteroventrally and articulated
either to the base of the skull or between the neural
spines of the first and second vertebrae; 9 4-11 vertebrae
in a highly arched column; decidedly more soft dorsal
and anal fin rays than basal pterygiophores. Two
species: Eoplectus bloti Tyler (1973b) and
Zignoichthys oblongus (Zigno 1874a).

The Recent species are divided into two subfamilies
mainly on the basis of the shape of the posterior portion
of the skull, the placement of the epiotics and the shape
of the pelvis, none of which features are known for the
fossil species. Comparative diagnoses of the two Recent
subfamilies follows:

Hollardiinae. The dorsal surface of the supraoccipital
entirely domelike, without a broad flat expanse, the pos-
terior surface of the dome convex; epiotics not meeting
medially on the dorsal surface of the skull, separated
there by the supraoccipital and meeting medially only on
the posterior surface of the skull, and articulating
anteriorly with the frontals; pelvis a sturdy shaft more or
less triangular or broadly heart-shaped in cross section;
first anal fin basal pterygiophore tending to be -t- -shaped
in cross section, with an anteromedial process; teeth con-
ical. Four western Atlantic species and one from Hawaii,
in two genera (Hollardia and Parahollardia) .

Comparative diagnoses of subfamilies (Spinacan-
thinae, Eoplectinae, Hollardiinae, Triacanthodinae).

—As discussed in greater detail by Tyler (1968, 1973b),
there are four subfamilies of triacanthodids, two fossil,
and two Recent, whose most pertinent diagnostic
features are summarized below. See Sorbini (1968) and
Blot (1969) for the upper portion of the lower Eocene
age of the Monte Bolca, Italy, fish beds.

The Eocene Spinacanthinae differ from both the
Eocene Eoplectinae and the Recent species (Hollardiinae
and Triacanthodinae) in having: enormously elongate
dorsal fin spines (50 to 90% SL) in an exceptionally long-
based fin (slightly over 50% SL), with the soft dorsal and
anal fins exceptionally short-based (about 11 to 12% SL);
the spiny dorsal fin origin over the level of the middle or
front of the eye; the eye small (5 to 7% SL) and placed
high in the head; a probably only poorly developed

Triacanthodinae. The dorsal surface of the supraoc-
cipital with a broad flat expanse, anteromedially on
which is an upraised area varying in shape from a small
dome to a laterally compressed thin crest; epiotics
meeting medially on the dorsal surface of the skull as
well as on the posterior surface, and separated anteriorly
from the frontals by the sphenotics; pelvis a flat basin
with upturned edges; first anal fin basal pterygiophore
T-shaped in cross section, without an anteromedial
process; teeth conical or wider than thick and more or
less truncate distally. Of the 14 species in 9 genera, 13
species belong to 8 genera (Triacanthodes, Mephisto,
Paratriacanthodes, Atrophacanthus, Bathyphylax,
Tydemania, Macrorhamphosodes, Halimochirurgus) in
the Indo-western Pacific, and 1 species belongs to a
monotypic genus (Johnsonina) in the western Atlantic.

Anatomical diversity. — The anatomical diversity of
the two Recent subfamilies has been treated in detail by

Tyler (1968), while the Eoplectinae were described sub-
sequently (Tyler 1973b), and the features of these three
subfamilies need be only briefly summarized here. The
Spinacanthinae were discussed by Tyler (1968) on the
basis of the literature alone, but the holotypes and only
known specimens of the two species of this subfamily
have been examined for the present work and are
redescribed below.

Protobalistum imperiale (Massalongo 1857:775), in
counterpart, MCSNV T9 (head left)-10, 522 mm SL, and
655 mm TL (total length), is gigantic by comparison to
the other members of the superfamily Triacanthoidea,

^i*"
<^^<

.e*^-^'

the largest triacanthodid otherwise known being a 174
mm SL specimen of Hollardia hollardi and the largest
triacanthid a 272 mm SL specimen of Triacanthus
biaculeatus (Tyler 1968:64; 1970f:5). The spiny dorsal fin
is enormous, with five well-developed strong spines
decreasing in length posteriorly in the series, with the
possibility that a much smaller sixth spine is present just
behind the base of the fifth spine but hidden there by a
slightly misplaced scale plate. All of the spines are
covered with the same kind of granulations as those of

-Parahollardia Uneata
omposite based on several
45.7-86.1 mm SL, Gulf of
Mexico to South Carolina.

lateral view
I specimens.

maxillary
premaxillary

pterosphenoid
parasphenoid
cartilage prefrontal
palatine \ ethmoid

frontal

supraoccipital

the scale plates of the body. The first dorsal spine, in-
cluding the groove representing at least part of the miss-
ing distal end of the spine, is at least 279 mm (53.6% SL),
the second at least 273 mm (52.4% SL), the third at least
210 mm (40.3% SL), the fourth at least 139 mm (26.6%
SL), and fifth, which seems complete even dis-
tally, 90.3 mm (17.3't SL). There are about nine dorsal
fin rays, of which the fourth is the longest, 79.5 mm
(15.2% SL). The anal fin is directly below the soft dorsal
fin and likewise has about nine rays, of which the third is
the longest, 91.3 mm (17.5'c SL). The spiny dorsal fin
base is about 276 mm (53'^c SL) long, while the bases of
the soft dorsal and anal fins are both about 57 mm (11%
SL) long. All but the first ray in the soft dorsal and anal
fins are extensively branched.

The broadly rounded caudal fin is 133 mm (25.6% SL)
and has 12 rays, of which the uppermost ray and the
lowermost ray are unbranched and the intervening 10
rays extensively branched. The length of the caudal
peduncle (from end of anal fin base to middle of caudal
fin base) is 126 mm (24.1% SL) and its least depth 55.0
mm (10.5% SL). The greatest body depth is about 210
mm (40% SL). The eye is placed high in the head just
below the base of the first dorsal spine and has an ap-
proximate diameter of 39 mm (7% SL).

There are remains of pectoral fin rays, but the number
cannot be counted. It is impossible to state whether a
pelvic fin was present or not. In redescribing the species,

Figure H.—Parahollardia lineata:

dorsal (left) and ventral (right)

views of skull; composite based on

several specimens. 45.7-86.1 mm SL,

Gulf of Mexico to South Carolina.

Figure 9.—Parahollardia lineata: posterior view of

orbit (left) (cross section of skull; dashed lines
represent cut surfaces of frontals and parasphenoid);
posterior view (right) of skull: below, posterior
and lateral views of first abdominal vertebra:
composite based on several specimens, 45.7-86.1
mm SL, Gulf of Mexico to South Carolina.

Zigno (1887a:4) said that there was a pelvic spine of 30
mm length, bent backward along the side of the body.
Along the lower edge of the body at the level between the
bases of the second and fourth dorsal fin spines, there is a
ridge of bone with the same type of surface granulation as
found on the scale plates, described below. This ridge is
about 173 mm long (33% SL) but is discontinuous, with
some segments showing surface granulations and others
the bone beneath the granulations. Zigno may have had
one of these granular segments in mind for the pelvic
spine. If this ridge does represent a pelvic spine, then it
was almost as well developed as the more anterior dorsal
fin spines. However, it is at least equally likely that this
ridge is simply the ventrolateral curved edge of the
carapace of scale plates that fully covers this region of
the body, as described below. In short, it is not known at
present whether Protobalistum had pelvic spines.

The teeth are well-preserved. They are large and
relatively massive, some with straight or rounded distal
edges as in incisors but others with a distinct cusp or dis-
tal constricted region. The teeth with more rounded dis-
tal edges may represent those toward the middle of the
jaws that have been worn down through use. There are
about four or five teeth in each half of the upper and

JL'^ — Tag's ^N

ox'S"*-"'!--
S .Sf X -2 -ii -S « "S

Sillily

a " o S .

.1 -1 -5 .- -3

1^*

Figure 15.— Protobalistum imperiale: top. lateral

views of the two counterparts; middle left, lateral

view of head, showing pharyngeal and jaw teeth;

lower right, lateral view of caudal region;

bottom, lateral view of middle of body (head to right),

showing remains of scale plates, pectoral fin

rays, coracoid, and pharyngeal teeth; holotype,

522 mm SL, Eocene of Monte Bolca, Italy.

lower jaws, but it cannot be determined if they occurred
in a single series or in a major outer series internal to
which were a few other teeth. Pharyngeal teeth are pres-
ent in the appropriate area at the intersection of the
levels of the horizontal through the middle of the mouth
and the vertical to the base of the second dorsal fin spine.
They are small and more elongate versions of the cusped
teeth of the mouth.

The anterior half of the body exclusive of the head
(i.e., in the region below the bases of the second to fifth
dorsal spines) is completely (taking into account missing
pieces) covered by relatively huge, thick bony plates with
numerous rounded bumps or granulations on their sur-
face. Many of these plates are distinctly hexagonal, and
all of them articulate to one another along all of their
edges of apposition by intermeshed delicate denticula-
tions, all of this being remarkably similar to the scale
plates in ostracioids. Measurements of a few of the more
complete scale plates are as follows: a plate below the
bases of the fourth and fifth dorsal spines is 56.9 mm
(lO.g'^o SL) deep and 64.3 mm (12.3% SL) wide; a plate
just below the former is 51.6 mm (9.9% SL) deep and 52.3
mm (10.0% SL) wide; one of the few scale plates on the
head, approximately midway between the eye and
mouth, is about 31 mm (5.9% SL) deep and 27 mm (5.2%
SL) wide. While a continuous ostracioidlike carapace ap-
parently is present only on the anterior half of the body
exclusive of the head, there are numerous, usually much

smaller, isolated scale plates with the same kind of sur-
face granulations on the head, especially dorsally and
ventrally, and on the body posterior to the carapace in
the region between the soft dorsal and anal fins and on
the caudal peduncle. The scale plates, and apparently
thick skin, on the caudal peduncle completely obscure
the vertebral column.

The possible phylogenetic implications of the presence
of ostracioidlike scales in the triacanthodid
Protobalistum is discussed in the section on subfamilial
relationships.

Spinacanthus cuneiformis (Blainville 1818:359), a
single plate, MNHN (LP) 10918, 104 mm SL, presently is
a very incomplete and poorly preserved specimen which
appears to have lost some of its surface and bony struc-
ture since it was studied by Volta (1796), Blainville
(1818), and Agassiz (1839 illustration, 1844c description).
Much less can be seen in the specimen today than in the
illustrations of it by Volta and Agassiz. The spiny dorsal
fin is even more enormous than in Protobalistum, with
the first spine, even with a slight amount of the distal tip
missing and not measurable, being 95.7 mm (92.0' r
SL). The second to fourth dorsal spines are missing too
much of their distal ends to be measureable, but ap-
parently they were almost as long as the first spine. The
fifth spine is about 35 mm (34% SL) long, and a smaller
sixth spine, if present, cannot be detected. There are
large retrorse thomlike barbs along the anterior edge of
at least the basal third of the first dorsal spine. There
appear to have been about 10 fin rays in both the ap-
posed soft dorsal and anal fins. The slightly rounded
caudal fin is poorly preserved, but it had about 12 rays
(Agassiz thought 7 or 8 branched rays plus an un-
branched ray above and below). The pectoral fin has
about 12 or 13 rays (Agassiz thought 12). Agassiz es-
timated the number of vertebrae to be no more than 20,

and I would guess at about 20 judging on the basis of the
size of those that show and of the space available for
those that do not show. The length of the caudal pedun-
cle is 26.0 mm (24.9% SL) and its depth at least 7.0 mm
(6.7% SL). The greatest body depth is 39.0 mm (37.4%
SL). The eye is placed high in the head just below the
base of the first dorsal spine and has a diameter of 7.2
mm (6.9% SL).

The branchiostegal rays from both sides are evident,
and in the more clearly seen set there appear to be six
rays. The prefrontal is one of the few bones that are rela-
tively well preserved, and it has an appropriately
moderate size. Agassiz thought that a pelvic spine was
present and displaced in the region of the branchios-
tegals. At least at present I find no structure in the
branchiostegal region that could possibly be interpreted
as a pelvic spine. However, just posterior to the branchi-
ostegals and below the pectoral fin is a vague indication
of a long rodlike strut of bone oriented vertically. It is
possible that this is the structure seen by Agassiz, and it
might possibly be a thin, weakly developed barbless
pelvic spine. It would appear to me more likely to be a
part of the pectoral girdle, perhaps the thickened pos-
terior edge of the coracoid. As with Protobalistum, it is
not known with surety whether Spinacanthus had a
pelvic fin.

The teeth in the lower jaw are well-preserved. There
appear to have been about seven of them in the left den-
tary. A few of these teeth are especially fully exposed and
can be seen clearly in three dimensions. One of these is
2.0 mm (1.9% SL) long from its flat circular base to its
distal tip. The tooth tapers only very slightly from basal-
ly to distally throughout most of its length, but in the

Figure 16.— Spinacanthus cuneiformis:

lateral view of holotype, 104 mm SL,

Eocene of Monte Bolca, Italy.

distal one-fifth of the length the tooth is strongly con-
stricted into a blunt nipple. In the distal halves of their
lengths the teeth bear about 10 annularlike rings, which
I doubt are artifacts. In the upper jaw, grooves represent
where teeth once were present, and perhaps of their
sockets. More than 20 pharyngeal teeth are present in the
region in front of the pectoral fin base, these being
similar to those of the jaws but much smaller, the largest
being about 0.5 mm (0.5% SL) long.

The body bears numerous, mostly isolated, scale
plates of moderate size. Each plate has a strong rela-
tively high central spine peripheral to which are much
smaller and lower rounded bumps or granulations similar
to those of Protobalistum. The more or less isolated
plates, which probably covered only about one-fourth of
the surface of the body, seem to become more numerous
ventrally, where they may have been placed in more or
less regular horizontal rows, with some contact between
individual plates. The plates are largest and most
regularly arranged along the ventral edge of the body
from the region of the branchiostegals posteriorly to the
anal fin base, the largest plates being in the abdominal
area. The largest plates have a diameter of about 5.0 mm
(4.8% SL). Very few scale plates are evident on the head.

the only one clearly seen being on the profile of the snout
about midway between the mouth and eye. The scale
covering of Spinacanthus is comparable to that found on
Protobalistum anterior and posterior to its carapace of
articulated hexagonal plates.

The anatomical features of the Eoplectinae, recently
described by Tyler (1973b), are summarized briefly
below.

The Eoplectinae are represented by two species known
only from their holotypes, only one of which (Eoplectus)
is well preserved.

In most of its features Eoplectus bloti Tyler (1973b) is
a generalized plectognath at the triacanthodid level of
organization. It has at least five and probably six dorsal
fin spines, a rounded caudal fin of 12 rays (i, 10, i), a
generalized caudal skeleton, a short caudal peduncle,
and the most generalized condition of the pelvic fin
known among the plectognaths, this being a spine and
four well-developed branched rays. The overall con-
figuration of the body is also triacanthodid. However,
Eoplectus has the jaws and dentition modified into a
crushing beak, with numerous small rounded dental
units incorporated into the matrix of the premaxillary
and dentary. This condition otherwise is found among

Figure U .—Eoplectus bloti: lateral view of holotype,
65.2 mm SL, Eocene of Monte Bolea, Italy.

the plectognaths only in the suborder Gymnodontes
(Tetraodontoidei), and it is closely similar to that of
Triodon, the monotypic representative of the Triodon-
tidae and the most generalized member of its suborder.

Eoplectus is similar to Recent triacanthodids and
differs from Triodon in the following ways: 1) Recent
triacanthodids have the most fully developed pelvic fin
among plectognaths, in more generalized forms con-
sisting of a spine and two unbranched rays, while in
Eoplectus this fin is even better developed as a spine and
four branched rays, the better development in Eoplectus
being less reduced than in triacanthodids from the
ancestral perciform condition of one spine and five
branched rays; 2) the rounded i, 10, i caudal fin, without
procurrent fin rays; 3) the short caudal peduncle; 4) six
dorsal fin spines borne on five well-developed basal
pterygiophores; 5) first basal pterygiophore of the spiny
dorsal fin articulated low on the base of the skull and
bearing the first two spines; 6) third to fifth basal
pterygiophores of the spiny dorsal fin with their shafts
directed ventrally or posteroventrally in relation to the
vertebral axis and articulated with the neural spines of
the fifth to eighth vertebrae; 7) 20 vertebrae; 8) probably
no ribs; 9) small prefrontal not extensively sutured to the
frontal, parasphenoid and ethmoid; 10) moderately large
eye located about midway between the snout and spiny
dorsal fin origin; 11) relatively large numbers of dorsal
and anal fin rays.

Eoplectus is similar to Triodon and differs from the
other triacanthodids in the following ways: 1) the small
individual rounded dental units intimately associated
with the matrix of the premaxillaries and dentaries to
form a crushing beak; 2) premaxillary probably im-
movably held by suturing to the maxillary, the ascending
process of the premaxillary poorly developed and the up-
per jaw probably nonprotrusible, rotating around a ball
and socket joint between the palatine and the maxillary;
3) the palatine probably immovably sutured to the
ethmoid-vomerine region; 4) vertebrae in a 9 -I- 11 ar-
rangement.

Eoplectus is similar to both the other triacanthodids
and to Triodon in the following ways: 1) the pelvis with
a long anterodorsal shaft probably held between the
anteromedial edges of the cleithra; 2) a generalized
caudal skeleton with the parhypural, numerous
hypurals, and single epural probably not fused to one
another or to the last centrum.

Eoplectus differs from both the other triacanthodids
and Triodon in the following ways: 1) probably no
uroneurals, or only very small ones at the most; 2) the
great depth of the caudal peduncle; 3) the placement of
the well-developed shaft of the second basal pterygio-
phore of the spiny dorsal fin, directed strongly antero-
ventrally and articulating either with the base of the
skull or between the neural spines of the first and second
vertebrae; 4) the spiny dorsal fin base equal in length to
the soft dorsal fin base rather than either much greater or
much less than it; 5) soft dorsal and anal fin rays far
more numerous than the basal pterygiophores supporting
them; 6) the extremely well developed ventral flange of

Figure iS.—Zignoichthys oblongus:

lateral view of holotype, ca. 161 mm SL,

Eocene of Monte Bolca, Italy.

the parasphenoid versus weakly to moderately
developed; 7) the highly arched vertebral column; 8) the
body with an incomplete covering of scales, many of
which are isolated and nonoverlapping.

The other species of Eoplectinae is the extremely poor-
ly known Zignoichthys oblongus (Zigno 1874a, b) from
the same strata of the Eocene of Monte Bolca, Italy, in
which Eoplectus occurs. Of its few known features, re-
described by Tyler (1973b) and briefly summarized here,
the most important is the huge jaw bone composed of
two equal sides fully fused together in the midline with
no evidence of any kind of separation or suture between
the right and left halves. The biting edge of the beaklike
jaw is not exposed, but the teeth can be expected to be
small dental units incorporated into the matrix of the
bone much as in Eoplectus, Triodon, and some of the
other gymnodonts. The body is covered with numerous
small scales with numerous upright spinules, and the
caudal fin has about 12 principal rays in addition to at
least two or three procurrent rays both above and below.
Among plectognaths, procurrent rays are otherwise
known only in Triodon. It is not known with surety
whether dorsal fin spines were present, for most of the
appropriate region of the body is missing, but it is
probably best to assume that a spiny dorsal fin of un-
known size was present. The region where one would ex-
pect to find a pelvic fin if one were present is relatively
well preserved, and no pelvic fin is apparent, and it can
probably be assumed that the pelvic spine, if present at
all, was of reduced size.

Too few critical features are known of Zignoichthys to
allow a precise placement of it among the plectognaths,
and it has very tentatively been associated (Tyler 1973b)
with Eoplectus and the Eoplectinae of the Triacanthodi-
dae for the following reasons. It has a generalized condi-
tion of the caudal fin and caudal peduncle, and of the
soft dorsal and anal fins and scales consistant with the
triacanthodids, and, if a spiny dorsal fin was present, it
would probably have been longer based than the soft dor-
sal fin, judging from the space available for it, while the
jaw that is exposed is similar to that of Eoplectus.

The Eoplectinae differ from the other fossil subfamily
of triacanthodids, the Spinacanthinae, mainly by having
a larger eye located about midway between the snout and
spiny dorsal fin origin, the lesser development of the
spiny dorsal fin and its shorter base and more posterior
position, better developed and longer based soft dorsal
and anal fins, and the massive beaklike jaws.

The anatomical diversity of the two Recent sub-
families of triacanthodids is summarized below from
Tyler (1968). Of greatest phylogenetic significance is the
arrangement and form of the bones of the posterodorsai
region of the skull and in the shape of the pelvis, which
features divide the Recent species into different sub-
families.

In the Hollardiinae, which form a compact group of
five species in two genera, the supraoccipital is domelike
and separates the epiotics on the dorsal surface of the
skull, while the epiotics directly contact the frontals and
the pelvis is a sturdy shaft with a more or less triangular
shape in cross-section. In the Triacanthodinae, whose 14
species in 9 genera range from relatively normal species
through intermediates to weirdly specialized forms with
extremely long tubular snouts, the supraoccipital
basically is a broad flat bone with a variously developed
crest and does not separate the epiotics on the dorsal sur-
face of the skull, while the epiotics are separated from
the frontals by the sphenotics and the pelvis is a broad
flat basin with upturned edges.

The two genera of Hollardiinae differ mainly in that
Parahollardia retains a small number of teeth internal to
the major outer series while the inner teeth on the
premaxillary and dentary are lost in Hollardia.

Similarly, in the Triacanthodinae, one genus
(Triacanthodes) retains inner series teeth which are lost
in the other eight genera. In the Triacanthodinae the
teeth are usually conical, as they are also in the Hollar-
diinae, but in a few genera the teeth are large, thin, and
wide, much compressed from front to back, and either
truncate (Tydemania) or truncate to rounded or pointed
(Macrorhamphosodes) distally. In Johnsonina the con-
ical teeth are smaller and far more numerous than in the
other genera of triacanthodins, while in Halimochi-
rurgus the teeth are small and conical but of low to
moderate (3 to 15) number in both jaws. In one species,
Macrorhamphosodes platycheilus, the few upper jaw
teeth are lost in adults.

The snout in most triacanthodins is of short to
moderate length and normal shape, but in one of the two
species of Bathyphylax, the snout, although of only
moderate length, is decidedly concave in front of the
eyes, giving it a tubular appearance, while in Macro-
rhamphosodes and Halimochirurgus the snout is enor-
mously (26 to 49% SL; least so in young) elongate and
tubular. The mouth is more or less terminal in Triacan-
thodes, Mephisto, and Paratriacanthodes, as it is in the
Hollardiinae, but it is slightly supraterminal in John-
sonina and Atrophacanthus, slightly to decidedly supra-
terminal in Bathyphylax, and decidedly supraterminal
in Tydemania, Macrorhamphosodes, and Halimo-
chirurgus.

The bones of the long tubular snouts of Macrorham-
phosodes and Halimochirurgus are strangely rearranged
from the condition of other members of the family. In
both species of Halimochirurgus and in M. uradoi the up-
per three-fourths of the tube is composed, from anteriorly
to posteriorly, more or less successively by the ecto-
pterygoids, palatines, and vomer, which are thin, curv-
ed, and elongate plates. In M. platycheilus, however, the
vomer extends further forward than in the others to form
the roof of the anterior region of the snout to the exclu-
sion of the ectopterygoids and palatines. The ventro-
lateral region of the tube in all four species is formed by
the elongate quadrate and the long slender anterior
prolongation of the preoperculum. The premaxillary
pedicels of M. uradoi are prolonged posteriorly as thin
filaments but in none of the four long-snouted species do
the pedicels extend posteriorly to the region of the
ethmoid, as they do in other triacanthodids. The bones of
the snout of Halimochirurgus more completely roof over
the tube than in Macrorhamphosodes, and the tube is
narrower and more elongate in Halimochirurgus than in
Macrorhamphosodes.

In Halimochirurgus the lips are of normal
triacanthodid size and shape, while in Macrorham-
phosodes the lips form a wide disklike structure around
the mouth. The size of the lips in Tydemania is more or
less intermediate between that of Macrorhamphosodes
and other triacanthodids. In Macrorhamphosodes the
mouth becomes twisted to the left or right, usually
progressively so with increasing specimen size, while in
Halimochirurgus it is symmetrically placed at all sizes.

The spiny dorsal fin is relatively well developed, with
all six spines protruding through the surface and of
gradually decreasing length posteriorly in the series in
Triacanthodes, Mephisto, Paratriacanthodes, and John-
sonina, as in the Hollardiinae, while in Bathyphylax the
fourth spine is much shorter than the third but protrudes
through the surface and is clearly seen, with the fifth and
sixth spines only slightly shorter than the fourth and
protruding only slightly less. In Atrophacanthus,
Tydemania, Macrorhamphosodes, and Halimo-
chirurgus the last three dorsal spines are rudimentary
and usually only barely, if at all, protrude through the
surface, while in Halimochirurgus the third spine also
tends to be rudimentary.

Triacanthodes, Johnsonina, and Paratriacanthodes
herrei usually have two pelvic fin rays, although in large
specimens the second may tend to become a buried rudi-
ment and be resorbed, while in P. retrospinis, Mephisto,
Atrophacanthus, Bathyphylax, Tydemania, Macro-
rhamphosodes, and Halimochirurgus there is only a single
pelvic fin ray.

The size of the gill opening varies from relatively huge
for a plectognath fish in Mephisto, in which the slit ex-
tends down well below the level of the lower edge of the
pectoral fin base, to simply large, as in Triacanthodes,
Halimochirurgus centriscoides, and Macrorham-
phosodes platycheilus, in which the slit extends down to
or around the level of the lower edge of the pectoral fin
base, to moderate or short, as in all the other species, in

which it extends to somewhere between the level of about
one-third to three-fourths down the pectoral fin base, the
slit being especially short in Halimochirurgus alcocki
and Macrorhamphosodes uradoi.

The length of the pseudobranch, more or less corre-
lated with its number of lamellae, varies from especially
long, as in Triacanthodes, in which it extends below the
level of the lower edge of the pectoral fin base, to short, as
in Mephisto, Bathyphylax, and Macrorhamphosodes, in
which it extends to the level of the upper edge of the pec-
toral fin base, with the other genera variously in-
termediate.

The greatest depth of the body in adults of Triacan-
thodes, Mephisto, Paratriacanthodes, Johnsonina, and

Bathyphylax, as in the hollardiins, is about 40 to 65%
SL, while in Atrophacanthus and Tydemania it is about
30 to 35% SL, and in Macrorhamphosodes and Halimo-
chirurgus about 15 to 25% SL. The eye of Johnsonina is
larger than in any other triacanthodid at comparable
size, and it is the only genus to feature a large ocellated
eye spot on the body, just below the soft dorsal fin.

In Triacanthodes, Mephisto, Johnsonina, and Para-
triacanthodes herrei, as in hollardiins, the basal flange of
the pelvic spine and the region of the pelvis against
which it slides are smooth, allowing for only a single posi-
tion of locking the erected spine against the pelvis, while
in P. retrospinis, Atrophacanthus, Bathyphylax, Tyde-
mania, Macrorhamphosodes, and Halimochirurgus these
two regions are grooved, allowing for multiple positions
of locking the erected spine against the pelvis.

The crest on the otherwise flattened supraoccipital of
triacanthodins is best developed in Triacanthodes and
Johnsonina, being high and relatively wide at the
anterior edge, while in all the other genera the crest is low
and thin. Other features of the anatomical diversity of
the triacanthodids are discussed by Tyler (1968), and for
the muscles by Winterbottom (1974).

Generic relationships. — Parahollardia and Hollardia
are obviously very closely related, with the former slight-
ly the more generalized of the two, based on the evidence
of its retention of a series of teeth internal to the major

branchiostegal rays

row, while the similarity in the configuration of the pos-
terodorsal region of the skull and of the pelvis clearly
indicates that these two genera, forming the hoUardiins,
represent a line of triacanthodid evolution distinct from
that of the other nine Recent genera, the triacanthodins.
The relationships of these two Recent subfamilies to the
two fossil subfamilies are discussed in the following sec-
tion on subfamilial relationships.

Among the triacanthodins, as summarized from Tyler
(1968:29-31), the deeper bodied genera with well-
developed spiny dorsal fins are more generalized than the
others, and of these generalized genera, Triacanthodes,
which retains a series of teeth internal to the major outer
series and which has a deep body, well-developed spiny
dorsal fin, and long gill opening, is the most generalized
of all. Mephisto and Paratriacanthodes are close deriva-
tives of the same ancestral stock which gave rise to
Triacanthodes and are simply progressively more
modified from it. Johnsonina is also a derivative of a
Triacanthodes-like ancestral stock which has indepen-
dently further specialized in most ways from it beyond
the levels of Mephisto and Paratriacanthodes.

Atrophacanthus, Bathyphylax, and Tydemania are
probably independent derivatives of a stock of Paratria-

Figure 22.— Hollardia hollardi: lateral view

of heads of moderate-sized specimen, 62.7 mm SL, (above)

and extremely large specimen. 174 mm SL

(the largest Recent triacanthodid fish recorded),

Caribbean, showing the change in configuration,

increased suturing in the occipital-otic

region and decreased amount of cartilage

visible externally in large specimens.

canthodes-like level of organization. An Atrophacan-
thus-like form, with the last three dorsal spines rudi-
mentary, the second pelvic ray absent, and the mouth
slightly but distinctly supraterminal, is probably not
ancestral to any other living triacanthodid, for the
tendencies it shows are for the snout to slightly decrease
in length and the conical teeth to increase in number
while becoming smaller and more sharply pointed. These
tendencies are not compatible to ancestry of any of the
other moderately or highly specialized genera.

However, a Tydemania-Vike form is ancestral to
Macrorhamphosodes, while a Bathyphylax-Uke one is
ancestral to Halimochirurgus. Thus, the two long-
snouted genera which are superficially closely related in
actuality have independently evolved long snouts and
are not as closely related to one another as to other
genera.

Tydemania and Macrorhamphosodes are the only two
genera of triacanthodids in which the teeth are wide and
thin and variously pointed to rounded to truncate distal-
ly. Tydemania has a snout of moderate length, a distinct-
ly supraterminal mouth of moderate width with wide
fleshy lips, the truncate teeth well developed in both
jaws but slightly fewer in the upper than in the lower jaw,
the last three dorsal spines rudimentary, and the second
pelvic ray absent. These are all characteristics that one
would expect to find in an ancestor of Macrorham-
phosodes, in which the elongate snout bears a distinctly
supraterminal mouth in a wide fleshy disk, the teeth are
more or less truncate and well developed in the lower jaw
but few in number or absent in the upper jaw, the last
three dorsal spines are rudimentary, and the second
pelvic fin ray is absent.

In Halimochirurgus the distinctly supraterminal
mouth has thin lips and feeble teeth in reduced number
in both jaws at the end of an extremely long tubular

Figure 23. — ParahoUardia schmidti
lateral view of head;
62.0 mm SL. Nicaragua.

Figure 24.—Triacanthodes anomalus:
lateral view of head;
71.5 mm SL, Japan.

Figure 25.—Paratriacanthodes
retrospinis: lateral view of head;
85.3 mm SL, Mozambique.

Figure 2^.— Johnson
eriomma: lateral
head; 67.0 mm SL,
Puerto Rico.

Figure 21 .—Atrophacanthus
japonicus: lateral view of head
4.3.1 mm SL. Celebes.

Figure 2». — Tydemania navigatoris
lateral view of head: H4.4 mm SL,
Bay of Bengal.

72

tu pro occipital

ipiwnoric

Figure 29.—Macrorhamphosodes uradoi:
lateral view of head; 69.9 mm SL, Japan.

branchiottagal rays
intaropwculum
•pihyal

snout, the last three dorsal spines rudimentary and the
second pelvic fin ray absent, all of which features are
foreshadowed in Bathyphylax, in which the snout is of
moderate width, the mouth slightly to distinctly supra-
terminal, the lips thin, the teeth conical and slightly
reduced in number in both jaws, the last two or three dor-
sal spines rudimentary, and the second pelvic fin ray ab-
sent.

Subfamilial relationships, and relationships to the
Triacanthidae and Gymnodontes.— Since the shape of
the posterodorsal region of the skull and of the portion of
the pelvis posterior to the pelvic spines, the primary dis-
tinguishing features of the two Recent subfamilies, is not
known in the two fossil subfamilies, a pertinent question
about the basal genus in each of the two Recent sub-
families is which comes closest in basic structure to the
ancestral stock which gave rise to them: was it a stock

with a domelike supraoccipital separating the epiotics on
the dorsal surface of the skull and with a shaftlike pelvis,
or a stock with a flattened supraoccipital bearing a
medial crest and allowing the epiotics to meet on the dor-
sal surface of the skull and with a basinlike pelvis, or
perhaps a stock with intermediate conditions? In the
absence of a solution from the fossil triacanthodids there
is but one clue as to which condition is more likely to
have been that of the early triacanthodids, and that is
the condition of the supraoccipital and epiotics in the
acanthurids, which are often considered to have evolved
in the Eocene from the same Cretaceous stock (perhaps
the Pharmacichthyidae, according to Patterson 1964) as
that which gave rise in the Eocene to the triacanthodids
and other plectognaths.

The supraoccipital of acanthurids is basically dome-
like and without a flattened basal portion, while the
epiotics are separated from each other medially on the

Figure 30.— Halimochirurgiu centrUcoidee:
lateral view of head; 99.2 mm SL, Bay of Bengal.

dorsal surface of the skull, basically the same condition
as found in the hollardiins and in contrast to that of the
triacanthodins. The acanthurid epiotic is separated for a
short distance from the frontal by a parietal, a bone lost
in all plectognaths, but the epiotic in acanthurids ap-
proaches the frontal closely enough to be more similar to
the condition in hollardiins than to that in triacan-
thodins. The indication from the acanthurids is that the
domelike supraoccipital excluding the epiotics from the
midline of the dorsal surface of the skull in hollardiins is
the more generalized of the two and ancestral to the
flattened condition of the supraoccipital found in
triacanthodins.

The phylogenetic implications of the shaftlike pelvis of
hollardiins versus the basinlike pelvis of triacanthodins
are unknown, for these differentially shaped portions of
the pelvis are long extensions behind the level of the
pelvic fins, and most fishes, including acanthurids, have
no such structure with which that of triacanthodids can
be compared.

While the triacanthodins are more speciose and
anatomically diversified than the other Recent subfami-
ly, from which they were derived, they apparently are not
ancestral to any other group of plectognaths, for it is the
hollardiins that are ancestral to the triacanthids, and
that same line probably is ancestral to the balistoids and
ostracioids as well, while the gymnodonts are derived
from the eoplectins, as discussed below. The hoUardiin
ancestry of the triacanthids is attested to by the similari-
ty in the domelike shape of the supraoccipital, which ex-
cludes the epiotics from meeting on the dorsal surface of
the skull, and in the shaftlike nature of the pelvis pos-
terior to the pelvic spines found in this subfamily of
triacanthodids and in the derived triacanthids.

Since the nature of the supraoccipital and pelvis is not
known in the fossil Eoplectinae and Spinacanthinae, it is
impossible to say whether one was more closely related
to the hollardiin or triacanthodin line of triacanthodid
radiation than the other. However, it is clear that the
eoplectins were ancestral to the gymnodonts and that the
spinacanthins were perhaps an unsuccessful experiment
of the early triacanthodids that gave rise to at least
one species with some superficial similarities to os-
tracioids but which became extinct without derivatives
alive today.

The phylogenetic significance of the Eoplectinae is dis-
cussed in detail by Tyler {1973b) and need be only briefly
summarized here. It is assumed to be extremely unlikely
that the highly specialized condition of small and
numerous dental units intimately incorporated into the
matrix of the premaxillaries and dentaries, forming a
parrotlike crushing beak, has arisen independently in
two or more lines of plectognath radiation. It is more par-
simonius to assume that this highly specialized condition
arose in only one basic line of plectognaths, that leading
to the gymnodonts, whose most generalized family is the
Triodontidae, and that all plectognaths with this condi-
tion are phylogenetically related. Eoplectus basically is
triacanthodid and generalized (general body shape, well-
developed pelvic and spiny dorsal fins, rounded i, 10, i

caudal fin) in most of its structure, but has a distinctly
triodontid and highly specialized dentition and jaw
structure. It represents almost precisely the kind of
Eocene ancestral line that has been speculated (Tyler
1962a:793) between the triacanthodids and triodontids,
and thus between the basal scleroderms and the deriva-
tive gymnodonts.

The conversion of an Eoplectus-like form into a
Triodon-Vike one involves mostly reductive tendencies,
which are well-known to be of great importance in the
evolutionary diversification of the plectognaths, as
follows: 1) complete loss of pelvic fin but retention of
pelvis; 2) reduction of spiny dorsal fin and its basal
pterygiophores into a posteriorly migrated rudimentary
structure; 3) reduction in number of soft dorsal and anal
fin rays; 4) great increase in length and decrease in depth
of caudal peduncle concomitant with conversion of
rounded into forked caudal fin; 5) decrease in height of
neural and haemal spines of caudal peduncular vertebrae
and development of anterolateral flanges on neural
spines, associated again with elongation and slenderiza-
tion of peduncle for more powerful swimming with forked
caudal fin; 6) increased covering of scales with more
elaborate surface spinulation and development of ex-
pansible dewlap of skin between finless pelvis and anal
region; 7) increase in size of prefrontal and of its sutural
connections to surrounding snout bones for stronger but-
tressing of crushing beak; 8) probable similar increase in
size of palatine for same reason; 9) decrease in depth of
ventral flange of parasphenoid; 10) straightening of
vertebral column associated with elongation of body.

Triodon has only two anatomical features known to be
more generalized than in Eoplectus: the presence of
procurrent caudal fin rays and of well-developed ribs and
epipleurals. It is unlikely that these are de novo acquisi-
tions of Triodon, and the ancestry of Triodon should have
these structures. In fact, the Eocene Zignoichthys, ap-
parently closely related to Eoplectus, does have
procurrent caudal fin rays, indicating that these were
simply lost by Eoplectus after it gave rise to the line
leading to Triodon. It is possible that small ribs and
epipleurals were present on the obscured anterior ab-
dominal vertebrae of Eoplectus, but it still must be
assumed that well -developed ribs and epipleurals were
lost by Eoplectus only after it gave rise to the Triodon-
like line. Thus, E. bloti must be considered a slightly
specialized member of the Eoplectinae line which gave
rise to the Triodon line, and not the immediate direct
ancestor of Triodon.

The number of dorsal and anal fin rays in Eoplectus is
in the high part of the range for triacanthodids, but the
number of basal pterygiophore supports is in the low part
of the range, and even lower for the dorsal pterygio-
phores. The fact that there are many more rays than
pterygiophores, especially in the posterior regions of
these fins, may indicate a prelude to an eventual reduc-
tion in the number of rays from posteriorly to anteriorly
in the series leading to the Triodon condition. This
reduction in number of rays would also tend to lengthen
the caudal peduncle, which would probably continue to

lengthen by posterior elongation as less reliance was
placed on the soft dorsal and anal fins for locomotion and
greater emphasis was placed on the caudal fin as it
gradually became forked and more heavily muscled in
the tapering peduncular region. Perhaps as it became a
more sustained rapid swimmer there was less advantage
to having a large defensive spiny dorsal fin, which even-
tually was reduced to a rudimentary structure, such as in
Triodon, in which it is sometimes absent altogether.

The spiny dorsal fin of Eoplectus has one important
characteristic which may shed light on the phylogeny of
the other scleroderras, and that is the placement of the
second basal pterygiophore, perhaps seemingly trivial.
The long shaft of this pterygiophore is oriented distinctly
anteroventrally and articulates either between the neural
spines of the first and second vertebrae or between the
neural spine of the first vertebra and the base of the
skull just behind the shaft of the first pterygiophore. This
is a decidedly nontriacanthodid arrangement, in which
the shaft of the second pterygiophore is otherwise always
directed ventrally or posteroventrally and articulates
with the neural spine of the fourth vertebra. In direct
contrast to the triacanthodids, the short shaft of the sec-
ond pterygiophore of triacanthids is directed slightly to
distinctly anteroventrally toward the region between the
neural spines of the first and second vertebrae. This is
similar to the situation in Eoplectus, except that the
shaft as well as the rest of the pterygiophore is of much
reduced size, as could be expected by the much smaller
size of the spine it bears in triacanthids relative to
Eoplectus. A forward shift in the orientation of the shaft
of the second pterygiophore (and of those posterior to it)
occurred somewhere in the line of evolution between the
ancestral triacanthodids and the derived triacanthids,
and Eoplectus shows such a shift of the second pterygio-
phore (but not of those posterior to it).

It is entirely possible that a pre-£op/ectus-like fish,
before the jaws and dentition had become 7>wdon-like,
was on the line of hollardiin triacanthodids that gave rise
to the triacanthids, undoubtedly through a line of evolu-
tion including the Eocene Protacanthodes, the most
generalized or triacanthodidlike of all the triacanthids.
Protacanthodes has a reduced spiny dorsal fin whose
base is much shorter than that of the long-based soft dor-
sal fin, an apparently shaftlike pelvis with a well-
developed spine and other features typical of triacan-
thids, while retaining from its triacanthodid ancestry a
deep and only slightly tapered caudal peduncle and a
large well-rounded caudal fin.

A pre-Eoplectus-like fish with: 1) a hollardiin domed
posterodorsal region of the skull; 2) shaftlike pelvis;
3) generalized triacanthodid jaws with well-developed
ascending premaxillary processes and discrete conical
teeth in sockets; 4) approximately one to one ratio of dor-
sal and anal fin rays to their basal pterygiophores; 5) uro-
neurals; 6) full-scale covering; 7) only moderately
developed parasphenoid ventral flange; 8) 8 -f 12 ar-
rangement of vertebrae; etc., could be on a line ancestral
to the beak -jawed Eoplectus and hence to Triodon and
the other gymnodonts on the one hand, and on the other

hand to the line ancestral to the Recent hoUardiins,
from which latter line early arose the Eocene Protacan-
thodes and its decendent Recent triacanthids.

This hypothetical generalized triacanthodid ancestral
line of pre-Eoplectus-like configuration could well have
diverged into two distinct radiations: one retained dis-
tinct well-developed individual teeth protruding from
sockets in the jaws, as well as relatively well-developed
spiny dorsal and pelvic fins, and the 8-1-12 vertebral
arrangement; the other specialized the jaws and denti-
tion into a crushing beak, while eventually losing the
spiny dorsal and pelvic fins and converting to a 9 -(- 11
vertebral arrangement.

The evolution of Protacanthodes from a conically
toothed pre-£op/ectus-like line would involve: 1) an in-
crease in the number of soft dorsal fin rays and of their
basal pterygiophores; 2) an elongation of the soft dorsal
fin base concomitant with a reduction in the size of the
second and subsequent dorsal spines and of their basal
pterygiophores; 3) a shortening of the spiny dorsal fin
base; 4) a slight elongation of the caudal peduncle; 5) a
reduction in the number of pelvic fin rays and of their
size. All of this is well within the range of possibility.
Somewhere along the line between the pre-Eoplectus-
like form, Protacanthodes, and the Recent triacanthids,
the enlargement of the prefrontal and its anterior exten-
sion alongside the ethmoid and vomer, and other such
typical triacanthid specializations, could have taken
place, while the caudal peduncle structure and caudal fin
shape would not have changed until between the
Protacanthodes and Recent triacanthid stage.

The configuration of the first three basal pterygio-
phores of the spiny dorsal fin in Eoplectus is also of in-
terest to the phylogeny of the balistids, for it in some
ways represents what one might expect a balistid
ancestral group to have, at least in this one respect. That
is, there is a large open space between the second
(anteroventrally directed shaft) and third (postero-
ventrally directed shaft) pterygiophores. The third
pterygiophore already seems to be acting partially as a
strut supporting the posterodorsal end of the second
pterygiophore and has the ventral end of its shaft in con-
tact with the neural spine of the fifth vertebra, as in
balistids. If the first two pterygiophores in Eoplectus had
greatly reduced ventral shafts so that the first pterygio-
phore articulated high on the rear of the skull and the
second pterygiophore to the posterior end of the first,
with the latter concurrently becoming anteriorly elongate
to reach the rear of the skull, then a prototype of the
balistid structure is achieved. The spine of the third
pterygiophore would become lost and the sole function of
this pterygiophore would be to brace the developing
specialized carina accommodating the complex locking
mechanism of the first two spines, with the third spine,
borne on the small second pterygiophore forming the
rear end of the carina, becoming reduced in size. At the
same time that these transformations would be taking
place, the fourth and fifth pterygiophores and their
spines would also become reduced in size and eventually
entirely lost. Such is a reasonable scenario for the de-

velopment of the balistid spiny dorsal fin apparatus.

The above hypothesis in no way implies that Eoplectus
is directly ancestral to the balistids, for the balistids
clearly are derived from triacanthids. However, fishes
like the Recent triacanthids are too specialized to be con-
sidered as the ancestral group to the balistids, and it is
simply suggested that balistids could have evolved from
an early line of Protacanthodes-like triacanthids in the
Eocene still having an Eoplectus-hke arrangement of the
first three basal pterygiophores of the spiny dorsal fin.

The Spinacanthinae appear to me to represent an ex-
tinct lineage of triacanthodids not ancestral to any of the
Recent groups of plectognaths, although with our present
negligible knowledge of the internal anatomy of the two
species of this subfamily, almost any phylogenetic state-
ment about them is mostly speculation. If the
Spinacanthinae were evolutionary dead ends, this sub-
family of triacanthodids joins the Cryptobalistinae sub-
family of triacanthids as the only two groups of fossil
plectognaths not clearly on evolutionary lines leading to
Recent groups of plectognaths. The reasons for con-
sidering the Cryptobalistinae as an evolutionary dead
end are discussed at length by Tyler (1968:243-249) and
for the Spinacanthinae below.

The Spinacanthinae seem to have been an unsuccess-
ful experiment in specialization from a more normal
basal triacanthodid configuration. These specializa-
tions include: 1) the enormous elongation of the dorsal
fin spines and increased length of the fin base; 2) the
reduction in number of fin rays in the soft dorsal and
anal fins and the shortening of the fin bases; 3) the ex-
tremely high placement of the small eye in the head;
4) the forward migration of the spiny dorsal fin origin to
above the eye; 5) the probably reduced size of the pelvic
fin spine, if present at all; 6) the steep profile of the
snout; 7) the immense size of the body of one of the
species; 8) the development of enormous more or less
hexagonal scale plates forming a complete carapace over
the anterior part of the body behind the head in one of
the species.

The trend for the elaboration of the dorsal fin spines
and lengthening of its base concomitant with a reduction
in the number of soft dorsal and anal fin rays and a
shortening of their bases is not compatible with a possi-
ble ancestry of the triacanthids or balistids, both of
which have reduced the spiny dorsal fin and elaborated
the soft dorsal and anal fins. The position of the eye and
the origin of the spiny dorsal fin in spinacanthins are,
respectively, higher and further forward than in any
triacanthids or balistids, and the snout is much steeper
than in those two families. Additionally, there is good
evidence that the hollardiin triacanthodids gave rise to
the triacanthids through a Protacanthodes-\ike line, and
that the balistids were derived from the early triacanthids
before they became as specialized as the Recent species.

All of this would seem to me to exclude the
spinacanthins from further consideration as possible
ancestors of the triacanthids and balistids, and, even
neglecting the role of the eoplectins as the ancestors of
the gymnodonts, there is nothing compelling in the com-

position of spinacanthins to think of them as possibly
related to the gymnodonts, other than that the pelvic fin
may have been reduced in size or absent and that the soft
dorsal and anal fins are reduced in size.

This leaves but one group of plectognaths to be con-
sidered as hypothetically derivable from spinacan-
thins: the ostracioids. Ostracioids, like spinacanthins,
have short-based soft dorsal and anal fins and have com-
pletely lost the pelvis and pelvic fin, while in spinacan-
thins the pelvic fin may have been of reduced size or ab-
sent, and the condition of the pelvis is unknown. The
scales in one of the two genera of spinacanthins, Proto-
balistum, are large thick hexagonal plates which form a
continuous cuirass around the anterior half of the body
behind the head, the plates interdigitating along their
edges of contact and having surface granulations. In os-
tracioids the plates of the carapace are remarkably
similar to those of Protobalistum, although they cover
the head as well as most of the body. It is easy to envision
a transition from the mostly isolated and approximately
circular scale plates of Spinacanthus, with surface
granulations and a higher central spine, to the hexagonal
carapace plates in the partial body cuirass of
Protobalistum, with surface granulations and interdigi-
tated edges, to the fuller carapace covering of ostracioids.
While the teeth of Protobalistum are too large and
heavy to be conveniently ancestral to those of os-
tracioids, the smaller more conical teeth of Spinacanthus
could easily be ancestral to those of ostracioids. The
profile of the snout of some ostracioids is as equally steep
as in spinacanthins, and the eye of ostracioids, although
somewhat larger than in the two species of
spinacanthins, is located high in the head. However, it is
difficult to compare eye position between two groups in
which one has a spiny dorsal fin and the other does not.
Weighing against any phylogenetic significance to
these similarities between spinacanthins and ostracioids
are three factors: 1) the presence of an enormously well-
developed spiny dorsal fin in spinacanthins, which one
would expect to be at least no larger than in other
triacanthodids if not of greatly reduced size in a line
leading to the ostracioids; 2) enlarged scale plates,
whether or not with granulations and hexagonal pat-
tern, are not unique to Protobalistum and ostracioids
among the plectognaths, but are found in some
tetraodontids and molids as well, and the scales of
balistids can easily be envisioned as ancestral to those of
ostracioids; 3) the osteological evidence from Recent
species indicates that the ostracioids are derived from
the same ancestral line as that which gave rise to the
balistids, while it is discussed above why the spinacan-
thins are unlikely to be ancestral to the triacanthids and
balistids.

The low number of fin rays and short base of the soft
dorsal fin in spinacanthins is probably correlated with
the trend for the enormous enlargement of the spiny dor-
sal fin, while for hydrodynamic reasons the apposed anal
fin would similarly become reduced in size to match its
equivalent above it. Thus, the short-based soft dorsal
and anal fins in spinacanthins could have no phy-

logenetic relationship to the short-based soft dorsal and
anal fins of ostracioids. Nevertheless, considering the
reductive tendencies in the spiny dorsal fin in the plec-
tognaths, it is not inconceivable that a group like the
spinacanthins at one time early in their history greatly
enlarged the spiny dorsal fin, an experiment which even-
tually proved a failure and gave way to the more usual
trend of reduction, leading to a spinacanthin line with
the spiny dorsal fin reduced in size and eventually lost,
leaving a hypothetical form much more suitable for con-
sideration as an ancestral line to the ostracioids than are
the two presently known species.

The scales in the tetraodontid Ephippion guttifer are
relatively normal prickles in the young, with two or three
radiations from the basal plate supporting the projecting
spine. By about 100 mm SL, however, only the prickles
on the belly retain the normal shape, those of the body
having the basal plate enlarged and elongate, and the
projecting spine reduced in size. By 200 mm SL the basal
plates of the scales between the levels of somewhat
behind the pectoral fin base and the soft dorsal and anal
fin origins are further enlarged into irregularly rounded
plates, with irregularly granular surfaces, closely held
together by numerous interdigitations along their edges
of contact. The scale plates anterior and posterior to the
girdle of sutured plates are progressively more distantly
spaced from one another and elongate rather than round-
ed. The scales of the belly, below the carapace girdle, re-
main as normal prickles. At larger sizes, as illustrated for
a 325 mm specimen, the scales of the girdle become
progressively larger and thicker and the girdle itself ex-
tended further anteriorly and posteriorly. The surface
sculpturing of the plates becomes more regularly
granular and the firmness of the interdigitation between
the plates along all their edges of contact stronger and
less flexible, the largest and heaviest plates mostly in the
ventral region of the girdle. The plates have relatively ir-
regular outlines, but many range from triangular to hex-
agonal.

This girdle of interdigitated thickened scale plates in
Ephippion adults is fully as strong and solid as that of
ostracioids, and only slightly more flexible, mainly
because it does not completely enclose the body, the bel-
ly always retaining normal prickles, and the girdle not
extending onto the head. In four specimens, the size of
the largest scale plates increased as follows: 3.6% SL at
101 mm; 4.3% SL at 232 mm; 7.1% SL at 325 mm; 9.2%
SL at 391 mm. In ostracioids the largest scale plates
range from about 6 to 13% SL relatively independent of
specimen size beyond juvenile stages.

Among the molids the individual scale plates are
relatively small. In Mola and Masturus the basal plates
are more or less rounded and flexibly articulated with
one another by delicate denticulations. In Ramania,
however, the basal plates become thicker with in-
creasing specimen size and the extent of the interdigita-
tion also increases so that adults have nearly the entire
body covered by an only slightly flexible carapace of
small (in two specimens, 0.8% SL at 65.1 mm and 1.2%
SL at 493 mm) minutely denticulated and irregularly

geometrically arranged basal plates with only low and in-
conspicuous surface sculpturing, ranging from 3 to 12
sided, but with some distinctly hexagonal. This carapace
of scale plates never attains the thickness of that of os-
tracioids or Ephippion.

In balistids the thick scale plates are usually rhom-
boidal with slightly overlapping edges (plates not over-
lapping in Canthidermis) and usually with low surface
sculpturing of wide variety, while some plates (es-
pecially on the caudal peduncle) may bear higher spiny
processes. Although usually rhomboidal, the anterior
and posterior apices sometimes are flattened, the plate
then having a decidedly hexagonal shape. Scale plates
such as found in balistids could easily be ancestral to
those of ostracioids, by an increase in thickness and sur-
face area, more usual hexagonal shape, and conversion
from slightly overlapping edges to apposed inter-
digitated edges.

The strongest reason, however, for not considering the
spinacanthins as ancestral to the ostracioids is that the
osteological evidence from Recent species indicates that
the balistids and ostracioids evolved from a common
triacanthid ancestral stock, as discussed fully under the
relationships of the ostracioids, but basically because
balistids and ostracioids have a well-developed prootic
shelf, the hyomandibular supported by the prootic and
pterotic but not by the sphenotic, an expanded
parasphenoid, and similarities in the shape or size of the
interoperculum, operculum, suboperculum, premaxil-
lary, and maxillary.

Thus, for the present at least and until more is known
of the internal anatomy of the spinacanthins, it seems to
me reasonable to tentatively consider the spinacanthins
an extinct line of early triacanthodid experimentation
not ancestral to any of the other known subgroups of
plectognaths.

In summary, there is good evidence that the HoUardi-
inae are ancestral to the Triacanthidae and the Eoplec-
tinae to the Tetraodontoidei, while probable that the
Triacanthodinae are derived from the hollardiins and
possible that the Spinacanthinae were an evolutionary
dead end.

In fact, the Eoplectinae are so obviously ancestral to
the Gymnodontes that the eoplectins could be con-
sidered as the most basal family of that suborder, even
more generalized than the Triodontidae, rather than as a
subfamily of the Triacanthodidae with highly specialized
jaws. Similarly, the Protacanthodinae are so obviously
ancestral to the Recent Triacanthidae and intermediate
between them and the Triacanthodidae that the
protacanthodins could be considered a specialized sub-
family of the Triacanthodidae rather than as the most
generalized subfamily of the Triacanthidae. It is on the
basis of my own personal assessment of the overall
similarities and differences between most triacan-
thodids, Eoplectus, Zignoichthys, and triodontids and
between triacanthodids, Protacanthodes, and most
triacanthids that I subjectively feel that Eoplectus and
Zignoichthys are best placed in the Triacanthodidae and
Protacanthodes in the Triacanthidae.

Family Triacanthidae

Comparative dia^osis (contrast with that of the
Triacanthodidae) (modified from Tyler 1968:234-
236). — A compact, strongly sutured skull and a mus-
cular body built for active swimming near the bottom;
very little cartilage visible on the external surface of the
skull between the regions of apposition of the bones in
the otic and occipital regions, the bones usually strongly
sutured to one another along all edges of proximity; an
outer series of about 8 to 10 heavy incisor teeth in each
jaw, internal to which are several more or less molariform
teeth, usually four (two in Trixiphichthys weberi) in

Figure 31.— Typical body form in the Recent
Triacanthidae: Paeudotriacanthus strigilifer.

the upper jaw and two in the lower; a large sturdy eth-
moid-frontal complex for the support of the massive den-
tition; the prefrontal with a long anterior extension,
sutured medially to the ethmoid and anteriorly to the
posterolateral extensions of the vomer; premaxillary
pedicels when fully retracted reaching only to the an-
terior basal region of the ethmoid, well-separated from
the frontals (probably least so in the Eocene
Protacanthodes); supraoccipital domelike, its posterior
surface concave; epiotics separated on the dorsal surface
of the skull by the supraoccipital and meeting medially
only on the posterior surface of the skull and ar-
ticulating anterolaterally with the frontals; ptero-
sphenoids meeting and suturing in the midline of the
posterior wall of the orbit; neural processes of the first
vertebra meeting and suturing in the midline above the
neural canal, forming with the exoccipitals a completely
enclosed bony well (with a bony bottom) in which the
shaftlike end of the first basal pterygiophore of the spiny
dorsal fin is immovably held (a different arrangement
present in the Oligocene Cryptobalistes); parasphenoid
in region of orbit with a dorsal arch and a well-developed
ventral flange below the orbit, the flange about 2 or 3
times as deep as the upper shaftlike portion of the bone
(except in the Eocene Protacanthodes, the shaft relative-
ly straight and the flange not much deeper than the
shaft); hyomandibular with a well-developed groove and
crest along its lateral surface more or less transversely;
pterotic with a ventral process overlying and firmly ar-
ticulating with the upper posterior portion of the
hyomandibular; supracleithrum placed vertically to the

horizontal axis of the skull, nearly its entire length over-
lying the cleithrum; mesopterygoid absent (at least in
adults) in all but one species {Trixiphichthys weberi),
and when present small and difficult to distinguish from
the metapterygoid, to which it is mostly fused; olfactory
cavity between the ethmoid and prefrontal well defined,
with distinct bony boundaries; basihyal absent; lower
two branchiostegal rays enlarged (except in the Eocene
Protacanthodes), much wider than the upper branchios-
tegal rays; operculum elongate, usually widest in the
middle, not triangular; air bladder thick walled,
somewhat elongate, large and extending posteriorly al-
most the entire length of the abdominal cavity; pelvis a
sturdy shaft; the two halves of the pelvis variously fused
or extensively interdigitated, like a railroad rail in cross
section just behind the level of the pelvic spine (except in
the Oligocene Cryptobalistes, which has a basinlike pel-
vis); a single oblique crest on the side of the pelvis at the
level of the flange of the pelvic spine, allowing for two,
and only two, positions of locking the erected spine
(probably a different arrangement in the Oligocene
Cryptobalistes); the ventrolateral surface of the pelvis at
the base of the spine without a complete foramen and the
two sides of the base of the spine not meeting medially;
all haemal arches and spines of adults fully fused to their
centra; only the epural and uppermost hypural as
separate elements articulated by fibrous tissue to each
other and to the centrum, the other hypurals and the
parhypural fully fused to one another and to the cen-
trum (probably some separation of the middle hypurals
in the Eocene Protacanthodes); a single pair of
uroneurals present; thick epipleurals present from the
third or fourth abdominal vertebra to the first or second,
sometimes the third and rarely the fourth, caudal
vertebra; fifth basal pterygiophore of the spiny dorsal fin
usually absent, rarely present as a tiny buried bony
splint beneath the usually rudimentary sixth dorsal
spine; first basal pterygiophore of the spiny dorsal fin a
stout shaft (except very short in the Oligocene Crypto-
balistes) with poorly developed anterior and posterior
medial flanges, the flanges never wider than the shaft;
first basal pterygiophore of spiny dorsal fin with a medial
flange dorsally not completely enclosing a foramen; sec-
ond to fourth basal pterygiophores of spiny dorsal fin
reduced in size, their shafts not reaching ventrally to
between the tips of the neural spines (but associated with
those of the first to third vertebrae, except in the Eocene
Protacanthodes, in which the shafts reach to the distal
ends of the neural spines of the second to fourth
vertebrae) and the fourth pterygiophore occasionally
missing; none of the basal pterygiophores of the spiny
dorsal fin sutured to one another distally; spiny dorsal fin
base (including rudiments) much shorter than soft dor-
sal fin base (except the Eocene Protacanthodes, with
spiny dorsal base slightly longer than soft dorsal base);
only four or five (six in the Eocene Protacanthodes and
perhaps in the Oligocene Cryptobalistes) abdominal
vertebrae anterior to the first basal pterygiophore of the
soft dorsal fin; first anal fin basal pterygiophore with a
sturdy medial flange anterior to its lateral flanges, the

pterygiophore + -shaped in cross section; well developed
lateral flanges for muscle attachment present along most
of the length of most of the basal pterygiophores of the
soft dorsal and anal fins; a sturdy lateral flange present
horizontally along the middle of the posterior half of the
last centrum and the anterior region of the hypural
plate; soft dorsal fin basal pterygiophores 20-26, anal fin
basal pterygiophores 14-19, most of these pterygiophores
sutured to one another distally; distal pterygiophores of
soft dorsal and anal fins often unossified, the ossifica-
tion, when present, always as a single piece and not as a
pair of elements to either side of the midline; sixth dor-
sal spine nearly always present as a rudiment, in all but
very young specimens usually as a splint of bone buried
beneath the skin, but sometimes protruding to the sur-
face even in adults; the fifth spine usually very short, but
nearly always protruding at least a short distance
through the skin; the first four spines always visible ex-
ternally, but the fifth and sixth spines sometimes absent
in one Recent species and perhaps in several fossil forms;
dorsal fin rays 19-26; anal fin rays 13-22; pelvic fin with a
large spine followed in some species by a single ray which
becomes a buried rudiment in adults; adults never with a
protruding pelvic ray; dorsal and pelvic spines with
numerous shallow lengthwise grooves (except in the
Miocene Marosichthys, in which the grooves seem to be
deep), mostly obscured by the overlying scale plates ex-
cept at the naked distal end (one-tenth or less of the
length); uppermost pectoral fin ray nubbinlike, the two
halves of the ray about equally short or the lateral half
shorter, the basal region of the medial half always im-
mensely larger than that of the lateral half; the slightly
overlapping basal plates of the scales of the body bearing
either an emarginate cruciform ridge or an anterior to
posterior series of vertical emarginate ridges (except in
the Eocene Protacanthodes, with numerous upright
spinules, and in the Oligocene Cryptobalistes, without
elaborate ornamentation); peritoneum pale, often with
silvery overtones; coloration basically silvery-gray, with
yellowish or greenish overtones; lateral line relatively
conspicuous (least so in Trixiphichthys weberi); scaly
skin forming a definite low sheath along the bases of the
soft dorsal and anal fins; olfactory lamellae 23-54,
relatively thin; anterior nostril with an upraised rim pos-
teriorly, essentially flush with the surface anteriorly, the
rim not tubelike; posterior nostril with an upraised rim
anteriorly; gill rakers laterally on first arch relatively
short, shorter than the width of the fleshy arch; usually
only one, rarely two, rakers laterally on the upper limb of
the first arch above the angle; caudal peduncle relatively
long, 16 to 31% SL, distinctly tapering to a narrow trans-
versely indented region above and below just in front of
the caudal fin base (except in the Eocene
Protacanthodes, which has a relatively deep and only
slightly tapered peduncle); least depth of the caudal
peduncle, between the precaudal grooves, 2 to 5% SL (ex-
cept in the Eocene Protacanthodes, the depth 15% SL,
and in the Oligocene Acanthopleurus collettei, the depth
5 to 8% SL, versus 4 to 5% SL in the Oligocene A. ser-
ratus and 2 to 4% SL in Recent species); the caudal

peduncle wider than deep at this point (except in A. col-
lettei, the width probably about equal to the depth, and
in the Eocene Protacanthodes, the width unknown but
obviously less than the depth); caudal fin deeply forked
(except in the fossil genera Protacanthodes, with a
rounded caudal, and Cryptobalistes, with an almost
truncate caudal).

Figure 32.~Pseudotriacanthus strigilifer:

showing the course of the lateral line; lower

left, scales from upper middle region of

body, including two lateral line canal

bearing scales: lower right, nasal

region as seen externally (above) and

the olfactory lamellae as seen with the top

of the nasal sac removed.

Detailed description o{ Pseudotriacanthus atrigilifer.

Material examined. — Five cleared and stained speci-
mens, 79.0-145 mm.

Occipital Region.

Basioccipital. — A short column, dorsolateral^
expanded; cartilage filled along its anterior and antero-
dorsal edges; articulates by extensive interdigitation
posterolaterally with the exoccipitals, anterolaterally
with the prootics and anteroventrally with the overlying
posterior end of the parasphenoid. The rim of the round
concave posterior end of the basioccipital articulates by
fibrous tissue with the rim of the concave anterior face of
the centrum of the first vertebra. A deeply concave
medial channel is present on the anterior three-fourths of
the ventral surface of the basioccipital. The anterior end
of this channel is mostly hidden from view by the overly-
ing parasphenoid. Posteriorly this channel is open to the

exterior at the base of the posterior bifurcation of the
parasphenoid, while anteriorly it opens into the
myodome where the anterior end of the basioccipital
forms the posterodorsal and posterolateral walls of the
myodome.

Exocclpital. — Cartilage filled at its dorsal, lateral,
and ventral edges; articulates by interdigitation dorsally
with the epiotic, anteroventrally with the prootic, ventro-
medially with the basioccipital and laterally with the
pterotic, while dorsomedially it interdigitates with its op-
posite member above the foramen magnum. The
foramen is entirely surrounded by the exoccipitals, ex-
cept ventromedially where it is bounded by the upper
surface of the basioccipital. Posteromedially the exoc-
cipitals interdigitate with the neural spine of the first
vertebra. An exoccipital condyle is present only in the'
form of a slightly posteriorly elongate region of the
posteroventral end of the exoccipital, the short condyle
overlying the anterolateral surface of the lower neural
arch region of the vertebra just above the region of the
centrum. The neural processes from either side of the
first vertebra meet and interdigitate in the midline above
the neural canal, forming, in conjunction with the exoc-
cipitals to which they are also sutured, a completely
enclosed bony well in which the shaftlike end of the first
basal pterygiophore of the spiny dorsal fin is immovably
held by fibrous tissue. The bony bottom of this well is
formed mainly by the exoccipitals where they meet
medially to roof over the foramen magnum.

Supraoccipital. — Dome-shaped; its stout rounded
anterodorsal edge forming the apex of the cranium and
its posterior surface concave; cartilage filled along most
of its ventral edges; articulates through interdigitation
posteroventrally with the epiotics and anteroventrally
with the frontals, the frontals overlying the anterolateral
edges of the supraoccipital.

Otic Region.

Pterotic. — Cartilage filled along most of its edges
of articulation with the other cranial bones; articulates
through extensive interdigitation posterodorsally with
the epiotic, posteromedially and ventromedially with the
exoccipital, anteromedially with the prootic, and antero-
laterally and anterodorsally with the sphenotic. The
anteroventral end of the pterotic is the main support for
the hyomandibular, its extreme anteroventral edge abut-
ting against the rear half of the dorsal end of the hyoman-
dibular while a sturdy ventral flange from its ventro-
lateral edge broadly overlies the posterodorsal region of
the lateral surface of the hyomandibular. While the ar-
ticulation between the pterotic and hyomandibular is
through fibrous tissue, it is very firm and relatively im-
movable, especially so because of the presence of the ven-
tral flange of the pterotic broadly overlying the hyoman-
dibular. Much of the posterolateral surface of the
pterotic is overlaid by the posttemporal, to which it is
firmly interdigitated.

Sphenotic. — Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates by
extensive interdigitation anteromedially with the
prootic, pterosphenoid, and frontal in the posterior wall
of the orbit, respectively from below to above. Laterally
on the skull the sphenotic articulates by interdigitation
dorsally with the frontal and epiotic and posteriorly with
the pterotic. Posterodorsally the sphenotic inter-
digitates for a short distance with the anterodorsal edge
of the posttemporal. Anterolaterally the sphenotic helps
form the concave groove that supports the dorsal edge of
the hyomandibular. The sphenotic forms the antero-
lateral part of the groove, the anteromedial part being
formed by the prootic and the posterior half by the
pterotic. The articulation of all these bones with the
hyomandibular is through fibrous tissue.

Epiotic. — Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates
through extensive interdigitation posterodorsally with
the supraoccipital and anterodorsally with the frontals.
Medially the edges of the two epiotics interdigitate with
one another, while posteroventrally they interdigitate
with the exoccipital and pterotic. Anteroventrally the
epiotic interdigitates with the posterodorsal region of the
sphenotic, while the extreme posterolateral edge of the
epiotic is broadly overlain by the upper end of the post-
temporal, with which it interdigitates. Ventromedially
on the posterior surface the epiotic helps form a very
short portion of the upper anterior region of the bony well
supporting the shaft of the first basal pterygiophore of
the spiny dorsal fin, while the main support for this shaft
is the portion of the well formed by the exoccipitals and
neural spine of the first vertebra.

Prootic. —Cartilage filled along all of its edges of
articulation with the other cranial bones, except an-
teriorly; articulates by extensive interdigitation antero-
dorsally with the pterosphenoid, anterolaterally with the
sphenotic, posterolaterally with the pterotic, posteriorly
with the exoccipital, and posteromedially with the
basioccipital. The prootic interdigitates along all of its
length medially with the parasphenoid, while laterally
the prootic helps support the dorsal head of the hyoman-
dibular. The lateral walls of the myodome are formed by
the ventromedial surfaces of the prootics, while the dor-
sal walls of the myodome are formed by the medially
directed horizontal shelves, which are attached to one
another through extensive interdigitation, from the ven-
tromedial surfaces of the two prootics. The prootics also
form the upper two-thirds to three-fourths of the anterior
wall of the myodome.

Orbital Region.

Frontal. — Wider posteriorly than anteriorly where
it tapers to a blunt point above the posterior end of the
ethmoid and medial to the prefrontals; articulates by ex-
tensive interdigitation posteromedially with the supraoc-
cipital, whose anterior edge it broadly overlies, pos-

teriorly with the epiotic, posterolaterally with the
sphenotic, anteriorly with the ethmoid, and antero-
laterally with the prefrontal. Posteromedially on its ven-
tral surface in the rear of the orbit the frontal inter-
digitates medially with the pterosphenoid and ventrally
with the sphenotic. Medially the two frontals are in close
apposition with one another and articulate either
through fibrous tissue or interdigitation, but at one or
more places along their lengths there is usually a small
amount of cartilage visible between the two halves. The
frontal interdigitates anteriorly with the ethmoid and
anterolaterally with the prefrontal, both of which bones
it overlies.

Prefrontal. — Cartilage filled along most of its
broad medial surface except for the long anterior strut
that lies alongside the ethmoid. Posteromedially the
edges of the two prefrontals interdigitate with one
another but anteromedially the prefrontals articulate
through cartilage with the ethmoid, except for the long
anterior strut which lies alongside and interdigitates
with the lateral surface of the anterior half of the eth-
moid, the shaft at its extreme anterior end also inter-
digitating with the posterolateral wing of the vomer.
Posteroventrally the prefrontal articulates through car-
tilage with the dorsal surface of the parasphenoid below.

Parasphenoid. — An elongate shaft with a well-
developed keel along its ventral surface in the region of
the orbit. A deep, anteriorly directed cleft is present in
the posterior portion of the parasphenoid, giving rise to
the forked region which broadly overlies and inter-
digitates with the basioccipital and at the same time
covers over the medial groove on the basioccipital
leading to the rear of the myodome. About two-thirds of
the way back the parasphenoid possesses paired dorso-
lateral wings which interdigitate with the anteroventral
edges of the prootics and thus form the lower part of the
anterior edge of the myodome, while more posteriorly the
parasphenoid forms the floor of the myodome. Posterior
to its articulation with the prootics, the parasphenoid
overlies and interdigitates with the basioccipital. Just
below its region of articulation with the anteroventral
ends of the prootics, the lateral surface of the para-
sphenoid is slightly expanded into a knob, anterior to
which the keeled portion of the bone is slightly concave.
It is to this concavity that the upper elements of the
branchial arches are held by fibrous tissue, especially the
epibranchials and pharyngobranchials. The first
pharyngobranchial or suspensory element articulates
through fibrous tissue to the lateral surface of the para-
sphenoid in this region. In the region of the orbit the
parasphenoid is slightly arched dorsally and the keel is 2
to 3 times the depth of the more rounded shaftlike por-
tion above it. At its extreme anterior end the ventral sur-
face of the parasphenoid is somewhat concave to receive
and interdigitate with the posterior shaftlike portion of
the vomer, while dorsally at its anterior end the para-
sphenoid articulates through cartilage with the pre-
frontals £md ethmoid.

Pterosphenoid. — Cartilage filled along its lateral
edges; articulates by interdigitation dorsally with the
frontals, posteriorly with the sphenotic, and ventrally
with the prootic. Along the middle of its medial edge the
pterosphenoid possesses a medial extension which meets
with that of its opposite member to interdigitate in the
midline.

Ethmoid Region.

Ethmoid. — The ethmoid remains largely car-
tilaginous, the ossification being mostly restricted to the
surface regions. However, the ossification is more ex-
tensive than in triacanthodids. The bone is widest along
the middle of its lateral surface where it interdigitates
with the long forward extensions of the prefrontals.
Posterodorsally the ethmoid interdigitates with the over-
lying anterior end of the frontal. Posterolaterally it ar-
ticulates through cartilage and some interdigitation with
the prefrontals, and ventrally through cartilage with the
dorsal surface of the parasphenoid and shaftlike portion
of the vomer. Anteromedially the ethmoid has a pro-
longation which interdigitates with the posteromedial
region of the dorsal surface of the vomer. The surface of
the medial region of the anterior one-fourth of the eth-
moid is somewhat thickened, this being the region over
which the premaxillary pedicel slides in the opening and
closing of the mouth.

Vomer. — The vomer is expanded laterally at its
anterior end, while its flattened shaftlike posterior ex-
tension fits into and interdigitates with a concave area on
the ventral surface of the petrasphenoid. Posterolaterally
the vomer has a pair of extensions which extensively in-
terdigitate with the anterior ends of the extensions from
the prefrontals alongside the anterior end of the eth-
moid. The dorsal surface of the vomer helps support the
premaxillary pedicel, while laterally the expanded an-
terior region of the vomer articulates by tough fibrous tis-
sue with the medial surface of the middle region of the
palatine. Posteromedially the vomer interdigitates with
the anterior end of the parasphenoid, while the dorsal
surface of the posterior shaft of the vomer abuts against
the ethmoid cartilage above it.

Mandibular Region.

Hyomandibular. — A stout rectangular shaft whose
dorsal end firmly articulates by fibrous tissue with the
groove on the lateral surface of the underside of the skull
formed anterolaterally by the sphenotic, anteromedially
by the prootic, and posteriorly by the pterotic, the latter
bone also having a ventral flange which broadly overlies
the posterodorsal region of the lateral surface of the
hyomandibular. A well-developed crest for muscle at-
tachment is present about midway along the lateral sur-
face of the hyomandibular. Posteriorly the hyoman-
dibular articulates by fibrous tissue along most of its
length to the anterior edge of the upper region of the
preoperculum, just above the end of which latter bone

the hyomandibular articulates by fibrous tissue with the
anterodorsal head of the operculum. Anteriorly the
hyomandibular ends in the cartilaginous plate between it
and the metapterygoid.

Quadrate. — Wide posteriorly, tapering anteriorly
to a head for articulation through fibrous tissue with the
articular of the lower jaw; cartilage filled along posterior
edge; with a long posteriorly directed process from its
posteroventral region which articulates by inter-
digitation and fibrous tissue with the anteroventral edge
of the symplectic in large specimens, whereas in smaller
specimens intimate contact between the symplectic and
posterior extension of the quadrate is not made and the
articulation is purely by fibrous tissue. Along the pos-
terior edge the quadrate articulates through cartilage
with the anterior edge of the metapterygoid, at this
region being cartilage filled. Ventrally the quadrate ar-
ticulates through fibrous tissue with the dorsal surface of
the anterior end of the preoperculum.

Metapterygoid. — A more or less flat plate; cartilage
filled along its anterior, ventral, and posterior edges; ar-
ticulates through cartilage anteriorly with the quadrate,
while along its ventral edge it extensively interdigitates
with the symplectic in large specimens but only ar-
ticulates to it through fibrous tissue in smaller
specimens. Posteriorly the metapterygoid articulates
through cartilage with the hyomandibular.

Symplectic. — Long and rod-shaped; cartilage filled
at its posterior end; articulates dorsally by inter-
digitation (in large specimens, see above) with the
metapterygoid and anteriorly variously through inter-
digitation and fibrous tissue with the posteroventral end
of the quadrate. Posteriorly the symplectic is in contact
with the cartilaginous plate between the metapterygoid
and hyomandibular. The upper end of the epihyal is at-
tached by fibrous tissue to this cartilaginous plate im-
mediately behind the posterior end of the symplectic.
Ventrally the symplectic articulates by fibrous tissue
with the dorsal surface of the preoperculum.

Palato-Pterygoid Region.

Palatine. — Cartilage filled at its extreme posterior
end; articulates by fibrous tissue anteriorly with the dor-
solateral surface of the maxillary, medially with the
lateral surface of the expanded portion of the vomer,
posteroventrally by fibrous tissue to the posterior end of
the ectopterygoid, and directly posteriorly through car-
tilage and fibrous tissue to the anterodorsal region of the
metapterygoid. The posterodorsal wing of the palatine
connects to a broad band of fibrous tissue that runs over
and above the premaxillary pedicels to join with the
posterodorsal wing of the palatine on the other side. Al-
though not sutured to any of the surrounding bones the
palatine is in effect firmly held in place by its extensive
fibrous tissue articulations.

Ectopterygoid. — Somewhat in the shape of an
elongate rectangle; articulates along all of its ventral
edges by interdigitation with the quadrate. Dorsally it
articulates by fibrous tissue with the posteroventral
region of the palatine and posteriorly through cartilage to
the anterodorsal region of the metapterygoid.

Mesopterygoid. — Absent in this and all other species
of triacanthids with the exception of Trixiphichthys
weberi, in which it is a small plate of bone closely inter-
digitated to the posterior third of the dorsal edge of the
far larger metapterygoid.

Opercular Region.

Operculum. — A thin sheet of bone broadest in the
middle and tapering to a point ventrally; with a short
dorsally prolonged flange above its anterodorsal head
which articulates by fibrous tissue with a groove on the
posterior edge of the hyomandibular just above the dor-
sal tip of the preoperculum. Ventrally the operculum
slightly overlies and articulates by fibrous tissue with the
suboperculum.

Suboperculum. — Rounded anteriorly, tapering to a
dorsally directed point posteriorly; articulates by fibrous
tissue dorsally with the operculum and anteriorly with
the interoperculum.

Interoperculum. — Very elongate and flat; rounded
posteriorly but tapering to a point anteriorly; articulates
by fibrous tissue posteriorly with the suboperculum and
anteriorly ligamentously with the angular in the lower
jaw. In the posterior half of its length the dorsal edge of
the interoperculum connects by a band of fibrous tissue
to the epihyal in the region where the epihyal articulates
with the interhyal.

Preoperculum. — Slightly expanded posteroven-
trally in the region below the hyomandibular; the dorsal
edge along the anterior half of the bone slightly thicken-
ed for articulation by fibrous tissue with the quadrate;
articulates dorsally in about the middle of its length
variously by fibrous tissue and through cartilage with the
symplectic, metapterygoid, and anterior end of the
hyomandibular, whereas the main articulation to the
hyomandibular is by fibrous tissue along the posterodor-
sal edge of the preoperculum.

Upper Jaw.

Premaxillary. — L-shaped, with the long dorsal arm
(pedicel) much longer than the ventral arm and movably
articulated by fibrous tissue along the dorsal surfaces of
the vomer and the extreme anterior end of the ethmoid,
allowing for a slight protraction of the upper jaw. When
fully retracted, the pedicel reaches to the anterior basal
region of the ethmoid. The shorter ventral arm of the pre-
maxillary forms the upper two-thirds of the anterior edge
of the upper jaw, the rest of the border being formed by

the maxillary. The premaxillary articulates by fibrous
tissue along the middle of the lateral surface of the pos-
terior arm with the medial surface of the upper portion of
the maxillary, while along the medial surface of its pos-
terior arm the premaxillary articulates through fibrous
tissue with its opposite member. The ventral arm of the
premaxillary articulates through fibrous tissue along its
posteroventral edge with the anterior edge of the maxil-
lary. There are usually five incisorlike teeth, decreasing
in size laterally, borne in shallow sockets along the an-
terior edge of the premaxillary. Behind and internal to
this outer row of teeth there are usually two additional
teeth, the more medial of the two being the largest, of a
size and shape similar to that of the large most medial
tooth in the outer row, and the lateral inner tooth similar
to the small outermost tooth in the outer row. The teeth
are replaced by new ones developing in new sockets
which have pores to the exterior just above and external
to the sockets of the outer row of teeth. After a new tooth
erupts through the surface of the premaxillary, it
gradually migrates slightly forward and downward to the
area of the socket of the older tooth which it is replacing.

Maxillary. — A more or less heavy rectangular
bone, rounded ventrally and less so dorsally, and slightly
constricted above the middle region; articulates by fi-
brous tissue dorsally along the lateral surface of its round-
ed portion with the anterior end of the palatine, while
along the medial surface of its rounded portion it at-
taches to the lateral and ventral surfaces of the posterior
arm of the premaxillary. Ventrally the medial surface of
the maxillary articulates by tough fibrous tissue with the
lateral surface of the posterodorsal part of the dentary,
while along the middle of its anterior edge the maxillary
articulates by similar tissue with the ventral arm of the
premaxillary.

Lower Jaw.

Dentary. — The posterior end concave on the
medial surface to accommodate the anterior end of the
articular, which it broadly overlies laterally but only
slightly overlies medially and to which it articulates by
fibrous tissue; articulates anteromedially by fibrous tis-
sue to its opposite member in the midline and postero-
laterally by interdigitation with the small angular. Along
the posterodorsal region of its lateral surface the dentary
articulates by tough fibrous tissue with the medial sur-
face of the ventral portion of the maxillary. There are
usually five incisorlike teeth borne in shallow sockets
along the anterior edge of the dentary and usually a
single tooth behind and internal to the outer series, the
single inner tooth being of a size comparable to that of
the two larger more medially placed teeth in the outer
series. The lower jaw teeth have the same form as the up-
per jaw teeth and are replaced by teeth developing in the
same manner as described for the premaxillary.

Articular. — More or less triangular in shape;
cartilage filled at its anterior end, where it is continuous

with the remains of Meckel's cartilage; articulates by fi-
brous tissue dorsally, ventrally, and laterally with the
broadly overlying dentary, posteroventrally by inter-
digitation with the angular, and posteriorly at the groove
on its posterior edge with the knob at the anterior end of
the quadrate. The sesamoid articular is a small ossifica-
tion held by fibrous tissue to the region of juncture of
Meckel's cartilage and the anteromedial surface of the
articular.

Angular. — A small block of bone; articulates by
interdigitation dorsally with the articular and anteriorly
with the dentary. Posteriorly the angular connects by
ligament to the anterior end of the interoperculum.

BRANCHIAL APPARATUS.

Hyoid Arch, Branchiostegal Rays, and Urohyal.

Hypohyals. —-Both hypohyal elements well de-
veloped; dorsal hypohyal cartilage filled along its ven-
tral and posterior edges, the ventral hypohyal cartilage
filled at its dorsal and posterior edges. The dorsal and
ventral hypohyals articulate through cartilage with one
another and with the ceratohyal, while they articulate by
fibrous tissue anteromedially with their opposite mem-
bers, and in the case of the dorsal hypohyal with the an-
terior third of the first basibranchial, which is held
between the posterior halves of the medial edges of the
apposed dorsal hypohyals. The ventral edge of the ven-
tral hypohyal articulates by fibrous tissue with the
urohyal. In large specimens there is sometimes a slight
amount of interdigitation between the dorsal and ven-
tral hypohyals.

Ceratohyal. — A large plate expanded posteriorly;
cartilage filled at its anterior and posterior edges; ar-
ticulates through cartilage anteriorly with both of the
hypohyals and posteriorly with the epihyal, the ar-
ticulation with the epihyal sometimes strengthened by
interdigitation in large specimens. The six branchios-
tegal rays articulate by fibrous tissue with the cera-
tohyal; the first two rays along the ventral edge of
the posterior half of the ceratohyal and the last four rays
to the posterodorsal edge of the ceratohyal and, in the
case of the uppermost one or two rays, to the lower lateral
surface of the epihyal.

Epihyal. — Large; cartilage filled at its anterior and
ventral edges; articulates anteriorly through cartilage
with the ceratohyal, this articulation sometimes being
strengthened by interdigitation, while posterodorsally
it articulates by fibrous tissue with the interhyal.

Interhyal. — Short, columnar; cartilage filled at its
dorsal and ventral edges; articulates by fibrous tissue
ventrally with the epihyal and dorsally with the car-
tilaginous plate between the symplectic and hyoman-
dibular immediately behind the posterior end of the sym-
plectic.

Branchiostegal rays. — Six in number; increasing
only slightly in length posteriorly in the series. The
branchiostegal rays are mostly flat plates of decreasing
width posteriorly in the series and articulate anteriorly
by fibrous tissue variously to the posterior end of the
ceratohyal and the lower lateral surface of the epihyal, as
described above. The decreasing width of the rays pos-
teriorly in the series makes the last few rays somewhat
rodlike rather than flat plates.

Urohyal. — Laterally expanded and thickened
along its anterior edge and, to a lesser extent, for a short
distance anterodorsally. Otherwise a flat, thin plate ar-
ticulating by fibrous tissue anterodorsally primarily with
the undersurfaces of the ventral hypohyals.

Branchial Arches. — All of the elements are cartilage -
filled at their edges of articulation with the other
elements of the series, and the articulations are usually
through cartilage and fibrous tissue. The branchial
arches are composed of three basibranchials, three pairs
of hypobranchials, five pairs of ceratobranchials, four
pairs of epibranchials, and four pairs of pharyngo-
branchials. Four gills are present, with a slit between the
fourth arch and the lower pharyngeal.

First arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. First basibranchial
the shortest of the three basibranchial elements, laterally
compressed in the anterior half of its length, where it is
held by fibrous tissue between the medial edges of the
dorsal hypohyal, while posteriorly it articulates mainly
with the anterior end of the second basibranchial and
only secondarily posterolaterally with the first hypo-
branchials. First hypobranchial about the same size as
the second hypobranchial but more rectangular in form;
articulates ventrally with an indented region along the
lateral surface of the anterior half of the second basi-
branchial. First ceratobranchial slightly the longest of
the ceratobranchial elements, which decrease slightly in
size or length posteriorly in the series; a more or less
columnar shaft of bone without a prominent ventral keel.
First epibranchial a more or less rectangular block of
bone except distally where it slightly bifurcates into two
short projections; the anterior projection articulating
with the base of the first pharyngobranchial, or suspen-
sory element, and the posterior projection articulating
with the second pharyngobranchial. First pharyngo-
branchial (suspensory pharyngeal) a short sturdy rod of
bone articulating ventrally with the anterior projection of
the first epibranchial and dorsally by fibrous tissue to the
lateral surface of the ventral keel of the parasphenoid in
the region just below the articulation of the anteroven-
tral edge of the prootic with the parasphenoid.

Second arch.— Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Second basi-
branchial slightly the longest of the three basibranchial
elements, with concave lateral surfaces in the anterior
half of its length for articulation with the ventral ends of

the first hypobranchials; articulates anteriorly with the
first basibranchial, posteriorly with the third basi-
branchial and posterolaterally with the second hypo-
branchials. Second hypobranchial expanded ventro-
medially where it articulates with the posterior end of the
second basibranchial and the anterior end of the third
basibranchial; articulates dorsally with the second
ceratobranchial, which in turn articulates dorsally with
the laterally expanded basal portion of the second epi-
branchial. The second pharyngobranchial is the first of
the tooth bearing pharyngobranchials, having six or
seven teeth, set in deep sockets, in a single row along its
ventral edge. The teeth are laterally compressed and
taper to blunt points distally and are much smaller than
those in the jaws. They are replaced by teeth which
develop in new sockets just anterior to the sockets of the
old teeth, much in the same manner as in the replace-
ment of the teeth in the jaws.

Third arch.— Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Third basi-
branchial articulates anteriorly with the posterior end
of the second basibranchial, anterolaterally with the
second hypobranchials, posterolaterally with the third
hypobranchials, and posteriorly, more distantly, with the
fourth ceratobranchials. Third hypobranchial with an
anteroventral process which articulates by fibrous tissue
with the posterodorsal edge of the urohyal; at its postero-
dorsal end it articulates medially with the posterior end
of the second basibranchial, anterolaterally with the
end of the third ceratobranchial. Third ceratobranchial
articulates dorsally with the third epibranchial, which is
about equal in size to the second pharyngobranchial.
Third pharyngobranchial bears six to seven teeth in a
single row along its ventral edge, similar to those on the
second pharyngobranchial. The three tooth bearing
pharyngobranchial elements (second to fourth) are more
or less closely held to one another by fibrous tissue.

Fourth arch. —Cerato-, epi-, and pharyngo-
branchial elements present. In the absence of a fourth
basibranchial and fourth hypobranchial, the fourth
ceratobranchial articulates ventrally with the posterior
ends of the third basibranchial and hypobranchials.
Fourth epibranchial a more or less rodlike element,
slightly expanded ventrally; articulates ventrally with
the fourth ceratobranchial and dorsally with the small
fourth pharyngobranchial. The fourth pharyngo-
branchial is the smallest of the three tooth bearing
elements, and bears four or five teeth in a single row
along its ventral edge, these teeth being smaller than
those on the second and third pharyngobranchials but of
similar shape.

Fifth arch. —Ceratobranchial (lower pharyngeal)
element only. Fifth ceratobranchial wider in the middle
than anteriorly where it is generally rounded or pos-
teriorly where it tapers to a short shaft, bearing teeth
more or less placed in three rows, with those of the pos-
terior row the largest. All of the teeth tend to be more

84

bluntly rounded than are the teeth of the upper pharyn-
geals. There are up to 15 teeth in the anterior series and
only slightly fewer on the average in the larger posterior
row, with 5 to 8 teeth in the middle row. Ventrally the
fifth ceratobranchial articulates with the base of the
fourth ceratobranchial.

PAIRED FIN GIRDLES.

Pectoral Fin.

Posttemporal. — A straight sheift of bone without
evidence of a forked condition, broadly overlying and
strongly interdigitated to the pterotic in the lower half of
its length and to the epiotic above, and, to a much lesser
extent, with the posterodorsal part of the sphenotic. The
rounded ventral head of the posttemporal articulates by
fibrous tissue with the concave dorsal end of the supra-
cleithrum.

Supracleithrum. — Located more or less vertically
in relation to the axis of the body; articulates by fibrous
tissue dorsally with the ventral head of the posttemporal
and ventrally with the cleithrum, which it broadly over-
lies for its entire length.

Cleithrum. — Laterally expanded along the ventral
two-thirds of its length; articulates by fibrous tissue dor-
solaterally with the overlying supracleithrum, while dor-
somedially it overlies the dorsal one-sixth or one-fifth of
the postcleithrum. Along the middle of its posterior sur-
face the cleithrum articulates by fibrous tissue with the
scapula and the anterodorsal region of the coracoid,
which it slightly overlies and mildly interdigitates with
in larger specimens. Ventromedially the cleithrum ar-
ticulates by tough fibrous tissue with its opposite mem-
ber, while just behind the top of this region the anterior
end of the pelvis is firmly attached to the rear edge of the
cleithra.

Postcleithrum. — The postcleithrum forms a long
posteroventrally directed, sturdy, more or less flattened
shaft from the dorsomedial region of the cleithrum along
the abdominal wall musculature to the region over the
posterior third to one-half of the posterior half of the pel-
vis. The postcleithrum is a single piece without evidence,
even in small specimens, of being divided into an upper
and lower portion. The middle region of the postcleith-
rum is strengthened on the inner surface by a thick shaft.

Coracoid. — Rounded dorsally, tapering ventrally to
a narrow shaft; produced posterodorsally into a prong
below the lowermost actinost; cartilage filled along its
dorsal edge and at the extreme anteroventral end; ar-
ticulates anterodorsally by fibrous tissue and slight inter-
digitation with the posterior edge of the cleithrum, which
slightly overlies it; articulates dorsally through cartilage
with the scapula and by fibrous tissue with the base of
the lowermost actinost.

Scapula. — Completely encloses the scapular
foramen; concave and cartilage filled at its anterior and
ventral edges; articulates anteriorly by fibrous tissue
with the cleithrum which overlies its extreme anterior
region, while ventrally it articulates through cartilage
with the coracoid. Posteriorly the scapula articulates by
fibrous tissue with the following elements, in order from
dorsal to ventral; the first pectoral fin ray borne on a long
knoblike projection and just behind it the small first or
uppermost actinost followed by the second and third ac-
tinosts which articulate along the lower third of the pos-
terior edge of the scapula.

Actinosts. — Four elements; cartilage filled at the
dorsal end and, to a lesser extent, at the ventral end; the
three uppermost actinosts articulating with the scapula,
the lower actinost being supported by the posterodorsal
edge of the coracoid; distally the actinosts support all of
the pectoral fin rays except for the first. The actinosts in-
crease in size from the first to the fourth.

Fin rays. — Usually 13 or 14 fin rays; the first ray
very short, only about one-tenth the length of the second
ray, the medial half much larger than the lateral half and
articulated directly with the scapula rather than with the
actinosts as are the other fin rays; the first two rays and
the last or lowermost ray unbranched, the intervening
rays branched. The small first ray without cross-stria-
tions; all other rays with cross-striations.

Pelvic Fin.

Pelvis. — A sturdy shaft, the two halves either
variously fused or extensively interdigitated together so
that the impression is that of a solid bone like a raiboad
rail in cross section behind the level of the pelvic spine. A
single oblique crest on the side of the pelvis at the level of
the flange of the pelvic spine allowing for two and only
two positions of locking the erected spine. The shaft of
the pelvis tapering to a point posteriorly. Anterior to the
level of the pelvic spine, the anteroventral end of the pel-
vis bifurcates into a concave region for muscle attach-
ment, extending up along the anterior base of the antero-
dorsal shaft of the pelvis. The anterodorsal shaft of the
pelvis articulates by strong fibrous tissue just above the
region of close medial articulation of the two cleithra.

Pelvic spine. — Large and strong; the medial end ar-
ticulating to the pelvis in much the same manner as
previously described for the triacanthodid Parahollardia
lineata. However, in the Triacanthodidae the side of the
pelvis at the level of the flange of the pelvic spine is
either smooth and thus allowing for a single position of
erection of the spine or it has numerous small grooves
allowing for numerous, continuous positions of erection
of the spine, rather than as in the Triacanthidae, ex-
emplified by Pseudotriacanthus strigilifer, in which
there is a single relatively large oblique crest allowing for
only two positions of erection. The rotation of the spine is

the same as described for Parahollardia lineata, the dif-
ference being only that when the erected spine is return-
ed to its unerected position it can have its flange, after
the latter is out of contact with the ventrolateral edge of
the pelvis, stopped by the oblique groove on the lateral
surface of the pelvis, permitting the second or partial
position of erection of the spine.

Fin rays. — No pelvic fin rays have been found in
any of the cleared and stained specimens or in other
specimens examined closely in the pelvic axil but not
cleared and stained. However, this is perhaps partially
attributable to the relatively large size of the specimens
examined, as explained in Tyler (1968:290), for a single
rudimentary fin ray is present in the young of some
species of triacanthids, the ray being resorbed in adults.

VERTEBRAL COLUMN. —All vertebrae with bicon-
cave centra, except the last, which ends posterodorsally
in the urostyle.

Abdominal Vertebrae.

First vertebra. — Neural spine thick, the left and
right halves meeting above the neural canal to come into
close apposition and interdigitation posteromedially;
from this point of meeting the neural spine spreads
anterolaterally to form the posterior half of the walls of
the bony well surrounding the ventral end of the stout
shaft of the first basal pterygiophore of the spiny dorsal
fin, the well also forming a bony roof over the neural
canal in this region. The anterolateral edge of the neural
arch and lower region of the neural spine articulate by
extensive interdigitation with the exoccipital and, in
some cases, with the extreme posteroventral edge of the
epiotic as well, but the articulation is mainly to the exoc-
cipital. Over the anterior third of the neural arch region
just above the centrum, the first vertebra extensively in-
terdigitates with the short and irregularly shaped exoc-
cipital condyles that overlie this region. The rim of the
concave anterior end of the centrum of the first vertebra
articulates with the rim of the concave posterior end of
the basioccipital. Posteriorly the first vertebra ar-
ticulates with the second vertebra by apposition of the
rims of their centra and by the short and bluntly rounded
neural postzygapophysis of the first vertebra slightly
overlying the neural prezygapophyseal area of the second
vertebra. No haemal zygapophyses are present.

Other abdominal vertebrae. — In 21 specimens the
abdominal vertebrae numbered eight. All of the ab-
dominal vertebrae, as well as the caudal vertebrae, have
a bony roof over the neural canal and a single undivided
neural spine. The neural spine of the second abdominal
vertebra is of about the same size as that of the first,
while the neural spines of the third and subsequent
vertebrae are longer than those of the first two ab-
dominal vertebrae and of approximately the same size.
The last abdominal vertebra has its neural prezy-
gapophysis overlying the neural postzygapophyseal area

of the preceding vertebra. No haemal pre- or post-
zygapophyses are present. Each neural arch has a neural
foramen along the middle of its lateral surface. The first
three abdominal vertebrae have no transverse or haemal
processes, but the fourth to last abdominal vertebrae
variously have transverse processes or haemal arches and
spines which bear the intermuscular bones. These trans-
verse processes become more ventrally directed until
they eventually form complete haemal arches. The
processes of the fourth to sixth vertebrae are incomplete,
whereas those of the seventh and eighth vertebrae form
complete haemal arches with very short spines. Eight
epipleurals or intermuscular bones are present in the
myocommata between the epaxial and hypaxial mus-
culature, five borne on the fourth to eighth abdominal
vertebrae, followed in series by three others on the first
three caudal vertebrae. The epipleurals articulate basal-
ly by fibrous tissue with the lateral surfaces of the trans-
verse processes of the fourth to sixth abdominal
vertebrae and of the haemal arches and spines of the
seventh abdominal to third caudal vertebrae.

Caudal Vertebrae. — In 21 specimens the caudal
vertebrae numbered 12. The haemal spine of the first
caudal vertebra articulates by fibrous tissue with the an-
terior surface of the dorsal one-fifth of the first anal fin
basal pterygiophore. The neural spine of the first caudal
vertebra is similar to that of the preceding vertebra, but
the other caudal vertebrae have progressively slightly
less long neural spines. The haemal spines similarly
decrease slightly in size from the second caudal vertebra
to the penultimate vertebra, which has its neural and
haemal spines of slightly increased length. The neural
prezygapophyses of the more Euiterior of the caudal
vertebrae overlie the neural postzygapophyseal area of
the preceding vertebrae. The haemal spines of the caudal
vertebrae support the anal fin basal pterygiophores by
fibrous tissue but do not make close contact with them.
The same is true of the neural spines of the abdominal
and caudal vertebrae that support the basal pteryg-
iophores of the soft dorsal fin. With the exception of the
last vertebra, each neural arch has a completely en-
closed neural foramen along the middle of the neural
arch region, and all have a bony roof over the neural and
haemal canals. All of the haemal arches and spines are
fused to their centra.

Caudal Skeleton. —The parhypural and all but the
uppermost hypural are indistinguishably fused to the
last vertebral centrum. There is a free epural and a uro-
neural represented by a single solid piece, the two halves
of the uroneural apparently having fused together. The
low neural arch of the vertebral centrum bifurcates an-
teriorly but comes together in the midline over the neural
canal posteriorly. It is to this anterior bifurcate region
that the ventral end of the epural articulates by fibrous
tissue. The single free hypural is rodlike and articulates
along the oblique upper anterior edge of the fused hy-
pural plate, the anterior end of the free hypural being
just behind the posterior end of the short neural arch of

the last vertebral centrum. The uroneural is located
between the epural and free hypural and has a tube
through its length enclosing the neural canal. The tube
also extends through the distal region of the epural to
exit near to the tip of the neural spine of the penul-
timate vertebra. A large foramen just below the middle of
the fused hypural plate represents the region of fusion of
the parhypural with the hypurals above it, but other-
wise the hypurals (except for the uppermost free ele-
ment) and the parhypural are indistinguishably fused
with the last vertebral centrum. As explained by Tyler
{1970b: 10) the single uroneural probably represents the
second pair of uroneurals of triacanthodids, in which the
first of the two pairs usually is much smaller and more
variable in occurrence than is the second pair. A promi-
nent horizontal crest is present on the fused centrum-
parhypural-hypural plate for muscle attachment.

Caudal fin rays. —Twelve in number; the upper-
most ray and the lowermost ray unbranched, the others
becoming increasingly branched toward the two middle
rays, which are branched in triple to quadruple
dichotomies. The upper unbranched ray has its bifur-
cate base articulated with the distal end of the free
hypural and the extreme dorsal end of the fused hypural
plate. The uppermost two or three branched rays have
their bifurcate bases overlapping the fused hypural
plate, whereas the middle four to five rays to not overlap
the plate. Similarly, the lower unbranched ray slightly
overlaps the fused parhypural-hypural plate, and the
branched rays above increasingly less so.

DORSAL AND ANAL FINS.

Dorsal Fin.

Spines and pterygiophores. — Six spines of greatly
decreasing length posteriorly in the series and only four
basal pterygiophores, without the intervention of os-
sified distal pterygial elements. The first two spines are
borne on the first basal pterygiophore, whereas the third,
fourth, and fifth spines are borne on their own indivi-
dual basal pterygiophores. There is no basal pterygio-
phore supporting the minute rudimentary sixth spine
which is buried beneath the skin. The concave medial
region of the base of the first spine rotates over a low
medial flange at the dorsal surface of the first basal pte-
rygiophore. However, this medial flange more posteriorly
becomes a tall, somewhat anteriorly directed prong
which fits into a transverse hole through the base of the
spine. It is around this dorsal flange of the pterygiophore
that the first spine rotates. The spine can be erected and
locked in any of a continuous series of positions by the
close apposition of the base of the first spine with the dor-
sal articular surface of the first pterygiophore on either
side of its medial flange, both surfaces being somewhat
rough and irregular. This mechanism is basically the
same as that described for the triacanthodid Parahol-
lardia lineata, although the medial flange of the basal

pterygiophore in the case of triacanthids is much less
elaborately developed than it is in triacanthodids (see
Tyler 1968:fig. 7 for a comparative illustration of the
structure in the two families of triacanthoids). The sec-
ond spine articulates basally with the dorsal surface of
the first basal pterygiophore just behind the posterior
edge of the anterodorsal flange around which the first
spine rotates, at a slight concavity on the surface of the
pterygiophore in this region. This slight concavity has a
very small medial flange fitting into the concave base of
the second spine. The third to the fifth spines articulate
at their generally rounded basal ends with simple shal-
low concavities in the dorsal surfaces of their individual
pterygiophores. The sixth spine is rudimentary and has
no basal pterygiophore supporting it.

The first basal pterygiophore is a very stout shaft of
bone with only narrow anteromedial and posteromedial
flanges for muscle attachment. Basally the stout shaft is
firmly held in place by fibrous tissue within the deep
bony well formed posteriorly by the neural spine of the
first abdominal vertebra and anteriorly by the exoc-
cipitals. This is a relatively immovable articulation. The
second to fourth basal pterygiophores are much smaller
than is the stout first pterygiophore and each bears its
own individual dorsal spine of decreasing length in the
series. The four basal pterygiophores articulate to one
another by fibrous tissue or, in some cases, by a very
slight interdigitation of their apposed surfaces. The sec-
ond to fourth pterygial elements are cartilage filled at
their extreme ventral ends, at the end of the vertical
shaft along their lateral surfaces. The second and third
pterygiophores also often remain cartilaginous at the ex-
treme posterodorsal tip.

Fin rays and pterygiophores. — Usually 21 or 22 fin
rays are present; the first 1 or 2 rays unbranched, the
others branched in single to double dichotomies. No os-
sified distal pterygiophores are present between the
bifurcate bases of the rays in the specimens of P.
strigilifer studied, but in some other species of
triacanthids ossified distal pterygiophores are present, at
least in large adult specimens (see Tyler 1968:268). When
ossified distal pterygiophores are present they are a
single piece in the midline rather than a pair of pieces as
found in triacanthodids. Basally the fin rays are sup-
ported by a number of basal pterygiophores similar in
number to that of the fin rays. The pterygiophores ex-
tensively interdigitate with one another in their upper
regions just below the apposed concavities in their
anterior and posterior edges which give rise to the round
holes between the distal heads of the pterygiophores that
are present at least anteriorly in the series. Lateral
flanges for muscle attachment are very well developed
along the lengths of the pterygiophores. These flanges are
widest toward the distal end of the pterygiophore where
they form prominent hooklike processes. The first basal
pterygiophore of the soft dorsal fin usually is located
between the neural spines of the fourth and fifth ab-
dominal vertebrae, while the last few basal pterygio-
phores lie between the neural spines of the fifth and sixth

caudal vertebrae. The pterygial elements are cartilage
filled at their dorsal and ventral ends.

Anal Fin.

Fin rays and pterygiophores. — Usually 15 or 16 fin
rays are present; the first ray and sometimes the second
ray unbranched, the others branched in single to double
dichotomies. No ossified distal pterygiophores are pres-
ent between the bifurcate bases of the rays. Basally the
fin rays are supported by a number of basal pterygio-
phores equal to or sometimes one less than the number of
rays. The pterygiophores articulate to one another by ex-
tensive interdigitation in a manner similar to that
described for the basal pterygiophores of the soft dorsal
fin and possess holes distally between the apposed
anterior and posterior edges of the pterygiophores, at
least anteriorly in the series. Well-developed lateral
flanges with hooklike processes are present just as with
the basal pterygiophores of the soft dorsal fin. The first
basal pterygiophore of the anal fin is by far the largest of
the series, the others decreasing in length posteriorly in
the series. There is a prominent crest anterodistally on
the first basal pterygiophore of the anal fin in addition to
the lateral flange similar to that which is present on the
other basal pterygiophores. The first basal pterygio-
phore is held immovably by fibrous tissue against the
posterior edge of the haemal spine of the ninth vertebra,
making the latter by definition the first caudal
vertebra. The last several basal pterygiophores of the
anal fin articulate by fibrous tissue between the neural
spines of the fifth and sixth caudal vertebrae, just as the
last few basal pterygiophores of the soft dorsal fin are
held between the neural spines of the same two
vertebrae.

Comparative diagnoses of subfamilies (Protacan-
thodinae, Cryptobalistinae, Triacanthinae). — There
are three subfamilies of triacanthids: two exclusively
fossil (one from the Eocene and one from the Oligocene)
and one with both fossil (Oligocene and Miocene) and
Recent species. These three subfamilies are compara-
tively diagnosed as follows, modified and expanded
from Tyler (1968), especially on the basis of the reexami-
nation of the holotype of the Eocene Protacanthodes.

The Eocene Protacanthodinae have: the pelvis
shaftlike posterior to the region of the pelvic spines and
probably not bifurcate anterior to the spines and of un-
known cross-sectional shape; the spiny dorsal fin base
slightly longer than the soft dorsal fin base; six ab-
dominal vertebrae anterior to the first basal pterygio-
phore of the soft dorsal fin; second to fourth basal
pterygiophores of the spiny dorsal fin with well-
developed ventral shafts oriented more or less vertically
in relation to the vertebral axis and with their proximal
ends in contact with the distal ends of the neural spines
of the second to fourth abdominal vertebrae; ventrally
directed portion of the first basal pterygiophore of the
spiny dorsal fin long and well developed, articulated

with the middle to basal region of the posterior surface of
the skull; most neural spines relatively oblique; skeletal
appearance similar to that of the Recent triacanthids
and not to that of balistids; caudal peduncle only slightly
tapering toward the tail, deeper than wide at the place of
least depth; caudal fin extremely long (53% SL) and
rounded; scales with numerous upright spinules. A single
species, Protacanthodes ombonii (Zigno 1887a).

The Oligocene Cryptobalistinae have: the pelvis
basinlike from the region of the pelvic spines posteriorly,
but bifurcate anterior to the spines; pelvis in cross section
behind the pelvic spines flat, with upturned edges form-
ing a concave dorsal surface, wider than deep; spiny dor-
sal fin base much shorter than the soft dorsal fin base;
probably six abdominal vertebrae anterior to the first
basal pterygiophore of the soft dorsal fin; second to third
(fourth, if present, unknown) basal pterygiophores of the
spiny dorsal fin essentially without ventral shafts, these
pterygiophores far removed from the distal ends of the
neural spines of the abdominal vertebrae; ventrally
directed portion of the first basal pterygiophore of the
spiny dorsal fin very short, articulated with the upper
region of the posterior surface of the skull; most neural
spines relatively vertical; skeletal appearance generally
similar to that of the balistids rather than to that of Re-
cent triacanthids; caudal peduncle probably tapered to a
constricted region in front of the caudal fin, and probably
about as wide as deep; caudal fin of moderate length
(31% SL), basically truncate, with a slight medial con-
cavity; scales apparently without elaborate ornamenta-
tion. A single species, Cryptobalistes brevis (Rath 1859).

The Oligocene to Recent Triacanthinae have: the
pelvis shaftlike from the region of the pelvic spines pos-
teriorly and not bifurcate anterior to the spines; pelvis in
cross section behind the pelvic spines like a railroad rail,
deeper than wide or at least about as deep as wide, the
lateral edges not upturned; spiny dorsal fin base much
shorter than the soft dorsal fin base; four or five ab-
dominal vertebrae anterior to the first basal pterygio-
phore of the soft dorsal fin; second to fifth basal pterygio-
phores of the spiny dorsal fin with moderately developed
ventral shafts, usually directed anteroventrally toward
the region between the neural spines of the first to third
abdominal vertebrae but not in direct contact with them,
usually well separated from them; ventrally directed
shaft of the first basal pterygiophore of the spiny dorsal
fin long and well developed, articulated with the middle
to basal region of the posterior surface of the skull; most
neural spines relatively oblique; skeletal appearance
similar to that of the Recent triacanthids and not to that
of balistids; caudal peduncle tapered to a narrow trans-
versely indented region above and below in front of the
caudal fin, wider than deep at this point; caudal fin of
short to moderate length (15 to 30% SL), deeply forked;
scales with low emarginate ridges. Seven Recent species
in four genera (Triacanthus, Trixiphichthys, Pseudotria-
canthus, Tripodichthys) in the Indo-western Pacific, two
fossil species in a single Oligocene genus (Acantho-
pleurus), and one species from the Miocene (Maro-
sichthys).

Figure 34.— Paeudotriacanthus strigilifer:
lateral view of head; 79.0 mm SL, India.

Figure 35.— Pseudotriacanthus strigilifer:

above, dorsal (left) and ventral (right) views of

skull; below, posterior view of skull;

79.0 mm SL, India, with a few details

from 130 mm SL, Thailand.

in.

^^-^^\ ,,w —

-4J

Figure Se.—Paeudotriacanthus strigilifer:

posterior view of orbit (cross section

of skull; dashed lines represent cut

surfaces of frontals and parasphenoid);

130 mm SL, Thailand.

90

basibranchials

epibranchials
pharyngobranchials

iterhyal hi

epihyal

hypobranchials
ceratobranchials
urohyal

dorsal hypohyal

Figure 3S.—Triacanthiu biaculeatut:

lateral (left) and anterior (right) views

of first two dorsal spines (erected),

with basal pterygiophore shown in lateral

view in both cases, 1 13 mm SL, Borneo.

Figure 37 .—Tripodichthys angustifrona:
dorsal view of branchial arches (extended

on lower side); lateral view of hyoid
arch and urohyal; 137 mm SL, Australia.

Figure 39.— A, Parahollardia lineata, 45.7 mm SL,

Florida, and B, Triacanthus biaculeatus,

SL, Borneo: lateral views of pelvis

and pelvic fin in place (above) and of

pelvis with pelvic fin removed (middle);

anterior views of the base of the pelvic spine (below),

show the structural differences in the locking

mechanism of the Triacanthodidae (indefinite

number of positions of erection) and

Triacanthidae (two positions of erection).

Anatomical diversity.— The anatomical diversity of
this family has been treated in detail by Tyler (1968) and
the major features need be only briefly summarized here.

The Recent species form a very slightly diversified
group of four rather finely split genera that in the near
past were all accommodated in one genus. The heavy in-
cisorlike teeth of Triacanthus, Pseudotriacanthus, and
Tripodichthys occur in an outer series of 10 in both jaws,

the upper jaw with an inner series of four molariform
teeth and the lower jaw with two. In Trixiphichthys,
however, the teeth are smaller and the number of inner
series teeth in the upper jaw is reduced to two, while the
snout is longer and the mouth is narrower than in the
other three genera. A mesopterygoid is absent except in
Trixiphichthys, in which it is small. The caudal peduncle
is much shorter and the lateral line slightly less con-

Figure i\ .—Trixiphichthya weberi: latei
head, 101 mm SL. Bay of Bengal

spicuous in Trixiphichthys than in the other three
genera. The second dorsal fin spine is best developed in
Pseudotriacanthus (greater than half the length of the
first spine), next best developed in Trixiphichthys
(about half the length of the first spine) and relatively
short in the other two genera (about one-third or less the
length of the first spine). The scales in Pseudotria-
canthus have an anterior to posterior series of high thin
distally emarginate vertical ridges, while in the other
three genera the scales have a low, distally emarginate
cruciform ridge. The shaft of the pelvis posterior to the
pelvic spines tapers to a sharp point posteriorly in
Pseudotriacanthus and Tripodichthys, but does not
taper to a sharp point in Triacanthus and Trixi-
phichthys. The length of the anal fin base is shorter
(usually about 2 times in the soft dorsal fin base) in
Pseudotriacanthus than in the other three genera (usual-
ly about 1.4 to 1.7 times in the soft dorsal fin base). The
lateral edge of the sagitta is relatively straight or
somewhat convex in Triacanthus, but has a distinct in-
dentation or cavity in the other three genera. The base of
the pseudobranch has a forward pointed apex about one-
third down its length in Trixiphichthys but is an even
curve without an apex in the other three genera, which
usually have a shorter base to the pseudobranch and
fewer lamellae. There are fewer dorsal, anal, and pectoral
fin rays in Pseudotriacanthus than in the other three
genera. The sixth, and last, dorsal fin spine is more often
completely lost in Triacanthus biaculeatus than in the
other six species. Externally visible rudiments of the
pelvic fin ray buried in the skin are retained longer in
Tripodichthys oxycephalus than in the other six species.
The fossil triacanthids include at least one genus
highly similar to the Recent species, another genus

highly dissimilar to the Recent species, and one genus
obviously intermediate in many respects between the
triacanthodids and triacanthids. These were treated by
Tyler ( 1968) mainly on the basis of the literature, but
some of the specimens have subsequently been re-
examined, and notes are given here on their anatomy.

The Eocene Protacanthodes ombonii, the only repre-
sentative of the Protacanthodinae, is known from a single
specimen which has recently been reexamined. The
holotypic specimen, in counterpart, IGUP 10.901 (head
to left) - 902, from the upper portion of the lower Eocene
of Monte Bolca, Italy, is 112 mm SL, an excellent im-
pression with well-preserved scales that obscure many
osteological features.

Proportional measurements (percent of standard
length in parentheses) for P. ombonii are: greatest
depth 45.7 mm (40.7% SL); snout to base of first dorsal
spine 44.2 mm (39.3% SL); eye 11.4 mm (10.1% SL);
spiny dorsal fin base 25.8 mm (23.0% SL); soft dorsal fin
base 23.5 mm (20.9% SL); anal fin base 12.4 mm (11.0%
SL); daudal peduncle length 26.2 mm (23.3% SL); least
caudal peduncle depth 16.4 mm (14.6% SL); caudal fin
length 50.6 mm (45.0% SL); length of pelvis posterior to
the base of the pelvic spine 17.9 mm (15.9% SL); length
of pelvic spine 27.9 mm (24.8% SL); soft dorsal fin height
10.5 mm (9.3% SL); anal fin height 10.0 mm (8.9% SL);
and the approximate lengths of the dorsal fin spines are
as follows: first 36.5 mm (2.5% SL), second 10.5 mm
(9.3% SL), third 6.0 mm (5.3% SL), fourth 5.0 mm (4.4%
SL), fifth 1.9 mm (1.7% SL), sixth 1.4 mm (1.2% SL).

The six dorsal spines, which decrease greatly in length
posteriorly in the series, are followed by 19 fin rays, the
distal ends of those that show being branched. The dis-
tance between the last spine and the first ray (5.1 mm,

(.

Figure 42.— Protocont/iodes ombonii: of the heads of the two counterparts; of the two counterparts, to show the
above, lateral views of the two below, lateral views of the dorsal six dorsal spines; holot.vpe, 112 mm

counterparts; middle, lateral views parts of the head and anterior body SL, Eocene of Monte Boica, Italy.

4.5% SL) is slightly greater than that between the last
few spines (4.1 mm, 3.6% SL). The anal fin has 14 rays,
the distal ends of those that show being branched. The
elongate rounded caudal fin has 12 rays, the uppermost
ray and the lowermost ray unbranched and the inter-
vening 10 rays branched. The pectoral fin rays cannot be
counted, nor are any pelvic fin rays evident, although a
small ray or two could well be present but obscured by
the pelvic spine. The branchiostegal rays are 2 + 4 = 6,
and the first two rays are not much wider than the others.
There are clearly 12 caudal vertebrae and probably 8 ab-
dominal vertebrae. The supracleithrum is placed ver-
tically in relation to the axis of the skull. The
parasphenoid is relatively straight in the region of the or-
bit, and the ventral flange seems to have been only
moderately developed, probably only slightly deeper
than the depth of the shaftlike portion of the bone above
it.

The prefrontal is very large and sturdy, and has at
least a moderate anteroventral extension alongside the
lower region of the ethmoid, although it is not clear
whether this extension took the form of a long process
which also sutured anteriorly with the posterolateral
wing of the vomer, as in Recent triacanthids. The upper
half of the prefrontal apparently was well separated from
the upper half of the ethmoid by a block of cartilage
larger than in the Recent species. The ethmoid appears
to have been large and sturdy, with a steeply oblique
anterior face that then tapered posterodorsally to suture
with the frontal and prefrontal, much as illustrated by
Tyler (1968:361, fig. 206) for Triacanthus biaculeatus,
but with the frontal extending a little further forward.

The upper jaw apparently is unnaturally displaced
posteriorly, for the tip of the upper jaw is well behind
that of the lower jaw, a functionally unlikely arrange-
ment in this family. As presently placed, the posterior
end of the premaxillary pedicel abuts against the top of
the oblique anterior face of the ethmoid just anterior to
the anterior end of the frontal. In life, however, the
pedicel undoubtedly never was retracted this far postero-
dorsally and probably slid along the anterior basal region
of the ethmoid when the mouth was protracted, as in the
Recent species. The moderate-sized eye is located just
above the middle of the distance between the snout and
the spiny dorsal origin. The pelvis posterior to the pelvic
spines appears to be a stout shaft of bone. The anterior
and lateral surfaces of the first dorsal spine and the
anterior, dorsal, and ventral surfaces of the pelvic spine
are covered with low spiny processes, probably asperities
on the scales covering these spines.

The caudal fin supporting structures are not well
preserved, but grooves indicate that there was probably a
good separation between what in generalized tria-
canthoids are the second and third hypurals. The condi-
tion of the parhypural, epurals, and uroneural elements,
if present and separate, is unclear. The first basal
pterygiophore of the spiny dorsal fin (supporting the first
two spines) has a strong ventral shaft directed ventrally
to between the middle to basal region of the skull and
what is probably the neural spine of the first vertebra
closely applied to the skull. The second basal pterygio-

HI

-\.)r

Figure 43.— Protocant/iodes ombonii: lateral
view of abdominal region (head to left) to
show the pelvis and pelvic spine (one from
each side), holotype, 112 mm SL, Eocene of
Monte Bolea. Italy.

phore (supporting the third spine) has its shaft angled
slightly anteroventrally and in contact with the distal
end of the neural spine of the second vertebra. The third
and fourth basal pterygiophores have their shafts
directed ventrally, about vertical to the vertebral
column, and in contact, respectively, with the distal ends
of the neural spines of the third and fourth vertebrae. If
a fifth basal pterygiophore supporting the slender,
minute sixth spine was present, it apparently was very
small, for it does not show in either plate. The first basal
pterygiophore of the soft dorsal fin is placed between the
neural spines of the sixth and seventh abdominal
vertebrae, and there appears to have been an approxi-
mately one to one ratio between the number of dorsal and
anal fin rays and their basal pterygiophore supports.

The teeth are relatively large incisors, about five in
each half of each jaw judging from the size of the
somewhat scattered elements. It is not possible to tell
whether an inner series of teeth was present in addition
to the major outer series. The largest teeth, antero-
medially in each jaw, have stout bases toward the sockets
but taper into blunt nipplelike caps distally, while the
teeth more laterally in each jaw have lesser cusps and, in
some cases, almost rounded distal edges.

Protacanthodes differs from the other triacanthids
mainly by having: 1) a deep and only slightly tapered
caudal peduncle; 2) an extremely elongate rounded
caudal fin; 3) the spiny dorsal fin base slightly longer
than the soft dorsal fin base; 4) better developed spiny
dorsal fin basal pterygiophores whose ventral shafts are
slightly differently arranged; 5) the upper shaftlike por-
tion of the parasphenoid in the region of the orbit rela-
tively straight and its ventral flange only slightly deeper
than the depth of the shaftlike portion; 6) the lower two
branchiostegal rays not enlarged; 7) possibly less fusion
of the hypurals; 8) one more abdominal vertebra anterior
to the first basal pterygiophore of the soft dorsal fin;
9) and in having scales with a row of upright spinules. All
of these features of difference between Protacanthodes
and Recent triacanthids relate Protacanthodes to the

Figure ii.~Acanthopleurus serratus:

lateral view of entire specimen, syntype,

ca. 95 mm SL, Oligocene of Canton Glarus,

Switzerland (Agassiz 1842:pl. 75).

triacanthodids and establish it as an intermediate
between the two famiHes, as discussed more fully in the
preceding section on the relationships of the Triacan-
thodidae.

The fossil species of Triacanthinae are represented
mainly by several dozen specimens of the Oligocene
Acanthopleurus. These specimens are little more than
impressions, although in some cases rather complete, left
in the black schist of Canton Glarus, Switzerland, and 18
of them have recently been reexamined for this work.
The general appearance of all the specimens is decidedly
like that of Recent triacanthids. The specimens of
Acanthopleurus examined fall naturally into two
groups: 1) moderately elongate, like most of the Recent
species; and 2) decidedly elongate, much more so than
any of the Recent species. The moderately elongate
specimens differ from all the Recent species by having a

slightly deeper (5-8% SL vs. 2-4% SL) caudal peduncle,
i.e., less constricted just in front of the caudal fin base.
Only one specific name is available in Acanthopleurus,
this being serratus Agassiz (1842 illustration, 1844b
description), based on decidedly elongate specimens.

The deeper bodied Acanthopleurus is described here as
collettei in honor of Bruce B. CoUette, National Marine
Fisheries Service, National Systematics Laboratory, who
is a valued colleague, diving companion, editor of this
monograph, and a responsible agent for having it
published. It is diagnosed on the basis of the depth of its
body and caudal peduncle, as specified above and in the
measurements given below and in Figure 46. The seven
type specimens of collettei are designated in the list of
fossil specimens examined for this work.

Proportional measurements as percent of standard
length for A. collettei are as follows, with the number of
specimens on which data were recorded for the character
preceding in parenthesis the range, which is followed by
the average value in parenthesis when ap-
propriate: greatest depth (7) 24.0-31.4 (27.7)% SL;
least caudal peduncle depth (6) 5.5-8.3 (6.7)% SL; length
of pelvic spine (6) 20.0-30.8 (25.5)% SL; length of first
dorsal spine (6) 20.3-35.2 (27.1)% SL; length of second
dorsal spine (6) 6.1-10.3 (8.2)% SL; length of third dorsal
spine (5) 2.6-3.9 (3.7)% SL; length of fourth dorsal spine

Figure iH.— Acanthopleurus serratus (above), lateral
views of entire specimens, left, 153 mm SL, and,
right, 127 mm SL. to contrast with the deeper bodied
A. collettei (below), left, ca. 120 mm SL, paratype, an(
right, 96.2 mm SL, paratype: all specimens from
the Oligocene of Canton Glarus, Switzerland.

150

•

140-

'

130-

• !

120

D

110-
100

\ It,

D

\^

D
D

go-

\

so

'•

a

DD

70

S.L.
ISO-

IS 20 2S

BODY DEPTH AS

CAUDAL PEDUNCLE DEPTH AS

Figure 46. — Charts show two proportional

measurements as percent of standard

length that differentiate Acanthopleurus

aerratus and A. collettei.

(2) 1.5-3.9% SL; distance between tip of snout and spiny
dorsal fin origin (1) 24.1'^c SL; length of pelvis posterior
to region of axil of pelvic spine (3) 14.7-15.6 (15.1)% SL.

Comparable proportions of A. serratus are as follows:
greatest depth (9) 16.0-21.8 (17.3)% SL; least caudal
peduncle depth (5) 3.8-4.7(4.4)% SL; length of pelvic
spine (7) 19.7-28.2 (22.9)^p SL; length of first dorsal
spine (9) 19.7-29.5 (23.7)% SL; length of second dorsal
spine (1) 8.6% SL (third and subsequent dorsal fin
spines not measurable in any of the specimens exam-
ined); distance between tip of snout and spiny dorsal fin
origin (1) 31.1' < SL; length of pelvis posterior to region
nf axil of pelvic spine (2) 17.6-17.8'^r SL.

In reviewing the Oligocene fishes of the black schist of
Canton Glarus, Wettstein (1886) believed that the dif-
ferences in depth of the specimens of Acanthopleurus
were due to distortion. However, many of the specimens
of both A. serratus and A. collettei seem to be in ex-
cellent proportion, and the impressions of the scales and
of their orientation and spacing do not seem to me to be
distorted, as one would expect them to be if the slender
specimens were simply artificially stretched out versions

of the deep-bodied specimens. Moreover, in the few
specimens of both species in which the sculpturing of the
scales can be at least partially deciphered, it appears
that A. collettei had a series of vertical emarginate
ridges, approximately like those of the Recent Pseudo-
triacanthus, while A. serratus had more of a cruciform
emarginate ridge, approximately like that of the other
three Recent genera of triacanthids. It seems clear that
at least two species of Acanthopleurus were present in
the Oligocene seas of Europe.

Both species of Acanthopleurus have the first dorsal
and pelvic spines as well developed as in Recent species,
and with a similarly roughened surface. As in most of the
Recent species, the second dorsal spine is less than half
(about one-third) the length of the first spine, with the
third and fourth spines progressively smaller. None of
the impressions of Acanthopleurus show a fifth or sixth
spine, but such could easily have been present but not
indicated in the impressions at such small size.
Epipleurals appear to have been as well developed on the
abdominal and more anterior of the caudal vertebrae as
they are in the Recent species. The dentition is not clear
in any of the specimens of Acanthopleurus, but some
specimens clearly show the presence of 12 caudal rays in
a deeply forked fin. There appear to have been 12 caudal
vertebrae in the few specimens that show this region rela-
tively well, while the abdominal vertebrae cannot be

counted in any of them, although they presumedly num-
bered eight as in the Eocene Protacanthodes, the Oligo-
cene Cryptobalistes, and the Recent species of the family.

On the basis of the configuration o( Acanthopleurus,
the Triacanthinae have changed very little from at least
the Oligocene.

The other fossil species of Triacanthinae is the very
poorly known Marosichthys huismani (de Beaufort
1926), based on a single very incomplete specimen from
the Miocene of the Celebes comprising the head and up-
per anterior part of the body. The dorsal fin spines are
better developed than in Acanthopleurus, and the body
deeper, but Marosichthys can be considered to be at the
same level of organization as Acanthopleurus, as discuss-
ed by Tyler (1968:241-242).

The Oligocene Cryptobalistes brevis (Rath 1859), the
only representative of the Cryptobalistinae, is known
from a single specimen (perhaps two, see Winterbottom
1974:96) from the same black schist of Canton Glarus,
Switzerland, in which the apparently thoroughly modem
Acanthopleurus is found. The specimen of Cryptoba-
listes was not reexamined for this work, for the holotype
cannot be located, and the following very briefly sum-
marizes the long phylogenetic discussion of the species
presented by Tyler (1968:243-249).

Cryptobalistes is an enigma. In general appearance it
is rather balistoid, although it has (probably) 20 rather
than 18 vertebrae, 12 of which are caudal, at least four
dorsal spines and no locking mechanism of the first by
the second, no supraneural strut bracing the last spiny dor-
sal fin basal pterygiophore, a large pair of pelvic spines,
and a relatively wide, dorsally concave, basinlike pelvis.

Although Cryptobalistes is in most ways an excellent
anatomical intermediate between the triacanthids and
balistids, the shape of its pelvis negates the other
evidence, and the balistoid appearance of Cryptobalistes
is presumed to be due to parallelism. Cryptobalistes
represents an extinct lineage, for in the same Oligocene
strata in which it is found there are also thoroughly
modem balistids (Balistomorphus) not much different
from genera alive today, and these Oligocene balistids
already posses a stout shaftlike pelvis with a dorsal
lobe, three dorsal fin spines with a locking mechanism,
a carina with a supraneural supporting strut, and no
pelvic spines. While Cryptobalistes represents an
evolutionary dead end, it greatly increases the known
anatomical diversity of the Triacanthidae and proves
that the triacanthids were capable to giving rise to
balistidlike forms.

Generic relationships. — As discussed by Tyler
(1968:250), the four Recent genera fall naturally into two
groups on the basis of the structure of the posterior shaft
of the pelvis, Triacanthus and Trixiphichthys having a
relatively untapered pelvis and Pseudotriacanthus and
Tripodichthys a distinctly tapering pelvis ending in a
point. On the meager evidence of the better development
of the second dorsal spine and shorter anal fin base in
Pseudotriacanthus, conditions closer to those of the
ancestral triacanthodids, Pseudotriacanthus can be con-
sidered slightly more generalized than Tripodichthys. On
the basis mainly of the specialized reduced dentition and
of the elongate and narrow snout, Trixiphichthys is less
generalized than Triacanthus. The relationship between
the two more generalized genera in the two basic subdivi-
sions of the Recent species is not clear, but both are cer-
tainly closely related to the Oligocene Acanthopleurus,
with the Miocene Marosichthys of unknown relationship
to Acanthopleurus. The Oligocene Cryptobalistes is of
uncertain relationship with the other triacanthids, while
the Eocene Protacanthodes is clearly on the line directly
intermediate between the ancestral triacanthodids and
the derived triacanthids, as discussed in the preceding
section on the relationships of the Triacanthodidae.

The relationship of the Triacanthidae to the Balistidae
is discussed under the latter, with the conclusion being
that Protacanthodes is not far removed from the line of
early triacanthids that gave rise to the balistids and
aracanids, and thus, respectively, to the monacanthids
and ostraciids as well.

In spite of its name, Protriacanthus gortanii d'Erasmo
(1946:116), from the upper Cretaceous, is not a plec-
tognath (see Patterson 1964:429-432).

Figure il .—Marosichthys huismani: lateral
view of anterior part of bod.v of holot.vpe,
ca. 36 mm head length, Miocene of the Celebes
(de Beaufort 1926; fig. 5 of pi. 5).

Figure iS.—Cryptobalisles brevis: lateral
view of holotype, with insets from left to
right showing the spiny dorsal fin, the
pelvis in dorsal view, scale plates, and

the pelvic spine, ca. 88 mm SL,

Oligocene of Canton Glarus, Switzerland

(Rath I859:fig. 4 of pi. 5).

Infraorder Balistoideo

Comparative diagnosis (contrast with that of the

Triacanthoideo).— Premaxillary without a pedicel or
ascending process, the blunt posterodorsal end of the
premaxillary rotating around the anterior ends of the
ethmoid and vomer; premtixillary immovably articu-
lated with the maxillary, often by suture, and the upper
jaw not even slightly protractile; maxillary only slightly
if at all indented dorsally for articulation with the
anterior end of the palatine; palatine varying from T-
shaped (as seen laterally) to a simple rod and not sutured
to other bones, or as a bony column sutured to the
palatopterygoid arch; ethmoid with a laterally expanded
dorsal or dorsolateral region, at least anteriorly, and
never narrower dorsally than ventrally, forming a strong
buttress for the rotation of the upper jaw; a prootic shelf
of varying size developed under the orbit in front of and
above the major articulation of the posterior region of the
parasphenoid with the prootics, except secondarily lost
in two closely related genera of highly specialized mona-
canthids; dorsal end of the hyomandibular articulated
only with the prootic and pterotic, not in direct contact
with the sphenotic; interoperculum a short rod extending
posteriorly no further than the level of the epihyal and
interhyal, connecting to the suboperculum or operculum
around the anterior edge of the region of overlap of these
two bones only by way of a long unossified ligament; dor-
sal fin spines three, two, one, or absent altogether, if pres-
ent supported by one or two basal pterygiophores; the
first dorsal spine capable of being locked in an erected
position through the agency of the second spine, if pres-
ent, but without an independent locking mechanism
between the base of the first spine and its basal pterygio-
phore; first basal pterygiophore of spiny dorsal fin
without a high dorsomedial flange and no anteropos-

terior canal present in the basal region of the first dorsal
spine; pelvic fin present or absent, when present never as
a prominent erectile and lockable spine at about the
middle of the pelvis, but as a rudiment at the extreme
posterior end of the pelvis consisting of delicate, poorly
ossified, partially fused together flexible filaments or
bony nubbins mostly, or entirely, hidden from external
view by modified scales; pelvis present or absent, when
present laterally compressed and shaftlike throughout its
length, and with or without a dorsal lobe posteriorly;
pelvis usually slightly to greatly rotatable around its
anterior articulation with the cleithra and a slightly to
greatly expansible dewlap of skin usually present
between the posterior end of the pelvis and the anus;
uroneurals rarely present; vertebrae (6-10) -I- (8-23) = 18-
30, but rarely, if ever, 8 + 12 or 9 -t- 11; no more than two
pharyngobranchials with prominent large protruding
teeth; posteromedial edges of epiotics not inturned and
neither they nor the exoccipitals and neural spine of the
first vertebra associated with the articulation of the first
basal pterygiophore of the spiny dorsal fin, when present;
articulation of the mesopterygoid varying from direct
contact with both the quadrate and ectopterygoid to in-
direct contact with both of them through the agency of,
respectively, the symplectic and mesopterygoid.

SUPERFAMILY BALISTOIDEA

Comparative diagnosis (contrast with that of the
Ostracioidea). — Head and entire body (except in the
relatively naked monacanthid Paratuteres prionurus and
in the snout region of several balistids) covered with a
more or less continuous field of scales whose edges are
either slightly separated from one another (the balistid
Canthidermis and several monacanthids) or slightly
overlapping, but never with apposed interdigitated
edges, the basal plates thick and usually rhomboidal in
balistids but thin and variously rounded to rectilinear in
monacanthids; body outline in cross section a simple
gently rounded and laterally compressed oblong; a spiny
dorsal fin of one to three spines; soft dorsal and anal fins

long-based, with 23 to 52 dorsal rays and 20 to 66 anal
rays, and usually a slightly lesser number of basal
pterygiophores; caudal fin with 12 rays; pelvis always
present but a modified rudimentary pelvic fin either pres-
ent (balistids and some monacanthids) or absent at the
posterior end of the pelvis; teeth relatively large and
compressed, more or less notched and incisorlike, six to
eight in an outer series and four to six in an inner series in
the upper jaw and four to eight in a single series in the
lower jaw; lateral line associated with grooves or spiny
processes of the scale plates (with the possible exception
of the monacanthid Pseudaluteres nasicornis); branchios-
tegals usually 2-1-4, but sometimes 1 -I- 4 or 1 -I- 3, with
none of the rays in the posterior division as broad and
laterally compressed as those in the anterior division;
distal end of last branchiostegal ray usually not closely
held to inner surface of suboperculum; elements of the
hyoid arch less compacted together anteroposteriorly and
most elements more elongate or less deep bodied;
urohyal relatively large and more or less L-shaped, with a
ventral flange of varying degrees of development; fifth
ceratobranchial either toothed or toothless; pharyngo-
branchials consisting of two elements bearing large
protruding teeth, with an anterior toothless suspensory
element present or absent; epibranchials usually four,
but three in the monacanthid Psilocephalus; gill rakers
present or absent along anterior edge of fifth cerato-
branchial (posterior edge of last gill slit); caudal fin sup-
porting structures only moderately consolidated, always
having at least the epural and parhypural autogenous
and usually having an autogenous uppermost hypural as
well; haemal spine of penultimate vertebra autogenous;
neural and haemal arches of the last vertebra in-
complete, the neural and haemal canals at least partially
open respectively above and below; uroneurals rarely
present; vertebrae normally 7 -I- 11 = 18 (balistids) or (6-
8) -I- (12-23) = 19-30 (monacanthids), but never nine or
more abdominal vertebrae; the first vertebra of normal
size and firmly sutured but not fused to the rear of the
skull; five (balistids) or four to eight (monacanthids) ab-
dominal vertebrae with neural spines anterior to the first
basal pterygiophore of the soft dorsal fin; four to six
caudal vertebrae posterior to the last basal pterygio-
phores of both the soft dorsal and anal fins; at least three
and usually four soft dorsal fin basal pterygiophores plac-
ed between at least some of the successive neural spines;
most neural spines positioned only slightly obliquely in
relation to the axis of the vertebral column; haemal
spines above the anal fin basal pterygiophores with rela-
tively long stout shafts similar to those of the neural
spines and penetrating into the proximal region of the
series of anal fin basal pterygiophores; soft dorsal fin
basal pterygiophores 22 to 51; anal fin basal pterygio-
phores 20 to 64; prominent thin lateral flanges present
along at least a part of the length of the soft dorsal and
anal fin basal pterygiophores; distal pterygiophores pres-
ent as ossified single or paired pieces in both the soft
dorsal and anal fins, except often absent or unossified
under the more posterior rays in both fins; all anal fin
basal pterygiophores in the midline of the body; one or

two basal pterygiophores supporting the anteriorly placed
spiny dorsal fin of one to three spines, but no supraneural
element present anteriorly from the dorsal end of the first
basal pterygiophore of the solt dorsal fin; epipleurals
always present on some of the abdominal vertebrae and
sometimes on some of the more anterior caudal
vertebrae, while ribs are rarely present; uppermost pec-
toral fin ray never of two relatively well -developed halves
of about equal length, either a single piece without a
foramen or a larger medial half and a much smaller
lateral nubbin; actinosts flexibly articulated with the
scapula and coracoid, not sutured to them or to one
another; coracoid and cleithrum not especially enlarged;
coracoid not expanded ventrally, much less wide there
than dorsally, without prominent flanges; coracoid
always with a well-developed posterior prong just below
the lowermost actinost; scapular foramen usually com-
plete, incomplete only in a few monacanthids; cleithrum
without an anterior flange from its lower anteromedial
edge; postcleithrum usually as a single piece, rarely with
two halves, in the form of a long sturdy rod directed
strongly obliquely toward the end of the pelvis; at least
the proximal end of the supracleithrum not in contact
with the cleithrum, and usually slightly less firmly held
to the cleithrum and posttemporal; Baudelot's ligament
not ossified; posttemporal usually relatively small;
palatine varying from T-shaped to a simple rod of bone
representing the top of the T and not sutured to any of
the bones with which it articulates, the foot of the T con-
nected by ligament to the anterodorsal edge of the ecto-
pterygoid and the top of the T primarily connected to the
maxillary and ethmoid; vomer not especially enlarged,
having a moderately laterally expanded anterior end and
a long posterior tapering shaft fitting into a deep concavi-
ty in the parasphenoid; ventral edge of the ventral flange
of the parasphenoid not at all laterally expanded; ventral
flange of parasphenoid with a deep indentation at about
the level of the prefrontal; parasphenoid without a dorsal
flange in the medial septum of the orbit, the para-
sphenoid and pterosphenoids not being in contact there
and there being no anteroventral extensions of the ptero-
sphenoids into the medial septum; the medial edges of
the pterosphenoids in contact in the posterior wall of the
orbit, usually in light contact dorsally and broader
sutural contact ventrally; myodome large; apposed sur-
faces of parasphenoid and basioccipital excavated to
form a canal leading anteriorly into the myodome cavity;
epiotics broadly in contact and sutured to one another
medially on the posterior surface of the skull, not
separated there by the more anteriorly placed supraoc-
cipital; supraoccipital never with a posterior crest, but
with or without an anterior crest on the surface of the
main body of the bone; prootic shelf usually relatively
smaller, never with a prominent ventral or ventrolateral
flange from its lateral edge; the major foramen in the
prootic shelf relatively small and rounded, and complete-
ly enclosed by the prootic alone; anterior edge of upper
part of preoperculum articulated only along the rear edge
of the hyomandibular; hyomandibular a more or less
flattened shaft, not greatly expanded.

Family Balistidae

Comparative diagnosis (contrast with that of the
Monacanthidae).— Teeth very sturdy, more or less in-
cisorlike, developed more for crushing than nibbling, four
outer and three inner on each premaxillary and four in a
single series on each dentary; pharyngobranchials con-
sisting of a toothless suspensory element and two uni-
serially toothed elements; fifth ceratobranchial toothed
but without gill rakers along its anterior edge (posterior
edge of last gill slit); three dorsal spines, the second more
than one-half the length of the first and ending ventrally
in large sturdy ventrally directed processes from either
side; dorsal fin spines supported by two basal pterygio-
phores and a supraneural strut; dorsal spines and their
basal pterygiophores placed behind the posterodorsal
apex of the skull, the anterior edge of the first basal
pterygiophore abutting against the rear of the postero-
dorsal surface of the skull; first basal pterygiophore with
a large lateral foramen to accommodate the ventral prong
of the second dorsal spine; supraoccipital large and stur-
dy, with a variously high (usually) to low (Rhinecanthus)
but sturdy vertical crest medially throughout its length,
the crest expanded laterally at its posterior end to help
support the anterior end of the first basal pterygiophore;
posterior region of the epiotic expanded dorsally, the
medial edges of the two epiotics well-separated dorsally,
bounding the posterolateral region of the deep foramen
in the skull between the epiotics and the supraoccipital
in which the short anteroventral shaft of the first basal
pterygiophore is held; palatine T-shaped, the long foot of
the T directed ventrally or posteroventrally to connect by
a short ligament with the anterior edge of the ec-
topterygoid; exoccipital meeting its opposite member in
the midline above the foramen magnum between the
medial edges of the bifid neural spine of the first
vertebra; prefrontal relatively large and heavy, firmly ar-
ticulated, often by interdigitation, with the frontal,
ethmoid, and parasphenoid; parasphenoid only slightly
expanded laterally just behind the orbit, the medial edge
of the pterotic on the ventral surface of the skull only
narrowly separated from its opposite member there by
the basioccipital and parasphenoid; parasphenoid much
expanded dorsally in front of the orbit, broadly overlying
the ethmoid and contacting the prefrontal; the laterally
expanded dorsal region of the ethmoid wide and thick.

Figure 49.— Range of diversity in body

form in the Recent Balistidae: Balistapus

undulatus, left; Odonus niger, right.

about as wide as or wider than the depth of the ventral
platelike portion; ventral platelike flange of the ethmoid
thick and sturdy, relatively shallow, not much if at all
deeper posteriorly than anteriorly, and broadly overlain
by the parasphenoid; posttemporal held in a deep groove
on the lateral surface of the pterotic; medial edge of den-
tary either denticulate or smooth; prootic shelf well
developed, extending forward at least to the level of the
middle of the orbit; posterior edge of supracleithrum
with a posteriorly directed process or hump; post-
cleithrum with a dorsal spur from its upper dorsal edge
supporting the ventral region of the tympanum
regardless of whether enlarged scales are present there or
not; postcleithrum either a single piece or divided into
dorsal and ventral segments; branchiostegal rays always
2-1-4; pelvis always with a dorsal lobe posteriorly on its
dorsal surface; encasing scales at the posterior end of the
pelvis in four segments, always flexible (least and only
slightly so in Canthidermis); a relatively well-developed
rudimentary pelvic fin element always present, running
the length of the last two segments and protruding to the
exterior through a foramen in the fourth segment (except
in Melichthys and Xanthichthys, in which the element is
reduced in size and does not protrude, and in Canthi-
dermis, in which the ray is an even smaller nubbin of
bone not protruding to the exterior); vertebrae 18 (7 +
11); vertebrae anterior to the first basal pterygiophore of
the soft dorsal fin always five, those posterior to the last
basal pterygiophores of both the soft dorsal and anal fins
always four; neural and haemal arches of the last cen-
trum relatively well developed although incomplete; the
last centrum with a vertical crest for muscle attach-
ment; uroneurals rarely present; an upper free hypural
always present; lateral flanges on the soft dorsal and anal
fin basal pterygiophores not ending distally in prominent
hooklike lateral expansions; the basal pterygiophores
extending proximally relatively closely to the neural and
haemal arches throughout the length of the series; the
great majority of the dorsal, anal, and pectoral fin rays
branched, only the first few rays, and sometimes the last,
being unbranched; scales large, with thick basal plates
and relative coarse spinulation, the basal plates broadly
overlapping (except in Canthidermis) and more or less
regularly arranged; scales above the pectoral fin base
usually enlarged and more or less separated, forming a
flexible covering to the tympanum (scales unmodified in
Canthidermis and only slightly modified but unenlarged
in Xanthichthys); species relatively large, adults usually
reaching well over 200 mm.

Figure 50.— Balistapug undulatusi upper

left, scale plates of the tympanum region

just behind the gill slit and above the

pectoral fin base; upper middle, scales from

upper middle region of body, including

lateral line canal bearing scales;

upper right, ventral view of encasing scales

at end of pelvis (anterior to left), with

arrow indicating major region of flexibility

(two terminal branches of modified fin-r>y

element protrude posteriorly);

lower left, nasal region as seen externally (above)

and the olfactory lamellae as seen with

the top of the nasal sac removed.

Detailed description of Balistapus undulatua.

Material examined.— Two cleared and stained
specimens, 122-124 mm, and three wet partially disar-
ticulated skeletons of approximately the same size as the
two previous specimens, prepared by maceration.

Occipital Region.

Basioccipital. —A short column, expanded antero-
dorsally; cartilage filled at its anterior and dorsal edges;
articulates by interdigitation dorsally with the exoc-
cipitals, anterolaterally with the prootics, and antero-
medially with the overlying posterior end of the para-
sphenoid. The rim of the round concave posterior end of
the basioccipital articulates by fibrous tissue with the
concave anterior face of the centrum of the first vertebra.
The ventromedial surface of the basioccipital is deeply
concave throughout its length. This concave channel is
mostly hidden from view by the overlying parasphenoid,
but the channel opens to the exterior posteriorly at the
base of the posterior bifurcation of the parasphenoid.
Anteriorly the channel opens into the myodome where
the anterior end of the basioccipital forms the postero-
lateral and posterodorsal walls of the myodome.

Exoccipital.— Cartilage filled at its ventral edge;
articulates by interdigitation dorsally with the epiotic,
laterally with the pterotic and ventromedially with the
basioccipital. Posteriorly the exoccipitals form the
lateral and ventral walls of the foramen magnum, while
dorsally the foramen is closed by the thin piece of car-
tilage between the dorsomedial edges of the two exoc-
cipitals. The posterior surface of the exoccipitals firmly
articulates by fibrous tissue and slight interdigitation
with the anterior surface of the bifid neural spine of the
first vertebra. The posteromedial portion of the exoc-
cipital, which forms the lateral wall of the myodome and
the firmest place of attachment of the neural spine of the

first vertebra, is stout and thickened, probably repre-
senting the much modified region of the exoccipital con-
dyle.

Supraoccipital. — Wide posteriorly, tapering to a
point anteriorly; expanded dorsomedially throughout its
length into a low crest; cartilage filled at its edges of ar-
ticulation with the other cranial bones; articulates by in-
terdigitation anterolaterally with the slightly overlying
frontals and posterolaterally with the epiotics. The
posterodorsal end of the supraoccipital is expanded into
a thickened articular face with a medial concavity to
support the anteroventral shaft of the first basal
pterygiophore of the spiny dorsal fin. This medial con-
cavity is partially bordered by the epiotics. The antero-
ventral shaft of the pterygial element is held by tough
fibrous tissue to both the supraoccipital and the epiotics.

Otic Region.

Pterotic. — Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates by
interdigitation dorsomedially with the epiotic, antero-
laterally with the sphenotic, anteroventrally with the
prootic, and anteromedially on its ventral surface with
the basioccipital. More or less transversely along its ven-
tral surface, the pterotic is produced into a sturdy flange
whose depth is greatly increased in about the middle of
its length. The anterior surface of this stout and elongate
portion of the ventral flange of the pterotic articulates by
fibrous tissue with the posterodorsal edge of the hyoman-
dibular. Just anterior to its ventral flange, the pterotic
bears a slight depression which continues anteriorly over
the prootic and into which the dorsal head of the hyo-
mandibular is held by fibrous tissue. In about the middle
of its lateral surface, the pterotic bears a deep vertical
groove into which the posttemporal fits and is held im-
movably by extensive interdigitation.

Sphenotic. — Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates by
interdigitation anterodorsally with the frontal and
pterosphenoid, posterodorsally with the epiotic, postero-
ventrally with the pterotic and posttemporal, and antero-
ventrally with the prootic.

Epiotic. — Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates by
interdigitation dorsally with the supraoccipital, dorso-
laterally with the frontal, laterally with the sphenotic,
ventrolaterally with the pterotic and a short section of
the posttemporal, posteroventrally with the exoccipital,
and medially with its opposite member. The posterodor-
sal edge of the epiotic is deeply concave where it inter-
digitates with the supraoccipital, forming, with the latter
bone, the concavity in the skull in which is held by
fibrous tissue the anteroventral shaft of the first basal
pterygiophore of the spiny dorsal fin.

Prootic. — Cartilage filled along all of its edges of
articulation with the other cranial bones, except antero-
medially; articulates by interdigitation medially with
the lateral surface of the parasphenoid, anteromedially
with the ascending fork of the parasphenoid, anterodor-
sally with the pterosphenoid, anterolaterally with the
sphenotic, and posteriorly with the pterotic. A concavity
is present in the middle of the ventral surface of the
prootic in which is held by fibrous tissue the anterior
half of the dorsal head of the hyomandibular. The medial
edges of the prootics form the lateral walls of the
myodome, while medially directed shelves from the
medial edges of the prootics interdigitate with one
another in the midline to form the dorsal roof of the
myodome. The anterior edge of the myodome is formed
by the prootics, except ventrally where it is formed by
the parasphenoid. The anterior end of the prootic is
prolonged anteriorly as a subocular shelf which serves for
muscle attachment.

Orbital Region.

Frontal. —Widest posteriorly, tapering to a point
anteriorly; its lateral edge above the orbit slightly uprais-
ed and thickened; articulates by interdigitation postero-
medially with the supraoccipital, which it slightly
overlies, posterolaterally with the sphenotic and pos-
teriorly with the epiotic. Posteriorly in the rear of the or-
bit it interdigitates with the pterosphenoid. The frontal
interdigitates anteriorly with the ethmoid and antero-
laterally with the prefrontal. On its ventral surface, the
medial edges of the frontals are narrowly separated by
the cartilaginous mass which is continuous anteriorly
with the remains of the ethmoid cartilage.

Prefrontal. — Large and elongate; cartilage filled
along its medial edge; articulates by interdigitation dor-
somedially with the frontal, ventrally with the para-
sphenoid, and anteriorly with the ethmoid. Along most

of its medial edge, the prefrontal is continuous with the
remains of the ethmoid cartilage.

Parasphenoid. — Elongate, running almost the
entire length of the skull; ventrally expanded into a thin
keel along the anterior half of its length. The bifurcate
posterior end of the parasphenoid broadly overlies and
interdigitates with the basioccipital, roofing over the
longitudinal concavity in the ventral surface of the
basioccipital. Under the posterior region of the orbit, the
parasphenoid possesses a pair of short, slightly forked,
dorsolateral projections which interdigitate with the
prootics and form the anterior edge of the ventral region
of the myodome. Posterior to these dorsal wings, the
parasphenoid interdigitates with the ventromedial edges
of the prootics and, more posteriorly, with the ventral
surface of the basioccipital. Just anterior to the orbit, the
parasphenoid interdigitates with the base of the prefron-
tals. Anterior to the level of its articulation with the
prefrontals, the dorsal edge of the parasphenoid becomes
concave and fits around the ventral edge of the ethmoid
keel. At its anterior end the parasphenoid possesses a
short, narrow concavity into which fits and interdigitates
the posterior shaftlike portion of the vomer.

Pterosphenoid. — Cartilage filled along all of its
edges of articulation with the other cranial bones; ar-
ticulates by slight interdigitation dorsally with the fron-
tal, laterally with the sphenotic, and ventrally with the
prootic. Medially the pterosphenoid articulates broadly
through cartilage and interdigitation with its opposite
member.

Ethmoid Region.

Ethmoid. — Wide and elongate; slightly expanded
laterally and ventrally at its anterior end; produced ven-
trally into a keel which is relatively low along most of its
length, but which increases in depth at its anterior end
above the shaftlike portion of the vomer; cartilage filled
at its posteroventral edge where it is continuous with the
remains of the ethmoid cartilage. The ventral edge of the
ventral flange of the ethmoid articulates by fibrous tissue
with the concave dorsal edge of the platelike portion of
the parasphenoid. At its posterior end the ethmoid ar-
ticulates by interdigitation laterally with the prefrontal
and dorsally with the frontal. The ethmoid articulates by
fibrous tissue anteroventrally with the anterior edge of
the vomer, while directly anteriorly the ethmoid supports
the posterodorsal edge of the upper jaw.

Vomer. — Short; laterally expanded anteriorly but
tapering to a shaft posteriorly; articulates posteriorly by
slight interdigitation of its posterior shaftlike portion
with the concavity at the anterior end of the para-
sphenoid; articulates by fibrous tissue dorsally with the
anteroventral surface of the ethmoid and anterolaterally
with the medial surfaces of the palatines and the postero-
dorsal edges of the upper jaw.

Mandibular Region.

Hyomandlbular. — Somewhat expanded dorsally,
but tapering to a stout shaft ventrally; cartilage filled at
its posterodorsal and anteroventral edges; articulates by
fibrous tissue dorsally with the longitudinal groove on
the ventral surfaces of the prootic and pterotic, while
posterodorsally the hyomandlbular abuts against and is
firmly held by fibrous tissue along the medial surface of
the elongate portion of the ventral flange of the pterotic.
Along the lower three-fourths of its posterior edge the
hyomandlbular articulates by fibrous tissue with the
preoperculum. Just posterior to the dorsal end of the
preoperculum, the ventral edge of the hyomandlbular is
thickened and bears a concavity with which the dorsal
end of the operculum articulates by fibrous tissue. The
anterior end of the hyomandlbular articulates variously
through cartilage and fibrous tissue with the metapter-
ygoid, symplectic, interhyal, and preoperculum.

Quadrate. — Widest posteriorly, tapering to a knob
anteriorly for articulation with the articular in the lower
jaw; cartilage filled at its posterior edge; deeply cleft
along its lower posterior edge to accommodate the
anterior end of the symplectic; articulates by slight inter-
digitation dorsally with the ectopterygoid, while it ar-
ticulates by fibrous tissue posteriorly with the
metapterygoid and symplectic, both of which somewhat
overlie the quadrate. Ventrally the quadrate articulates
by fibrous tissue with the preoperculum.

Metapterygoid. — Large; a more or less rounded flat
plate; cartilage filled at its anterior edge; articulates by
fibrous tissue anteriorly with the quadrate, posteriorly
with the interhyal and hyomandlbular; articulates by
slight interdigitation dorsally with the mesopterygoid,
anterodorsally with the ectopterygoid, and ventrally with
the symplectic.

Symplectic. — Large; cartilage filled at its anterior
and posterior edges; dorsally expanded in the middle of
its length; articulates by fibrous tissue anterodorsally
and anteroventrally with the quadrate; articulates by
slight interdigitation posterodorsally with the
metapterygoid, while along its posteroventral edge the
symplectic is attached to the fibrous tissue sheet that is
present between the preoperculum, hyomandlbular, and
metapterygoid.

Palato-Pterygoid Region.

Palatine. — T-shaped; articulates by fibrous tissue
ventrally with the ectopterygoid, anterodorsally with the
premaxillary and maxillary, and posterodorsally with the
laterally expanded anterior ends of the ethmoid and

Ectopterygoid. —Elongate; articulates by slight
interdigitation anteroventrally with the quadrate, which

it somewhat overlies, posteroventrally with the
metapterygoid, and posterodorsally with the mesoptery-
goid. Along the middle of its anterior edge it articulates
by tough fibrous tissue with the base of the palatine.

Mesopterygoid. — Small; articulates by inter-
digitation anteriorly with the ectopterygoid and ven-
trally with the metapterygoid.

Opercular Region.

Operculiun. — Thin and flat, except dorsally where
it thickens into a rounded articular facet; articulates by
fibrous tissue with the slightly upraised area on the ven-
tral surface of the posterodorsal region of the hyoman-
dlbular, while ventrally it overlies and articulates by fi-
brous tissue with the suboperculum.

Suboperculum. — Very thin and flat, slightly wider
anteriorly than posteriorly; held to the overlying oper-
culum by fibrous tissue.

Interoperculum. — A straight rod, slightly wider
posteriorly than anteriorly; extends from the region of
the interhyal to about the anterior end of the preoper-
culum; articulates by a tough ligament anteriorly with
the angular in the lower jaw, while at its posterior end
two ligaments are present. One of these ligaments is
short and connects with the dorsal surface of the epi-
hyal, while the other one is long and runs posteriorly to
connect with the anterior edge of the operculum just
above the point where the operculum begins to overlie
the suboperculum.

Preoperculum. — Not greatly expanded in its mid-
dle region; its dorsal surface only slightly flattened for ar-
ticulation with the quadrate; articulates by fibrous tis-
sue along the posterior portion of its dorsal edge with the
hyomandlbular, along the middle portion of its dorsal
edge with the symplectic, metapterygoid, and interhyal,
and along the anterior portion of its dorsal edge with the
quadrate.

Upper Jaw.

Premaxillary. — A slightly curved plate, wider dor-
sally than ventrally; its posterodorsal edge slightly con-
cave to articulate by fibrous tissue with the anterior ends
of the ethmoid and vomer. It also articulates by fibrous
tissue laterally with the medial surface of the palatine.
The anterior edge of the premaxillary forms the border of
the upper jaw, except for a short distance ventrally where
the maxillary does so. The premaxillary is in close ap-
position with the maxillary and articulates immovably
with it by fibrous tissue and slight interdigitation. The
flat dorsomedial edges of the two premaxillaries are held
closely to one another by fibrous tissue. Each premaxil-
lary bears seven teeth, four in an outer row and three in
an inner row. Both of these rows are in close contact with

one another. The teeth are borne in very shallow depres-
sions on the surface of the premaxillary in their fully
formed condition. They develop, however, from large
deep sockets which open to the surface just anterior to
the base of the old tooth which they will replace. As the
tooth erupts to the surface to replace an old tooth, the
socket from which it arose seems to become overgrown or
filled in with bone and to disappear so that the tooth
rests flatly against the surface of the premaxillary. Most
of the interior of the premaxillary is given over to the
dental pulp cavity, this cavity communicating with the
exterior through a foramen in the posterodorsal surface of
the bone as well as through numerous smaller holes on
the inner and outer surfaces of the bone.

Maxillary. — Widest ventrally, becoming narrower
dorsally; articulates firmly with the premaxillary by fi-
brous tissue and slight interdigitation. The medial sur-
face of the ventral region of the maxillary articulates by
fibrous tissue with the lateral surface of the upper por-
tion of the dentary. The posterodorsal surface of the
maxillary articulates by fibrous tissue with the lateral
expansions of the ethmoid and vomer and with the
medial surface of the palatine.

Lower Jaw.

Dentary. —Wide posteriorly; its posterior edge con-
cave to accommodate the anterior portion of the ar-
ticular, with which it articulates by interdigitation.
Posteroventrally the dentary articulates by inter-
digitation with the angular. From the lateral surface of
its posterodorsal region the dentary articulates by fi-
brous tissue with the medial surface of the maxillary.
Ventromedially the flat surface of the dentary articulates
by fibrous tissue with its opposite member. Each den-
tary bears four teeth in a single row, corresponding to the
outer row of teeth in the premaxillary. The teeth are
borne flush against the surface and are replaced by new
teeth developing in deep sockets, just as described for the
teeth of the upper jaw. Most of the interior of the dentary
is filled with the dental pulp, the cavity opening to the
exterior at the concave posterior edge of the dentary, as
well as at pores on the inner and outer surface of the
bone.

Articular. — Wide along its posterior edge, with a
concave facet for articulation by fibrous tissue with the
anterior knob of the quadrate; articulates by inter-
digitation with the concave posteromedial surface of the
dentary. Along its ventral edge the articular inter-
digitates with the angular. The sesamoid articular is a
small but thick nodule of bone closely held by fibrous tis-
sue to the medial surface of the articular just in front of
the concave articular facet of that bone.

Angular. — Small, squarish; articulates by inter-
digitation dorsally with the articular and anteriorly with
the dentary. Posteriorly the angular connects by liga-
ment with the interoperculum.

BRANCHIAL APPARATUS.

Hyoid Arch, Branchiostegal Rays, and Urohyal.

Hypohyals. —Both hypohyal elements well de-
veloped, the ventral element considerably larger than
the dorsal element; the dorsal element cartilage filled at
its dorsal Eind posterior edges; the two elements ar-
ticulate with one another through the cartilage that lies
between them; the dorsal and ventral elements ar-
ticulate by fibrous tissue at their anteromedial edges
with their opposite members. The dorsal hypohyal ar-
ticulates by fibrous tissue at its dorsal edge with the
lateral surface of the first basibranchial. The ventral
hypohyal articulates through cartilage at its posterior
edge with the ceratohyal, and by fibrous tissue from its
ventromedial surface with the urohyal.

CeratohyaL — A short shaft, wider posteriorly than
anteriorly; cartilage filled at its anterior and posterior
edges; articulates through cartilage anteriorly with the
ventral hypohyal; articulates posteriorly through car-
tilage and interdigitation with the epihyal. The first two
branchiostegal rays articulate with slight depressions on
the ventral edge of the ceratohyal. The ventralmost one
or two rays of the remaining four rays have their fibrous
tissue articulation principally with the ceratohyal.

EpihyaL — Large; cartilage filled at its ventral
edge; articulates through cartilage and interdigitation
ventrally with the ceratohyal, while dorsally it ar-
ticulates by fibrous tissue with the interhyal and inter-
operculum.

InterhyaL- A short, thick rod; cartilage filled at
both ends; articulates by fibrous tissue ventrally with the
epihyal and dorsally with the symplectic.

Branchiostegal rays. — Six in number; increasing
slightly in length posteriorly in the series; first ray con-
siderably flatter and wider than the others. The first two
rays articulate by fibrous tissue with shallow grooves on
the ventral edge of the posterior half of the ceratohyal.
The other four rays articulate by fibrous tissue with the
ceratohyal and epihyal. The more ventral of these four
rays have their connective tissue fibers attached prin-
cipally to the ceratohyal, while the more dorsal rays have
them attached principally to the epihyal.

Urohyal. — Thick along its dorsal and anterior
edges, otherwise a thin plate; articulates by fibrous tis-
sue anteroventrally with the medial surface of the ven-
tral hypohyal, while posteriorly it articulates with the
ventral surface of the first two basibranchial elements.

Branchial Arches. — All the elements are cartilage
filled at their edges of articulation with the other
elements of the series, and the articulations are usually

105

through cartilage and fibrous tissue. The branchial
arches are composed of three basibranchials, three pairs
of hypobranchials, five pairs of ceratobranchials, four
pairs of epibranchials, and three pairs of pharyngo-
branchials. Four gills are present, with a slit between the
fourth arch and the lower pharyngeal.

branchial; about 12 to 15 teeth present in a single row,
the teeth like those of the outer series of the jaws, but
much smaller and with sharper points. The teeth
decrease in size posterolaterally in the series and are
replaced by new teeth developing in sockets below the
bases of the old teeth.

First arch.— Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. First basibranchial
short, wider posteriorly than anteriorly, displaced for-
ward so that it articulates posteriorly with the second
basibranchial and posterolaterally with the first hypo-
branchials. First hypobranchial very wide, more so
posterodorsally than ventrally; the largest of the hypo-
branchial elements, which decrease in size posteriorly in
the series; articulates ventrally with the region of ar-
ticulation between the first and second basibranchials
and dorsally with the first ceratobranchial. First cerato-
branchial a sturdy rod; the longest of the cerato-
branchial elements, which decrease in length posteriorly
in the series; no ventrally directed flange present on the
ceratobranchials, although from the second to the last
ceratobranchial the ventral ends of the elements become
increasingly enlarged; articulates dorsally with the first
epibranchial. First epibranchial with its ventral end
rounded and its dorsal end forming two divergent stubby
projections; its rounded ventral end articulates with the
first ceratobranchial, while the anterior projection of its
dorsal end articulates with the first pharyngobranchial,
and the posterior projection of its dorsal end with the
dorsal end of the second epibranchial and the ventral end
of the second pharyngobranchial. First pharyngo-
branchial (suspensory pharyngeal) short, slightly wider
ventrally than dorsally; articulates ventrally with the
first epibranchial and dorsally by fibrous tissue with the
ventral edge of the keel of the parasphenoid at the level
of the middle of the orbit.

Third arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Third basi-
branchial narrow anteriorly and posteriorly, but ex-
panded in the middle; articulates anteriorly with the sec-
ond basibranchial, anterolaterally with the second
hypobranchials, posterolaterally with the third hypo-
branchials, and posteriorly with the fourth cerato-
branchials. Third hypobranchial rounded posteriorly but
with a tapering anteroventral process that articulates by
fibrous tissue anteriorly with the ventral surface of the
more anterior basibranchials; articulates postero-
laterally with the third ceratobranchial and posteromedi-
ally with the posterior end of the third basibranchial.
Third ceratobranchial with a much expanded ventral
end; articulates ventrally with its opposite member
and with the third hypobranchial, and dorsally with
the third epibranchial. Third epibranchial a short rod
with a slightly expanded base and a short projection pos-
teriorly from the basal part of its posterior edge; ar-
ticulates dorsally with the third pharyngobranchial; the
projection from the basal portion of its posterior edge ar-
ticulates by fibrous tissue with a similar projection from
the anterior edge of the fourth epibranchial. Third
pharyngobranchial like the second pharyngobranchial,
except slightly smaller and with its ventral arm shorter
and wider; teeth in about the same number and arrange-
ment as described for the second pharyngobranchial; ar-
ticulates with the dorsal ends of the third and fourth epi-
branchials. The two toothed pharyngobranchial
elements are held to one another by fibrous tissue.

Second arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Second basi-
branchial longer than the first basibranchial and about
the same length as the third basibranchial, but wider
than either of the other two; articulates anteriorly with
the first basibranchial, anterolaterally with the first
hypobranchials, posteriorly with the third basibranchial
and posterolaterally with the second hypobranchials. Sec-
ond hypobranchial squarish, with a short posterior ex-
tension; articulates ventrally with the area of ar-
ticulation between the second and third basibranchial
and dorsally with the second ceratobranchial. Second
ceratobranchial a long rod with a slightly expanded ven-
tral end; articulates dorsally with the second epi-
branchial. Second epibranchial a short rod; articulates
dorsally with the second pharyngobranchial and the
posterodorsal arm of the first epibranchial. Second
pharyngobranchial with an elongate tooth bearing por-
tion and a short arm for articulation with the second epi-
branchial and the posterodorsal arm of the first epi-

Fourth arch. — Cerato- and epibranchial elements
only. Fourth ceratobranchial much expanded ventrally;
articulates ventrally with the ventral end of its opposite
member and with that of the third ceratobranchial.
Fourth epibranchial rodlike; longer than the second and
third epibranchials; articulates ventrally with the fourth
ceratobranchial and dorsally with the base of the third
pharyngobranchial, and, from a short projection on its
lower anterior edge, with the third epibranchial.

Fifth arch. — Ceratobranchial (lower pharyngeal)
element only. Fifth ceratobranchial the shortest of the
ceratobranchial elements; wider ventrally than dorsally,
with a short anterior projection from its anteroventral
edge; articulates ventrally with the base of the fourth
ceratobranchial; bearing two rows of teeth, the posterior
row much larger than the anterior row; teeth in both rows
similar to those of the pharyngobranchials, and becom-
ing smaller dorsally; about 25 teeth in the anterior row
and 15 in the posterior row.

PAIRED FIN GIRDLES.

Pectoral Fin.

Posttemporal. — A slender wedge of bone ventrally
where it is held in a deep concavity on the lateral surface
of the pterotic, but slightly expanded posterodorsally
where it contacts the sphenotic and, to a lesser extent,
the epiotic. It is firmly and immovably held to all three of
these bones, and especially to the pterotic, by a com-
bination of fibrous tissue and interdigitation. Ventrally
the posttemporal articulates by fibrous tissue with the
supracleithrum.

Supracleithrum. — More or less vertical in position
in relation to the axis of the body; articulates im-
movably by fibrous tissue dorsally with the rounded ven-
tral head of the posttemporal and ventrally with the
cleithrum, which it broadly overlies.

Cleithrum. — Laterally expanded at its anterior
edge, particularly in the middle one-third of its length;
articulates by fibrous tissue dorsally with the overlying
supracleithrum and with the dorsal postcleithrum, which
it overlies. Along its posterior edge the cleithrum broadly
overlies and articulates by fibrous tissue with the an-
terior edges of the scapula and coracoid. Ventromedially
the cleithrum articulates by fibrous tissue with its op-
posite member, and just above this region the anterior
end of the shaftlike pelvis is held between the cleithra.

Postcleithra. — The postcleithra form a long
posteroventrally directed strut from the dorsomedial sur
face of the cleithrum along the abdominal wall muscula
ture to the area just above the dorsal lobe of the pelvis
The dorsal postcleithrum is wide anteriorly where it ar
ticulates by fibrous tissue with the overlying cleithrum
with a short dorsal projection from about the middle of
its length supporting the posteroventral edge of the en-
larged scales of the tympanum; tapering to a point pos-
teriorly where it broadly overlies and articulates by fi-
brous tissue with the anterodorsal region of the ventral
postcleithrum. The ventral postcleithrum a simple shaft,
tapering to a point posteriorly.

Coracoid. — Wide dorsally, tapering to a blunt
point ventrally; with a short spinelike process under the
lowermost actinost from its posteroventral edge; the up-
per one-third of its posterior edge with an inturned flange
directed anteromedially, forming a broad basin for mus-
cle attachment; cartilage filled at its dorsal edge and for
a short distance at its anteroventral edge; articulates by
fibrous tissue anteriorly with the cleithrum, which
broadly overlies the dorsal one-half of its anterior edge;
articulates anterodorsally through cartilage and inter-
digitation with the scapula; articulates dorsally by fi-
brous tissue with the bases of the ventralmost two ac-
tinosts.

Scapula. — Completely encloses the scapular
foramen; cartilage filled at its anterior and ventral edges;
articulates anteriorly and anteroventrally by fibrous tis-
sue with the overlying cleithrum; articulates postero-
ventrally through cartilage and interdigitation with the
coracoid. Along its posterodorsal edge the scapula ar-
ticulates by fibrous tissue with the following elements, in
order from anterodorsal to posteroventral: the first pec-
toral fin ray, the small first actinost, the second actinost,
and the anterior half of the base of the third actinost.
The articulations with the first pectoral fin ray and the
first actinost occur at upraised areas on the dorsal edge of
the scapula, the stubby projection supporting the first
pectoral fin ray being slightly larger than that sup-
porting the first actinost.

Actinosts. — Four elements; all cartilage filled at
both ends; first actinost small, the others increasing
slightly in size posteroventrally in the series; first two ac-
tinosts articulated to the scapula as described in the
preceding section; third actinost articulated with the
dorsal edge of the articular area between the scapula and
coracoid and the fourth actinost with the coracoid; all
the articulations are by fibrous tissue. Distally the ac-
tinosts support by fibrous tissue all of the fin rays, ex-
cept for the first, which is supported by the scapula.

Fin rays. — Fourteen fin rays in most specimens,
with the first ray about one-fifth the length of the second
ray and articulated directly with the scapula, rather than
with the actinosts, as are the other rays. First ray with its
medial half enlarged and thickened at its basal region of
articulation with the scapula; its lateral half reduced to a
thin, short splint closely held to the medial half, or even
fused to it. First two rays and the last ray unbranched,
the others branched. First ray without cross-striations,
all the other rays cross-striated.

Pelvic Fin.

Pelvis. — A stout shaft; not divided into separated
right and left halves; anterior half of its lateral surface
with a concavity for muscle attachment; a large dorsal
lobe present from its posterodorsal region which serves as
a place of tough fibrous tissue attachment for the over-
lying skin of the distensible abdominal area or ventral
flap (dewlap). A series of enlarged scales encircles the
posterior end of the pelvis and obscures from view the
rudimentary fin-ray (or spinous) element present in the
midline just behind the end of the pelvis. The lateral sur-
face of the posterior end of the pelvis bears several ridges
for articulation with these scales. The scales occur in four
segments, the anteriormost of which is immovably held
to the end of the pelvis, while the other three segments
are movable in a dorsoventral plane against the scales of
the first segment and the end of the pelvis. The fin-ray
element is deeply bifurcate anteriorly into divergent
dorsal and ventral halves, as described in detail by Tyler
(1962:228-229, figs. 30-37). The anterior one-third of the

dorsal surface of the element has a groove in the midline
which superficially separates the element into right and
left halves, but the ventral surface shows no such groove.
Posteriorly the element becomes divided into four
slender rods, the tips of which project posteriorly a short
distance beyond the end of the encasing scales, through a
hole in the posterodorsal surface of the last scale seg-
ment. A large plug of cartilage intermediates between
the bifurcate anterior end of the fin-ray element and the
posterior end of the pelvis. The fin-ray element is
movable in a dorsoventral plane around the posterior end
of the pelvis, for ligaments are attached to the dorsal and
ventral halves of the anterior end of the element, and the
ligaments connect anteriorly to strong muscles. The ven-
tral ligament runs along a concavity on the ventral sur-
face of the pelvis to its muscle, while the dorsal ligament
runs anteriorly through a longitudinal tunnel in the basal
region of the dorsal lobe of the pelvis to make contact
with its muscle, which is housed in a concavity along the
dorsal surface of the pelvis anterior to the dorsal lobe.
The pelvis itself is movable in a dorsoventral plane
around its fibrous tissue attachment between the ventro-
medial edges of the cleithra.

VERTEBRAL COLUMN. —All vertebrae with bicon-
cave centra, except the last, which posteriorly ends in a
plate representing the fused centrum, hypurals, and
parhypural.

Abdominal Vertebrae.

First vertebra. — Neural spine enlarged, somewhat
laterally expanded, bifid to the centrum and thus not
forming a bony roof over the neural canal; articulates by
fibrous tissue and slight interdigitation along all of the
anterior face of its neural spine with the exoccipitals. The
rim of the concave anterior face of its centrum ar-
ticulates by fibrous tissue with the rim of the concave
posterior end of the basioccipital, while the rim of its pos-
terior face articulates similarly with the rim of the an-
terior face of the second vertebra. From about the mid-
dle of its posterolateral surface, the centrum possesses a
posteroventrally directed process which makes fibrous
tissue contact over the anterior half of the lower surface
of the centrum of the second vertebra. The neural pre-
zygapophysis of the second vertebra makes close fibrous
tissue contact with the dorsolateral surface of the first
vertebral centrum.

Other abdominal vertebrae. — In five specimens the
abdominal vertebrae numbered seven. Except for the
first vertebra, all the abdominal vertebrae, as well as all
of the caudal vertebrae except the last, possess bony
roofs over the neural canal and have single, undivided
neural spines. The neural spine of the third vertebra is
the largest of those of the abdominal vertebrae, for it is
much expanded dorsally. The basal regions of the neural
spines and the dorsal regions of the neural arches become
increasingly anteroposteriorly enlarged, from the second
to the last abdominal vertebra, into broad articular sur-

faces. These enlarged neural spines may have their an-
terior and posterior edges interdigitated with the
preceding and succeeding neural spines in large
specimens, but normally they articulate with one
another only by fibrous tissue. Each neural arch has a
neural foramen in its lateral surface. Short haemal post-
zygapophyses are present which slightly overlie the ven-
trolateral surface of the succeeding vertebral centra, but
there are no haemal prezygapophyses. The second to the
seventh abdominal vertebrae possess processes from
their centra which bear all but one of the seven
epipleural ribs. These processes become increasingly
longer and stouter posteriorly in the series. The second
vertebra bears a short lateral projection from its cen-
trum just below the anterior region of the neural arch,
and this projection articulates by fibrous tissue with the
first epipleural. The third to fifth vertebrae bear their
transverse processes progressively lower on their centra,
so that the fifth vertebra has its transverse process taking
its origin from the anteroventral edge of the centrum,
while the process itself is directed posteroventrally. The
third to the fifth vertebrae bear, respectively, the second
to fourth epipleurals from the dorsal surfaces of their
transverse processes. The processes of the sixth and
seventh vertebrae differ from those anterior to them in
that from the medial surface of the process on each side
of the centrum there is a medially directed projection
which meets and is continuous with that of its opposite
member, completely enclosing the haemal canal. The
haemal arch of the sixth vertebra does not have a haemal
spine, but that of the seventh vertebra does. The haemal
spine of the seventh vertebra, however, is not a single
medial piece. Rather, it is bifid and forms a short
posteroventrally directed process on either side of the
midline. The sixth and seventh vertebrae bear, respec-
tively, the fifth and sixth epipleurals from the lateral sur-
faces of their haemal arches. The neural spine of the fifth
vertebra articulates by fibrous tissue dorsally with the
supraneural strut that supports the posterior end of the
spiny dorsal fin pterygial elements, while posterodor-
sally it articulates with the first basal pterygial element
of the soft dorsal fin. The neural spines of the sixth and
seventh vertebrae articulate by fibrous tissue between,
respectively, the first and second and the second and
third basal pterygial elements of the soft dorsal fin.

Caudal Vertebrae. — The caudal vertebrae numbered
11 in five specimens. As with the more posterior of the
abdominal vertebrae, the first two caudal vertebrae have
their neural spines enlarged. From the third to the last
vertebra, however, the degree of enlargement progres-
sively decreases so that the more posterior caudal verte-
brae have relatively normal neural spines. All of the
caudal vertebrae, except the last, possess complete
neural and haemal arches. There are no haemal pre-
zygapophyses, and small haemal postzygapophyses are
only developed on a few of those vertebrae supporting the
anal fin. The first caudal vertebra has a thick haemal
spine that articulates firmly by fibrous tissue with the
upper anterior surface of the enlarged first anal fin basal

pterygiophore. From the lateral surface of the haemal
spine of the first caudal vertebra, just below the level of
the haemal arch, the seventh or last epipleural ar-
ticulates by fibrous tissue. The second caudal vertebra
has the largest haemal spine of any of the vertebrae, and
posterior to this vertebra the length of the haemal spines
progressively decreases until the 10th caudal vertebra is
reached. The 10th caudal vertebra has its haemal spine
longer than those just anterior to it, and in contrast to all
the others, its haemal spine is autogenous. The lengths of
the neural spines decrease from the first to the 10th
caudal vertebra. The 10th caudal vertebra has its neural
spine longer than those just anterior to it. The haemal
and neural spines of the last caudal vertebra are de-
scribed below.

Caudal Skeleton. — The caudal complex consists of an
epural, a free upper hypural, a free parhypural, and a
large plate composed of the centrum fused to most of the
hypural elements. The epural is a narrow shaft dorsally
but an expanded plate ventrally, where it articulates by
fibrous tissue with the neural arch of the centrum, some-
times more closely than that in the illustrated specimen.
The free upper hypural is wider dorsally than ventrally
and is held by fibrous tissue between the edges of the
epural and the posterior half of the dorsal edge of the fus-
ed hypural plate. The parhypural is a flattened rod
similarly held between the edges of the posteroventral
half of the autogenous haemal spine of the penultimate
vertebra and the ventral lobe of the fused hypural plate.
The large hypural plate is the centrum fused to what in
more generalized plectognaths are the first to fourth
hypurals, the deep indentation on the posterior edge of
the plate representing what would be the division
between the second and third hypurals. The centrum
region is thickened into a broad vertical crest for muscle
attachment. The two halves of the neural arch of the last
centrum diverge anteriorly, but more posteriorly they
sometimes approach each other more or less closely in
the midline, although they do not interdigitate or fuse.
The neural canal is thus relatively open above, never be-
ing more than very partially roofed over. Similarly, the
two halves of the haemal arch closely approach one
another ventromedially, but do not interdigitate or fuse
and leave the haemal canal only partially roofed over.
The haemal canal courses through the haemal arch to
exit at the deep notch in the anteroventral end of the fus-
ed hypural plate. The parhypural thus represents only
the haemal spine of the last vertebra, and not both the
spine and arch.

As discussed and illustrated by Tyler (1970b:14, fig. 22)
three of the five study specimens show no evidence of
uroneurals, nor do any of the other species of balistids ex-
amined, but two of the specimens of Balistapus un-
dulatus do. In one of these specimens there is a pair of
small nubbins of bone resting on the dorsal surface of the
upper free hypural and directly behind the posterior edge
of the epural, these apparently being the two halves of a
rudimentary uroneural. In the other specimen, the "uro-
neural" elements are much further forward, being a pair

of ossifications just above and behind the posterodorsal
region of the neural arch of the last centrum. Whether
these elements are homologous to the uroneurals of the
other specimen is questionable, for they could also be un-
consolidated neural arch material, as sometimes found in
triacanthodids, which have relatively well-developed
uroneurals.

Caudal fin rays. — Twelve in number; the upper-
most ray and the lowermost ray unbranched, the others
becoming increasingly branched toward the middle fin
rays, which are branched in triple dichotomies. The bifid
bases of the fin rays articulate by fibrous tissue with the
caudal skeleton as follows: the uppermost two rays to
the upper free hypural; the lowermost ray to the parhy-
pural and the remaining eight rays to the lobes of the
fused hypural plate, four to the dorsal lobe and four to
the ventral, with the lowermost branched ray receiving
partial support from the parhypural as well.

DORSAL AND ANAL FINS.

Spines and pterygiophores. — Three spines borne on
two basal pterygiophores. First spine long and stout; sec-
ond spine slightly shorter and narrower; third spine
about one-fourth the length of the first spine. The first
two spines are borne on the enlarged first pterygiophore
and the third spine on the much smaller second pteryg-
iophore. The mechanism by which the bifid bases of the
first two dorsal spines rotate over the medial upraised
portion of the dorsal surface of their pterygiophore and
lock themselves in an erected position, with the postero-
ventral edge of the first spine resting against the antero-
ventral edge of the second spine, has been described so
often in the past that a detailed redescription of it here
seems superfluous. For such descriptions see Hollard
(1853:102), Bnihl (1856:59; 1880; 1891:pl. 24), Mayer
(1864:fig. 4 of pi. 7), Bliss (1872:10), Thilo (1879:12;
1896b:291; 1902), Klein (1881:350), S(<rensen (1884:50;
1897), Mohr (1928), Jacobs (1935:156-157), Clothier
(1939), Smith (1949a:406), and Monod (1950, 1960). The
papers of S</rensen, Clothier, and Monod are par-
ticularly recommended. Anteriorly the first pterygio-
phore has a short, bluntly rounded, ventral process which
is firmly held by fibrous tissue in the concavity in the
midline of the posterodorsal region of the skull formed
dorsally by the supraoccipital and laterally and ven-
trally by the epiotics. Posteriorly the first pterygiophore
is broadly overlain by the anterior region of the second
pterygiophore, with which it interdigitates. On the ven-
tral edge of the second pterygiophore, there is a shallow
indentation where fibrous tissue articulation is establish-
ed with the thick shaftlike supraneural element. The
supraneural slants posteroventrally from its articulation
with the second pterygiophore to make fibrous tissue
contact with the dorsal edge of the neural spine of the
fifth abdominal vertebra and with the lower anterior
edge of the first basal pterygiophore of the soft dorsal fin.

109

Fin rays and ptery^ophores. — Twenty-six fin rays
are present in most of the study specimens; the first ray
unbranched, the others branched in single or double
dichotomies. Each fin ray has a small, unpaired, distal
pterygiophore between the bifurcate base of the ray.
Basally the fin rays are supported by 27 basal pterygio-
phores. Each basal pterygiophore has a well-developed
lateral flange throughout its length, except at the very
proximal and distal ends of the element, for muscle at-
tachment. The lateral flanges decrease slightly in height
posteriorly in the series. The first basal pterygiophore is
somewhat shorter, but stouter, than those just posterior
to it, and articulates by fibrous tissue ventrally between
the neural spines of the fifth and sixth abdominal verte-
brae, while anteroventrally it supports the base of the
supraneural. Posterodorsally the first basal pterygio-
phore interdigitates with the second basal pterygio-
phore, and all of the subsequent basal pterygiophores ex-
tensively interdigitate to one another along their edges of
contact, the more posterior elements only slightly less ex-
tensively so than more interiorly. The last few basal
pterygiophores articulate between the neural spines of
the seventh and eighth caudal vertebrae. The degree of
interdigitation between the pterygiophores, and their
closeness of apposition with the neural spines, increases
with increased size of the individual, so that in very large
specimens a massive, almost solid plate is present
between the vertebrae and the soft dorsal fin. The basal
pterygial elements are cartilage filled at their dorsal and
ventral edges. In all but one of the five study specimens
the number and arrangement of the basal pterygial
elements is as described above, but slight variation in the
number of elements is to be expected.

Fin rays and pterygiophores. — Twenty-three fin
rays are present in most of the study specimens; first ray
unbranched, the others branched in single or double
dichotomies. Each fin ray with a small, unpaired, distal
pterygiophore between its bifurcate base. Basally the
rays are supported by 22 basal pterygiophores, like those
of the dorsal fin. The first basal pterygiophore is by far
the largest of the series and has its anterior edge ex-
panded laterally. It articulates by fibrous tissue dorsally
between the haemal spines of the first and second caudal
vertebrae, while posteroventrally it interdigitates with
the second basal pterygiophore, and all of the sub-
sequent basal pterygiophores are extensively inter-
digitated to one another, as in the soft dorsal fin, with the
interdigitation and closeness of apposition to the haemal
spines increasing with increasing specimen size. The last
few basal pterygiophores articulate between the haemal
spines of the seventh and eighth caudal vertebrae.

Anatomical diversity.— The balistids are one of the
least anatomically and otherwise diversified families of
plectognaths, of a degree of diversity comparable to that
of the Triacanthidae, Aracanidae, and Diodontidae.
Internally the balistids differ remarkably little from one

another in fundamental plan. The vertebral column (7 -(-
11), basal pterygiophore support system of the soft dor-
sal and anal fins, and the caudal fin supporting struc-
tures are basically the same in all species. The skulls
differ mainly in the degree of development and massive-
ness of the supraoccipital and epiotic supports of the
basal pterygiophores of the spiny dorsal fin (least
developed in Rhinecanthus), perhaps correlated at least
in part with the size of the third spine, and of the degree
of anterior displacement of the suspensorium (most dis-
placed in Xanthichthys and, especially, Odonus), cor-
related with the degree of upturning of the mouth.

The genera of balistids are, in fact, distinguishable
mainly by external features such as: the degree of
development of a cusp or enlargement on the medial end
of the biting edge of the teeth (least developed in Melich-
thys, in which at least the more medial of the cusped
teeth of young specimens become truncate true incisors
with increasing size; best developed in Xanthichthys
and, especially, Odonus, in both of which the most
medial tooth of the premaxillary is smaller than that just
lateral to it, whereas it is the medial tooth that is the
largest in all other balistids; the enlarged cusp of the sec-
ond most medial tooth in Odonus projecting far beyond
the others and resting on the lower lip as a fang when the
jaws are closed); the presence or absence of a deep groove
in front of the eye and below the nostrils of unknown
functional significance (absent in Balistapus and Rhine-
canthus, and only weakly present in Pseudobalistes); the
presence or absence of enlarged, rounded, or otherwise
modified nonoverlapping and flexible scales covering the
tympanum (not enlarged in Canthidermis and
Xanthichthys); scales and spinulation patterns on the
head and body; degree of development of the third dor-
sal spine (least developed in Melichthys, Rhinecanthus,
and Xanthichthys; becoming minute in large specimens
of Canthidermis and Balistoides conspicillum) , these
diagnostic features supplemented by the shape of the
caudal peduncle and caudal fin. These and other features
in the anatomical diversity of the Recent species are dis-
cussed further in the section on generic relationships.

The balistids have changed little since the Oligocene.
In the black schist of Canton Glarus, Switzerland, there
are three species of balistids placed in the genus Balisto-
morphus that are, from what little is known of them from
their impressions, thoroughly modem in appearance.
The holotypes and a few additional specimens of these
three species have recently been reexamined and data
on them is given below.

Balistomorphus ovalis (Agassiz 1842 illustration, 1844b
description) is the most slender bodied of the three
species in the genus. The holotype, IGUN uncatalogued
(but XXXVH scratched on the back of the plate), a
single impression (head left) is 121 mm SL. The depth of
the body between the soft dorsal and anal fin origins is
45.3 mm (37.3% SL). The least depth of the caudal
peduncle is 12.8 mm (10.5% SL). The length of the first
dorsal spine is 25.7 mm (21.2% SL). The second dorsal
spine cannot be measured, but the third spine is 4.5 mm
(3.7% SL). There are 11 caudal vertebrae, as in Recent

prefrontal frontal pterosphenoid supraoccipital

metapterygoid hyomondibular

ethmoid

frontal

prootic pterotic porosphenoid
operculum

ethmoid

parasphenoid

prefrontal

sphenotic

C2) " pterotic

posttemporal

supraoccipital "s^' \l

epiotic

pterotic / ^

posttemporal

basioccipital

' exoccipital

exoccipital

epiotic

112

Figure 52. (opp. page, upper)— Baliatapus

undulatus: lateral view of head,

composite ba§ed on several specimens.

ca. 120-124 mm SL, western Pacific.

posttemporal

\ ^/ V / /r, ^

posterior myodome cavity
parasphenoid

pital basioccipital foramen magnur

neural spine

Figure 53. (opp. page, \ovieT)—BalUtapus
undulatus: dorsal (left) and ventral (right) vi<
of skull, composite based on several specimen
ca. 120-124 mm SL, western Pacific.

Figure 54. — Balistapus undulatus: posterior
view of orbit (left) (cross section of skull;

dashed lines represent cut surfaces of

fronlals, supraoccipital, and parasphenoid)

and posterior view of skull (right); below,

posterior and lateral views of first

abdominal vertebra; composite based on several

specimens, ca. 120-124 mm SL, western Pacific.

113

ilil

8 S « .

2 f' = J

llli

...sl.-l... llli

1 1 1 .1

0,0-- « a:

Figure 5><.—Balistapus undulatus:

lateral view of spiny dorsal

fin and its pterygial supports,

composite based on several specimens,

ca. 120-124 mm SL, western Pacific.

basal pterygiophore

r~ 3 fin I

Figure 59.—Balistapus undulatus:
left, dorsal view of pterygial supports

of spiny dorsal fin, with the spines

removed; right, posterior view of first

dorsal spine and anterior view of

second dorsal spine to show their

articular facets; composite based on

several specimens, ca. 120-124 mm SL,

western Pacific.

Figure SO.—Balistapus undulatus:

lateral view of caudal fin supporting structures,

composite based on several specimens,

ca. 120-124 mm SL, western Pacific.

caudal vertebrae

Figure 61.— External features of other

representative balistid genera: Abalistes

etetlaris— upper left, scale plates of the

tympanum region just behind the gill slit and above

the pectoral fin base; upper middle, scales

from upper middle region of body, including

lateral line canal bearing scales; upper

right, ventral view of encasing scales at

end of pelvis (anterior to left), with arrow

indicating major region of flexibility

(two terminal branches of modified

fin-ray element protrude posteriorly); lower

left, nasal region as seen externally

(above) and the olfactory lamellae as seen

with the top of the nasal sac removed.

Figure 62.— External features of other

representative balistid genera:

Canthidermis maculatus— upper left, scales of

the relatively unmodified tympanum region

just behind the gill slit and above the

pectoral fin base; upper middle, scales

from upper middle region of body, including

lateral line canal bearing scales; upper

right, ventral view of encasing scales at

end of pelvis (anterior to left), with

arrow indicating major region of Hexibility

(modified fin-ray element not protruding);

lower left, nasal region as seen

externally (above) and the olfactory

lamellae as seen with the top of the

nasal sac removed.

Figure 63.— External features of

other representative balistid

genera: Xanthichthya ringens —

upper left, scales of tympanum

region just behind the gill slit

and above the pectoral fin

base; upper middle, scales from

upper middle region of body,

including lateral line canal

bearing scales; upper right,

ventral view of encasing scales

at end of pelvis (anterior to

left), with arrow indicating

major region of flexibility

(three terminal branches of

modified fin-ray element protrude

posteriorly); lower left,

nasal region as seen externally

(above) and the olfactory

lamellae as seen with the top

of the nasal sac removed.

Figure 64.— External features of other

representative balistid genera : Odonua niger—

upper left, scale plates of the tympanum

region just behind the gill slit and above the

pectoral fin base; upper middle,

scales from upper middle region of body,

including lateral line canal bearing scales;

upper right, ventral view of encasing

scales at end of pelvis (anterior to left),

with arrow indicating major region of

fiexibility (two terminal branches of

modified fin-ray element protrude posteriorly);

lower left, nasal region as

seen externally (above) and the olfactory

lamellae as seen with the top of the

nasal sac removed.

species, and room for about seven abdominal vertebrae,
also as in Recent species, although only the last four ab-
dominal vertebrae can be seen relatively clearly. There
appear to be about 12 caudal fin rays in a slightly round-
ed fin. There is a well-developed supraneural strut sup-
porting the rear end of the carina. The basal pterygio-
phores that are best shown, anteriorly in the soft dorsal
fin, have well-developed lateral flanges. The form and
number of the teeth do not show. The stout pelvis has a
small but distinct dorsal lobe posteriorly. The scales,
best seen posteriorly on the body, bear one or more low
longitudinal ridges along the basal plate, those in the
region of the caudal peduncle being larger and longer
than elsewhere and oriented in longitudinal rows, just as
in many Recent species.

An additional specimen from the black schist of Can-
ton Glarus, NSKG 178b, 119 mm SL, can be assigned to
Balistomorphus oualis, mainly on the basis of its slender
body, the depth between the soft dorsal and anal fin
origins being about 36 mm (30% SL). It is a poor impres-
sion that at one time was in counterpart, but the head to
the right plate (presumably numbered 178a) cannot be
found. The caudal peduncle depth is 12.4 mm (10.4%
SL). The first dorsal spine is 22.8 mm (19.2% SL), the
second 9.5 mm (8.09; SL), and the third 5.1 mm (4.3%
SL), all of which measurements are compatible with that
of the holotype of B. oualis from the same strata, as are
the well-developed supraneural strut supporting the car-
ina and the approximately 11 caudal vertebrae. Little
else of interest can be seen.

Balistomorphus spinosus (Agassiz 1842 illustration,
1844b description) is a moderately deep bodied species.
The holotype, and only relatively complete specimen
available, BMNH P. 3973, a single impression (head
right) is 92.0 mm SL. The depth of the body between the
soft dorsal and anal fin origins is 46.0 mm (50% SL). The
least depth of the caudal peduncle is 12.2 mm (13.3%
SL). The length of the first dorsal spine, which bears low
asperities laterally, is 21.0 mm (22.8% SL). The second
dorsal spine cannot be measured, but the third spine is

5.0 mm (5.4% SL). The vertebral column can be inter-
preted either as 7 + 11, as in Recent species, or as 8 + 10.
The caudal fin appears to have 12 rays. There is a well-
developed supraneural strut supporting the rear end of
the carina. The basal pterygiophores have well-
developed lateral flanges. The form and number of the
teeth do not show. There is a slight indication that the
stout pelvis had a low dorsal lobe posteriorly. The scales
are not as well indicated as in B. oualis, but low granula-
tions of some sort were present on the basal plates. The
anal fin base seems to have been substantially shorter
than the soft dorsal fin base, more so than in B. oualis.

An additional specimen from the black schist of Can-
ton Glarus, NSKG 189a and b, which is too poor an im-
pression in counterpart to accurately measure, can
probably be assigned to B. spinosus on the basis of its
moderately deep body, as can another not listed in the
material examined (BMNH P. 1819) from the same
strata whose bones are too disorganized to measure.
These additional specimens add little to our knowledge
of the species, except that the first dorsal spine in one of
them bears even better developed asperities laterally,
and apparently anteriorly, than does the holotype of B.
spinosus.

Balistomorphus orbiculatus (Heer 1865, see also 1876)
is an extremely deep bodied species. The holotype, and
only known specimen, NSKG 2688, a single impression
(head left), is 65.6 mm SL. Heer's illustration of the
specimen shows the head facing right, so either the illus-
tration was reversed or, less likely, the specimen was
originally in counterpart and only the head to left half
remains in the Glarus collection. Woodward (1901:568)
believed the great depth of the specimen to be due to dis-
tortion, but, as with the other specimens of Balistomor-
phus (and of the Oligocene triacanthid Acanthopleurus,
which see) from the black schist of Canton Glarus, the
neural and haemal spines of the vertebrae have a normal
relationship with the basal pterygiophores of the soft dor-
sal and anal fins and the patterns of the scales are nor-
mal, neither of which would be the case if B. orbiculatus

Figure 65. — Fossil balistids:
A. Balistomorphus ovalU, lateral view
of holotype, 121 mm SL, Oligocene of
Canton Glarus, Switzerland (Agassiz

1842:pl. 75); B, B. ovalis, lateral

view of entire specimen, 119 mm SL,

Oligocene of Canton Glarus, Switzerland:

C, B. spinosus. lateral view of holotype,

ca. 9U mm SL, Oligocene of Canton
Glarus, Switzerland (Agassiz 1842:pl. 75);

D, B. orbiculatus, lateral view of

holotype, 65. G mm SL, Oligocene of Canton

Glarus, Switzerland; E, Oligobalistes

robustus, lateral view of syntype,

ca. 60 mm SL, Maikop deposits of the

Oligocene of the Caucasus (Danil'chenko

1960:175, fig. 32).

were unnaturally drawn out vertically. The depth of the
body between the soft dorsal and anal fin origins is 63.6
mm (97.0% SL). The least depth of the caudal peduncle
is 20.1 mm (30.6% SL), this being substantially deeper
than in the other two species of the genus. The length of
the first dorsal spine is 24.6 mm (37.5% SL). The second
dorsal spine cannot be measured, but the third spine is
3.9 mm (5.9% SL). The number of dorsal and anal fin
rays appears to have been similar, and between about 20
and 25. The vertebral column cannot be seen in the ab-
dominal region, but the caudal series can be interpreted
as either 11, as in Recent species, or 12. The caudal fin
appears to have 12 rays. There is a well-developed, es-
pecially obliquely placed, supraneural strut supporting
the rear end of the carina. The basal pterygiophores have
exceptionally well-developed lateral flanges. The form
and number of the teeth are poorly indicated as large in-
cisors, with at least three probably located in an outer
series in the upper jaw. The stout pelvis has a low but
distinct dorsal lobe posteriorly, and there is some in-
dication that the scales between the posterior end of the
pelvis and the anus were enlarged and elongate to sup-
port a small fan or dewlap in this region, as in most Re-
cent species. The scale pattern is relatively well preserv-
ed, and the basal plates appear to bear two or three more
or less vertically oriented low emarginate ridges, sub-
stantially different from the condition of spinulation in
the other two species of the genus.

In addition to the three species of Balistomorphus, one
other species of balistid is known from the Oligocene seas
of Europe. Oligobalistes robustus Danil'chenko (1960),
known from three specimens of 27 to 60 mm SL from the
Maikop deposits of the Caucasus, is the best preserved
fossil balistid. It has not been reexamined for this work,
and the following is based on the description by
Danil'chenko. Oligobalistes differs from Balisto-
morphus by having no surface sculpturing on the scale
plates, and by the better development of the second and
third dorsal spines (respectively 15 and 10% SL in
Oligobalistes versus 8 and 4 to 6% SL in Balisto-
morphus), although the first dorsal spine of Oligobalistes
is slightly less well developed (30% SL) than in one of the
species of Balistomorphus (B. orbiculatus, 37.5% SL).
DanH'chenko believed Oligobalistes to differ from all of
the Recent genera by the large size of the first dorsal
spine and the presence of longitudinal rows of spinules
along its anterior edge. The first dorsal spine is slightly
longer, but not stouter, than in the adults of any Recent
genera, some of which (e.g., Rhinecanthus) have
longitudinal rows of spinules along the anterior and
lateral edges of the scales that are as well developed as in
Oligobalistes. If the scales of Oligobalistes are com-
pletely smooth then it also differs in this respect from the
Recent genera, all of which have at least low emar-
ginations or spinules.

The teeth in Oligobalistes were described as num-
bering four to six in each jaw, but presumedly any num-
ber above the four in an outer series (as found in all Re-
cent species) represent misplaced teeth from another
outer series or from any of the inner series. The vertebral

column was described as 7 -I- 10, but the illustration
clearly shows 11 caudal vertebrae, and a 7 -I- 11 column,
as in all Recent species.

Danil'chenko thought Oligobalistes most closely
related to the Recent Balistes, but because of the ab-
sence of a preocular groove, Oligobalistes is probably
more closely related to Balistapus and Rhinecanthus, as
discussed under Generic Relationships.

Generic relationships.— The generic relationships
within the Balistidae are not especially clear on the basis
of the present work. Features that a priori seem general-
ized on the basis of the evolution of balistids from tria-
canthids are a terminal mouth, well -developed third dor-
sal spine and overlapping scale plates.

The relatively straight-edged or gently curved heavy
crushing incisors of triacanthids contrast to the slightly
fewer in number, but at least superficially larger, notch-
ed teeth basic to balistids. While slightly larger in sur-
face area, balistid teeth are thinner edged and less heavy
than in triacanthids, except in very large individuals of
most species of balistids, in which the teeth become just
as massive if not more so than in triacanthids, while still
retaining at least some degree of notching. The presence
of unnotched edges on the more medial teeth in adult
Melichthys is not an indication of a generalized con-
dition, for Berry and Baldwin (1966:449) have shown that
the teeth are normally cusped in juveniles and only
gradually do the more medial teeth become worn down to
a relatively straight edge in adults, definite remnants of
the notching always being present on the more lateral
teeth.

However, the dentition of Xanthichthys and Odonus
can surely be considered a specialization, with the en-
largement of the second most medial tooth in the
premaxillary and, in Odonus, of its cusp, into a fang. The
second most medial tooth in the dentary is also often
slightly larger in Xanthichthys and Odonus than it
would be in other balistids. In contrast to all other
balistids, which have white teeth, the teeth oi Xanthich-
thys have a dark tinge, at least as adults (Berry and
Baldwin 1966:456), and those of Odonus are a deep red.
Also in contrast to other balistids, the teeth in Xanthich-
thys and Odonus, as well as in Canthidermis, the only
truly oceanic balistid, do not become exceedingly mas-
sive and thickened in large adults, the dietary changes
between juveniles and large adults in these three genera
apparently not being at all as extensive as in other

The specialized condition of the anterior displace-
ment of the suspensorium and lower jaw is found most
prominently in Xanthichthys and, especially, Odonus,
the mouth being distinctly supraterminal. The mouth in
Melichthys is slightly supraterminal, but in other
balistid genera it is essentially terminal.

Probably not much importance should be placed on
the size of the third dorsal spine as an indicator of
specialization and relationship. The third spine is not in-
volved in the locking mechanism of the first two spines,
and it has neither a complex basal articulation

120

■ Odonus
Xanthichthys

Melichthys

Sufflamen

Pseudobalistes

Balistomorphus
Oligobalistes
(Oligocene)

Figure 66.— Hypothesized phylogenetic
relationships of the genera of Balistidae.

mechanism with its basal pterygiophore nor a complex
extensive musculature. The third spine functionally
seems rather insignificant, and great variation in such a
structure can be expected, the variation from genus to
genus or within a genus being of little value to deter-
mining relationships. Thus, the minute third dorsal
spine found in Rhinecanthus, Melichthys, and
Xanthichthys probably does not indicate close relation-
ship between these three otherwise relatively dissimilar
genera, nor does it necessEirily indicate that they are
specialized in the face of almost any other evidence to the
contrary. The third dorsal spine is well developed in the
Oligocene genera Balistomorphus and, especially,
Oligobalistes.

The significance of the presence of a preocular groove
and of specialized flexibly articulated nonoverlapping
scales over the tympanal region above the pectoral fin
base is difficult to interpret. Neither structure is present
in triacanthids. They occur only in balistids and not in
the derived monacanthids. The fossil record of the balistids,
first known from the Oligocene, consists of less than a
dozen specimens, many of them rather imperfect. The
specimens assigned to the three species of Balisto-
morphus from the Oligocene of Switzerland, including
the holotypes of each, have recently been reexamined,
and there is no indication of a preocular groove or of
specialized tympanal scales, but these regions are not
well shown on any of the specimens. However, in de-
scribing the well-preserved Oligobalistes robustus from
the Oligocene of Russia, Danil'chenko (1960:173-174)
specifically says that a preocular groove is absent, while
his statement that there is a "small body plate with
radial sculpture situated directly behind gill slit, level

Figure 67.— Dentition of representative balistids:

A. Baliatea polylepis, 56.1 mm SL, Galapagos;
. Xanthichthys lineopunctatus, 181 mm SL, Hawaii

C, Odonug niger, 173 mm SL, New Guinea:
3, Melichthys niger, 147 mm SL, locality unknown.

with dorsal end of latter" would seem to describe a
modified tympanal scale.

The two Recent genera which lack large tympanal
scales, Canthidermis and Xanthichthys, are otherwise
very different in structure and surely are not closely
related. It is simply impossible to say precisely at present
whether the generalized condition of balistids is the
presence of enlarged plates which have subsequently
been lost by a few genera, or whether the basal balistids
lacked specialized tympanal scales which were sub-
sequently developed by most genera. I think the former
view to be the more likely, and that the development of
specialized tympanal scales aiding in and refining sound
production was one of the major features of the basal
balistids which led to their differentiation from the
triacanthids, concomitant with the reduction of the
spiny dorsal fin and development of a complex locking
mechanism of the first two spines, the reduction of the
pelvic fin and modification of the pelvis into a stout
rotatable shaft, and increased sturdiness of the skull to
support the nonprotrusible jaws with a change in den-
tition.

In this view, Canthidermis lacks specialized tym-
panal scales simply because all of its scales have become
relatively small and nonoverlapping, perhaps associated
with the more flexible body covering necessary for lateral
flexion of the caudal peduncle in a strong swimming and
truly oceanic species. Xanthichthys, on the other hand,
seems closely related to Odonus, which has large
specialized tympanal scales, and I have no explanation
for the lack of them in Xanthichthys, except that in
Xanthichthys the scales in the tympanal region are
smaller than those surrounding them, but they are not
overlapping and are just about as flexible as those of the
tympanal region of other balistids. It would seem that

Xanthichthys has not truly lost the specialized tym-
panal scales, but simply reduced their size. It would be of
interest to know if the sound productions associated with
the drumming of the pectoral fin against the side of the
body (and thereby against the anterolateral end of the
swim bladder) differ in Canthidermis and Xanthichthys
from other balistids.

If the supposed absence of a preocular groove in the
fossil species of balistids, of which relatively complete
specimens are known only from the Oligocene, is cor-
rect, then that structure is a more recent acquisition by
all genera except BaUstapus and Rhinecanthus. The
latter two genera are very closely related, differing only
by Rhinecanthus having a lower supraoccipital crest, a
much smaller third dorsal spine and a more constricted
caudal peduncle with fewer rows of spine bearing scales.
In fact, BaUstapus could serve as a model for a generaliz-
ed balistid. The mouth is terminal, the third dorsal spine
well developed, the large specialized tympanal scales
well developed, the encasing scales at the end of the pel-
vis with great flexibility (although with a slightly reduc-
ed number of scales in one of the segments) and well-
cusped teeth at all sizes and the teeth of decreasing size
laterally in both jaws. It is possible, but hardly proven
here, that BaUstapus is a relative generalized balistid
and that its lack of a preocular groove may also represent
the generalized condition.

One final character must be considered, that of the
degree of development and massiveness of the supraoc-
cipital and epiotic supports of the basal pterygiophores of
the spiny dorsal fin. Is a high sturdy supraoccipital crest,
such as found in most balistids, the generalized con-
dition? In the early development of balistids from
triacanthids, it is reasonable to expect that the process of
conversion of the domelike uncrested supraoccipital of

^W

Figure 68.— Scales just above tip of
pectoral fin of Balistoides viridescens ,

114 mm SL, Indonesia, to show

the hexagonal shape often obtained by

the scale plates in adults as the

anterior and posterior edges of the

rhomboid plate become flattened.

^*

>£

triacanthids into the flatter and crested supraoccipital of
balistids passed through a low crest stage such as is
found today in Rhinecanthus before achieving the higher
and sturdier buttressing crest found in most other Recent
balistids. However, the low crest of Rhinecanthus could
well be a secondary reduction of the crest from a high
crested ancestral stock and simply be associated with the
low nape of Rhinecanthus.

In short, the generic relationships within the
Balistidae are not clear on the basis of the osteological
data presented here, nor are many of them clear on the
basis of external features. However, Xanthichthys and,
especially, Odonus, are two of the most specialized
genera and are probably closely related, the former an-
cestral to the latter. Canthidermis is also specialized, in
a much different way, for an oceanic existence, the scales
all being reduced in size and nonoverlapping and the
body thus more flexible. Canthidermis is also specialized
in having the full ossification of the skeleton much
delayed, as well as in having the most rudimentary pel-
vic apparatus among the balistids. Instead of the firm os-
sification and sturdiness of the skeleton of all other
balistids, the bones of specimens of Canthidermis under
about 100 mm remain poorly ossifed and slightly spongy,
although not to the extent found in the molids Mola and
Masturus, but less ossifed than in the molid Ranzania.
Only in fully adult specimens of Canthidermis is the
skeleton fully ossified. The pelvic fm ray of Canthidermis
is discussed below.

Balistapus and Rhinecanthus are also obviously close-
ly related, as explained above, the latter probably being
a specialized offshoot of the former. Likewise, Batistes
(including Nematobalistes and Verrunculus, properly
synonymized with Batistes by Berry and Baldwin
1966:435) and Pseudobalistes seem closely related, the
latter a derivitive of the former with reduced cheek scales
and preocular groove. Sufftamen may be related to
Batistes, and Metichthys to Xanthichthys, hnt Abatistes
and Batistoides are not clearly (to me) related closely to
any of the groups of genera discussed above. I suppose
that the basic question is whether the Batistes-like
genera are more generalized than the Batistapus -like
genera. I suspect the latter are the more generalized, but
have no firm evidence of it.

Fraser-Brunner (1935a:660-661) thought Canthi-
dermis to be the most primitive genus, "since there is no
evidence that it is derived from any of the forms having
ossified plates behind the gill opening." With this I do
not agree, Canthidermis simply having lost the overlap-
ping body scales found in all other genera, but from what
stock Canthidermis is derived I cannot say, although I
would suspect a Batistes-like origin.

Fraser-Brunner thought Abatistes to be a slightly
modified derivative of Batistes on a line leading to
Batistoides, Sufftamen, and Batistapus, with Rhine-
canthus a derivative of the latter. He also thought
Batistoides to be ancestral to Pseudobalistes on the one
hand and to Metichthys and hence Xanthichthys on the
other, with Odonus a derivative of Metichthys. I can add
nothing to these unproven speculations.

Relationship to the Monacanthidae.— Balistids are

clearly derived from triacanthids, with the major
changes involving the sturdier, stronger skull structure
associated with the nonprotrusible jaws with less mas-
sive crushing teeth, and heavier scales with less body
flexibility but more delicacy of swimming movements
associated with the sacrifice of speed for defensive
features, including the development of the complex
locking mechanism of the first two spines and a rotatable
pelvis. Balistids are likewise clearly ancestral to mona-
canthids, the major changes involving a reduction
in massiveness of the skull associated with the thinner,
more sharply edged teeth serving a more nibbling
function with finer food and thinner, less regularly over-
lapping scales associated with a sacrifice in defensive
features such as the reduction of the spiny dorsal fin and
less swimming strength for a more secretive and often
cryptic existence, blending into the environment rather
than standing out conspicuously from it. The more
secretive monacanthids also tend to be much smaller in
size than balistids, which are among the largest of plec-
tognaths (exclusive of molids), except that the four
species oi Atutera all reach sizes in excess of 200 mm;
one, A. monoceros, reaching over 5(X) mm SL (Berry
and Vogele 1961).

The more important osteological differences between
the balistids and monacanthids, as pointed out in the
comparative diagnosis of each, are mostly concerned
with the reduction in size and forward displacement of
the spiny dorsal fin and its basal pterygiophores and the
features associated with the more delicate teeth and jaws
and their supporting structures.

The heavy crested supraoccipital of balistids sup-
porting the front end of the first basal pterygiophore of
the spiny dorsal fin becomes, at least posteriorly, a
delicate flattened plate in monacanthids upon which the
ventral surface of the basal pterygiophore rests, at least
in part. However, a few genera of monacanthids retain a
vestige of the large vertical crest of balistids as a smaller
crest on an anterior prolongation of the bone above the
orbit, the crest supporting the ventral edge of the ver-
tical platelike anterior region of the basal pterygiophore.
The deep hole in the balistid skull surrounded by the
supraoccipital and epiotics in which is held the short
anteroventral shaft of the first basal pterygiophore is lost
in monacanthids, and the epiotics in the latter no longer
have the dorsal extension that in balistids helps buttress
the apex of the skull for the support of the anterior end of
the first basal pterygiophore.

The second basal pterygiophore of balistids sup-
porting the third spine is lost by monacanthids along
with the third spine and the supraneural strut that in
balistids supports the posterior end of the carina against
the vertebral column. The second dorsal spine of
balistids becomes much reduced in size in
monacanthids, but the locking mechanism, involving
only the basal regions of the two spines, remains the
same. The massive first dorsal spine of balistids becomes
more slender in monacanthids, and usually more highly
ornamented with spinules and barbs. However, the first

124

125

Mltl

dorsal spine in some balistids (especially Rhinecanthus)
has its anterior face well ornamented with relatively
large barbs, while some monacanthids have the first dor-
sal spine absolutely smooth. The first dorsal spine of
balistids serves, among other defensive functions, to
wedge the fishes into holes and cavities that they seek
out in danger and in rest. I suspect that the first dorsal
spine in most monacanthids is used less in this manner
and more as a rapid vibrator in communication during
agonistic and breeding behavior. In some monacanthids
the spine can be vibrated back and forth with such speed
that it blurs before the eyes.

The typical crushing dentition of balistids has been
converted in monacanthids to more of a nibbling form,
while the massive supporting structures of the balistid
jaws give way in monacanthids to a much more delicate
suspensorium, the palatine being reduced in size and less
intimately connected with the ectopterygoid, and the
ethmoid and parasphenoid far less massive. The pharyn-
geal apparatus of balistids is changed in monacanthids
by the loss of the suspensory pharyngobranchial, the loss
of teeth on the fifth ceratobranchial, and the usual loss of
the second branchiostegal ray.

A balistidlike pelvis is found in the more generalized
monacanthids, although it usually is not quite so mas-
sive as in balistids. However, the main difference
between the pelvic apparatus in balistids and generaliz-
ed monacanthids is in the reduction in the size of the
rudimentary fin ray (or spinous element) at the end of
the pelvis and of the number of series of specialized
scales surrounding it. In all balistids there are four series
of encasing scales, and in all but Canthidermis these
series have a dorsoventral flexibility. In the great majori-
ty of balistids the rodlike ossification extending pos-
teriorly from the end of the pelvis, representing a com-
pound formed from a right and left rudimentary and
highly modified fin ray or spine (see Tyler 1962b) of the
otherwise absent pelvic fin, exits through a foramen in
the fourth and last segment of encasing scales to
protrude to the exterior. The compound rod usually is
branched distally, although it has lost its segmentation.
This is the condition found in the various species ex-
amined of Batistes, Balistoides, Abalistes, Hemibalistes,
Balistapus, Rhinecanthus, Sufflamen, and Odonus. Of
the two species of Xanthichthys examined, the ray is like
that in most other balistids in X. ringens, but in X.
lineopunctatus the ray is shorter, and, while branched
distally, it does not protrude to the exterior and there is
no foramen in the last series of encasing scales in the
single large (181 mm) specimen examined. Since the ray
itself is similar in X. ringens and X. lineopunctatus other
than the shorter distal portion in the latter, it is possible
that small specimens of X. lineopunctatus have the ray
protruding through a foramen which eventually closes
over while the distal portion of the ray is resorbed.
However, I have no small specimens of the relatively
rarely collected X. lineopunctatus to examine for this. In
both of the two species of Melichthys examined the fin-
ray element is a large cone-shaped ossification represent-
ing only the anterior end of the element as found in most

Figure 74.— Lateral views of

relatively normal balistid skulls:

above, Batistes capriscus. ca. 360 mm SL

Gulf of Mexico; below, B. polytepis,

ca. 390 mm SL, Mexico.

Other balistids. It does not project to the exterior and
there is no foramen in the last series of scales. In Melich-
thys the encasing scales are flexible in young specimens
but tend to become less flexible in large adults (Berry
and Baldwin 1966). In Canthidermis, in which the en-
casing scales are only slightly flexible even in young
specimens and more or less immovable in large adults,
the fin-ray element is even more reduced than in Melich-
thys, being only a small conelike ossification at the end of
the pelvis which is far too short to protrude to the ex-
terior even if a nonexistant foramen in the last series of
encasing scales were present.

In the more generalized monacanthids there are three
series of encasing scales, and there is always at least one
region of great dorsoventral flexibility between them.
One of the encasing scale series of balistids has been lost,

but the flexibility retained, as is a dorsal lobe of the pel-
vis to strengthen the support of the modified enlarged
scales of the distensible skin between the end of the pel-
vis and the anus. In these generalized monacanthids the
fin-ray element is much more rudimentary than in
balistids, and, in contrast to the single rodlike piece of
balistids (except a stubby cone in Melichthys and
Canthidermis), the element in monacanthids is always
further reduced in size and divided into a dorsal and ven-
tral half widely separated by the plug of cartilage on
which they rest at the end of the pelvis. These two splints
represent only the anterodorsal and anteroventral ends of
the element as found in balistids. As discussed under the
anatomical diversity of the Monacanthidae, about half of
the species of that family examined have the above de-
scribed generalized condition of the pelvic apparatus, a

supraoccipito

reduced but functionally similar version of that of
balistids, but the other monacanthids have further
reduced the apparatus variously to the point of the total
absence of a fin -ray element, encasing scales and dorsal
pelvic lobe. It is clear, however, that it was a balistid
group with a flexible series of encasing scales that gave
rise to the monacanthids, the most generalized of which
also have a flexible series of scales.

Probably associated with a greater body flexibility in
monacanthids, the constant 7 + 11 = 18 vertebral for-
mula of balistids is changed in monacanthids by an in-
crease of at least one vertebra in the caudal series, and
often more. In balistids the basal pterygiophores of the
soft dorsal and anal fins bear a flange for muscle attach-
ment along their lengths. This is true also of
monacanthids, but in the latter family the crest is fur-
ther elaborated, particularly distally, where it is es-
pecially expanded laterally into a more or less hooklike
process partially separating the inclinator, erector, and
depressor muscles of each fin ray from those of the
preceding and succeeding rays. The distal pterygio-
phores are much less different between the two groups.
Those of most balistids, like their triacanthid ancestral
group, have the right and left halves fused into a single
piece, although far posteriorly in the soft dorsal and anal
fins the right and left halves may remain separate. The
distal pterygiophores of balistids may also be expected to
be in separate halves in very young specimens. Of the

Figure 75.— Lateral view of most specialized

balistid skull and suspensorium:

OdonuB niger, 173 mm SL, New Guinea.

balistids examined, only Balistes polylepis, B. ca-
priscus, B. ueluta, and Sufflamen verres have most of
the distal pterygiophores as separate halves, but the
specimens of these are all relatively small, and the distal
pterygiophores can be expected to be fused into a single
piece in adult specimens of these species. Only in
Canthidermis maculatus do the distal pterygiophores re-
main as separate halves even in large specimens. In an
80.1 mm specimen nearly all the distal pterygiophores
have separate halves, while in a large 217 mm adult
those in approximately the anterior one-third of the soft
dorsal and anal fins have the two halves fused into a
single piece while more posteriorly they remain separate.
The retention of mostly separate halves in the distal
pterygiophores of Canthidermis is probably related in
some way both to its oceanic existence and late develop-
ment of full ossification.

In the derived monacanthids, nearly all of the distal
pterygiophores are single pieces even in small specimens,
and always so in large specimens.

Figure 76.— Batistes polylepis: ventral

(left) and dorsal (right) views of skull,

ca. 390 mm SL, Mexico.

Relationship to the Triacanthidae.— The presence
in balistids of such features as a well-developed spiny
dorsal fin, a well-developed pelvis with a pelvic fin
(even though highly modified), discrete teeth in the jaws,
a caudal fin with 12 rays, supported by a complex
with a free epural and parhypural, and a free palatine
obviously relates the balistids to the triacanthoids,
while a number of characteristics shared by balistids
and triacanthids but not by triacanthodids indicates
that the triacanthids are the most closely related ances-
tral group of the balistids.

The main features of similarity between balistids and
triacanthids, followed by the dissimilar condition in
triacanthodids, as mostly summarized from Tyler (1968),
are: 1) large heavy incisorlike teeth in an outer series
and more molariform teeth in the inner series versus
relatively smaller, conical to wide and thin in an outer
series £md, when present, of the same type in an inner
series; 2) hyomandibular with a well-developed crest
across its outer surface versus no such crest; 3) pterotic
with a large ventral flange over the posterodorsal end of
the hyomandibular versus no such flange; 4) operculum

expanded in the middle about equally anteriorly and
posteriorly versus triangular in shape; 5) basihyal ab-
sent versus present; 6) pterosphenoids suturing to one
another in the midline of the posterior wall of the orbit
versus not in contact there; 7) spiny dorsal fin base much
shorter than soft dorsal fin base (except Protacanthodes,
the bases about equal) versus much longer (except
Eoplectus, the bases about equal); 8) epipleurals
relatively thick versus thin; 9) basal pterygiophores of
soft dorsal and anal fins with lateral flanges along their
lengths and sutured to one another distally versus
without flanges and not sutured; 10) a single free hypural
(probably more in Protacanthodes) versus three or more
free hypurals; 11) uppermost pectoral fin ray very short,
the lateral half smaller than the medial, versus
moderately short, the two halves of about the same size.
It is extremely unlikely that this great array of.
similarities between balistids and triacanthids in con-

il spine of Ist vertebra
of 1st vertebra
centrum of 2nd vertebra

Figure 77. — BatUtes capriscus: ventral view
of sltull, ca. 360 mm SL, Gulf of Mexico.

tradistinction to triacanthodids has been arrived at in-
dependently.

There are only three important ways in which balistids
are more similar to triacanthodids than to
triacanthids: 1) the parhypural is free; 2) the haemal
spine of the penultimate vertebra is autogenous; 3) and
the caudal peduncle is relatively shorter in balistids and
triacanthodids than in triacanthids, except
Protacanthodes. These specializations of the Recent
triacanthids are associated with their more rapid sus-
tained swimming in comparison to balistids and
triacanthodids, and it must be assumed that balistids
evolved from triacanthids at a level of organization less
specialized than that of the Recent species, probably at
about the level of Protacanthodes.

Although balistids (and, by implication, the
monacanthids and ostracioids, which see) are clearly
derived from triacanthids, the number of ways in which
balistids differ from triacanthids is impressive. Most of
these differences are associated with the development in
balistids of: 1) a locking mechanism of the first dorsal
spine through the agency of the second spine; 2) an even
heavier skull structure for greater buttressing of non-
protractile jaws adapted to an even heavier crushing or
biting function of the teeth than found in triacanthids; 3)
a rudimentary but complex pelvic fin-ray element; 4)
heavier scales and less flexible body for less powerful sus-
tained swimming but with greater delicacy of move-
ment.

The principal differences between balistids and
triacanthids, summarized and modified from Tyler
(1968), are as follows: There are usually six, and rarely
only five, dorsal fin spines and usually four, rarely three
or five, basal pterygiophores supporting them in
triacanthids, while in balistids there are only three
spines supported by two basal pterygiophores, although
the supraneural element of balistids is homologous to the
third basal pterygiophore of triacanthids, as discussed
subsequently.

In triacanthids there is an autogenous mechanism of
locking the first dorsal spine in an erected position in-
volving the apposition of its roughened ventral end
against the similarly roughened dorsal surface of the first
basal pterygiophore below it, the first spine having an
anteroposterior canal through its base which rotates
around a high dorsal pronglike flange on the basal pteryg-
iophore, the locking mechanism not directly involving
the second spine. In balistids the locking of the first spine
is not autogenous, depending on a roughened area on the
slightly convex anteroventral surface of the second spine
fitting into a roughened medial concavity on the postero-
ventral surface of the first spine, while the first spine has
no anteroposterior canal through it and the basal pteryg-
iophore no pronglike flange.

In triacanthids the first basal pterygiophore is of sim-
ple structure, being a stout vertical shaft articulated
between the lower rear of the skull and the neural spine
of the first vertebra, except in the Oligocene Crypto-
balistes in which it has only a short ventral shaft and ar-
ticulates high on the rear of the skull and apparently out

132

of contact with the first vertebra. In balistids the first
basal pterygiophore is a highly complex, horizontally
elongate structure articulated high on the rear of the
skull by a short ventral shaft. The pterygiophore features
a concave dorsal surface in the region of the first two
spines that opens laterally through the wall of the pte-
rygiophore to admit the muscles controlling the locking
mechanism, there being nothing similar to this in tria-
canthids.

In triacanthids the second basal pterygiophore, bear-
ing the third spine, is relatively small, has a relatively
flat dorsal surface, and is not sutured to the first basal
pterygiophore, while in balistids it is relatively large,
deeply concave dorsally, and is firmly sutured to the
first. The third basal pterygiophore of triacanthids, bear-
ing the fourth spine, is probably the pterygiophore
modified in balistids into the supraneural strut bracing
the rear end of the carina (formed by the first two pteryg-
iophores) against the tip of the neural spine of the fifth
vertebra and the anteroventral edge of the first basal
pterygiophore of the soft dorsal fin.

The articulation of the carina high on the rear of the
skull in balistids has drastically changed the configura-
tion of the skull in this region, but it still bears evidence
of the triacanthid ancestry of the balistids. In both the
triacanthids and balistids the posterior surface of the
supraoccipital is concave, and while the supraoccipital of
balistids is so modified that it is no longer a simple dome,
it does exclude the epiotics from meeting one another on
the dorsal surface of the skull, and the epiotics contact
the frontals, just as in triacanthids and their ancestral
hoUardiin triacanthodids, and not as in the tria-
canthodin triacanthodids. The face that one Oligocene
triacanthid {Cryptobalistes) has the otherwise
triacanthidlike spiny dorsal fin articulated high on the
rear of the skull by a short ventral shaft, even though the
more numerous basal pterygiophores are small and do
not form a carina, and there is neither a supraneural
strut nor a locking mechanism between the first two
spines, shows the potentiality and preadaptation of the
triacanthids to give rise to the balistid-type spiny dorsal
fin and pterygiophores.

In triacanthids the premaxillary pedicels are well
developed and help allow the protusibility of the upper
jaw, while the premaxillary and maxillary are movably
articulated to one another through fibrous tissue. In
balistids there are no premaxillary pedicels, the upper
jaw is not protrusible and the premaxillary and maxil-
lary are immovably held to one another, often by sutur-
ing. The palatine in triacanthids is a flat squarish bone
with an anterior prong articulating with the deeply
recessed dorsolateral surface of the maxillary, while in
balistids the palatine is T-shaped, the anterior end of the
cross bar articulated with a slight concavity on the dorso-
lateral surface of the maxillary. In triacanthids there are
usually five teeth in an outer series in each half of each
jaw, internal to which are one or two teeth in each half of
the upper jaw and one in each half of the lower jaw. In
balistids there are always four teeth in an outer series in
each half of each jaw, and three inner series teeth in each

half of the upper jaw but none in the lower jaw. Thus,
balistids have slightly reduced the number of teeth as
found in triacanthids, and also modified their form. In
triacanthids the edges of the teeth are straighter than in
the typically notched teeth of balistids, and while the
size of the teeth in balistids is usually larger than that of
triacanthids, they are thinner and less massive, except in
most large adult balistids in which the teeth become just
as massive as in triacanthids. The teeth in the balistids
Canthidermis, and, to a lesser extent, Odonus and
Xanthichthys, do not tend to become so massive in large
adults.

In triacanthids the pelvis is shaftlike only posterior to
the level of the pelvic spines in about the middle of its
length and cannot be rotated around its anterior ar-
ticulation with the cleithra. In balistids the pelvis is
shaftlike throughout its length and can be rotated in a
vertical plane around its anterior articulation with the
cleithra, while the pelvic fin is represented by a rudimen-
tary but complex structure representing the fused fins
from each side at the posterior end of the pelvis, sur-
rounded by enlarged encasing scales. Whether this com-
plex fusion product of the pelvic fins represents the
spines or rays of triacanthids is problematical. One sus-
pects that because the fin rays in triacanthids are
relatively rudimentary and often lost while the spines are
always well developed, that, in the process of reduction
of the pelvic fin from that of triacanthids to that of
balistids, the rays would be irrevocably lost early in the
process, leaving the spine alone to become further reduc-
ed in size and eventually fused with its opposite member
at the rear of the pelvis. As the spine became of relatively
minute size and less fully ossifed, it may have secondari-
ly become branched distally, as in many balistids, this
branching of the unstriated element perhaps not at all
homologous to that of rays.

It is possible that the reduced spine migrated pos-
teriorly along the shaftlike posterior portion of the pelvis
eventually to arrive at its extreme posterior end while at
the same time the anterior half of the pelvis became
shaftlike and rotatable. This seems more likely to me
than to account for the position of the balistid pelvic fin
as a result of the loss of the shaftlike posterior half of the
triacanthid pelvis at the same time that the spine was
reduced and the anterior half of the pelvis was becoming
shaftlike and rotatable and also greatly elongate so that
the stationary spine was moved posteriorly to the anal
region by the growth of the anterior half of the pelvis
rather than by migration.

The shaftlike form of at least the posterior half of the
pelvis in balistids, triacanthids, and hoUardiin
triacanthodids is additional evidence of the phylo-
genetic validity of their relationship.

The vertebrae of triacanthids are 8 -f 12, while those of
balistids are 7 -(- 11. In triacanthids there are three tooth-
ed pharyngobranchials following the toothless suspen-
sory element, but only two toothed elements in balistids.
In triacanthids the hyomandibular is supported dorsally
by the prootic, pterotic, and sphenotic, but in balistids
only by the prootic and pterotic. In triacanthids there is

133

no prootic shelf such as is present in balistids. In
triacanthids the interoperculum extends posteriorly to
the suboperculum, but in balistids it is shorter and more
rodlike, not extending posteriorly beyond the epihyal. In
triacanthids the parasphenoid is a relatively narrow
shaft in front of the orbit, but it is greatly expanded there
into a thick plate in balistids. In triacanthids the vomer
is larger and less concave anteriorly than in balistids,
which lack the posterolateral wings that in triacanthids
suture to the long forward extensions of the prefrontal
(with the possible exception of Protacanthodes) . In
triacanthids the posttemporal is larger and more super-
ficially held to the skull than in balistids.

The above outline of the similarities and differences
between triacanthids and balistids clearly indicates that
while balistids are highly differentiated from
triacanthids, the latter are the closest ancestral group of
the balistids, and that the level of organization of the Re-
cent triacanthids (and their close Oligocene relatives) is
too specialized in a few ways for their consideration as
being ancestral to balistids. Moreover, the same line of
triacanthids that gave rise to the balistids also gave rise
to the a^acanids, and the few features of both balistids
and aracanids that are more generalized than in Recent
triacanthids must have been present in the early
triacanthids leading to the balistids and aracanids.
These few generalized features required of the early
triacanthids are: 1) a free parhypural; 2) haemal spine
of penultimate vertebra autogenous; 3) caudal peduncle
relatively deep and not yet elongate and highly tapered
and the caudal fin not deeply forked; 4) soft dorsal fin
base probably not yet highly elongate; 5) the uppermost
pectoral fin ray only moderately reduced in size and the
two halves still of about equal length; 6) the teeth not yet
enlarged to massive incisors but somewhat intermediate
between conical and incisorlike; 7) the prefrontal not yet
with a long forward extension alongside the ethmoid
sutured anteriorly with posterolateral wings of the
vomer; 8) the second and third basal pterygiophores of
the spiny dorsal fin not yet greatly reduced in size.

Among the fossil triacanthids, the Oligocene Acantho-
pleurus and Marosichthys are so like the Recent
triacanthids that they are unlikely to be closely related to
the line of early triacanthids giving rise to the balistids
and aracanids, while the Oligocene Cryptobalistes has
certain balistid features in its configuration which shows
the possibility of triacanthids being able to give rise to
certain balistidlike features, even though Cryptobalistes
seems to have been an evolutionary dead end unrelated
to the line of triacanthids which were ancestral to the
balistids. The Eocene Protacanthodes, which is inter-
mediate between the triacanthodids and triacanthids, is
the only known example of the early triacanthids, and of
those features of its anatomy that are exposed it is
generalized enough to encompass some of the critical
generalized features required of the triacanthid line
leading to the balistids and aracanids as listed above.

In Protacanthodes the condition of the uppermost pec-
toral fin ray, the parhypural and the haemal spine of the
penultimate vertebra are not known, but the caudal

peduncle is relatively deep and of only moderate length
and the caudal fin is rounded. These are conditions to be
expected in the balistid-aracanid ancestral line, except
that the caudal peduncle of Protacanthodes is probably
already somewhat too long for that line. The more medial
of the teeth in the jaws aie large incisors with con-
stricted distal ends, and thus somewhat notched, while
more laterally the edges were relatively straighter. While
the teeth in Protacanthodes may have been slightly
smaller than in Recent species (except Trixiphichthys),
they are probably already slightly too enlarged and in-
cisorlike to be conveniently ancestral to those of the
balistid-aracanid line, even assuming that early balistids
had less massive teeth than at present and that those of
early aracanids were larger and less conical than at pres-
ent. The prefrontal in Protacanthodes is large and has
at least a moderate anteroventral extension alongside the
ethmoid, although there is no evidence that it formed a
long process articulating anteriorly with the vomer, and
it is better separated anterodorsally from the ethmoid
than in Recent triacanthids. Thus, the prefrontal of
Protacanthodes is not known to have been much more
specialized than in the balistid-aracanid line, although it
may have been slightly more specialized. The soft dorsal
fin base in Protacanthodes is not yet greatly elongate,
and the second and third basal pterygiophores of the
spiny dorsal fin are still moderately large, even though
the spiny dorsal fin base is much shorter than in the an-
cestral triacanthodids. The second basal pterygiophore
articulates ventrally with the tip of the neural spine of
the second vertebra and the third basal pterygiophore to
that of the third, so that the articulations of both of these
pterygiophores has already shifted forward relative to the
condition in triacanthodids. In the line ancestral to
balistids and aracanids the third basal pterygiophore
would be expected to retain an articulation with a more
posteriorly located neural spine, as in triacanthodids,
and especially as in Eoplectus, as discussed in detail
there. Thus, the second and third basal pterygiophores of
Protacanthodes are already slightly too specialized to be
conveniently ancestral to the condition postulated to be
necessary in the balistid-aracanid line.

While Protacanthodes is by far the most generalized of
the triacanthids, a number of its characteristics (caudal
peduncle length, tooth size and shape, placement of
third basal pterygiophore of spiny dorsal fin) are already
slightly too specialized for it to have given rise to the
balistids and aracanids. The divergence of the balistid-
aracanid line must have taken place among the early
triacanthids at a slightly more generalized (that is, more
triacanthodidlike) level of organization than as seen in
certain anatomical regions of Protacanthodes. In this
light the line of early triacanthids derived of a hollardiin-
like triacanthodid ancestry diverged into one line leading
with little change to Protacanthodes and to another line
which after far greater change gave rise to the balistids
and aracanids, the latter diverging the furthest from the
common triacanthid ancestral stock, and the
Protacanthodes line continuing on to give rise to the
Triacanthinae and probably Cryptobalistinae as well.

Family Monacanthidae

Comparative dia^osis (contrast with that of the
Balistidae). — The more or less incisorlike teeth relative-
ly more delicate, developed for nibbling, three outer and
two inner on each premaxillary, and three, occasionally
only two, in a single series on each dentary; pharyngo-
branchials consisting only of two uniserially toothed
elements; fifth ceratobranchial toothless but with a
series of gillrakers along its anterior edge (posterior edge
of last gill slit); usually two dorsal spines, sometimes
only one, the second spine, when present, not more than
about one-third the length of the first, and usually much
smaller, ending ventrally in delicate ventrolaterally
directed prongs from either side; dorsal fin spines sup-
ported by a single basal pterygiophore, without a supra-
neural strut; dorsal spines and at least most of the length
of the basal pterygiophore placed over the top of the
skull; no lateral foramen in the basal pterygiophore to ac-
commodate the venterolateral prongs of the second dor-
sal spine; at least the posterior region and usually all of
the supraoccipital a flat or slightly rounded thin plate
without a vertical crest and broadly overlain (only slight-
ly in Psilocephalus) by the first basal pterygiophore
above it, while in some genera an anterior prolongation of
the supraoccipital above the orbit bears a vertical crest
for articulation with the ventral edge of the vertical
platelike portion of the anterior region of the first basal
pter>-giophore; posterior region of the epiotic not ex-
panded dorsally, the medial edges of the two epiotics in
contact throughout their lengths, there being no large
foramen in the skull between the epiotics and the supra-
occipital; palatine more or less rodlike, never distinctly
T-shaped, but sometimes with a bulge or short process on
its ventral edge representing a rudiment of the long foot
of the T as found in balistids, usually connected by a long
ligament with the anterior edge of the ectoptergiod; ex-
occipital not meeting its opposite member in the midline
above the foramen magnum, only slightly if at all

protruding medially beyond the medial edge of the bifid
neural spine of the first vertebra; prefrontal relatively
small and delicate, articulated by fibrous tissue mainly
with the frontal and sometimes, to a much lesser extent,
with the ethmoid, but never directly contacting the para-
sphenoid; parasphenoid moderately to much expanded
laterally just behind the orbit (except in the elongate
Psilocephalus), the medial edge of the pterotic on the
ventral surface of the skull broadly separated from its op-
posite member there by the basioccipital and para-
sphenoid; parasphenoid not expanded or only very
slightly expanded dorsally in front of the orbit, only
slightly if at all overlying the ethmoid and not con-
tacting the prefrontal; the laterally expanded dorsal
region of the ethmoid relatively narrow and thin, much
less wide than the depth of the ventral platelike portion
ventral platelike flange of the ethmoid thin and delicate
relatively deep, much deeper posteriorly than an
teriorly, and only slightly if at all overlain by the para
sphenoid; posttemporal (absent in Psilocephalus) held
more superficially and in a much less deep groove on the
lateral surface of the pterotic; medial edge of dentary al-
ways denticulate; prootic shelf less well developed (and
sometimes absent), not extending forward to the level of
the middle of the orbit, except in Psilocephalus in which
it reaches the front of the orbit; posterior edge of supra-
cleithrum relatively straight and not posteriorly ex-
panded; postcleithrum without a dorsal spur on its dor-
sal edge and no specialized tympanal region present;
postcleithrum always a single piece; branchiostegal rays
2 -(- 4, 1 -I- 4, or 1 -I- 3; pelvis with or without a dorsal lobe
posteriorly on its dorsal surface; encasing scales at the
posterior end of the pelvis in three or fewer segments, or
of indeterminate number or absent, when present either
flexible or fixed; a rudimentary pelvic fin element either
present or absent; vertebrae 19 (usually 7 -I- 12, sometimes
6 + 13) or more (7 or 8 -I- 13 to 23), up to 30; vertebrae

Figure 78.— Range of diversity in

body form in the Monacanthidae:

A. Monacanthus ciliatus;

B. Amansea acopas;

C. Brachaluteres trossulus;

D, Pnlocephalua barbatua.

anterior to the first basal pterygiophore of the soft dorsal
fin four to eight, those posterior to the last basal pteryg-
iophores of both the soft dorsal and anal fins four to six;
neural and haemal arches of the last centrum poorly
developed and nearly always incomplete (complete in
Brachaluteres trossulus); the last centrum with a
horizontal crest for muscle attachment (reduced to a
prong in B. trossulus); uroneurals never present; an
upper free hypural occasionally absent; lateral flanges
on the soft dorsal and anal fin basal pterygiophores end-
ing distally in prominent hooklike lateral expansions,
except in Chaetoderma spinosissimus and Rudarius
minutus; the basal pterygiophores extending proximally
relatively closely to the neural and haemal arches
usually only anteriorly in the length of the series; all
of the dorsal, anal, and pectoral fin rays unbranched;
scales small, with thin basal plates and usually fine
spinulation, the basal plates either overlapping or well
separated but usually not regularly arranged; scales
above the pectoral fin base unmodified; most species
relatively small, adults usually being 2(X) mm or smaller,
except for some species of a few genera (e.g., Canther-
hines, Nauodon) that have larger adults and, especially,
Alutera. at least one species (monoceros) of which
reaches over 5(K) mm (Berry and Vogele 1961:66).

Detailed description of Monacanthua ciliatua.

Material examined.— Twelve cleared and stained
specimens, 38.3-119 mm.

SKULL.

Occipital Region.

Basioccipital. — A short column, expanded antero-
dorsally; cartilage filled at its anterior and anterodorsal
edges; articulates by interdigitation dorsally with the
exoccipitals, anteroventrally with the prootics, and
anteromedially with the overlying posterior end of the
parasphenoid. The rim of the round concave posterior
end of the basioccipital articulates by fibrous tissue with

the concave anterior face of the centrum of the first
vertebra. The ventromedial surface of the basioccipital is
deeply concave throughout its length. This concave
channel is mostly hidden from view by the overlying
parasphenoid, but the channel opens to the exterior pos-
teriorly at the base of the posterior bifurcation of the
parasphenoid. Anteriorly the channel opens into the
myodome where the anterior end of the basioccipital
forms the posterolateral and posterodorsal walls of the
myodome.

Exoccipital. — Cartilage filled along most of its ven-
tral, anterior, and dorsal edges; articulates through car-
tilage and slight interdigitation dorsally with the epiotic,
laterally through cartilage and interdigitation with the
pterotic, and ventromedially mainly through inter-
digitation with the basioccipital. Posteriorly the exoc-
cipitals form the lateral and ventral walls of the foramen
magnum, while dorsally the foramen is closed by the
epiotics. The posterior surface of the exoccipitals is
firmly attached by fibrous tissue and interdigitation with
the anterior surface of the bifid neural spine of the first
vertebra. A short posterior process from the postero-
medial portion of the exoccipital overlies the upper an-
terior region of the centrum of the first vertebra, this pos-
terior extention representing the modified exoccipital
condyle.

Supraoccipital. — Widest posteriorly, where its
relatively flat dorsal surface is mostly obscured from ex-
ternal view by the large basal pterygiophore of the spiny
dorsal fin, but with a long forward extension above the
orbit with a vertical crest intervening between the fron-
tals and the anterior third of the basal pterygiophore;
cartilage filled along all of the edges of its posterior flat-
tened surface; articulates dorsally by interdigitation with
the basal pterygiophore of the spiny dorsal fin, pos-
teriorly through cartilage and interdigitation with the
epiotics, laterally and anteriorly, and ventrally on its an-
terior prolongation, through interdigitation with the
frontals, while the extreme anterior end of the anterior
extension reaches to the posterior end of the ethmoid in
large specimens.

Figure 79.— Monacanthua ciliatus:

upper left, nasal region as seen

externally (far left) and the olfactory

lamella as seen with the top of the nasal

sac removed; middle left, scales from

upper middle region of body, including

two lateral line canal bearing scales:

lower left, ventral view of encasing scales

at end of pelvis (anterior to left).

Otic Region.

Pterotic. — Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates
through cartilage and interdigitation dorsally with the
epiotic, posteriorly with the exoccipital, ventromedially
on the ventral surface of the skull with the basioccipital
and parasphenoid, anterolaterally with the sphenotic,
laterally through extensive interdigitation with the
posttemporal which overlies a portion of its lateral
surface, and anteroraedially in the region of the orbit
by interdigitation with the prootic. The anteroventral
region of the pterotic is prolonged ventrally as a shaft -
like projection overlying and extending ventral to the
posterodorsal end of the hyomandibular, to which it
articulates by fibrous tissue. Just anteromedial to its
ventral shaft the pterotic bears a deep depression which
continues over the prootic and to which depression the
dorsal head of the hyomandibular is held by fibrous tis-

Sphenotic. —Cartilage filled along all of its edges of
articulation with the other cranial bones except for its
anteroventral prong; articulates in the orbit through car-
tilage and interdigitation anteromedially with the ptero-
sphenoid, prootic, and pterotic; articulates antero-
laterally by extensive interdigitation with the frontal,
posterodorsally through cartilage and slight inter-
digitation with the epiotic, and posteroventrally with the
pterotic.

Epiotic. — Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates dor-
somedially by extensive interdigitation with the over-
lying basal pterygiophore of the spiny dorsal fin. Below
this articulation with the basal pterygiophore the dor-
somedial edge of the epiotic articulates through cartilage
and interdigitation with its opposite member in the
midline. Anteriorly the epiotic articulates through
cartilage and interdigitation dorsally with a short
portion of the posterior edge of the supraoccipital and
through extensive interdigitation along most of its
anterior edge with the frontal, while anteroventrally
it articulates through cartilage and interdigitation
with the posterodorsal edge of the sphenotic. Ventrally
the epiotic articulates through cartilage and inter-
digitation with the pterotic and exoccipital, while the
posterolateral surface of the lower half of the epiotic
is broadly overlain and articulates by interdigitation
with the bifid neural spine of the first abdominal
vertebra.

Prootic. — Cartilage filled along all of its edges of
articulation with the other cranial bones, except antero-
medially; articulates by interdigitation medially with
the lateral surface of the parasphenoid, anteromedially
with the ascending fork of the parasphenoid, through
cartilage and interdigitation anterodorsally with the
pterosphenoid, anterolaterally with the sphenotic, and
posteriorly with the pterotic. A deep concavity on the

ventral surface of the prootic articulates by fibrous tissue
with the anterior half of the dorsal head of the hyoman-
dibular. The medial edges of the prootics form the lateral
walls of the myodome, while medially directed shelves
from the medial edges of the prootics meet in the midline
and articulate through cartilage and fibrous tissue with
one another to form the dorsal roof of the myodome. The
anterior edge of the myodome is formed by the prootics,
except ventrally, where it is formed by the para-
sphenoid. The anterior end of the prootic is prolonged
anteriorly as a subocular shelf which serves for muscle at-
tachment.

Orbital Region.

Frontal. — Widest posteriorly, tapering to a point
anteriorly; its lateral edge above the orbit only slightly
upraised and thickened; articulates by interdigitation
posteromedially with the flattened posterior portion of
the supraoccipital, which it slightly overlies, posteriorly
with the epiotic, and posteroventrally with the sphenotic.
Posteriorly in the rear of the orbit it articulates by fibrous
tissue with the pterosphenoid and by interdigitation with
the sphenotic. Medially above the orbit the frontal
slightly overlies and articulates by fibrous tissue and
slight interdigitation with the ventrolateral edge of the
anterior extension of the supraoccipital, just posterior to
which the frontal is broadly overlain and articulates by
extensive interdigitation with the laterally expanded
middle region of the basal pterygiophore of the spiny dor-
sal fin. The frontal interdigitates anteriorly with the eth-
moid, whose posterodorsal surface it broadly overlies,
while laterally at the front of the orbit the frontal articu-
lates by interdigitation with the dorsolateral surface of
the prefrontal. On its ventral surface, the medial edges of
the frontals are separated by the cartilaginous mass
which is continuous anteriorly with the remains of the
ethmoid cartilage.

Prefrontal. — Expanded dorsolaterally;
filled along its medial edge below the frontal; articu-
lates by interdigitation dorsomedially with the frontal,
while ventromedially it is continuous with the ethmoid
cartilage lying between it and its opposite member and
the posterior end of the ethmoid.

Parasphenoid. — Elongate, expanded laterally only
posteriorly in the region below the rear of the orbit; ven-
trally expanded into a thin keel in the region below and
just anterior to the orbit. The bifurcate posterior end of
the parasphenoid broadly overlies the ventral surface of
the basioccipital in the region of the deep concave groove
in the latter, roofing over in this region the channel
leading into the myodome. Articulates posteriorly by in-
terdigitation with the basioccipital, pterotics, and
prootics. Under the posterior region of the orbit the para-
sphenoid possesses a pair of short, slightly forked, dorso-
lateral projections which interdigitate with the prootics
and form the anterior edge of the ventral region of the
myodome. Anterior to the orbit the dorsal surface of the

137

parasphenoid is concave and receives the rounded ven-
tral edge of the platelike portion of the ethmoid, to which
it articulates by fibrous tissue. The anterior end of the
parasphenoid is deeply concave to receive the posterior
shaftlike portion of the vomer, to which it is held by
fibrous tissue.

Pterosphenoid. — Cartilage filled along all of its
lateral edges of articulation with the other cranial bones;
articulates through cartilage and fibrous tissue with the
frontal, which broadly overlies it, and through cartilage
and interdigitation with the sphenotic and prootic. Me-
dially the pterosphenoid possesses a projection toward
the midline which interdigitates with a similar projection
from its opposite member, the pterosphenoids here
meeting in the midline. The pterosphenoids also are iri
contact through cartilage at their extreme dorsomedial
ends.

Ethmoid Region.

Ethmoid. —Elongate and deep; expanded laterally
along its dorsal region but a deep flattened plate ventral-
ly, the rounded edge of the ventral plate fitting into a
concavity along the dorsal surface of the anterior end of
the parasphenoid and held there by fibrous tissue. More
anteriorly the ventral edge of the ethmoid articulates by
fibrous tissue and light interdigitation with the dorsal
surface of the anterior half of the vomer. Postero-
laterally the dorsal surface of the ethmoid is overlain and
interdigitated to the extreme anterior ends of the fron-
tals. Posteriorly the ethmoid articulates through its car-
tilage filled posterior edge with the anteroventral car-
tilaginous ends of the prefrontals. The anterior end of the
ethmoid is somewhat more laterally expanded than is the
dorsal edge behind it, especially ventrally where a flange
protrudes anteroventrally and articulates through
fibrous tissue with the posterior end of the palatine. The
extreme anterior end of the ethmoid remains somewhat
cartilaginous centrally, and it is against this cartila-
ginous face that the rear edge of the premaxillary abuts
in rotation.

Vomer. — An elongate shaft except anteriorly where
it is laterally expanded; articulates posteriorly at its
shaftlike portion by fibrous tissue to the concavity at the
anterior end of the parasphenoid; articulates by fibrous
tissue and slight interdigitation dorsally with the antero-
ventral end of the ethmoid, while anteriorly it helps sup-
port the posterior end of the upper jaw.

Mandibular Region.

Hyomandibular. — A flat shaft for most of its
length, somewhat expanded dorsally; cartilage filled
anteroventrally; the posterior edge thickened for articu-
lation by fibrous tissue with the preoperculum and, just
above the dorsal end of the latter, with the anterodorsal
head of the operculum; articulates by fibrous tissue dor-
sally with the groove on the ventral surfaces of the

prootic and pterotic, while posterodorsally the hyoman-
dibular abuts against and is firmly held by fibrous tissue
along the medial surface of the elongate portion of the
ventral flange of the pterotic. The articulation with the
head of the operculum is at a slight bulge of the hyo-
mandibular whose concave face is cartilage filled and
abuts against the similarly cartilage filled head of the
operculum. The anteroventral end of the hyomandib-
ular articulates variously through cartilage and fibrous
tissue with the posterior end of the metapterygoid,
symplectic, interhyal, and preoperculum.

Quadrate. — Widest posteriorly, tapering to a knob
anteriorly for articulation with the articular in the lower
jaw; cartilage filled at its posterior edge; deeply cleft
along its posterior edge to accommodate the anterior end
of the symplectic; articulates dorsally by extensive inter-
digitation with the ectopterygoid, while posteriorly it ar-
ticulates through cartilage with the mesopterygoid and
metapterygoid; articulates by fibrous tissue and inter-
digitation with the symplectic. Ventrally the quadrate
articulates by fibrous tissue with the preoperculum.

Metapterygoid. —A large flat plate curving
medially along its dorsal edge; cartilage filled at its
anterior and posterior edges; articulates anteriorly
through cartilage with the quadrate while anterodor-
sally it is overlain and interdigitated with the
mesopterygoid and anteroventrally extensively inter-
digitated with the symplectic. Along the middle of its
ventral edge it articulates through fibrous tissue with the
dorsal end of the interhyal. Posteroventrally the
metapterygoid articulates through fibrous tissue and car-
tilage with the preoperculum and anteroventral edge of
the hyomandibular.

Symplectic. — A flattened shaft; relatively large;
cartilage filled at its posterior and, to a lesser extent,
anterior edges; articulates by extensive interdigitation
posterodorsally with the metapterygoid and anteriorly
with the indented region of the quadrate; ventrally ar-
ticulates by fibrous tissue and, in some cases, by slight
interdigitation with the dorsal edge of the preoperculum.
The posterior cartilaginous end of the symplectic is in
contact with the region of articulation between the in-
terhyal and metapterygoid.

Palato-Pterygoid Region.

Palatine. —A more or less simple short rod with a
slight bulge along its ventral edge for articulation by
fibrous tissue with the extreme anterodorsal edge of the
ectopterygoid; articulates anteriorly by fibrous tissue
with the dorsal head of the maxillary and posteriorly
with the lateral flanges of the ethmoid and, to a lesser ex-
tent, with the anterolateral expansion of the vomer.

Ectopterygoid. —Elongate; more or less triangular,
the apex of the triangle being anterior in the middle of its
length where the ectopterygoid articulates through

fibrous tissue with the middle of the ventral edge of the
palatine; articulates by extensive interdigitation ventral-
ly with the quadrate and posteriorly with the
mesopterygoid.

Mesopterygoid. — A small plate extensively inter-
digitated anteriorly with the ectopterygoid and, to a
slight extent, with the quadrate in some specimens, and
posteriorly with the metapterygoid; its anteroventral
edge in contact with the cartilaginous region between the
quadrate and the mesopterygoid.

Opercular Region.

Operculum. — More or less thin and flat, except
dorsally where it thickens into a rounded shaft with an
articular facet for articulation by fibrous tissue with the
posterior edge of the dorsal region of the hyomandibular,
while ventrally it overlies and articulates by fibrous
tissue with the suboperculum.

Suboperculum. — Very thin, flat, and oblong; held
to the overlying operculum by fibrous tissue.

Interoperculum. — A straight rod, tapering slightly
to a point anteriorly; extends from the region of the in-
terhyal to just beyond the anterior end of the preoper-
culum; articulates by a tough ligament anteriorly with
the angular in the lower jaw, while at its posterior end
two ligaments are present. One is short and connects
with the dorsal surface of the epihyal just below the ar-
ticulation of the latter with the interhyal, while the other
one is long and runs posteriorly to connect to the anterior
edge of the operculum just above the point where the
operculum begins to overlie the suboperculum.

Preoperculum. — Expanded in its middle region; its
dorsal surface only slightly flattened for articulation with
the quadrate; articulates along the posterior portion of
its dorsal edge by fibrous tissue with the hyomandib-
ular, along the middle portion of its dorsal edge through
cartilage and fibrous tissue with the symplectic,
metapterygoid, and interhyal, and along the anterior
portion of its dorsal edge by fibrous tissue with the
quadrate.

Upper Jaw.

Premaxillary. — A medially curved plate, wider
dorsally than ventrally; its posterodorsal edge slightly
concave to articulate by fibrous tissue with the anterior
ends of the ethmoid and vomer. It also articulates by
fibrous tissue laterally with the medial surface of the
palatine, although the main articulation of the latter is
with the maxillary. The anterior edge of the premaxillary
forms the entire anterior border of the upper jaw and only
shares the border with the maxillary ventrally. The
premaxillary is in close apposition with the maxillary
and articulates immovably with it by fibrous tissue and
slight interdigitation. The flat dorsomedial edges of the

two premaxillaries are held closely to one another by
fibrous tissue. Each premaxillary bears five teeth, three
in an outer row and two in an inner row. Both of these
rows are in close contact with one another. The teeth are
borne in very shallow depressions on the surface of the
premaxillary in their fully formed condition. They
develop, however, from large deep sockets which open to
the surface just at the base of the old tooth which they
will replace. As a tooth erupts to the surface to replace an
old tooth, the socket from which it arose apparently is
overgrown or filled in with bone and disappears and the
tooth rests only in a shallow depression on the surface of
the premaxillary. Most of the interior of the premaxillary
is given over to the dental pulp cavities, these cavities
communicating with the exterior through numerous
holes on the inner and outer surfaces of the bone.

Maxillary. — Widest ventrally, becoming narrow in
the middle, and expanded again dorsally where it ar-
ticulates with the anterior head of the palatine; articu-
lates firmly with the premaxillary by fibrous tissue and
slight interdigitation. The medial surfaces of the ventral
rti-ioiis of the maxillary and, to a lesser extent, of the
premaxillary articulate with the lateral surface of the up-
per portion of the dentary, by fibrous tissue. The postero-
dorsal surface of the maxillary articulates by fibrous
tissue with the lateral expansions of the ethmoid and
vomer and with the medial surface of the palatine.

Dentary.— Wide posteriorly; its posterior edge
concave to accommodate the anterior portion of the £tr-
ticular, with which it articulates by interdigitation.
Posteroventrally the dentary articulates by fibrous tissue
and slight interdigitation with the angular. From the
lateral surface of its posterodorsal region the dentary ar-
ticulates by fibrous tissue with the medial surfaces of the
maxillary and, to a lesser extent, of the premaxillary.
Ventromedially the dentary has broad denticulations
with which it articulates through fibrous tissue with its
opposite member. Each dentary bears three teeth in a
single row, corresponding to the outer row in the premax-
illary. The teeth are borne in shallow sockets on the sur-
face of the dentary and are replaced by new teeth
developing in deep sockets just as described for the teeth
of the upper jaw. Most of the interior of the dentary is
filled with the pulp cavity of the developing teeth, the
cavity opening to the exterior at the concave posterior
edge of the dentary as well as at pores on the inner and
outer surface of the bone.

Articular.— Wide along its posterior edge, with a
concave facet for articulation by fibrous tissue with the
anterior knob of the quadrate; articulates by interdigi-
tation with the concave posteromedial surface of the den-
tary. At its extreme posteroventral edge the articular
interdigitates slightly with the angular. The sesamoid ar-
ticular is a small nubbin of bone closely held by fibrous

tissue to the medial surface of the articular just in front
of and above the concave articular facet of that bone.

Angular. — A small wedge of bone articulated by
interdigitation dorsally with the articular and anteriorly
with the dentary. Posteriorly the angular connects by
ligament with the interoperculum.

BRANCHIAL APPARATUS.

Hyoid Arch, Branchiostegal Rays, and Urohyal.

Hypohyals. —Both hypohyal elements well de-
veloped, and of about equal size; the dorsal element
cartilage filled at its ventral and posterior edges, the ven-
tral element cartilage filled at its dorsal edge; the two
elements articulate with one another through the car-
tilage that lies between them; the dorsal and ventral
elements articulate by fibrous tissue at their antero-
medial edges with their opposite members. The dorsal
hypohyal articulates posteromedially by fibrous tissue
with the lateral surface of the first basibranchial. The
posterior edge of the ventral hypohyal and a short portion
of the posteroventral edge of the dorsal hypohyal articu-
late through cartilage and fibrous tissue with the
ceratohyal, while the medial surface of the ventral
hypohyal is held by fibrous tissue to the anterodorsal
region of the urohyal.

Ceratohyal. — A flat shaft, wider posteriorly than
anteriorly; cartilage filled at its lower posterior edge; ar-
ticulates through cartilage anteriorly with the ventral
hypohyal and, to a lesser extent, with the dorsal
hypohyal; articulates posteriorly through cartilage and
interdigitation with the epihyal. K six branchiostegal
rays are present, the first two rays articulate in about the
middle or just behind the middle of the ventral edge of
the ceratohyal. However, only five branchiostegal rays
are usually present and only one articulates to the ven-
tral edge of the ceratohyal. Of the four branchiostegal
rays always present in the posterior group, the lower two
or even three have their fibrous tissue articulation
primarily with the lateral surface of the posterior end of
the ceratohyal in the region immediately below the inter-
digitation with the epihyal.

Epihyal. — Large, about the same size as each of the
hypohyals; cartilage filled at its ventral edge; articu-
lates through cartilage and interdigitation anteriorly
with the ceratohyal, while dorsally it articulates by
fibrous tissue with the interhyal and interoperculum.

Interhyal.— A short, thick rod; cartilage filled at
both ends; articulates by fibrous tissue ventrally with the
epihyal and dorsally with the extreme posterior end of
the symplectic and the ventral edge of the metapterygoid
just behind the symplectic.

Branchiostegal rays. — Usually five in number. Of

the 12 specimens examined, 11 have five branchios-
tegals on each side, one articulated to the middle of the
ventral surface of the ceratohyal and four articulated to
the lateral surfaces of the posterior and ventral ends of,
respectively, the ceratohyal and epihyal. However, one
specimen has five branchiostegals on one side and six on
the other. The side with the five branchiostegals has a
typical arrangement of one ray articulated to the ventral
surface of the ceratohyal and four articulated to the
lateral surface of the ceratohyal and epihyal. On the one
side with six branchiostegals, two are articulated to the
ventral surface of the ceratohyal and the other four more
posteriorly. In the illustration of the branchial apparatus
of M. ciliatus. the additional branchiostegal ray occa-
sionally present is indicated by dashed lines, this being
the branchiostegal ray immediately behind the large
first branchiostegal articulated to the ventral surface
of the ceratohyal. The branchiostegal rays are all
articulated by fibrous tissue to the ceratohyal or epihyal.
The first branchiostegal is much wider than are the
succeeding rays, the more posterior of which are so
narrow that they are, in essence, slender rods.

Urohyal. — Thick along its anterior and anterodor-
sal edges, otherwise a relatively thin plate; articulates by
fibrous tissue anterodorsally between the medial edges of
the ventral hypohyal, while posteriorly it articulates with
the ventral surfaces of the posterior end of the first
basibranchial and the anterior end of the second
basibranchial.

Branchial Arches.— All the elements are cartilage
filled at their edges of articulation with the other
elements of the series, and the articulations are usually
through cartilage and fibrous tissue. The branchial
arches are composed of three basibranchials, three pairs
of hypobranchials, five pairs of ceratobranchials, four
pairs of epibranchials, and two pairs of pharyngo-
branchials. Four gills are present, with a slit between the
fourth arch and the lower pharyngeal.

First arch.— Basi-, hypo-, cerato-, and epibran-
chial elements present. First basibranchial short, wider
posteriorly than anteriorly; articulates posteriorly with
the second basibranchial and posterolaterally with the
first hypobranchial. First hypobranchial of increasing
width posterodorsally; the largest of the hypobranchial
elements, which decrease in size posteriorly in the series;
articulates ventrally with the region of articulation
between the first and second basibranchials and dorsally
with the first ceratobranchial. First ceratobranchial a
stout rod; the longest of the ceratobranchial elements,
which decrease in length posteriorly in the series; no ven-
trally directed flange present on the ceratobranchials, ex-
cept for the fourth, which is somewhat compressed and
has a small ventral flange basally; articulates ventrally
with the first hypobranchial and dorsally with the first
epibranchial. First epibranchial a flattened plate,
rounded ventrally where it articulates with the cerato-

hyal but then diverging into two prongs; the anterior
prong articulates by fibrous tissue with the ventral
keel of the parasphenoid just anterior to the middle of
the orbit, while the posterior prong articulates with
the anteriormost pharyngobranchial (that of the second
arch) and the dorsal end of the second epibranchial. The
pharyngobranchial element of the first arch is absent.

Second arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Second basi-
branchial the longest of the three elements of the series
and slightly wider posteriorly than anteriorly; articu-
lates anteriorly with the first basibranchial, antero-
laterally with the first hypobranchials, posteriorly with
the third basibranchial, and posterolaterally with the
second hypobranchials. Second hypobranchial wider
posterodorsally than anteroventrally; articulates ventral-
ly with the area of articulation between the second and
third basibranchials and dorsally with the second cerate-
branchial. Second ceratobranchial a long rod with a
slightly expanded ventral end; articulates dorsally with
the second epibranchial. Second epibranchial a short
rod; articulates dorsally with the second pharyngo-
branchial and, to a lesser extent, with the posterodorsal
arm of the first epibranchial. Second pharyngobranchial
(the first of the two remaining elements of the series) an
oblong block bearing on its ventral surface a row of about
five slender sharply pointed teeth and articulating ven-
trally with the dorsal end of the second epibranchial and
the posterodorsal projection of the first epibranchial.
New teeth develop in sockets just behind and below the
bases of the old teeth.

Third arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Third basibran-
chial a short column, widest in the middle; articulates
anteriorly with the second basibranchial, anterolaterally
with the second hypobranchials, posterolaterally with
the third hypobranchials, and posteriorly with the
fourth ceratobranchials. Third hypobranchial rounded
posteriorly but with a short tapering anteroventral
process that articulates by fibrous tissue anteriorly
with the ventral surface of the more anterior basi-
branchials and the posterior arm of the urohyal; arti-
culates posterolaterally with the third ceratobranchial
and posteromedially with the posterior end of the third
basibranchial. Third ceratobranchial expanded ventrally
but shaftlike more distally; articulates ventrally with its
opposite member and with the third hypobranchial, and
dorsally with the third epibranchial. Third epibranchial
rodlike distally and expanded in the middle region; ar-
ticulates dorsally with the third pharjoigobranchial; the
expanded middle region is directed anteriorly and ar-
ticulates with the basal region of the second
epibranchial. Third pharyngobranchial like the second
pharyngobranchial except smaller and usually with only
three teeth, the teeth slightly smaller than those of the
other pharyngobranchial; articulates ventrally with the
distal ends of the third and fourth epibranchials. The two

pharyngobranchial elements are held to one another by
fibrous tissue.

Fourth arch. — Cerato- and epibranchial elements
only. Fourth ceratobranchial more compressed than the
other ceratobranchials and with a narrow ventral flange;
articulates ventrally with the ventral end of its opposite
member and that of the third ceratobranchial. Fourth
epibranchial rodlike, only slightly larger in the middle
than at either end; articulates ventrally with the fourth
ceratobranchial and dorsally with the third pharyngo-
branchial.

Fifth arch. — Ceratobranchial (lower pharyngeal)
element only. Fifth ceratobranchial the shortest of the
ceratobranchial elements; ntirrowest dorsally, widest in
the middle, and with an expanded base articulating with
the base of the fourth ceratobranchial; toothless.

PAIRED FIN GIRDLES.

Pectoral Fin.

Posttemporal. — A short relatively straight shaft
broadly overlying the lateral surface of the pterotic and,
to a lesser extent, the sphenotic and extensively inter-
digitated to both, and therefore immovably held to the
skull; articulates ventrally through fibrous tissue with
the dorsal head of the supracleithrum.

Supracleithrum. — A straight shaft of bone oriented
more or less vertically to the axis of the body; most of its
medial surface broadly overlying and held by fibrous
tissue to the lateral surface of the upper end of the
cleithrum, while dorsally it articulates immovably
through fibrous tissue with the head of the posttemporal.

Cleithrum. — Laterally expanded at its anterior
edge in the middle of its length; articulates by fibrous
tissue dorsally with the overlying supracleithrum and
with the upper end of the postcleithrum, which it broadly
overlies. Along its posterior edge the cleithrum overlies
the anterior region of the scapula and coracoid and ar-
ticulates with both through fibrous tissue and slight in-
terdigitation. Along the ventral two-thirds of its length
the medial edges of the two cleithra are firmly held to one
another by fibrous tissue, while ventromedially they also
articulate through strong fibrous tissue with the anterior
end of the shaftlike portion of the pelvis which is closely
held between the two cleithra.

Postcleithrum. — The postcleithrum is a long
flattened strong strut not divided into dorsal and ventral
parts. The dorsal end of the postcleithrum articulates by
fibrous tissue with the dorsomedial surface of the cleith-
rum. The postcleithrum is angled obliquely down from
its upper point of attachment toward the region of
the abdominal wall in front of the dorsal lobe of the
pelvis.

Coracoid. — Widest dorsally, tapering to a shaft
ventrally; with a slender spinelike process under the
lowermost actinost from its posteroventral edge; the up-
per one-third of its posterior edge with an intumed flange
medially; cartilage filled at its dorsal end and at the tip
of its ventral shaft; articulates by fibrous tissue and in-
terdigitation anteriorly with the cleithrum, which
overlies the upper anterior edge of the coracoid; articu-
lates anterodorsally through cartilage with the scapula;
posterodorsally the edge of the coracoid supports the
lowermost and largest actinost and sometimes, to a les-
ser extent, the lower edge of the base of the third acti-
nost.

Scapula. — Completely encloses the scapular
foramen; cartilage filled at its anterior and ventral edges;
articulates anteriorly by fibrous tissue and slight inter-
digitation with the cleithrum, while ventrally it articu-
lates through cartilage with the coracoid. Along its pos-
terodorsal edge the scapula articulates by fibrous tissue
with the following elements, in order from anterior to
posterior or above to below: the first and uppermost
pectoral fin ray attached to a projection from the surface
of the scapula, immediately below which the scapula
supports on a small flange the small uppermost actinost,
with the second and third actinosts articulating along
the relatively unmodified posterior edge of the sca-
pula, although the posterodorsal edge of the coracoid
helps to support the lower basal region of the third acti-
nost.

Actinosts. — Four elements; all cartilage filled at
both ends; the first actinost small, the others increasing
slightly in size from above to below in the series; articu-
lated to the scapula and coracoid as described in the
preceding section. Distally the actinosts support by
fibrous tissue all of the fin rays except the first, which is
supported by the scapula.

Fin rays. — Twelve fin rays in most specimens, with
the first ray about one-tenth the length of the second ray
and articulated directly with the scapula, rather than
with the actinosts, as are the other rays. First ray with its
medial half enlarged and thickened at its basal region of
articulation with the scapula; its lateral half reduced to a
thin short splint closely held to the medial half. All of the
pectoral fin rays unbranched; the first ray without cross-
striations, all the other rays cross -striated.

Pelvic Fin.

Pelvis. — A stout shaft; not divided into separate
right and left halves; anterior three-fourths of its lateral
surface with a concavity for muscle attachment; a large
relatively slender dorsal lobe present from its postero-
dorsal region which serves as a place of tough fibrous
tissue attachment for the overlying skin of the highly dis-
tensible abdominal area or ventral flap (dewlap). A
series of enlarged scales encircle the posterior end of the

pelvis and obscure from view the rudimentary fin-ray
element present in the midline just behind the end of the
pelvis. The lateral surface of the posterior end of the
pelvis bears several ridges for articulation with these
scales. The scales occur in three segments, with only the
distal segment movably articulated, the major region of
flexibility in the series being between the second and
third series. The fin-ray element is reduced to a single
nubbin of bone above and below the plug of cartilage that
extends posteriorly from the end of the pelvis, as describ-
ed in detail by Tyler (1962b:229-232, figs. 38-45). These
bony nubbins, representing the remains of a fin-ray (or
spinous) element, are closely held by fibrous tissue to the
cartilaginous plug. Running anteriorly from each of the
nubbins is a tendon. The tendon from the dorsal nubbin
courses through a hole in the dorsal flange of the pelvis to
attach to a muscle housed in the concave region on the
dorsal surface of the pelvis, while the tendon from the
ventral nubbin runs along a similar concavity on the ven-
tral surface of the pelvis. The fin-ray nubbin and the at-
tached plug of cartilage at the end of the pelvis are
movable in a dorsoventral plane around the posterior end
of the pelvis by the contractions of the muscles on the
dorsal and ventral surface of the pelvis to which they are
connected by ligament. The pelvis itself is movable in a
dorsoventral plane around its fibrous tissue attachment
between the ventromedial edges of the cleithra.

VERTEBRAL COLUMN. —All vertebrae with
biconcave centra, except the last, which posteriorly ends
in a plate representing the fused centrum, hypurals, and
parhypural.

Abdominal Vertebrae.

First vertebra. — Neural spine shorter than those of
the succeeding vertebrae but sturdy; laterally expanded,
bifid to the centrum and thus not forming a bony roof
over the neural canal; articulates by extensive interdigi-
tation along most of its anterior face with the exoc-
cipitals and epiotics. The rim of the concave anterior face
of its centrum articulates by fibrous tissue with the rim
of the concave posterior end of the basioccipital, while
the rim of its posterior face articulates similarly with the
rim of the anterior face of the second vertebra. Anteriorly
between the region of the centrum and the neural arch,
the first vertebra is slightly overlain by the short exoc-
cipital condyle. From the lower region of its postero-
lateral surface the centrum possesses a short postero-
ventrally directed process which makes fibrous tissue
contact over the anterior fifth of the lower surface of the
centrum of the second vertebra, although this is mostly
obscured from lateral view by the lateral flange of the
second vertebra.

Other abdominal vertebrae. —In 12 specimens the
abdominal vertebrae numbered six. Except for the first
vertebra, all the abdominal vertebrae, as well as all of the

142

caudal vertebrae except the last, possess bony roofs over
the neural canal and have single, undivided neural
spines. The neural spines of the abdominal vertebrae are
all of about the same length but they are of increasing
width posteriorly in the series. Their edges of apposition
articulate with one another through fibrous tissue and, in
some cases, slight interdigitation. Each neural arch has a
neural foramen in its lateral surface. Haemal prezyga-
pophyses are weakly developed on a few vertebrae, but
mostly they are absent or insignificant. Except for the
first, all of the abdominal vertebrae have transverse
processes which bear epipleural ribs. These processes
become increasingly laterally expanded and stouter pos-
teriorly in the series. The transverse process of the second
vertebra arises in about the middle of the anterior region
of the centrum, but the transverse processes of the subse-
quent vertebrae originate low on the centrum. The
epipleurals articulate by fibrous tissue to the dorso-
lateral surfaces of the transverse processes. The trans-
verse processes of the fifth and sixth abdominal
vertebrae have posteromedial flanges from each side
which make contact with the haemal regions of the
succeeding vertebrae, roofing over the haemal canal un-
der the fifth and sixth vertebrae, while more anteriorly
the haemal canal is open. The first basal pterygiophore of
the soft dorsal fin articulates between the neural spines
of the fourth and fifth vertebrae, the articulation being
by fibrous tissue. The second basal pterygiophore of the
soft dorsal fin articulates similarly between the neural
spines of the fifth and sixth abdominal vertebrae.

Caudal Vertebrae. —The caudal vertebrae number 11
in 12 specimens. The neural spines of the caudal
vertebrae become less expanded anterodorsally and in-
creasingly shorter posteriorly in the series, except for the
last few vertebrae, behind the soft dorsal and anal fin
bases, which have the neural and haemal spines expand-
ed. All of the caudal vertebrae, except for the last,
possess complete neural and haemal arches and spines.
Haemal pre- and postzygapophyses are essentially ab-
sent, but neural pre- and postzygapophyses are distinctly
developed in both the caudal and abdominal series. The
sturdy haemal spine of the first caudal vertebra is
somewhat concave ventrally along its posterior surface
and slightly bifurcate at its extreme distal end. It is
against the posterior surface of this haemal spine that
the first basal pterygiophore of the anal fin is firmly held
by fibrous tissue. The haemal spines of the second and
third caudal vertebrae are progressively slightly longer
than that of the first, but more posteriorly the length of
the spines decreases until those of the 10th to 12th
vertebrae, which are of increased length. The length of
the neural spine similarly decreases posteriorly in the
series and then increases behind the soft dorsal fin base.
The haemal spine of the penultimate vertebra is auto-
genous. The haemal and neural spines of the last caudal
vertebra are described below.

Caudal Skeleton. — The caudal complex usually

consists of an epural, a free parhypural, and a large plate
composed of the centrum fused to the hypural elements.
The free epural and parhypural are of about equal size.
The neural arch of the last centrum is incomplete, the
neural canal not being roofed over. Similarly the haemal
canal is not roofed over and its presence is simply in-
dicated by an indentation in the anteroventral edge of
the hypural plate just below the centrum and above the
dorsal end of the parhypural. The fused hypural plate
bears a horizontal keel for muscle attachment on the rear
portion of the centrum area and anterior region of the
fused hypurals. There is also a deep indentation on the
middle of the posterior edge of the fused plate, this in-
dentation representing what would be the division
between the second and third hypurals in a more
generalized plectognath having five free hypurals. The
above is the condition found in 10 of the 12 study
specimens. However, as described and illustrated by
Tyler (1970b:16, fig. 31), 2 of the 12 specimens have an
upper free hypural, representing the fifth hypural of
generalized plectognaths. The presence of an upper free
hypural is the norm for the family, all species examined
having one with the exception of, usually, M. ciliatus,
and the several specimens of Rudarius ercodes and
R. minutus cleared and stained or radiographed, and
the single specimen of Amanses scopas cleared and
stained.

Caudal fin ra.vs. — Twelve in number; the uppermost

ray and the lowermost ray unbranched, the others
becoming increasingly branched toward the middle fin
rays, which are branched in triple dichotomies. The bifid
bases of the fin rays articulate by fibrous tissue with the
caudal skeleton as follows: the upper six rays to the up-
per half of the fused hypural plate and the lower six rays
to the lower half of the plate.

DORSAL AND ANAL FINS.

Dorsal Fin

Spines and pterygiophores. — Two spines borne on a
large and elongate basal pterygiophore. First spine long
and moderately stout; second spine much smaller, con-
sisting of a basal rounded region, a slender tapering dis-
tal end and, from the ventrolateral edge of the basal ex-
panded region, a laterally projecting flange for muscle
attachment. The concave ventromedial region of the first
spine rotates over a low medial flange on the pterygio-
phore, while posteromedially the first spine has a ven-
trally directed flange which rotates into a deep concavity
on the basal pterygiophore just behind the medial flange.
The thick posterior region of this posteromedial flange of
the first spine is roughened and makes contact with the
similarly roughened anterior edge of the basally expand-
ed portion of the second spine to form the locking
mechanism. The concave ventral surface of the second
spine rotates over a low medial flange on the pterygio-

143

phore just posterior to that supporting the first spine.
The second spine also has a very short and blunt pos-
teromedial expansion which is accommodated in a low
concavity on the pterygiophore just behind the medial
flange over which the base of the spine rotates. The lock-
ing mechanism of the two spines is basically similar to
that in the Balistidae and continuous positions of erec-
tion from partial to full are possible.

The ventral surface of the basal pterygiophore is
strongly interdigitated with the top of the skull, especial-
ly with the frontals, and extends forward to the level of
the prefrontals. In addition to the frontals, the basal
pterygiophore is interdigitated with the supraoccipital
and epiotics. The pterygiophore is especially sturdy and
laterally expanded in the region below the first dorsal
spine.

Fin rays and pterygiophores. — There are usually
between 30 and 36 fin rays; all of the rays unbranched,
but with cross-striations. In the illustrated specimen the
31 fin rays are borne on an equal number of basal
pterygiophores. Each fin ray has a small, unpaired, distal
pterygiophore between the bifurcate base of the ray.
Each basal pterygiophore, except for the last several, has
a well-developed lateral ridge along its length, except at
the very proximal and distal ends of the element, for
muscle attachment. The lateral ridges decrease slightly
in height posteriorly in the series. Distally the ridge is es-
pecially laterally expanded and curved slightly ventrally
to form a hooklike process. The first basal pterygiophore
is slightly the largest in the series, the other pterygio-
phores decreasing slightly in length posteriorly in the
series. The first basal pterygiophore articulates by
fibrous tissue between the neural spines of the fourth and
fifth abdominal vertebrae, while the last few basal
pterygiophores are between those of the 9th and 10th
caudal vertebrae. Posterodorsally the first basal
pterygiophore interdigitates with the second basal
pterygiophore, and all of the subsequent basal pterygio-
phores are extensively interdigitated to one another
along their edges of contact, except for the last few basal
pterygiophores which are held to one another mainly by
fibrous tissue instead of extensive interdigitation. Just
below their distal ends the successive pterygiophores are
not in close contact and a foramen is formed between
their apposed edges. The degree of interdigitation
between the pterygiophores decreases slightly in the
series posteriorly, as does their degree of contact with the
neural spines supporting them. The basal pterygial
elements are cartilage filled at their dorsal and ventral
edges.

Fin rays and pterygiophores. — There are usually
between 30 and 35 fin rays supported by an equal
number of basal pterygiophores or by only one or two less
than the number of rays; all of the rays unbranched but
with cross-striations. In the illustrated specimen the 31
fin rays are supported by 29 basal pterygiophores. Each

fin ray has a small, unpaired, distal pterygiophore
between its bifurcate base. The basal pterygiophores of
the anal fin are similar to those of the soft dorsal fin and
bear the same kind of lateral flanges for muscle at-
tachment along most of their lengths. The first two basal
pterygiophores of the anal fin articulate by fibrous tissue
between the haemal spines of the first and second caudal
vertebrae, while they articulate by extensive interdigi-
tation with one another, as do the other pterygiophores in
the series. This interdigitation decreases posteriorly in
the series to the extent that the last few pterygiophores
articulate to one another mainly by fibrous tissue. The
pterygiophores decrease slightly in length posteriorly in
the series, as does the height of their lateral flanges.
There are usually three to four anal fin basal pterygio-
phores articulated between successive haemal spines,
whereas in the dorsal fin there are usually only two or
three basal pterygiophores between successive neural
spines, this because the number of basal pterygiophores
in the dorsal and anal fins is similar, whereas the dorsal
fin is longer based and extends over more vertebrae than
does the anal fin.

Anatomical diversity. — The monacanthids are one of
the more diversified families of plectognaths, com-
parable in many ways to the degree of diversity found in
the Recent triacanthodids, although without as much
diversity in dentition and with no indication of two major
phyletic lines within the family such as is evident in the
Recent triacanthodids. Fossil species of monacanthids
are not yet known, although the ancestral balistids have
been found as early as the Oligocene. The mona-
canthids, with about 90 species, are one of the most
speciose families of plectognaths, only the tetraodontids
having more species, perhaps a few over 100. Even
though two subfamilies of tetraodontids (tetraodontins
and canthigasterins) are easily recognizable and none
such are in the monacanthids, the latter still are
probably more anatomically diversified than the former.
That is, there is no sharp break in the structural con-
tinuity of monacanthids, while there is in tetraodontids
between Canthigaster and the tetraodontins, only par-
tially bridged by Carinotetraodon.

Most monacanthids have an only slightly elongate
body form, similar to that of balistids, although they
tend to be thinner bodied (more laterally compressed)
than balistids. However, monacanthids range from those
of nearly perfectly rounded outline {Brachaluteres,
Chaetoderma), in which the greatest body depth, in-
cluding the normal pelvic region expansion, approaches
the standard length, to the extremely elongate Psiloce-
phaluti, whose depth is contained many times in the
standard length.

Although always over the top of the head, the place-
ment of the first dorsal spine is variable, ranging from
over the rear edge of the eye (many genera) to well
forward of the eye (Pseudaluteres only), while in most
genera it originates somewhere over the rear half of the
eye. Two dorsal spines are present in all genera except

basal pterygiophi

Figure »2.—Monacanthu» ciliatus: dorsal (left)
and ventral (right) views of skull, 51.3 mm SL, Colombia.

Brachaluteres and Psilocephalus, in which only the first
spine remains. The single spine of Brachaluteres is rela-
tively well developed, but that of Psilocephalus is very
short and weak, as is its pterygiophore. In most genera
the first dorsal spine is ornamented with downward and
backward projecting spiny processes, especially along the
two posterolateral edges, while the anterior surface is
usually ornamented with smaller and less regularly ar-
ranged barbs, these oriented either upward, downward,
or straight forward. However, a few genera have the
barbs on the anterior face of the spine as well developed,
or almost so, as on the posterolateral edges: Laputa,
Chaetoderma, Arotrolepis, and Acanthaluteres . Others
have a relatively low and plain ornamentation over most
of the surface of the spine which is not greatly rougher

than that of the skin: Amanses, Cantherhines, and
Alutera. In a few genera (Brachaluteres and Pseuda-
luteres) there are low barbs mainly on the anterior face,
these continued onto the posterior face only distally. In
two genera, Paraluteres and Psilocephalus, there are no
barbs at all. In Paraluteres the barbs have obviously
been lost in conjunction with the role of this filefish as a
mimic of the pufferfish Canthigaster valentini, the spine
being covered by smooth scaleless skin and probably not
fully erectile in life. The smoothness of the spine of Psilo-
cephalus is undoubtedly related to the reduction of the
spine in that genus.

The second dorsal spine is always much smaller than
the first and is absent in Brachaluteres and Psiloceph-
alus. The second spine always has a sturdy rounded basal

146

Figure S3.— Monacanthut cUiahu: posterior

views of skull (left) and
of orbit (right) (cross section of skull; dashed

lines represent cut surfaces of frontals,

supraoccipital. basal pterygiophore of spiny

Jorsal fin and parasphenoid), 81.3 mm SL, Florida.

region for locking contact with the posterior edge of the
posteromedial flange of the first spine and has a pair of
ventrolateral processes for muscle attachment, but the
length of the thin narrow distal shaft of the second spine
is variable. The distal shaft is usually so short that it is
difficult to see as it lies buried in the narrow band of skin
between the posterior edge of the first spine and the dor-
somedial edge of the body just behind it, and even when
relatively long, it never conspicuously protrudes through
the skin.

Figure 84. — Monacanthus ciliatus:
dorsal view of branchial arches
(extended on lower side); lateral
view of hyoid arch and urohyal;
75.7 mm SL, Florida.

^epibranchials
'^pharyngobranchials

rohyal

dorsal hypohyal

ceratohyal

branchiostegal rays
147

lateral flanges

dorsal tendon

dorsal portion o(
fin ray element

muscle attachment

ventral portion of
fin ray <

Figure S5.—Monacanthus ciliatus: above, lateral
view of pelvis and encasing scales at end of
pelvis which completely obscure from external
view the rudimentary fin-ray element; below,

detail of the end of the pelvis, with the
encasing scales removed to expose the rudimentary
rin-ray element and its cartilage plug and
tendons: 81.3 mm SL, Florida.

Figure 86.— External features of

Monacanthus chinenais. for comparison

with those of M. ciliatus:

' , nasal region as seen externally

(far left) and the olfactory lamella as

seen with the top of the nasal sac removed;

middle left, scales from upper middle region of

body, including two lateral line canal bearing

scales; lower left, ventral view of encasing

scales at end of pelvis (anterior to left).

Figure 87.— External features of

other representative monacanthid

genera: Stephanolepis cirrhifer—

upper left, nasal region as seen

externally (far left) and the

olfactory lamella as seen with the

top of the nasal sac removed;

middle left, scales from upper middle

region of body, including two

lateral line canal bearing scales; lower

left, ventral view of encasing scales

at end of pelvis (anterior to left).

-Hr%^"^^

Figure 88.— External features of

other representative monacanthid

genera: Chaetoderma spinosissimus-

upper left, scales from upper middle

region of body, including two

lateral line canal bearing scales;

middle left, nasal region as

seen externally (olfactory

lamella, if present, indistinct in

specimen examined); lower left,

ventral view of encasing scales at

end of pelvis

(anterior to left).

Figure 89.— External features of

other representative monacanthid

genera: Paramonacanthus curtorhynchua-

upper left, nasal region as seen

externally (far left) and the olfactory

lamella as seen with the top of the

nasal sac removed; middle left, scales

from upper middle region of body,

including ivio lateral line canal

bearing scales; lower left, ventral

view of encasing scales at end of

pelvis (anterior to left).

149

Figure 90.— External features of

other representative monacanthid

genera: Pervagor melanocephalus-

upper left, nasal region as seen

externally (far left) and the

olfactory lamella as seen with the

top of the nasal sac removed;

middle left, scales from upper

middle region of body, including

three lateral line canal bearing

scales; lower left, ventral view of

encasing scales at end of pelvis

(anterior to left).

Figure 91. — External features of other
representative monacanthid genera:
Rudarius ercodes — upper left, nasal
region as seen externally (olfactory
lamella, if present, indistinct in
specimen examined); middle left, scales

from upper middle region of body,

including two lateral line canal bearing

scales; lower left, ventral view of

encasing scales at end of pelvis

(anterior to left).

Figure 92.— External features of other

representative monacanthid genera:

Navodon hypocrepis — upper left, nasal

region as seen externally (far left) and

the olfactory lamellae as seen with the

top of the nasal sac removed; middle left,

scales from upper middle region of body,

including a lateral line canal bearing

scale; lower left, ventral view of

encaging scales at end of pelvis

(anterior to left).

150

Figure 93.— External features of other

representative monacanthid genera:

Eubalichthys spilomelanurus — upper

left, nasal region as seen externally

(far left) and the olfactory lamella as

seen with the top of the nasal sac removed;

lower left, scales from upper middle

region of body, including a lateral line

canal bearing scale; no encasing scales at

end of pelvis (or below it).

Figure 94.— External features of other

representative monacanthid genera:

Cantherhines pullus— upper left, nasal

region as seen externally (far left) and

the olfactory lamella as seen with the

top of the nasal sac removed: middle

left, scales from upper middle region of

body, including two lateral line canal

bearing scales; lower left, ventral view of

encasing scales at end of pelvis

(anterior to left).

Figure 95.— External features of other

representative monacanthid genera:

Amanses scopas— upper left, nasal

region as seen externally (far left) and

the olfactory lamellae as seen with

the top of the nasal sac removed; middle

left, scales from upper middle

region of body, none of which have

apparent lateral line canals or pores; lower

left, ventral view of encasing scales

at end of pelvis (anterior to left).

Figure 96.— External features of other
representative monacanthid genera:
Alutera monoceros — upper left, nasal
region as seen externally (far left) and
the olfactory lamella as seen with the top

of the nasal sac removed; low-er left,
scales from upper middle region of body,

including a lateral line canal bearing

scale; no encasing scales at end of pelvis

(or below it).

Figure 97.— External features of other

representative monacanthid genera

Pailocephalus barbalus— upper left,

nasal region as seen externally

(olfactory- lamella, if present,

indistinct in specimen examined); lower

left, scales from upper middle region of

body, including a series of numerou>>

lateral line canal bearing scales; no

encasing scales at end of pelvis (or

below it).

Figure 98.— External features of other

representative monacanthid genera

Oxymonacanthus longiroatris— upper left,

scales of upper middle region of body, includin

three lateral line canal bearing scales,

middle left, nasal region as seen externally

(olfactory lamella, if present, indistinct

in specimen examined); ventral view of

scales at end of pelvis (anterior

to left).

Figure 99.— External features of other

representative monacanthid genera:

Pseudaluterea nasicornU — upper left,

nasal region as seen externally

(olfactory lamella, if present,

indistinct in specimen examined); lower
left, scales of upper middle region of
body (specialized lateral line canal or

pore-bearing scales absent); no encasing
scales at end of pelvis (or below it).

Figure 100.— External features of other

representative monacanthid genera:
Paraluterea prionurus— upper left, nasal

region as seen externally (olfactory

lamella, if present, indistinct in specimen

examined); lower left, scales from upper middle

region of body, including a lateral line canal

bearing scale; no encasing scales at end of

pelvis (or below it).

152

Figure 101.— External features of other
representative monacanthid genera:
Brachaluteres trossulus — upper left, nasal
region as seen externally (olfactory lamella.
if present, indistinct in specimen examined):
lower left, scales from upper middle region

of body, including a lateral line canal

bearing scale; no encasing scales at end of

pelvis (or below it).

The size and shape of the basal pterygiophore sup-
porting the spiny dorsal fin are highly diverse among the
monacanthids. It is expectedly least developed in Psilo-
cephalus, in which only a single, short, thin rudi-
mentary spine is present. The basal pterygiophore of
Psilocephalus is a thin triangular bone (as seen in lateral
view) whose dorsal edge is only slightly expanded lateral-
ly and whose ventral edge is held by fibrous tissue only
anteriorly to the dorsal surface of the supraoccipital. The
dorsal surface of the pterygiophore at the place of articu-
lation of the spine is only slightly expanded laterally and
upraised to support the equally only slightly laterally ex-
panded base of the spine, the whole mechanism being far
less complex than in other monacanthids.

In all other monacanthids the basal pterygiophore is
relatively much larger and sturdier than in
Psilocephalus. Most typically, as illustrated for Mona-
canthus ciliatus, it is a large bone whose flattened or
gently curved ventral surface is closely applied by fibrous
tissue and sometimes interdigitation to the dorsal sur-
face of the skull from the level of the front or middle of
the eye posteriorly, obscuring much of the supraoc-
cipital from view. It is laterally expanded in about the
middle of its length into two thick buttresses supporting
the articular facets of the base of the first spine, while
posteriorly an upraised medial buttress supports the sec-
ond spine, posterior to which the basal pterygiophore
becomes a thick vertical plate with a laterally expanded,
slightly concave, dorsal edge. Anterior to the large lateral
expansion supporting the first spine, the pterygiophore
rapidly narrows into a thinner vertical plate whose dor-
sal edge is slightly expanded laterally and either flatten-
ed or convex. The region of the pterygiophore posterior to
the second spine is supported mainly by the epiotics,
while the middle region is supported by the supraoc-
cipital, and, especially in the case of the buttresses for
the first spine, by the frontals. Anterior to the first spine,
the laterally compressed portion of the pterygiophore is
supported by the frontals alone in most genera, but by
the supraoccipital as well in those three genera (Mona-
canthus, Stephanolepis, Paramonacanthus) in which

that bone extends far forward above the orbit as a low
vertical crest (the possible phylogenetic significance of
which is discussed under the section on generic relation-
ships).

The above applies well to most of the genera in which
the first dorsal spine is above or behind the middle of the
eye, with the basal pterygiophore extending from behind
the skull to about the level of the front of the eye. In
Paraluteres, with the dorsal spine distinctly in front of
the eye, the basal pterygiophore is enormously elongate,
reaching nearly the entire length of the skull, although it
retains the same basic makeup as in more normal mona-
canthids. The basal pterygiophore in Paraluteres articu-
lates posteriorly with the supraoccipital, which it broadly
overlies, and along the ventral edge of its laterally com-
pressed plate with the epiotics, while the middle two-
thirds of its length is firmly held to the frontals and its
anterior end to the ethmoid. Many of the specializations
of Paraluteres are associated with its mimetic rela-
tionship with a Canthlgaster pufferfish (Tyler 1966).

The basal pterygiophore in the especially deep-bodied
Chaetoderma is very high and domelike, while in the
species ofAlutera it is slightly smaller than in most other
monacanthids, having only a short anterior vertical crest
in front of the level of the first spine, and the pterygio-
phore is not as broadly held ventrally to the supraoc-
cipital and epiotics.

In two genera the basal pterygiophore is of more or less
normal size in lateral surface area, but much thinner
than in other monacanthids {Psilocephalus excepted). In
the deep-bodied Brachaluteres the pterygiophore sup-
porting the single but well-developed dorsal spine is es-
pecially high, but it is much thinner and less heavy than
normal, the buttresses for the spine being only slightly
expanded laterally and most of the surface of the ptery-
giophore being a thin compressed plate. In Paraluteres,
in which the two skin covered dorsal spines are probably
not fully erectible in life, the pterygiophore is of reduced
sturdiness and of relatively short length, although the
buttressing for the spines is more or less normal.

In comparison to the spiny dorsal fin, there is even

greater diversity in the pelvic apparatus of monacan-
thids. The number of segments of scales in the encasing
series ranges from three to two to one to none, and a rudi-
mentary fin ray represented by a pair of bony splints
above and below a plug of cartilage at the end of the
pelvis is variously present or absent. The encasing scales
are either flexible or fixed, the dorsal lobe to the pelvis
varies from greatly to only slightly developed or absent
altogether, and the pelvis itself ranges from relatively
massive and extending to the region of the anus to
slender and reduced in length.

In what can be considered the most generalized condi-
tion of the pelvic apparatus in monacanthids (i.e., the
least reduced from that of balistids), the pelvis is well-
developed and has at least a moderate dorsal lobe, the
encasing scales are flexible and in three series, and a fin-
ray element is present as two bony splints. This general-
ized condition is found in Monacanthus, Paramonacan-
thus, Stephanolepis, Pervagor, and Laputa. Tendons
attach to the splints and run forward to muscles on the
dorsal and ventral surface of the pelvis, the dorsal tendon
passing through a horizontal tube in the basal region of
the dorsal lobe, as described by Tyler (1962b) for Mona-
canthus ciliatus. The splints representing the fin-ray ele-
ment are of moderate size, about like that illustrated by
Tyler (1962b) for M. ciliatus, in all of the species examin-
ed of the above genera, except that they are longer in M.
mylii and M. chinensis and shorter in Paramonacanthus
barnardi. In one other genus a fin-ray element and flexi-
ble encasing scales are present, this being the monotypic
Chaetoderma, in which there are only two series of en-
casing scales and the fin-ray splints are minute. The only
genus not examined which has a flexible encasing scale
series, Arotrolepis, can be expected to have fin-ray
splints.

In all of the genera examined in which the encasing
scales are fixed or absent, the fin-ray splints are absent,
although in a few species tendons still attach to the small
cartilaginous plug at the end of the pelvis, as described
by Tyler (1962b) for Cantherhines sandwichiensis. Tyler
(1962b) attributed three series of encasing scales to that
species. Examination of additional specimens of C. sand-
wichiensis and of other species of Cantherhines makes it
unclear to me now whether the scales are in two or three
series; perhaps it should be counted as two and a half
series, the half series being a pair of scales scarcely if at
all meeting in the midline ventrally between the larger
anterior and posterior pair. Whether these represent two
or three series of scales is problematical, but, in any case,
when the encasing scale series is inflexible, the length of
the series is shorter than in those genera in which there
are clearly three series and in which flexibility is pres-
ent.

Of the genera examined with an inflexible series, there
are two and a half or clearly only two series of encasing
scales in Amanses, Cantherhines, Nauodon, and Acan-
thaluteres, and there is no fin-ray element. This is
probably also the case in the other genera with fixed en-
casing scales: Scobinichthys, Pseudomonacanthus,
and Meuschenia; and, questionably, in Eubalichtys, as

discussed below. In Rudarius there are two series of en-
casing scales and no fin-ray element in R. ercodes, but
only a single series and no ray in R. minutus, the reduc-
tion of the series into a single, although highly spinous,
series in the latter perhaps being associated with the ex-
tremely small adult size of that species. In Oxymona-
canthus there is no fin-ray element and no precisely
delineated series of encasing scales; rather, the scales
bordering the ventral surface of the posterior end of the
pelvis simply gradually become larger and more spinous
distally so that the encasing scales are of indeterminate
number.

The genus Alutera provides an example of the process
of loss of the encasing scales. No fin-ray element is pres-
ent, but Berry and Vogele (1961) have shown that in
small specimens of three of the four species (A.
heudelotii, A. scripta, and A. monoceros) there is a single
enlarged spinous scale present on the midline below the
ventral edge of the pelvis, this being the last remnant of
the encasing scales. This scale becomes unornamented
and then lost in large adults. In A. schoepfi there is no
rudimentary encasing scale at any size (Berry and Vogele
1961).

In Acanthaluteres, the specimens of the single species
here examined {spilomelanurus, if I have it properly
identified, and I am not sure) lack encasing scales,
whereas Fraser-Brunner (1941b: 178) gives as a key char-
acter for the genus "Pelvic shield present at all stages
(sometimes minute or inconspicuous)." It is possible that
there are allometric changes in the encasing scales, just
as in Alutera. This may also be true of Eubalichthys, of
which Fraser-Brunner (1941b: 180) says: "Pelvic shield
minute, inconspicuous."

There are no encasing scales or fin-ray elements in
Psilocephalus, Brachaluteres, Paraluteres, and Pseuda-
luteres, and, according to Fraser-Brunner (1941b:181), in
Blandowskius.

The structure of the pelvis itself is highly diversified in
monacanthids. In those species with flexible encasing
scales, the pelvis is always, relatively well developed and
has a dorsal lobe of at least moderate height. The dorsal
lobe is largest in the genus Monacanthus, the species of
which are noted for the especially large size of the fan of
distensible skin with enlarged scales that can be flared
out between the anus and pelvis when the latter is
rotated downward and forward. However, the presence of
a dorsal lobe is not absolutely correlated with the expan-
sion of the abdominal dewlap, for one genus (see below)
without a lobe has the ability to flare a remarkably large
fan.

In general, however, most of the genera (Pseuda-
luteres, Paraluteres, Brachaluteres, Alutera) without a
dorsal lobe also do not flare fans, the exception being
Psilocephalus, which not only has no dorsal lobe but also
the most slender and weakly developed pelvis in the
family, and yet can flare as large a fan as in Monacan-
thus. The presence of a dorsal lobe is probably more
strictly correlated with the presence of enlarged scales
along the posterior edge of the fan, these scales being
supported by the lobe. The scales along the posterior

154

t

%

edge of the fan in Psilocephalus are not much enlarged or
differentiated from those just anterior to them. The dor-
sal lobe is poorly developed in Oxymonacanthus, which
flares only a small or moderate sized fan and which has
no modified scales in the fan.

Of the genera with fixed encasing scales, the pelvis
ranges from very strongly (e.g., Amanses) to only moder-
ately (e.g., Oxymonacanthus) developed. Of the genera
with no encasing scales, none have a dorsal lobe, and the
size of the pelvis ranges from massive (e.g., Alutera), to
normal (Brachaluteres), to slightly reduced in size
{Pseudaluteres and Acanthaluteres), to short and
somewhat slender (Paraluteres) , and to long but very
weak and slender {Psilocephalus}.

In Pervagor alone the lateral surface of the anterior
region of the pelvis has a lateral knoblike expansion
which articulates with a similar expansion on the pos-
terior edge of the coracoid, forming the point of pivot
around which the pelvis is rotated downward and
forward to flare the fan.

There is little variation in the branchial apparatus in
the monacanthids, except for the number of branchios-
tegal rays. All species with the exception of Psiloceph-
alus barbatus have three basibranchials, three hypo-
branchials, five ceratobranchials (the last toothless),
four epibranchials, and two pharyngobranchials (both
toothed). In Psilocephalus there are only two basi-
branchials and three epibranchials, and all of the various
pieces in the arches are more or less reduced in
massiveness, although the ceratobranchials are more
elongate.

There are six branchiostegals in a 2 -(- 4 arrangement in
the various species examined of Nauodon, Acantha-
luteres, Alutera, Cantherhines, and Amanses, but only
four branchiostegals, in a 1 -(- 3 arrangement, in Psilo-
cephalus. The great majority of species examined,
however, have five branchiostegals, in a 1 -)- 4 arrange-
ment, these being of the genera Monacanthus, Stephano-
lepis, Paramonacanthus, Pervagor, Laputa, Rudarius,
Oxymonacanthus, Chaetoderma, Paraluteres, Bracha-
luteres, and Pseudaluteres. When five rays are present, it
is the second of the anterior group of two that has been
lost, while in Psilocephalus, with four rays, one has also
been lost from the posterior series of four. It is of interest
that of 12 specimens of Monacanthus ciliatus examined,
11 had the 1-1-4 arrangement on both sides, but one had
1 -(- 4 on one side and 2 -f 4 on the other, the second ray of
the anterior group being present on one side only.

There are great differences in the number, size, and
placement of epipleural bones in monacanthids, which
has led to confusion. Fraser-Brunner (1941b:176), in his
diagnoses of the Balistidae and Monacanthidae, stated
that in balistids epipleurals were found on no more than
the first two caudal vertebrae, while monacanthids had
epipleurals on the first four or five caudal vertebrae.
From the materials examined here, it appears that most
balistids have epipleurals on the second abdominal to
the first caudal vertebra, but in the species of the genus
Rhinecanthus they extend back from the second ab-
dominal to the fifth or even sixth caudal vertebra. In

most monacanthids the epipleurals begin on the second
abdominal vertebra and usually but not always extend
onto the caudal vertebrae. However, in a number of
species the epipleurals are confined to the abdominal
vertebrae. In the species of Alutera, for example, the
epipleurals extend from the second abdominal only to
the sixth (next to last) or, rarely, seventh (last) ab-
dominal vertebra, while in Pseudaluteres they are pres-
ent from the second to the eighth (next to last) ab-
dominal vertebra and in Psilocephalus from the second
to sixth (next to last) abdominal vertebra. As an ex-
amination of the lateral view illustrations of monacan-
thids presented here will show, the epipleurals of other
genera extend posteriorly variously anywhere from the
first to fifth caudal vertebrae, while in Rudarius (both er-
codes and minutus) and Brachaluteres they are present
only from the third abdominal to the second to fourth
caudal vertebrae, being absent on the second abdominal
vertebra. Thus, here is no absolute familial distinction
concerning the epipleurals between the balistids and
monacanthids.

The epipleurals in monacanthids are usually slender
splints, less sturdy than in most balistids, but they are
ver\' sturdy in at least the single large adult oi Amanses
examined, and relatively well developed with large sur-
face area in Brachaluteres and Alutera. In Alutera the
epipleurals become hyperostotic in large adults.

True ribs have previously been found among plec-
tognaths only in the primitive gymnodont Triodon
macropterus (see Tyler 1962a:798 for discussion),
and in the monacanthid scleroderm Pseudaluteres nasi-
cornis as well (see Tyler 1973b). The latter would be
a less amazing occurrence of ribs, in light of the ances-
tral balistids lacking true ribs, were Pseudaluteres at
least a primitive monacanthid rather than being, as
discussed under generic relationships, surely one of the
most specialized.

The genetic information on constructing ribs would
appear to have been retained in the otherwise ribless
monacanthids and their ancestral balistids and their
ancestral and equally ribless triacanthids and triacan-
thodids from the doubtless rib bearing basal group of
plectognaths from which both the scleroderms and gym-
nodonts evolved. What peculiarity in the functional life
of Pseudaluteres has made it advantageous to have ribs
when other monacanthids (and Recent scleroderms) do
not is a mystery to me, but it would not appear to be as-
sociated with its two other primary specializations; the
far forward placement of the dorsal fin spine in front of
the eye or of the lack of encasing scales and loss of the
rudimentary pelvic fin-ray element.

A posttemporal is present in all monacanthids examin-
ed except Psilocephalus barbatus, none of the three adult
specimens of which shows any evidence of one. Whether
the ossification center for the posttemporal is lost or in-
corporated indistinguishably with that of the pterotic
remains problematical. The dorsal head of the supra-
cleithrum in Psilocephalus articulates directly with the
lateral surface of the pterotic. This loss of the posttem-
poral is undoubtedly correlated with the greatly reduced

167

Figure Ma.—Stephanolepii
hUpidiu: lateral view of
50.4 mm SL, Florida.

basal pterygiophi

Figure W^.—Paramonacanthu

cryptodon: lateral view of hea

68.5 mm SL, Thailand.

Figure in.—Pervagorapiloeomus: lateral
view of head, 80.0 mm SL, Hawaii.

premaxillary

quadrate sy^plect

mefapterygoid

preoperculum

1st dorsal spine basal pterygiophore

2nd dorsal spine

Figure \19.—Alutera heudelotii
lateral view of head,
107 mm SL, Florida.

preroric
hyomandibula
operculum
suboperculum
preoperculum
metapterygoid

articular

angular

interoperculum

symplectic

mesopterygoid

ectopterygoid

Figure 120.— Psilocephalus barbatua: lateral
view of head, 137 mm SL, Singapore.

170

Figure 12\ .—Brachaluterea trosaulus: lateral
view of head, 55.5 mm SL, Australia.

Figure 122.— Oxymonacanthut
longirostris: lateral view of
head, 26.5 mm SL, Seychelles.

Figure \23.—Pseudaluteres
nasicornis: lateral view of
head, 108 mm SL, Philippines,
with inset above showing
detail of the snout region

size of the pectoral girdle and of the fin it supports, with
the rays reduced in number to eight (usually) or nine,
plus the uppermost rudimentary ray. This attests to the
far greater reliance in Psilocephalus on the especially
long-based and many-rayed soft dorsal (43-52 rays;
average of five specimens 49) and anal (53-66 rays,
average 58) fins for locomotion. Psilocephalus is the only
monacanthid regularly to have far more anal than dorsal
fin rays.

As in balistids, the scapula completely encloses its
foramen in most monacanthids, but in Pseudaluteres
and Psilocephalus the scapular foramen is incomplete
anteriorly, being bounded by the cleithrum. In
Psilocephalus this is probably correlated with the reduc-

ed size of the pectoral girdle and fin, but Pseudaluteres
has a well-developed pectoral girdle and a moderate
number (11-13) of rays.

Variation in the monacanthid caudal skeleton mainly
concerns the presence or absence of an upper free hy-
pural. There is always a single epural and usually a free
upper hypural in contrast to the other hypurals which are
fused to one another and to the centrum, and a free
parhypural. A free upper hypural is absent in the majori-
ty of specimens of Monacanthus ciliatus examined and
in those of Rudarius ercodes and R. minutus and in the
single specimen of Amanses scopas examined (Tyler
1970b: 15, erroneously implied that R. ercodes has an up-
per free hypural). Of 12 specimens of M. ciliatus examin-

ed, only 2 had a free upper hypural. Other variation in
the monacanthid caudal skeleton (discussed in Tyler
1970b) is that the usually horizontal crest for muscle at-
tachment on the last centrum is reduced to a prong in
Brachaluteres trossulus, and its neural and haemal
canals are better developed than in other monacanthids,
being roofed over distally rather than open. In
Psilocephalus only the middle 6 to 8 rays of the ex-
tremely elongate caudal fin are branched, rather than
having 10 branched rays as in all other monacanthids.
The posterior edge of the fused hypural plate is much
more convex in Psilocephalus than in all other monacan-
thids, while in Brachaluteres the cleft in the posterior
edge of the fused hypural plate, that marks the region of
fusion between what in generalized plectognaths like
triacanthodids are the second and third hypurals, is
much deeper than in all other monacanthids.

In balistids and in all but two genera of monacanthids
the side of the rear part of the cranium just above the
region of articulation of the upper end of the hyoman-
dibular is formed by the sphenotic anteriorly and by the
pterotic posteriorly, the hyomandibular articulating with
a groove across the surface of the prootic and pterotic.
The sphenotic usually has a well -developed (least
developed in Rudarius) ventrally or anteroventrally
directed prong from its lower edge and the pterotic a
larger ventrally directed flange that overlies the postero-
dorsal end of the hyomandibular. The anterolateral edge
of the sphenotic forms the lower rear margin of the orbit,
with the frontal forming the margin more dorsally.

The two exceptions to these conditions are Oxymona-
canthus and Pseudaluteres, in both of which the
sphenotic is displaced posteriorly by a ventral extension
of the frontal, the side of the rear of the cranium above
the articulation of the hyomandibular being formed by
the frontal anteriorly and the pterotic posteriorly, with
the sphenotic squeezed in between. The sphenotic
retains a short ventrally directed prong and the pterotic a
normally developed large flange overlying the postero-
dorsal end of the hyomandibular, which articulates in a
normal manner with a groove across the surface of the
prootic and pterotic. The frontal, which forms all of the
posterior margin of the orbit in these two genera to the
exclusion of the sphenotic, has a ventrally directed prong
from its anteroventral end in this region. This prong on
the frontal is comparable to that on the sphenotic of all
other monacanthids, where it forms the lower rear mar-
gin of the orbit.

Monacanthids always have the prootic shelf much less
strongly developed than in balistids. The shelf in
monacanthids does not extend as far forward as the level
of the middle of the orbit, except in Psilocephalus, in
which the shelf is long, thin, and delicate and reaches to
the front of the orbit. Within these limits, the prootic
shelf of monacanthids is variously developed, from
relatively well to only moderately, and in two genera it is
completely absent, these being Oxymonacanthus and
Pseudaluteres.

All monacanthids have an outer series of three and an
inner series of two teeth in each premaxillary, the inner

series closely applied to the inner (or under) surface of
the outer series and to the bone. In the dentary there is a
single series of teeth, corresponding to the outer series of
the upper jaw, usually three in number, but reduced to
two in Psilocephalus barbatus, Rudarius ercodes and R.
minutus, Brachaluteres trossulus and (according to
Clark and Gohar 1953:46) B. baueri and (according to
Scott 1969:40) B. wolfei, Paraluteres prionurus, and Oxy-
monacanthus longirostris (presumedly also only two
teeth in the closely related 0. halli Marshall as well).
While three teeth are present in the dentary of Pseuda-
luteres nasicornis, the outermost tooth is much smaller
than in other monacanthids with three dentary teeth. In
all monacanthids the medial edges of articulation of the
apposed dentaries are well-denticulated.

The teeth are relatively thinner and have more
delicate cutting edges than in balistids, but they are
usually notched in basically the same way in both
families. The least notched condition of the teeth among
the monacanthids examined is found in Acanthaluteres
spilomelanurus, in which the two most medial teeth in
both the upper and lower jaws (the most medial tooth of
both premaxillaries and both dentaries) are relatively
straight edged, forming a broad relatively sharp straight
nipping edge along the front of the mouth. The position
of the mouth tends to be terminal or only slightly supra-
terminal in most monacanthids, but it is distinctly
supraterminal in the elongate Psilocephalus barbatus.
The mouth of Oxymonacanthus is relatively small and
laterally compressed, with the teeth also relatively nar-
row but elongate, for what is probably a specialized mode
of feeding between the branches of coral of this strictly
reef-dwelling species. Among the monacanthids examin-
ed, the teeth are most massive in Amanses scopas, but
this could in large part be a function of the large size of
the single specimen examined. It will be of interest to
know the size of the teeth in juveniles and young adults
of Amanses, which have yet to be collected (Randall
1964:335).

While the number of vertebrae in the ancestral
balistids is almost invariably 7 -(- 11 = 18 (in contrast to
their ancestral triacanthoids in which it is 8 -I- 12 = 20),
the number of vertebrae in monacanthids is highly
variable, although the most frequent formula is 7 H- 12 =
19 and none of the species examined normally has as few
as the 18 vertebrae of balistids. Only a few species of
monacanthids have other than seven abdominal verte-
brae, most of the variation in total number being in the
caudal series. The only monacanthids normally with six
abdominal vertebrae are the two Atlantic species of
Monacanthus {ciliatus and tuckeri, the subgenus Lep-
rogaster of Fraser-Brunner, with 6 -I- 13 = 19 in contrast
to the 7 -I- 12 = 19 of the Indo-Pacific species) and
Acanthaluteres spilomelanurus (6 -I- 14 = 20). The only
monacanthids normally with eight abdominal vertebrae
are Oxymonacanthus longirostris (8 -(- 17 = 25) and
Pseudaluteres nasicornis (8 -I- 18 = 26).

When the number of vertebrae is increased beyond
what is probably the generalized number of 19, it is
usually by the addition of one or more vertebrae in the

173

caudal series (as well as, in Oxymonticanthiu and
Pseudaluteres, by one in the abdominal series), as ex-
emplified by Alutera, whose four species have normal
vertebral counts ranging from 20 to 23 but all of which re-
tain 7 abdominal vertebrae. The greatest increase in
number of vertebrae is in the extremely elongate body of
Psilocephalus barbatus, the increase again being con-
fined to additions in the caudal series. Four of the five
specimens of P. barbatus cleared and stained or radio-
graphed have 7 abdominal and 22 or 23 caudal verte-
brae, while one has 8 abdominal and 22 caudal verte-
brae.

The neural spines of the abdominal vertebrae anterior
to the first basal pterygiophore of the soft dorsal fin vary
from relatively narrow long slender shafts (e.g.,
Brachaluteres) to short slender shafts (e.g., Chaeto-
derma) to broadly anteroposteriorly expanded plates
(e.g., Pseudaluteres), with intermediate conditions
found in other genera. As in balistids, which always have
five vertebrae anterior to the first basal pterygiophore of
the soft dorsal fin, the great majority of monacanthids
also have five so placed. The only exceptions are the two
Atlantic species of Monacanthus (ciliatus and tuckeri),
the single species of Acanthaluteres examined, nd one
of the two species of Rudarius (minutus), all of which
have only four vertebrae anterior to the first soft dorsal

Figure 124.— Dorsal viewi of skull* of:

A, Pnudalutereg nasicomis, 108 mm SL,

Philippines, highly specialized;

B, Paraluteres prionurue, 46.4 mm SL,

Seychelles, moderately specialized;

C, Chaetoderma spinosiasimus, 33.0 mm SL,

Malaya, relatively generalized.

fin basal pterygiophore, and Pseudaluteres nasicornis
with six and Psilocephalus barbatus with eight.

As in balistids, the lateral line system in most mona-
canthids is inconspicuous, and can usually only be fol-
lowed in cleared and stained specimens examined under
a microscope. However, it can be relatively easily seen
with the unaided eye even in alcohol preserved speci-
mens of a few species of Monacanthus (especially chinen-
sis), Paramonacanthus, and Paraluteres prionurus, and,
to a lesser extent, in Psilocephalus barbatus.

The scales of the lateral line in monacanthids typically
have a foramen in the basal plate with an upright spiny
process like that of the other scales bordering the
foramen above and below. The only species examined in
which such lateral line scales could not be found was
Pseudaluteres nasicornis.

In all species except Paraluteres prionurus the head
and body are more or less fully covered with scales whose
upright spinulations usually give a shagreenlike quality

to the skin. In P. prionurus the skin is smooth to the
touch, except on the caudal peduncle of mature males
where there are two pairs of retrorse barbs and a patch of
short upright setae extending forward to the level of the
anal fin origin. The fact that there are very low weakly
developed spinulose scales on the head and ventral sur-
face from the mouth to the anus is only readily apparent
in cleared and stained specimens.

The scales of monacanthids vary greatly in the degree
and structure of the spinulation and in the size and
degree of overlap (if any) of the usually rounded to rec-
tilinear basal plates, but they are always much smaller
than in balistids. In most species the basal plates more or
less broadly but irregularly overlap, while in others (e.g.,
Chaetoderma) there is little if any overlapping of the
relatively large and more or less triangular plates. The
scales of Chaetoderma are more or less regularly ar-
ranged in rows, and, according to Fraser-Brunner
(1941b: 178), they are in distinct longitudinal tracts in

Figure 125.— From left to right:

A, Paraluteres prionurus. ventral view of

skull, 46.4 mm SL, Seychelles; B. Pervagor

apilosomus, ventral view of skull,

ca. 65 mm SL, Hawaii; C, Psilocephalus

barbatu^, dorsal and ventral views

of skull, 137 mm SL, Singapore.

Arotrolepis, while in the great majority of monacanthids
the arrangement is irregular. There is always at least one
upright spinule (least developed in Paraluteres, best in
Chaetoderma) per scale plate, and often many. The in-
dividual scale plates are largest in Chaetoderma, in
which many of the upright spinules support large der-
mal flaps, and smallest in Psilocephalus.

All monacanthids, like balistids, have two nostrils in a
scaleless area on both sides, each nostril usually at the
end of a short tube, but sometimes more or less flush with
the surface (especially Paraluteres and Psilocephalus).

basibranchials

:^^^ ' ' ^'" ' ' "^ -"^^v' " hypobranchials

^ - ceratobranchials

urohyal

-c:2ss.

S. f-^

branchiostegal rays

The nasal sac beneath the surface usually has a single
longitudinal upraised fold or lamella.

Generic relationships. — Given their evolution from
balistids, one would expect generalized monacanthids to
have a relatively well-developed pelvis with a dorsal lobe
and at least rudiments of the fin-ray element hidden
within a flexible series of encasing scales of a maximum
number of segments. Likewise, the basal pterygiophore
supporting the dorsal fin spines would be expected to be
well developed and over the rear part of the skull (i.e.,
the least distance migrated forward) and with the skull
the least flattened for articulation with it (i.e., with some
remnant of the medial supraoccipital crest of balistids).

Other generalized features to be expected a priori
would be an only slightly elongate body and terminal
mouth, six branchiostegals, a groove posterior to the dor-
sal spines into which they are folded more or less flush
with the surface when unerected, the least reduction in
the number of teeth, the least increase in the number of
vertebrae, and the presence of an upper free hypural.

Of the above features, I tend to place greatest em-
phasis on the pelvic apparatus and the form of the top of
the skull in relation to the basal pterygiophore of the
spiny dorsal fin. The supraoccipital is best developed in
Stephanolepis, Monacanthus, and Paramonacanthus,
reaching to the front of the eye, and only in these three
genera does the supraoccipital retain a vertical medial
crest anteriorly for articulation with the laterally com-
pressed platelike anterior region of the spiny dorsal fin
basal pterygiophore. I take this low anteriorly placed
supraoccipital crest as the remnant of the much sturdier

crest that is found more or less throughout the length of
the balistid supraoccipital and which ends posteriorly in
the thick lateral expansion which helps, along with the
epiotics, to support the anterior end of the first basal
pterygiophore of the spiny dorsal fin.

Stephanolepis, Monacanthus, and Paramonacanthus
also have the pelvic apparatus well developed, with the
dorsal lobe of the pelvis of moderate to large size and
with the maximum monacanthid number of three seg-
ments of flexible encasing scales surrounding a hidden
fin-ray element to which muscles attach through ten-
dons running along the dorsal and ventral surface of the
pelvis, the dorsal tendon coursing through a canal in the
base of the dorsal lobe. In all three of these genera the
dorsal spines are placed relatively posteriorly on the
skull, at a level behind the middle of the eye.

In most other genera of monacanthids examined (Per-
vagor, Amanses, Cantherhlnes, Acanthaluteres, Oxy-
monacanthus, Rudarius, Laputa, Navodon, Pseuda-
iuteres, Paraluteres, Psilocephalus) the supraoccipital
does not extend as far forward as in Stephanolepis,
Monacanthus, and Paramonacanthus and there is no
crest at all. With the exception of Laputa (and Chaeto-
derma as well), the former group of genera are also
specialized in having an inflexible series of encasing
scales, or none at all, and no fin-ray element, other than
sometimes a mere cartilaginous plug at the end of the
pelvis.

Laputa is the only genus with flexible encasing scales
and a rudimentary fin-ray element (small in the single
species examined) in which the supraoccipital does not
extend far forward and is without a crest, although the

176

Figure 126. (opp. page)— Dorsal

view of branchial arches

(extended on lower side) and

lateral view of hyoid arch and

urohyal of: left, Alutera heudelotii,

114 mm SL. Florida (labeled),

and right, Psilocephalug

barbatus, 1,39 mm SL.

Singapore (unlabeled).

Figure 127.— Medial views of

upper (left) and lower (right) jaws of:

A. Alutera heudelotii, 114 mm SL. Florida;

B, Monacanthu« ciliatus, 51.3 mm SL, Colombia;

C. PBilocephalus barbatus, 137 mm SL. Singapore;

D. Amanses scopas, 167 mm SL, Saipan;

E, Acanthaluteres apilomelanurus,

63.7 mm SL, Australia.

Figure 128.— Hypothesized phylogenetic relation-
ships of the genera of Monacanthidae.

dorsal spines are placed posteriorly on the head. Thus,
Laputa is as generalized as Stephanolepis, Mona-
canthus, and Paramonacanthus except for the shorter
and more flattened condition of the supraoccipital.

In Chaetoderma there are only two encasing scale seg-
ments, a slight specialization, but minute remnants of
the fin-ray element remain and the encasing scale
segments are flexible, the minuteness of the fin-ray ele-
ment also being a slightly specialized condition.
Moreover, Chaetoderma has a relatively well developed
forward extension of the supraoccipital, almost to the
front of the eye and only slightly shorter than in
Stephanolepis, Monacanthus, and Paramonacanthus.
The forward extension of the supraoccipital in Chaeto-
derma bears a thin low crest over the middle and pos-
terior regions of the eye which is probably a remnant of
that found in the three genera in which the crest is well
developed.

In two other genera besides Chaetoderma the supraoc-
cipital extends forward only slightly less than the front of
the eye, but both are obviously specialized by the total
loss of the pelvic fin-ray element and the dorsal lobe of
the pelvis, as well as by either the total loss of the en-
casing scale series or its reduction to a single scale that is
lost in large adults. These two genera are Alutera, in
which all four species have a low supraoccipital crest on
the forward extension, and the single species of
Brachaluteres examined (trossulus), in which there is no
crest in the smaller of the two specimens but a very low
crest present from the middle of the eye posteriorly in the
larger specimen.

One other genus must be mentioned in connection
with the shape of the supraoccipital, that being Psilo-
cephalus, in which there is no true long forward and
mainly medial extension of the supraoccipital as found in
the genera just discussed. In Psilocephalus the supraoc-
cipital is represented only by the relatively larger more

laterally expanded posterior portion. But probably
because of the relative shortening of the posterior part of
the skull and the great elongation of the snout, the an-
terior end of the supraoccipital in Psilocephalus does ex-
tend forward to the level of the dorsal end of the pre-
frontals, although still short of the front of the eye.

In my opinion, the combination of the most generaliz-
ed condition of the pelvic apparatus and that of the
supraoccipital supporting the basal pterygiophore and its
dorsal spines placed over the rear of the eye points to
Stephanolepis, Monacanthus, and Paramonacanthus as
representing the most generalized group of mona-
canthids. Of the other features listed in the opening
paragraph of this section as being considered generalized
a priori on the basis of the structure of the ancestral
balistids, these three genera have terminal mouths with
the most typical monacanthid number of teeth (the
minimum reduction from balistids), 19 vertebrae (the
minimum increase in number from balistids), and an up-
per free hypural, except that only a minority of
specimens of one of the species {ciliatus) of Mona-
canthus have a free upper hypural.

However, Stephanolepis, Monacanthus, and Para-
monacanthus are among those majority of mona-
canthids with 1-1-4 branchiostegals, rather than the 2 +
4 arrangement found in balistids and in Navodon,
Acanthaluteres, Alutera, Cantherhines, and Amanses
among the monacanthids, and in having no relatively
distinct and deep groove medially behind the basal
region of the dorsal spines into which the latter are
received and held more or less flush with the surface, as
found in balistids and in the monacanthids Acantha-
luteres, Cantherhines, Amanses, Meuschenia, Oxy-
monacanthus, Eubalichthys, and most but not all
species of Pervagor.

The 2 + 4 versus 1 + 4 number of branchiostegals and
the degree of development of a groove for the reception of
the dorsal spines seem in this case to be less important
indicators of the overall generalized condition of mona-
canthids than do the other outstanding features dis-
cussed above, this especially in light of the fact that, as
described here, Monacanthus ciliatus at least sometimes
has the 2 + 4 branchiostegal number and that among the
apparently closely related species of Pervagor, most have
a groove behind the dorsal fin, while at least one species
(tomentosus, according to Fraser-Brunner 1941b: 183)
does not.

The three most generalized genera, Stephanolepis,
Monacanthus, and Paramonacanthus, seem very closely
related, and only a single genus has often been recog-
nized for the species now placed in Stephanolepis and
Monacanthus. The latter two genera have been best dif-
ferentiated by Berry and Vogele (1961:63, 68), mostly on
the basis of the spinulation of the scales. Monacanthus
has unbranched and relatively fewer spinules per scale
than does Stephanolepis, which has branched spinules,
although the number of spinules increases greatly with
increasing specimen size in both genera, just as in most
other monacanthids. Berry and Vogele also pointed out
that the ventral dewlap between the end of the pelvis and

178

the anus is much larger in Monacanthus, in which it ex-
tends out far beyond the end of the encasing scales, than
in Stephanolepis, in which it extends out only to the dis-
tal end of the encasing scales.

In Paramonacanthus the ventral dewlap is especially
short, extending posteriorly only to the anterior edge of
the encasing scale series, so that the end of the pelvis
protrudes prominently from the ventral contour. The
scales in Paramonacanthus have unbranched spinules,
as in Monacanthus and the vast majority of other mona-
canthids, but the spinules are less coarse than in Mona-
canthus and the skin of Paramonacanthus thus has
much the same velvety shagreen consistency as found in
Stephanolepis, rather than the rougher consistency of
Monacanthus. In Paramonacanthus the scale spinules
are unmodified, while, as described in detail by Berry
and Vogele (1961), mature males of Stephanolepis
develop a patch of bristles on the caudal peduncle and
mature males of Monacanthus have bristles and retrorse
barbs, females of Monacanthus developing smaller pos-
teriorly pointed barbs on the caudal peduncle but very
little if any bristle patch.

In Monacanthus the dorsal lobe of the pelvis is very
large, the longest among monacanthids, being as-
sociated with support of the modified scales in the enor-
mous dewlap. In the various species of Stephanolepis and
Paramonacanthus the dorsal lobe is only moderately
developed, about that normal for the family when a dor-
sal lobe is present at all.

In Paramonacanthus and Stephanolepis there is
always a free uppermost hypural, as is also the case in
the Indo-Pacific species of Monacanthus. In the two
Atlantic species of Monacanthus, one species (tuckeri)
usually has a free upper hypural (present in three out of
four cleared and stained specimens) while the other
species iciliatus) usually lacks the free hypural (present
in 2 out of 12 cleared and stained specimens).

In Stephanolepis and Paramonacanthus epipleurals
usually extend back to the fifth caudal vertebra while in
Monacanthus they usually extend back only to the sec-
ond or third caudal vertebra (one specimen of M. chinen-
sis with epipleurals on the fourth caudal vertebra).

In Stephanolepis, Paramonacanthus and the Indo-
Pacific species of Monacanthus there are 7 abdominal
and 12 caudal vertebrae, with 5 of the abdominal verte-
brae anterior to the first basal pterygiophore of the soft
dorsal fin, this being the most common vertebral con-
dition in the family. In the two Atlantic species of
Monacanthus the total number of vertebrae remains 19,
but it is the haemal spine of the seventh rather than that
of the eighth vertebra that supports the anterodorsal end
of the first basal pterygiophore of the anal fin, and the
neural spine of the fourth rather than that of the fifth
vertebra supports the anteroventral end of the first basal
pterygiophore of the soft dorsal fin, M. ciliatus and
tuckeri thereby having only 6 rather than 7 abdominal
vertebrae and only 4 rather than 5 predorsal vertebrae.

With the similarities and differences discussed above
between the three genera here considered to be the most
generalized monacanthids, the question is whether one of

them can be considered closer to the ancestral group
than the other two. There is little to go on.

In balistids the ventral fan is only moderately
developed, with the expansible skin extending pos-
teriorly across most of the length of the encasing scale
series, ending at about the beginning of the most distal of
the four segments of scales. This condition is somewhat
intermediate between that of Paramonacanthus, with a
very short dewlap, and Stephanolepis, with a moderate
dewlap, but closer to that of Stephanolepis.

The moderate dorsal lobe of the pelvis of Paramona-
canthus and Stephanolepis is closer to that of balistids
than is the extremely large one of Monacanthus.

The upright spinules on small basal scale plates in
monacanthids is so different from the structure of the
scales of balistids that it is impossible to say whether the
branched spinules of Stephanolepis are more generalized
or specialized than the unbranched spinules in
Monacanthus and Paramonacanthus, and in any case
many of the scales of the head and extreme upper and
lower parts of the body in the Indo-Pacific species of
Monacanthus have branched spinules, while those of
most of the body are unbranched but sometimes notched
along their anterior edge.

The phylogenetic implications of the presence of en-
larged caudal peduncular scales forming a patch of
bristles or spines in mature males of Stephanolepis and
in mature males and, to a lesser extent, in mature
females of Monacanthus, but in neither sex of Para-
monacanthus, is difficult to interpret. Among balistids
only two genera (Balistapus and Rhinecanthus) have
well-developed caudal armature, in the form of large
spines, but numerous genera (e.g., Sufflamen, Melich-
thys, Xanthichthys, Abalistes, Balistoides) have
horizontal ridges formed by upraised areas on the suc-
cessive rows of scales on the caudal peduncle and pos-
terior region of the body, and only a few genera are total-
ly lacking these ridges (e.g., Balistes and Odonus).
Without knowing from what type of balistids the mona-
canthids have been derived, it cannot be said at present
whether the caudal armature present in a few mona-
canthids is a generalized holdover from their balistid an-
cestors or a relatively specialized feature.

I would guess that Stephanolepis is the most generaliz-
ed of the three genera, basically because of its moderate
dorsal lobe of the pelvis and moderate dewlap. Mona-
canthus would be a slightly specialized close derivative
of Stephanolepis whose evolutionary trend is for a great
increase in the size of the dewlap and of the dorsal lobe
supporting it, while in another direction Paramona-
canthus is an equally close relative of a Stephanolepis-
like stock, specialized by a great reduction in the size of
the dewlap but not of the dorsal lobe.

Fraser-Brunner (1941b: 181) considered Pervagor the
most generalized monacanthid, stating that: "Pervagor,
in which dorsal and pelvic spines are better developed
than in the succeeding [i.e., all the other] genera, ex-
hibits two features which are not found among the other
forms possessing a movable pelvic spine - the forward
position of the dorsal spine and the presence of a deep

groove in which it is received when depressed (the latter
character, however, being lacking in the subgenus
Acreichthys noted [described as new| below). On the one
hand it leads to the rest oi' the genera retaining the pel-
vic spine but having the dorsal spine behind the eye and
without groove; of these, Stephanolepis seems to be the
group round which the specialized forms such as Chaeto-
derma and Monacanthus are clustered. On the other
hand extends the large series of genera in which the pel-
vic spine has been lost. ..."

I do not find that the dorsal spines and encasing scales
in Pervagor are better developed than in most other
genera, and certainly they are no better developed than
in Stephanolepis, Monacanthus, and Paramona-
canthus. Moreover, I do not think that the somewhat for-
ward position of the dorsal spine (over the middle of the
eye) can be considered the generalized condition when
the ancestral balistids have the dorsal spines behind the
eye.

As discussed previously, the least forward migration of
the dorsal spines is the most generalized monacanthid
condition, with the spines over the rear of the eye. Per-
vagor does not possess the generalized monacanthid con-
dition of the supraoccipital, for it lacks the forward ex-
tension with vertical crest to the front of the eye that is a
remnant of the condition of the balistid supraoccipital.

In fact, the only definitely generalized feature of Per-
vagor to my way of thinking is its pelvic apparatus, with
it having a moderate dorsal lobe, three segments in the
flexible encasing scale series and moderately well-
developed rudiments of the fin-ray element. As dis-
cussed previously, a groove behind the dorsal spines
would seem a priori a generalized condition in mona-
canthids since the ancestral balistids have such a groove.
But of the genera of monacanthids with a groove behind
the dorsal spines, only Pervagor has a generalized pelvic
apparatus, all the other genera having the segments of
encasing scales reduced in number and inflexible or ab-
sent entirely, and without a fin-ray element.

All of the genera with the groove behind the dorsal
spines have the most specialized condition of the
supraoccipital, without a long or even moderate forward
extension and no crest. It seems as likely that the
monacanthid postdorsal groove is a de novo develop-
ment in moderately specialized forms as it is that it is a
hold over from the ancestry of the monacanthids, the
most generalized Recent species of which would then
have to be presumed to have subsequently lost it.

A similar situation occurs with the interpretation of
the 2-1-4 branchiostegal number found only in a few
genera of monacanthids, these all being genera with a
specialized pelvic apparatus in which the fin-ray element
is absent and the segments of encasing scales are reduced
in number and inflexible or absent entirely, a small dor-
sal lobe is present on the pelvis, and there is either no
forward supraoccipital extension (Nauodon, Acantha-
luteres, Cantherhines, Amanses) or only a small exten-
sion to above the middle of the eye {Alutera). Thus, on
all of the most important grounds, the genera with the 2
+ 4 number of branchiostegals are much more specializ-

ed in general than are many of the genera with the 1 -t- 4
number, including Stephanolepis, Monacanthus, and
Paramonacanthus.

Pervagor also possesses one unique feature found in
neither the other monacanthids nor in balistids. The
lateral surface of the anterior region of the pelvis has a
protuberance for articulation with a similar
protuberance from the posterior edge of the coracoid, the
pelvis rotating around this joint when the dewlap is ex-
panded.

In short, Pervagor cannot be considered one of the
most generalized monacanthids, but, rather, it is
probably a derivative of a Paramonacanthus-like stock.
In both Paramonacanthus and Pervagor the dewlap is
short, not extending out beyond the beginning of the en-
casing scale series, which protrude conspicuously from
the ventral contour. The specializations of Pervagor
beyond the Paramonacanthus level are the loss of the for-
ward extension of the supraoccipital, the more anterior
position of the dorsal spines, the greater elongation of the
body, and the development of the rotation processes on
the coracoid and pelvis.

Two other genera with a flexible encasing scale series
and rudiments of the fin-ray element are Laputa and
Chaetoderma, both of which also have the dorsal spines
placed in the generalized location over the rear of the eye.

Chaetoderma has a relatively well-developed forward
extension of the supraoccipital and at least a low weak
crest, the forward extension being not much less exten-
sive than in Stephanolepis, Monacanthus, and Para-
monacanthus. The most outstanding specializations of
Chaetoderma are that the encasing scale segments have
been reduced to two, while retaining their flexibility, and
that the two rudiments of the fin-ray element are
minute. The ventral flap of Chaetoderma is small and
similar to that of Paramonacanthus. Chaetoderma could
well be a derivative of Paramonacanthus, specialized by
the increased depth of the body, the great development
of dermal flaps on the skin, the reduction in the number
of segments of encasing scales and in the size of the rudi-
ments of the fin-ray element, and the slight reduction in
the size of the forward extension of the supraoccipital.

However, the scales of Chaetoderma are strongly
reminiscent of those of the rough skinned Monacanthus,
especially those of the Indo-Pacific species, in which the
strong central upright spinule is backwardly directed and
bears a notch along its anterior or outer edge. The scales
of Chaetoderma are an elaborate version of this, with a
deep cleft in the usually strong single spinule dividing
the spinule into a smaller upwardly or anteriorly directed
prong and a larger posteriorly directed prong.

I suspect that the Australasian Chaetoderma is deriv-
ed from the Indo-Pacific stock of Monacanthus, by all
the ways of specialization mentioned above in com-
parison to Paramonacanthus, and additionally by the
loss of caudal peduncle armature and the great reduc-
tion in the size of the dewlap probably associated in some
way with the camouflage value of the extensively
developed dermal flaps. Additional evidence for a close
Chaetoderma and Indo-Pacific Monacanthus relation-

ship is that at least some species of the latter have at
least poorly developed dermal flaps.

In Laputa there is a full complement of three seg-
ments in the flexible encasing scale series and the fin-ray
rudiments, while small, are not minute. The dorsal lobe
of the pelvis is of moderate size and the dewlap just as
short as in Paramonacanthus. The most prominent
specializations of Laputa in comparison to the other
genera with flexible encasing scales and fin-ray elements
are that the forward extension of the supraoccipital has
been lost and that the encasing scale series is somewhat
reduced in size. I suspect that Laputa is on the same line
as that leading from Paramonacanthus to Pervagor, with
Laputa being a derivative of the forerunner of Pervagor
in which the dorsal spine had not yet migrated forward to
above the middle of the eye.

In addition to Stephanolepis, Monacanthus, and Paro'
monacanthus, with fully developed forward extensions of
the supraoccipital bearing a relatively well-developed
crest, and Chaetoderma, with a somewhat shorter for-
ward extension and very shallow crest, two other genera
have a moderately developed forward extension and low
crest, a still rather generalized condition, and another
genus has the forward extension but no crest. These are
Alutera, Brachaluteres, and Psilocephalus, in all of
which the forward extension reaches to between the mid-
dle and almost the front of the eye, with the crest
relatively distinct in Alutera but only very poorly
developed in Brachaluteres and absent in Psilocephalus.
All three of these genera have highly specialized pelvic
apparatuses, with no dorsal lobe to the pelvis, no pelvic
fin-ray element, and either no encasing scales or a single
enlarged scale representing the remnant of the series
(some Alutera), even this being lost in large adults.

The first dorsal spine is placed over about the middle
of the eye in Alutera and Brachaluteres, and only slightly
more posteriorly in Psilocephalus, while Brachaluteres
and Psilocephalus are further specialized by having lost
the small second dorsal spine, and Psilocephalus by the
great elongation of the body, increased number of verte-
brae and development of a long chin barbel. In fact, of
the genera studied here, I suspect that at least the ma-
jority of those without a dorsal pelvic lobe, with the en-
casing scale series essentially absent and with no fin-ray
element, form a natural group representing two diver-
gent lines of diversification from a more or less Alutera-
like ancestral group, which itself evolved from some
derivative of the Stephanolepis- Paramonacanthus level
of organization, perhaps from an early Rudarius -like
group that still retained the forward extension of the
supraoccipital, and which had the dorsal spine placed
over the middle of the eye and a tendency for the reduc-
tion of the fixed encasing scale series, such as is evident
in the two Recent species, R. ercodes with two segments
and R. minutus with only one.

The Alutera-\ike ancestral group referred to above is
envisioned as being essentially like Recent Alutera, with
the exception of having a better developed spiny dorsal
fin basal pterygiophore more closely applied to the skull.
This ancestral group is seen as diverging into two lines.

One line leads to Alutera and Psilocephalus by a
decrease in the size and closeness of association of the
basal pterygiophore with the skull, reduction in the size
of the first dorsal spine, and elongation of the body and
increased number of vertebrae, as found in some Recent
Alutera (vertebrae 20 to 23) and carried to an extreme in
Psilocephalus (vertebrae 29 or 30). In Psilocephalus the
basal pterygiophore is greatly reduced in size and only
closely attached to the skull anteriorly, while the first
dorsal spine is slender and delicate and the second spine
is lost, and the pelvis is reduced to a thin narrow shaft.

The other line of evolution from the Alutera-Uke an-
cestral group would seem to lead to those genera with a
simplified pelvic apparatus (no dorsal lobe, no encasing
scales, no fin-ray element) but with a well-developed
spiny dorsal fin basal pterygiophore and a relatively well-
developed first dorsal spine. On this line is Bracha-
luteres, which retains the forward extension of the supra-
occipital and at least a remnant of the crest, as well as
the relatively low number of 20 vertebrae found also in
one species of Alutera, while the body depth is increased
rather than decreased. Part of the increased size of the
deep basal pterygiophore in Brachaluteres may be as-
sociated with this increased depth of the body and thus
of the muscle bands attaching to the sides of the pteryg-
iophore. The first dorsal spine of Brachaluteres is strong,
but the second spine is absent. Brachaluteres has the
same kind of elaboration and branching of the epi-
pleurals as found in some Alutera.

Undoubtedly also on this line and closely related to
Brachaluteres is Paraluteres, with an equally low num-
ber of vertebrae, an even more reduced pelvis and neither
a forward extension of the supraoccipital nor a crest. The
basal pterygiophore, while strong and closely held to the
skull, is placed a little further back on the skull than in
Brachaluteres, and the dorsai spine is over the rear
rather than the middle of the eye, perhaps in conjunc-
tion with the resculpturing of the skull that took place as
Paraluteres developed its mimicry of Canthigaster valen-
tini, with it being advantageous for the dorsal spine to be
able to be laid back inconspicuously against the body
rather than jutting out prominently from the head. In
fact, the dorsal spine in Paraluteres is covered with thick
scaleless skin and in life is probably not capable of being
erected at a full 90° to the dorsal profile, even though the
second dorsal spine and presumedly also the locking
mechanism are present.

Paraluteres probably evolved from the same ancestral
group as Brachaluteres, but before the second dorsal
spine was lost and the body depth so greatly increased.
From this same ancestral group from which Bracha-
luteres and Paraluteres evolved also probably came Oxy-
monacanthus and Pseudaluteres, with much the same
tendencies for elongation and increased number of verte-
brae as in the Alutera-Psilocephalus line, but with the
exact opposite trend with the basal pterygiophore and
dorsal spine, the pterygiophore remaining large and
closely associated with the skull, as in Brachaluteres and
Paraluteres.

In Oxymonacanthus the basal pterygiophore of the

spiny dorsal fin is prolonged anteriorly over the posterior
one-third of the surface of the ethmoid, and the first dor-
sal spine is placed over the front of the eye. In Pseuda-
luteres the basal pterygiophore is even further prolonged
anteriorly, being of such great length that it covers nearly
the entire length of the dorsal surface of the skull, the
dorsal spines being placed further anteriorly than in any
other monacanthid, distinctly in front of the eye.

As discussed before, there is no counterpart in any
other scleroderm plectognath fish of the true or pleural
ribs found in Pseudaluteres.

Oxymonacanthus and Pseudaluteres share a specializ-
ed condition of the sphenotic found in no other mona-
canthids. In other monacanthids the lower region of the
side of the cranium, just above the region of articulation
of the upper end of the hyomandibular, is formed by the
sphenotic anteriorly and by the pterotic posteriorly, the
former ending ventrally as a ventrally or anteroventrally
directed prong and the latter having a larger ventrally
directed prong that overlies the posterodorsal region of
the hyomandibular. In both Oxymonacanthus and
Pseudaluteres, as explained in detail in the section on
anatomical diversity, the sphenotic is displaced pos-
teriorly by a ventrally directed portion of the posterior
region of the frontal bearing its own ventrally directed
prong, the side of the cranium just above the ar-
ticulation of the hyomandibular thus being formed by
the frontal anteriorly and the pterotic posteriorly, with
the sphenotic squeezed in between and with a relatively
small ventral prong. Moreover, Oxymonacanthus and
Pseudaluteres are the only two genera of monacanthids
to have completely lost the prootic shelf.

Oxymonacanthus is speculated to be an early offshoot
of the line leading to Pseudaluteres, an offshoot of the
pie-Alutera group before the dorsal lobe of the pelvis was
lost and the encasing scales nearly entirely lost. In Oxy-
monacanthus the dorsal lobe of the pelvis is only slightly
developed and the encasing scales are reduced to an in-
determinate number of scales gradually and slightly in-
creased in size from those of the surrounding region,
while the number of vertebrae is increased to about 25
(only 1 less than in Pseudaluteres), and the forward ex-
tension of the supraoccipital is absent.

Except for its specialized small nipping mouth (with
reduced number of dentary teeth) and slightly elongate
snout, a form such as Oxymonacanthus could easily have
given rise to Pseudaluteres by the continued reduction in
the dorsal lobe and encasing scales and the further
migration forward of the dorsal spine and the anterior
elongation of the basal pterygiophore. The shared highly
specialized nature of the sphenotic in both genera is as-
surance of their close relationship.

All of the other genera of monacanthids examined
here, but not yet discussed, form the middle group of
moderately specialized genera intermediate in many
ways between the basal Stephanolepis-Monacanthus-
Paramonacanthus group and their close relatives and the
group of genera of several lines thought to be derived
from a pre-A/u(era-like form. The middle group con-
tains genera without a forward extension of the supraoc-

cipital and without a fin-ray element, but with encasing
scales usually present in an inflexible series or only
minutely developed, with 19 or 20 vertebrae and the body
not especially either elongate or deepened. These are
Amanses, Cantherhines, Rudarius, Navodon, and
Acanthaluteres.

Of these, Amanses and Cantherhines have two and a
half segments of encasing scales, Navodon two, Rudarius
two or one, and Acanthaluteres only minute series when
present. Amanses, Cantherhines, and Acanthaluteres
have a deep groove posterior to the dorsal spines which is
not present in Navodon and Rudarius, while the first dor-
sal spine is placed behind the middle of the eye in
Navodon, Rudarius, and Acanthaluteres, but over the
middle or front of the eye in Amanses and Cantherhines,
the latter two genera also having relatively plain dorsal
spine ornamentation.

Amanses and Cantherhines have long been considered
closely related, rightly so I think, and have been well
characterized by Randall (1964), who showed them to
differ primarily in spinulation {Amanses with coarser
scales and a patch of long quill-like spines on the side of
the body between the soft dorsal and anal fins) and a
deeper caudal peduncle and lower numbers of soft dor-
sal and anal fin rays. Amanses is obviously a slightly
specialized spinier scaled version of Cantherhines, and
Randall (1964:332) has suggested that Scobinichthys and
Meuschenia (neither studied here) are also closely
related to Cantherhines.

I suspect that Cantherhines and its close relatives are
derived from a Pervagor-like stock (prior to the develop-
ment of the special rotation joint between the pelvis and
coracoid), some species of which have a groove behind
the dorsal spines located above the middle of the eye, by
a decrease in the amount of ornamentation of the first
dorsal spine and sometimes of its slight forward migra-
tion, and by the reduction of the encasing scale series
from a flexible three segments around a fin-ray element
to an inflexible two and a half segments without a fin-ray
element.

Whether Acanthaluteres, with a deep groove behind
the dorsal spines, is related to the Cantherhines-hke
genera is difficult to say, for it has a more generalized
dorsal spine position over the rear of the eye but a more
specialized pelvic apparatus, the dorsal lobe being es-
sentially absent and the encasing scales either absent or
minute. I suspect that Acanthaluteres is an offshoot of
the line leading to the pre-/l/ufera-like group hy-
pothesized here as containing the Alutera-Psilocephalus
line and the Brachaluteres-Paraluteres and the Oxy-
monacanthus-Pseudaluteres lines.

Rudarius is probably a derivative of the line leading
from a Pervagor-\\ke form to those of the Cantherhines-
like genera, but with the placement of the dorsal spine
remaining a little more posterior, usually just behind the
middle of the eye, while the trend of reducing the in-
flexible encasing scale series continued further to only
two segments in one of the species and to one segment in
the other, while the dorsal lobe and ventral dewlap
remained of moderate to small size.

The dense patch of setae found on the caudal peduncle
of male Rudarius ercodes is similar to that found in
males of some species of Cantherhines (others have pairs
of retrorse spines; see Randall 1964 and Tyler 1970e),
while the much longer posteriorly directed spines on the
caudal peduncle of male R. minutus are undoubtedly a
specialized development from the ancestral R. ercodes-
like form and are not in any way related to the even
longer quills further forward on the body of Amanses.
Similarly, the large dermal flaps found in R. minutus
have a much smaller counterpart in some species of
Cantherhines (see Randall 1964).

Navodon shares many characteristics with Rudarius,
as pointed out by Fraser-Brunner (1941b:178-179), and
the two are probably closely related, both specialized in
their own way, Rudarius with a relatively deep small
body and only two teeth in each dentary and Navodon
with a larger more elongate body and a full complement
of three dentary teeth but a more forwardly placed gill
slit and reduced size of the ventral dewlap.

Four genera not examined for this work and not dis-
cussed above should be briefly mentioned, these being
Arotrolepis, Pseudomonacanthus, Blandowskius, and
Eubalichthys. According to the data in Fraser-Brunner
(1941b), these four genera can be characterized as fol-
lows:

Arotrolepis has the dorsal spines originating behind
the middle of the eye, a flexible encasing scale series
(presumedly of three segments and with a fin-ray ele-
ment) of small size, the ventral dewlap of no more than
moderate size, and the scales usually with a single strong
central spinule and arranged in distinct longitudinal
tracts. Fraser-Brunner thought Arotrolepis most closely
related to Monacanthus, although in his key to the
genera (which he cautions does not show the true rela-
tionships of some of the genera) it is coupled with
Chaetoderma. I suspect that he is right and that
Arotrolepis is on the Monacanthus-Chaetoderma line. It
obviously would be of great interest to know the form of
the supraoccipital in Arotrolepis.

Pseudomonacanthus (most data from Fraser-Brunner
1940c) has the dorsal spines originating over the middle
or behind the middle of the eye and an inflexible short
series of encasing scales (presumedly of less than three
segments and with the fin-ray element absent) and no
dorsal pelvic lobe.

Blandowskius, which Fraser-Brunner thought was
most closely related to Pseudomonacanthus, has no
encasing scales at all and the dorsal spines over the
rear of the eye, with both genera being intermediate
between the Rudarius and the Alutera levels of organiza-
tion.

Eubalichthys, with the dorsal spines over or in front of
the middle of the eye and followed by a groove for their
reception, and with the inflexible encasing scale series
much reduced in size until it is minute and incon-
spicuous, was thought by Fraser-Brunner to be a
derivative of the Cantherhines-Amanses line.

In short, many of the generic relationships within the
monacanthids remain unclear, and will probably con-
tinue to be unclear until many more species, represent-
ing all genera, have been examined osteologically than it
has been possible to do here. However, it does seem clear
that the most generalized forms are those with an
anteriorly elongate supraoccipital bearing a vertical crest
and with a well-developed series of flexible encasing
scales of three segments covering over a rudimentary fin-
ray element, and with the dorsal spine placed behind the
middle of the eye. The genera sharing these conditions
are Stephanolepis, Monacanthus, and Paramona-
canthus.

A Stephanolepis-like stock is probably ancestral to
both Monacanthus and Paramonacanthus. Chaeto-
derma (and probably Arotrolepis) seems a derivative of a
Monacanthus-like stock, while Laputa is probably deriv-
ed from Paramonacanthus along with Pervagor. It is sus-
pected that a Pervagor-\\ke stock is ancestral to the
various moderately specialized genera with inflexible en-
casing scale series and no fin-ray element. An Alutera-
like stock is suspected to have evolved from some
derivitive of a Paramonacanthus-like stock, perhaps
from a form like Rudarius or Navodon, and to have
diverged in two directions, one leading to Alutera and
Psilocephalus and the other to two lines leading to
Brachaluteres and Paraluteres and to Oxymonacanthus
and Pseudaluteres.

SUPERFAMILY OSTRACIOIDEA

Comparative diagnosis (contrast with that of the
Balistoidea). — Head and most of body encased in a
relatively inflexible (except on cheek in front of gill slit)
cuirass of especially thick and mostly hexagonal scale
plates whose apposed edges are articulated by minute in-
terdigitations, the portion of the body posterior to the
carapace with or without isolated scale plates of various
shape but never continuously covered by scales; body
outline in cross section less laterally compressed and
always with two or more angles or ridges; no spiny dorsal
fin; soft dorsal and anal fins short-based, with 9 to 13
rays and a slightly lesser number of basal pterygio-
phores; caudal fin with 10 or 11 rays; pelvis and pelvic fin
absent; teeth relatively small and more or less conical or
only slightly compressed basally but often constricted
distally, between 6 and 17 in a single series in both jaws;
lateral line not associated with grooves or spiny processes
on the scale plates; branchiostegals usually 2-1-4 but
sometimes 1 -I- 4 or 2 -I- 3, at least some of the rays in the
posterior division as broad and laterally compressed as
those in the anterior division; distal end of last bran-
chiostegal ray always articulated to the inner surface of
the suboperculum; elements of the hyoid arch more com-
pacted together anteroposteriorly and most elements less
elongate and more deep bodied; urohyal much reduced in
size, a more or less flattened plate or a slightly curved
rod, a ventral flange either absent or very poorly
developed; fifth ceratobranchial (lower pharyngeal) al-

ways toothless; pharyngobranchials consisting either of a
toothless suspensory element followed by three elements
bearing small but protruding or minute nonprotruding
teeth, or a toothless suspensory element (absent in some
Acanthostraclon) followed by two elements, one or both
of which bear minute nonprotruding teeth; epi-
branchials always four; gill rakers always present along
anterior edge of fifth ceratobranchial (posterior edge of
last gill slit); caudal fin supporting structures extremely
consolidated, the epural, hypural, parhypural, and cen-
trum elements fully fused into a single plate; haemal
spine of penultimate vertebra either autogenous or fused
to its centrum; neural arch of the last vertebra com-
plete, the canal accommodated in a tube through the
plate exiting on the posterior half of the dorsal edge,
while the haemal arch is either complete and the canal
accommodated in a tube through the plate or the arch is
essentially absent, the haemal region of the plate solid
and not pierced by the canal; uroneurals never present;
vertebrae normally 9 -I- 9 = 18 or 10 -I- 8 = 18, but 9 -I- 10
= 19 in one species, and always 9 or 10 abdominal verte-
brae; the first 2 to 5 vertebrae reduced in size at least
anteroposteriorly and at least partially fused to the rear
of the skull; 7, more rarely 8, abdominal vertebrae with
neural spines anterior to the first basal pterygiophore of
the soft dorsal fin; 3 to 5 caudal vertebrae posterior to the
last basal pterygiophores of both the soft dorsal and anal
fins; usually no more than 2, rarely 3, soft dorsal fin basal
pterygiophores placed between successive neural spines;
most neural spines positioned strongly obliquely in rela-
tion to the axis of the vertebral column; haemal spines
above the anal fin basal pterygiophores relatively
shorter, either rudimentary or well developed as antero-
posteriorly expanded plates but never long and shaftlike
and not penetrating into the proximal region of the series
of anal fin basal pterygiophores, and at least some of
these haemal spines markedly different in shape and size
from the neural spines above them; soft dorsal fin basal
pterygiophores 8 to 10; anal fin basal pterygiophores 8 to
10; prominent thin lateral flanges not present on the soft
dorsal and anal fin basal pterygiophores; distal pteryg-
iophores either absent or unossified in both the soft dor-
sal and anal fins; 4 to 6, or about half, of the anal fin
basal pterygiophores with only their distal ends placed in
the midline of the body, their proximal regions in-
creasingly divergent variously to the left and right of the
midline in the musculature bordering the rear of the ab-
dominal cavity; no spiny dorsal fin and hence no spiny
dorsal fin basal pterygiophores, but a supraneural ele-
ment present anteriorly from the dorsal end of the first
basal pterygiophore of the soft dorsal fin, this supraneural
element representing a modified basal pterygiophore, per-
haps the supraneural strut of balistids; epipleurals and
ribs never present; uppermost pectoral fin ray either of
two relatively well-developed halves of about equal length
(aracanids) or of a single piece bearing a foramen basally
(ostraciids); actinosts inflexible and articulated by
suturing to one another and to the scapula and coracoid;
coracoid and cleithrum enlarged; coracoid expanded
ventrally, as wide as or wider there than dorsally, with

prominent broad flanges; coracoid with or without a pos-
terior prong along its posterodorsal edge, the prong when
present not so well -developed or closely associated with
the lowermost actinost; scapular foramen always com-
plete; cleithrum with an anterior flange from its lower
anteromedial edge; postcleithrum usually with two dis-
tinct halves, rarely as a single piece, the bone directed
posteriorly or only slightly obliquely downward and
variously slightly to enormously expanded posteriorly
and ventrally into a flat relatively thin plate, but never
as a long sturdy rod; nearly the entire length of the supra-
cleithrum overlying the cleithrum, or cleithrum and
postcleithrum, and the supracleithrum very firmly held
to them and the posttemporal; Baudelot's ligament os-
sified as a stout heavy rod between the posterior region of
the parasphenoid and the region of articulation of the
supracleithrum, cleithrum, and posttemporal; posttem-
poral usually relatively large; palatine an elongate block
of bone representing an enlarged foot of the T-shaped
balistid palatine and firmly sutured along all but the dis-
tal end of its medial surface to the ectopterygoid and
mesopterygoid while its distal end is held by ligament
primarily to the vomer; vomer much enlarged, having
either a strongly laterally expanded anterior end and a
relatively short posterior tapering shaft fitting into a
shallow concavity in the parasphenoid (aracanids) or a
strongly expanded lateral surface throughout its length,
so that the bone is more or less square and has a flat-
tened posterior surface that sutures to the similarly flat-
tened anterior end of the parasphenoid (ostraciids); ven-
tral edge of the ventral flange of the parasphenoid
laterally expanded slightly to moderately and only par-
tially along its length (aracanids) or well expanded along
all of its length anterior to the level of about the pre-
frontal (ostraciids); ventral flange of the parasphenoid
without a deep indentation at about the level of the pre-
frontal; parasphenoid with a high dorsal flange in the
medial septum of the orbit which sutures broadly or only
slightly with long anteroventral extensions of the ptero-
sphenoids into the medial septum; the medial edges of
the pterosphenoids not in contact in the posterior wall of
the orbit, well separated there; myodome small or ab-
sent; apposed surfaces of the parasphenoid and basioc-
cipital not excavated and no canal between them leading
anteriorly into the myodome cavity; epiotics not in con-
tact medially on the posterior surface of the skull, widely
separated there by the more posteriorly placed supraoc-
cipital; supraoccipital either with or without a posterior
crest, but never with an anterior crest on the surface of
the main body of the bone; prootic shelf usually relative-
ly larger, always with a prominent ventral or ventro-
lateral flange from its lateral edge; the major foramen in
the prootic shelf relatively large and elongate, and not
completely enclosed by the prootic, being bounded
laterally by the prootic and medially by the para-
sphenoid; anterior edge of the upper part of the preoper-
culum articulated below along the rear edge of the
hyomandibular but above to a groove on the lateral sur-
face of the hyomandibular; hyomandibular a greatly ex-
panded plate, without much of a shaftlike portion.

Family Aracanidae

Comparative diagnosis (contrast with that of the
Ostraciidae). — Supraorbital region of the skull relative-
ly strong, the frontal relatively thick in this region and
not held extraordinarily firmly to the undersurface of the
carapace; frontal with a long forward extension broadly
overlying the dorsolateral surface of the ethmoid to, or
very nearly to, the extreme anterior end of the snout;
supraoccipital with a long, thin, vertical crest extending
posteriorly well beyond the level of the base of the skull;
epiotic much prolonged posterolaterally, well beyond the
level of the base of the skull, and beyond the level of the
distal end of the supraoccipital crest; prootic shelf with a
recurved wing from its lateral edge which makes contact
with the anterior edge of the laterally expanded pos-
terior region of the bone; myodome small but distinct;
four pharyngobranchials, the first a toothless suspen-
sory element, the second either toothless or with minute
teeth, the third and fourth with few but large teeth;
branchiostegals usually 2 -t- 4, but sometimes 1 -I- 4 or 2
-I- 3; ethmoid relatively deep and platelike and not much
laterally expanded anteriorly; pterosphenoid and para-
sphenoid with a broad region of suturing in the interor-
bital septum; parasphenoid anterior to its region of
suturing with the anteromedial edges of the prootic
shelves either not much expanded laterally or expanded
laterally only for a portion of its length there, the lateral-
ly expanded portion being either anteriorly near the
vomer or posteriorly near the prootic shelf, but not form-
ing a complete hard palate over the roof of the oral
cavity; posterodorsal region of the coracoid with a
moderately to well-developed posterior prong; two small
vertebrae, represented mostly by neural arch material, at
the front of the vertebral column variously fused and
sutured to one another and to the exoccipitals and
basioccipital, the fusion region without a ventral process
below the level of the centra; vertebrae immediately fol-
lowing the fusion region with long neural spines and
anteroposteriorly much compressed centra with well-
developed transverse processes; the centra of the first
several vertebrae anterior to the caudal plate not much
compressed anteroposteriorly; haemal arch and spine of
the penultimate vertebra relatively well developed and
autogenous; caudal fin rays i, 9, i (except the Eocene
Proaracana, with i, 8, i); postcleithrum (especially the
ventral piece) greatly expanded posteriorly into a deep
thin plate reaching the level of the anterodorsal ends of
the more anterior anal fin basal pterygiophores; supra-
neural long and well developed, with a relatively deep
ventral flange making contact with most or all of the
neural spines of the predorsal vertebrae (exclusive of
those of the small first two fused vertebrae), extending
anteriorly well beyond the level of the anteroventral end
of the first dorsal fin basal pterygiophore; haemal spines
on the more posterior abdominal and on most of the
caudal vertebrae well developed, especially those above
the anal fin basal pterygiophores which they support
through a connective tissue sheet rather than by close ap-
position; first anal fin basal pterygiophore without

Figure 129.— Typical body form in
the Recent Aracanidae: Aracana aurita.

anterolateral wings from its distal end; haemal canal en-
tirely straight, in the midline under the vertebral centra
and relatively well enclosed by well-developed haemal
arches and spines on all but the more anterior ab-
dominal vertebrae; none of the vertebrae posterior to the
anteriormost two fused vertebrae sutured to one another;
uppermost pectoral fin ray relatively well developed,
consisting of two halves with a few cross-striations dis-
tally, the two halves of about the same length or one half,
usually the lateral, only slightly shorter than the other;
body relatively deep, the distance between the distal
ends of the first dorsal and anal fin basal pterygiophores
being contained less than 3 times in the SL and the dis-
tance between the top of the rear of the cranium and the
ventralmost edge of the pectoral girdle being contained
about one and three-fourths to slightly more than twice
in the SL; carapace open behind the dorsal and anal fins;
carapace with a well to only slightly developed ventral
keel; caudal peduncle always with scale plates isolated
from the carapace proper, often almost completely, if not
completely, encircling it.

Detailed Description of Kentrocapros aculeatus.

Material examined.— Two cleared and stained
specimens; 90.0-90.7 mm.

Occipital Region.

Basioccipital. — A short column, slightly expanded
anterolaterally; cartilage filled anterodorsally; ar-
ticulates through cartilage and interdigitation dorsally
with the exoccipitals and anterolaterally with the
prootics, while its ventromedial surface is broadly over-
lain by and extensively interdigitated with the bifurcate
posterior end of the parasphenoid. The upper region of
the extreme posterolateral surface of the basioccipital is
more or less indistinguishably fused with the basal region
of the rudimentary first two abdominal vertebrae, as ex-
plained in the section on the vertebral column.

Figure 130.— Range of diversity in

body form and external features of

representative aracanid genera:

Strophiurichthya inermis (above) and

S. ro6u«(u«— in front of each, above,

nasal region as seen externally (far

left) and the olfactory lamellae as

seen with the top of the nasal sac

removed, and. below,

outline of cross section of middle of body.

Exoccipital. — Cartilage filled along all of its edges
of articulation with the other cranial bones; articulates
through cartilage posteromedially with its opposite
member and posterodorsally with the supraoccipital;
articulates through cartilage and extensive inter-
digitation posterolaterally with the epiotic, ventro-
medially with the basioccipital, and anteroventrally with
the prootic, while ventroiaterally it articulates mainly
through cartilage but with some interdigitation with the
pterotic. The lateral surface of the extreme posterior end
of the exoccipital is overlain by and interdigitated with
the neural spines of the rudimentary first two ab-
dominal vertebrae, as explained in the section on the
vertebral column. In this region the exoccipitals form the
lateral wails of the foramen magnum, while the dorsal
wall is formed by the cartilaginous sheet between the
posteromedial edges of the two exoccipitals; the ventral
surface of the foramen magnum is bounded by the basi-
occipital.

Supraoccipital. — Laterally expanded and with a
well-developed supraoccipital crest directed posteriorly;
cartilage filled along all of its edges of articulation with
the other cranial bones; articulates by extensive inter-
digitation anteriorly and anterolaterally with the fron-
tals, while it articulates through cartilage laterally with
the epiotics and ventrally with the exoccipitals.

Otic Region.

Pterotic. — Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates
through cartilage and interdigitation anteroventrally
with the prootic, posteroventrally with the exoccipital,
posterodorsally with the epiotic, and anterodorsally with
the sphenotic; the posterodorsal portion of the sphenotic
intervening between the pterotic and frontal so that the

latter two bones do not articulate directly with one
another. For a short distance medially along its antero-
ventral edge the pterotic supports through cartilage the
posterodorsal edge of the hyomandibular. The antero-
lateral region of the pterotic is prolonged ventrally into a
stout shaft whose medial edge articulates through fi-
brous tissue with the posterodorsal region of the hyoman-
dibular while laterally it is broadly overlain and exten-
sively interdigitated with the posttemporal and medially
it helps to support by fibrous tissue the expanded lateral
end of the ossified Baudelot's ligament.

Sphenotic. — Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates
through cartilage and interdigitation posterodorsally
with the pterotic and, to a lesser extent, with the epiotic,
while anterodorsally it interdigitates with the frontal;
medially in the posterior wall of the orbit the sphenotic
articulates through cartilage and interdigitation with the
pterosphenoid, prootic, and pterotic. The dorsolateral
surface of the sphenotic is broadly overlain by the fron-
tal.

Epiotic. — A rounded sturdy cone anteriorly but
prolonged posteroventrally into a thin flange; cartilage
filled along all of its edges of articulation with the other
cranial bones; articulates through cartilage and inter-
digitation dorsally with the frontal, anteroventrally
along its lateral edge with the sphenotic and pterotic and
posteroventrally with the exoccipital and medially,
through cartilage only, with the supraoccipital.

Prootic. —Cartilage filled along all of its edges of
articulation with the other cranial bones, except an-
teriorly and ventrally where it articulates with the para-
sphenoid; articulates by interdigitation ventromedially
with the laterally compressed keellike dorsal region of

Figure 131.— Range of diversity in body form and

external features of representative aracanid genera:

Aracana aurita— upper left, nasal region as

seen externally (olfactory lamellae, if present.

indistinct in specimen examined); lower left,

outline of cross section of middle of body.

Figure 132.— Range of diversity in body

form and external features of representative

aracanid genera: Anoplocapros amygdatoides —

upper left, nasal region as seen

externally (far left) and the olfactory lamellae as

seen with the top of the nasal sac removed: lower

left, outline of cross section of middle of body.

Figiire 133.— Range of diversity in body form

and external features of representative

aracanid genera: Capropygia untstnafa— uppei

left, nasal region as seen externally

(olfactory lamellae, if present, indistinct in

specimen examined); lower left, outline of

cross section of middle of body.

Figure 134.— Range of diversity in body

form and external features of representative

aracanid genera: Caprichthys gymnura— upper

left, nasal region as seen externally (olfactory

lamellae, if present, indistinct in specimen

examined); lower left, outline of cross

section of middle of body.

the parasphenoid; articulates through cartilage and in-
terdigitation ventrolaterally with the pterotic, postero-
ventrally with the basioccipital and exoccipital, dorso-
medially with the pterosphenoid, and dorsolaterally with
the sphenotic. Along a short distance of the anterior edge
of its laterally expanded posterior portion the prootic ar-
ticulates through cartilage with the posterodorsal end of
the hyomandibular. The anteromedial region of the
prootic possesses a long forward extension under the or-
bit which makes contact by interdigitation with the
parasphenoid at the level of the prefrontals. A recurved
portion of this prootic subocular shelf is directed postero-
laterally and makes contact with the anterolateral region
of the rear half of the prootic, the two parts of the bone
interdigitating in this region just medial to the ar-
ticulation of the prootic with the dorsal edge of the
hyomandibular. Just ventral to this region of contact
between the posterolateral wing of the prootic shelf with
the rest of the prootic, the wing is thickened and ar-
ticulates by fibrous tissue with the dorsal surface of the
lateral region of the ossified Baudelot's ligament. The
medial edges of the prootics form the lateral walls of the
myodome, while medially directed shelves from the
medial edges of the prootics more or less meet in the mid-
line and articulate through cartilage with one another to
form the dorsal roof of the myodome. The anterior edge
of the myodome is formed by the prootic, except ven-
trally where it is formed by the parasphenoid.

Orbital Region.

Frontal. — Wide in the posterior half of its length,
then rapidly tapering to a point anteriorly; extending
throughout almost the entire length of the skull and a
relatively sturdy bone throughout its length; articulates
through cartilage and interdigitation posteromedially
with the supraoccipital and posterolaterally with the
epiotic and sphenotic, broadly overlying the latter. In the
posterior wall of the orbit the frontal articulates through
cartilage and interdigitation with the pterosphenoid and
sphenotic. The anterior prolongation of the frontal
broadly overlies and is held by fibrous tissue to the dorso-
lateral surface of the ethmoid. Immediately posterior to
where it begins to overlie the ethmoid, the ventral sur-
face of the frontal is held by fibrous tissue to the dorsal
surface of the prefrontal.

Prefrontal. — Large and wedge-shaped; cartilage
filled along its medial edge where it is continuous with
the ethmoid cartilage; articulates by fibrous tissue dor-
sally with the broadly overlying frontal and ventrally
with the dorsal edge of the parasphenoid, the latter ar-
ticulation perhaps strengthened by slight inter-
digitation. Its major surface of articulation, however,
remains with the ethmoid cartilage, that broadly inter-
venes between the prefrontal and the ethmoid.

Parasphenoid. —An elongate, mostly laterally com-
pressed, slab of bone; expanded in the middle of its

length for articulation by interdigitation with the antero-
medial edges of the prootic shelves, and, more pos-
teriorly along its ventral edge, where it interdigitates
with the prootics and basioccipital. Dorsally in the region
of the orbit the parasphenoid has a large dorsal flange
which meets and interdigitates with the ventral flanges
of the pterosphenoids. Just anterior to this dorsal flange
the parasphenoid articulates by fibrous tissue and per-
haps slight interdigitation with the ventral end of the
prefrontal. Anterodorsally the parasphenoid articulates
by fibrous tissue and interdigitation with the ventral
edge of the ethmoid, while the extreme anteroventral
edge of the parasphenoid becomes concave for reception
of the posterior end of the vomer, with which it inter-
digitates. At its extreme posterior end the parasphenoid
bifurcates in the region that overlies the ventral surface
of the anterior half of the basioccipital. However, there is
no apparent opening into the myodome in this region
such as is found in triacanthoids and balistoids. Along its
laterally expanded ventral surface in the region of the
rear of the orbit the parasphenoid articulates by fibrous
tissue and perhaps slight interdigitation with the medial
ends of the ossified Baudelot's ligaments from either
side.

Pterosphenoid. — Cartilage filled along all of its
edges of articulation with the other cranial bones, ex-
cept for a very short distance anteroventrally; ar-
ticulates through cartilage and interdigitation postero-
dorsally with the frontal, which somewhat overlies it,
posteriorly with the sphenotic, and posteroventrally with
the prootic. The anterior end of the pterosphenoid is
prolonged anteroventrally into a process which inter-
digitates with the dorsal edge of the dorsal flange of the
parasphenoid in about the middle of the orbit.

Ethmoid Region.

Ethmoid. —Elongate; laterally expanded and
somewhat rounded dorsally, but ventrally becoming
laterally compressed where it articulates by inter-
digitation and fibrous tissue with the parasphenoid,
which in the posterior region of the articulation overlies
the ventrolateral edges of the ethmoid; cartilage filled at
its posterior edge, where it is continuous with the eth-
moid cartilage; articulates by fibrous tissue dorsally with
the anterior prolongation of the frontal, which broadly
overlies the dorsolateral surface of the ethmoid. Antero-
ventrally the ethmoid articulates through cartilage and
slight interdigitation with the dorsal surface of the
vomer. At its anterior edge the ethmoid helps support,
through fibrous tissue, the upper jaw.

Vomer. — A short, squarish block of bone with two
lateral expansions and a posterior tapering shaft; ar-
ticulates through cartilage and interdigitation dorsally
with the ethmoid and by interdigitation posteriorly
where its shaft fits into the concave anteroventral edge of
the parasphenoid. The anterior end of the vomer sup-

ports, through fibrous tissue, the upper jaw. The con-
cave region between the two lateral expansions of the
vomer articulates by strong fibrous tissue with the
medial surface of the palatine, holding it immovably to

Mandibular Region.

Hyomandibular. — Expanded posteriorly, and
slightly narrowed anteriorly where it ends bluntly; car-
tilage filled at its posterior and anterior edges; ar-
ticulates by fibrous tissue posteriorly with the extreme
lateral edge of the posterior portion of the prootic and
with the anteromedial edge of the pterotic, as well as
along the medial surface of the ventral shaft of the
pterotic. Dorsomedially the hyomandibular is firmly
held by fibrous tissue to the lateral edge of the prootic
shelf and its posterolateral wing. Anteriorly the hyo-
mandibular articulates along its dorsal edge by fibrous
tissue with the posteroventral end of the metapterygoid,
while its extreme anteroventral edge articulates through
cartilage and fibrous tissue to the fibrous tissue sheet
between the metapterygoid, symplectic, and preoper-
culum. In about the middle of its length the posterior
edge of the hyomandibular has a blunt process with a
concave head for articulation by fibrous tissue with the
head of the operculum. Just anterior to this articular
area, the hyomandibular bears an elongate groove along
its posterolateral surface into which the slender postero-
dorsal end of the preoperculum fits and is tightly held by
fibrous tissue.

Quadrate. — Wide posteriorly, tapering anteriorly
to a knob for articulation with the articular in the lower
jaw, while from its posteroventral edge it possesses a
well-developed posteriorly directed process; cartilage fill-
ed at its posterior edge; articulates by interdigitation
dorsally with the ectopterygoid, and, to a lesser extent,
with the extreme ventral edge of the mesopterygoid; ar-
ticulates posteriorly through cartilage with the metap-
terygoid; at the indented region on its lower posterior
edge the quadrate articulates by fibrous tissue and some
interdigitation with the anterior end of the symplectic,
which broadly overlies its lateral surface. Along most of
its ventral edge the quadrate articulates by fibrous tis-
sue with the preoperculum.

Metapterygoid. — A more or less elongate, rec-
tangular plate, somewhat wider anteriorly than pos-
teriorly; cartilage filled at its anterior edge; articulates
anterodorsally by interdigitation with the mesoptery-
goid, which it broadly overlies; articulates anteroven-
trally through cartilage with the quadrate, while along
the anterior half of its ventral edge it articulates by inter-
digitation with the symplectic. The posterior and
posteroventral edges of the metapterygoid are firmly held
by fibrous tissue to the anterodorsal edge of the hyoman-
dibular. The dorsal end of the interhyal articulates by
fibrous tissue with the ventral edge of the metapterygoid

just posterior to the posterodorsal end of the symplectic
and just in front of a ventral process of the metaptery-
goid.

Symplectic. — Slender and somewhat triangular in
shape; slightly filled with cartilage at its anterior and, to
a lesser extent, posterodorsal end; articulates by fibrous
tissue and interdigitation anteriorly with the quadrate,
which it broadly overlies, and posterodorsally by fibrous
tissue and extensive interdigitation with the anteroven-
tral surface of the metapterygoid, which it also overlies.
The posterior edge of the symplectic is in contact with
the fibrous tissue sheet between the symplectic, metap-
terygoid, hyomandibular, and preoperculum, and its
posterodorsal end is closely adjacent to the region where
the dorsal end of the interhyal articulates with the
metapterygoid.

Palato-Pterygoid Region.

Palatine. — A sturdy dome; articulates by exten-
sive interdigitation ventrally with the ectopterygoid and
mesopterygoid. Dorsomedially the palatine articulates
by fibrous tissue with the concave region on the lateral
surface of the vomer between the two laterally expanded
processes of the vomer.

Ectopterygoid. — Elongate, slightly widest in the
middle; articulates by extensive interdigitation dorsally
with the palatine, posteriorly with the mesopterygoid,
and ventrally with the quadrate, which broadly overlies
its posteroventral region.

Mesopterygoid. — A large triangular slab whose
lateral surface is broadly overlain by the metapterygoid,
to which it is extensively interdigitated; anteriorly the
mesopterygoid extensively interdigitates with the
palatine and ectopterygoid and, to a much lesser extent,
with the upper posterior edge of the quadrate.

Opercular Region.

Operculum. — A relatively flat plate except at its
upper end where it is expanded into a facet for ar-
ticulation by fibrous tissue with the slight concavity on
the ventrally directed knob of the posterior edge of the
hyomandibular. Ventrally the operculum broadly over-
lies and articulates by fibrous tissue with the suboper-
culum.

Suboperculura. — A thin plate, widest anteriorly, its
ventral region broadly rounded, tapering to a point pos-
teriorly; articulates by fibrous tissue dorsally with the
broadly overlying operculum. At the anterior edge of the
region where the operculum overlies the suboperculum, a
strong ligament coming from the interoperculum makes
contact with the suboperculum and, to a lesser extent,
with the operculum.

189

Interoperculum. — A short rod of somewhat ir-
regular shape but usually slightly wider in the middle
region than at either end; extends from the level of the
anterior end of the preoperculum to the level of the pos-
terior end of the epihyal; connects by a strong ligament
anteriorly to the angular in the lower jaw, while pos-
teriorly it connects by a short ligament to the epihyal and
by a longer more diffuse ligament to the anterior edge of
the suboperculum and, to a lesser extent, of the oper-
culum.

Preoperculum. — Relatively short; moderately ex-
panded in the middle region and tapering to narrow ends
anteriorly and posteriorly; articulates by fibrous tissue
anteriorly along its dorsal edge with the quadrate, while
posteriorly the narrow shaft of the preoperculum is held
by fibrous tissue along the ventral edge of the hyoman-
'dibular and, more dorsally, in the shallow groove on the
lateral surface of the hyomandibular. The middle of the
dorsal edge of the preoperculum also connects to the fi-
brous tissue sheet between the symplectic, hyoman-
dibular, and metapterygoid.

the vomer and ethmoid, and ventromedially with the
dorsolateral surface of the dentsu^.

Lower Jaw.

Dentary. —Wider posteriorly than anteriorly; its
posteromedial surface concave dorsally to accommodate
the articular, with which it interdigitates. Just below the
articular, the dentary interdigitates with the angular.
The ventromedial edges of the two dentaries are held
closely together by fibrous tissue. Laterally from its
posterodorsal region the dentary articulates by fibrous
tissue with the medial surface of the maxillary. Each
dentary bears four teeth, like those of the upper jaw, in
deep grooves on its outer surface, each of the tooth bear-
ing grooves ending posteriorly in a deep socket in which
new teeth develop. The sockets are in communication
with the large pulp cavity that fills most of the hollow in-
terior of the dentary. The pulp cavity communicates
with the exterior not only at the sockets but also at its
posterior concave region of articulation with the artic-
ular.

Upper Jaw.

Premaxillary. — A slightly curved plate, wider dor-
sally than ventrally; its posterodorsal region flattened for
articulation by fibrous tissue with the anterior edges of
the vomer and ethmoid; the anterior edge of the upper
jaw formed by the premaxillary, except for a short dis-
tance ventrally where it is formed by the maxillary; the
dorsomedial edges of the two premaxillaries held in close
apposition by fibrous tissue. Each premaxillary bears
four teeth in a single row in the two specimens examin-
ed. The teeth are borne in relatively deep and elongate
grooves on the outer surface of the premaxillary. The
lowermost teeth are shaftlike and taper to sharp points,
but more dorsally they are flattened and blunter at the
distal ends, except when they first erupt through the gum
as replacement teeth, at which time they too are sharp
pointed but apparently rapidly become worn down to
bluntness through use. At the posterior end of each tooth
bearing groove there is a deep socket in which new teeth
develop before moving forward to replace the old ones.
Most of the interior of the premaxillary contains the den-
tal pulp from which the new teeth develop. This pulp
cavity communicates with the exterior not only by the
deep sockets in which the new teeth develop but also by a
large hole in the posterodorsal surface of the maxillary.
The premaxillary articulates by extensive inter-
digitation with the maxillary along all of its posterior
edge, except for a short distance dorsally.

Maxillary. — Widest ventrally, with a deep in-
dentation along its lower posterior edge; articulates by
extensive interdigitation anteriorly along all of its length
except extremely ventrally with the premaxillary, which
it somewhat overlies. The maxillary articulates by fi-
brous tissue posterodorsally with the anterior edges of

Articular. — Small; its posterior edge with a con-
cavity for articulation by fibrous tissue with the anterior
knoblike process of the quadrate. Anteriorly the articular
interdigitates with the concave upper half of the postero-
medial surface of the dentary. On the medial side of its
ventral edge the articular interdigitates with the angular,
but the lateral surfaces of the two bones are not in con-
tact; rather, they are slightly separated by the dentary.
The sesamoid articular is a flattened nubbin of bone
mostly held to the medial surface of the anterior end of
the articular just behind the region where the upper
medial edge of the articular meets the medial surface of
the dentary.

Angular. — Small and mainly confined to the
medial surface of the lower jaw; articulates by inter-
digitation dorsomedially with the articular, while dorso-
laterally, anteriorly, and ventrally it interdigitates with
the dentary. Posteriorly the angular connects by liga-
ment with the anterior end of the interoperculum.

BRANCHIAL APPARATUS.

Hyoid Arch, Branchiostegal Rays, and Urohyal.

Hypohyals. — Both elements well developed; ven-
tral element larger than the dorsal element; dorsal ele-
ment cartilage filled at its ventral and posterior edges,
the ventral element cartilage filled at its dorsal and pos-
terior edges. The two elements articulate with one
another and with the ceratohyal mainly through car-
tilage, but on their medial surfaces the two hypohyals
have thin extensions which interdigitate with one
another. The anteromedial edges of both elements ar-
ticulate by fibrous tissue with their opposite members.
The dorsomedial edge of the dorsal hypohyal articulates

by fibrous tissue posteriorly with the first basibranchial
and anteriorly with the reduced urohyal, while the dorso-
medial edge of the ventral hypohyal also articulates by
fibrous tissue with the urohyal.

Ceratohyal. — A wide flat plate; shortened antero-
posteriorly and expanded dorsoventrally, particularly
posteriorly; cartilage filled along all of its edges except
for the indented regions dorsally and ventrally; ar-
ticulates through cartilage anteriorly with the dorsal and
ventral hypohyals, and posteriorly with the epihyal. The
first two branchiostegal rays articulate by fibrous tissue
with slight concavities on the anterior half of the ventral
edge of the ceratohyal. The next three branchiostegal
rays articulate by fibrous tissue with the posteroventral
edge of the ceratohyal, and the last ray with the ventral
region of the lateral face of the epihyal and, to a lesser ex-
tent, the lateral face of the extreme posterior end of the
ceratohyal.

Epihyal. — Large, elongate dorsoventrally; cartilage
filled at its anterior, anterodorsal, and ventral edges; ar-
ticulates through cartilage anteriorly with the cerato-
hyal, while posterodorsally it supports the interhyal by
fibrous tissue. Just anterior to its articulation with the
interhyal, the lateral surface of the epihyal articulates by
fibrous tissue with the posterior end of the interoper-
culum.

Interhyal. —A short column; cartilage filled at both
ends; articulates by fibrous tissue ventrally with the epi-
hyal and dorsally with the concavity on the ventral edge
of the metapter>'goid immediately behind the postero-
dorsal end of the symplectic.

Branchiostegal rays. — Six in number; the last ray
slightly longer than the others; articulates by fibrous tis-
sue to the ceratohyal and, to a lesser extent, the epihyal,
as explained under those bones.

Urohyal. —Reduced to a somewhat irregularly
curved rod without a ventral keel; articulates by fibrous
tissue anterodorsally with the anteromedial edge of the
dorsal hypohyal and the dorsomedial edge of the ventral
hypohyal; articulates by fibrous tissue posterodorsally
with the ventral surface of the first basibranchial.

branchial short, only slightly wider posteriorly than an-
teriorly; displaced forward so that it articulates pos-
teriorly with the second basibranchial and anteriorly
with the hypohyals and urohyal, but with no direct con-
nection with the first hypobranchials. First hypo-
branchial a square block of about the same size as the
second hypobranchial; articulates ventrally with the
lateral surface of the second basibranchial and dorsally
with the first ceratobranchial. First ceratobranchial a
sturdy rod; about equal in length to the other cerato-
branchial elements; somewhat greater in depth ven-
trally than dorsally, this increased depth at the ventral
end increasing from the first to the last ceratobranchial,
the last being widest in the middle region; no ventrally
directed flange present on any of the ceratobranchials;
articulates ventrally with the first hypobranchial and
dorsally with the first epibranchial. First epibranchial a
sturdy column; articulates anterodorsally with the base
of the first or suspensory pharyngobranchial. First
pharyngobranchial a toothless flattened plate; ar-
ticulates dorsally by fibrous tissue to the undersurface of
the extreme anterior end of the prootic suborbital shelf
just lateral to the region where the prootic interdigitates
with the laterally expanded portion of the parasphenoid
in the anterior region of the orbit.

Second arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Second basi-
branchial the largest of the three elements in the series;
articulates anteriorly with the first basibranchial,
laterally with the first hypobranchial, and posteriorly
with the third basibranchial. Second hypobranchial ar-
ticulated ventrally with area of articulation between the
second and third basibranchials and dorsally with the
second ceratobranchial. Second ceratobranchial ar-
ticulated dorsally with the second epibranchial. Second
epibranchial a wide heavy column; articulates dorsally
with the base of the second pharyngobranchial. Second
pharyngobranchial the largest of the four pharyngo-
branchial elements and the first of the tooth bearing
elements. The teeth are very small and difficult to see
and in life may largely be buried in the pharyngeal tis-
sues. The 15 to 20 minute teeth on the second pharyngo-
branchial are much shorter than those on the third and
fourth pharyngobranchials, and are grouped together in
the central region of the ventral face of the element.

Branchial Arches. — AH the elements are cartilage
filled at their edges of articulation with the other
elements in the series, and the articulations are usually
through cartilage and fibrous tissue. The branchial
arches are composed of three basibranchials, three pairs
of hypobranchials, five pairs of ceratobranchials, four
pairs of epibranchials, and four pairs of pharyngo-
branchials. Four gills are present, with a small slit
between the fourth arch and the lower pharyngeal.

First arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. First basi-

Third arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Third
basibranchial a short column, the shortest of the three
basibranchials; articulates anteriorly with the second
basibranchial, anterolaterally with the second
hypobranchials, posterolaterally with the third hypo-
branchials and third ceratobranchials, and posteriorly to
the region between the fourth certobranchials. Third
hypobranchial a long slender rod, directed anteroventral-
ly and with only its extreme posterodorsal end at the
level of the other hypobranchial elements; articulates
posterolaterally with the third ceratobranchial and

posteromedially with the third basibranchial, while its
ventraily directed anterior end articulates by fibrous tis-
sue with the undersurface of the more anterior branchial
elements, although I am unable in the two study
specimens to see exactly where the attachments end.
Third ceratobranchial articulated ventraily with the
posterior end of the third hypobranchial and third
basibranchial, and dorsally with the third epibranchial.
Third epibranchial slender dorsally and somewhat ex-
panded ventraily; articulates dorsally with the third
pharyngobranchial. Third pharyngobranchial smaller
than the second pharyngobranchial but larger than the
fourth; bearing a single row of four or five long slender
teeth with relatively sharp points on the compressed
ventral edge of the bone; articulates ventraily with the
third epibranchial and is held by fibrous tissue more or
less closely to the second and fourth pharyngobranchials.

Fourth arch. — Cerato-, epi-, and pharyngo-
branchial elements present. Fourth ceratobranchial ar-
ticulated ventraily with the cartilaginous area between
the third ceratobranchials and the third basibranchial,
and dorsally with the fourth epibranchial. Fourth epi-
branchial the longest of the epibranchial elements; rod-
like; articulates dorsally with the fourth pharyngo-
branchial. Fourth pharyngobranchial a very small plate
bearing a single row of two or three long slender teeth
much like those of the third pharyngobranchial but
slightly smaller.

Fifth arch. — Ceratobranchial (lower pharyngeal)
element only. Fifth ceratobranchial the widest and
shortest of the ceratobranchial elements; much ex-
panded in the middle region of its length; articulates
ventraily with the cartilaginous region between the bases
of the fourth ceratobranchials; toothless.

PAIRED FIN GIRDLES.

Pectoral Fin.

Posttemporal. —A large long shaft broadly over-
lying the lower half of the lateral surface of the pterotic,
to which it is firmly interdigitated; anterodorsally it also
interdigitates with the sphenotic, while its ventral head
is firmly held by fibrous tissue to the supracleithrum. On
its ventromedial surface the posttemporal helps, along
with the pterotic and, to a far lesser extent, the cleith-
rum, to support by fibrous tissue the expanded lateral
end of the ossified Baudelot's ligament.

Baudelot's ligament. — This ligament is fully os-
sified as a large stout laterally expanded shaft giving
great support to the pectoral girdle. It interdigitates in
the midline with its opposite member and is firmly held
by fibrous tissue at its medial end to the ventral surface
of the parasphenoid just below the level of the origin of
the prootic shelf. Its expanded lateral end is firmly held
by fibrous tissue mainly to the ventromedial surface of

the ventral flange of the pterotic and to the ventral end of
the posttemporal, and, to a much lesser extent, to the ex-
treme dorsal end of the cleithrum. Along its dorsal edge,
just lateral to its middle region, Baudelot's ligament con-
nects by fibrous tissue with the recurved portion of the
prootic shelf, while its ventral edge helps support by a
fibrous tissue sheet the dorsal edge of the medially ex-
panded platelike portion of the cleithrum.

Supracleithrum. —Located slightly obliquely
posterodorsally to anteroventrally in relation to the axis
of the body; relatively short and overlain for a short dis-
tance anterodorsally by the ventral end of the posttem-
poral, to which it is articulated firmly and relatively im-
movably by fibrous tissue. The medial surface of the
supracleithrum broadly overlies and is firmly held by
fibrous tissue to the anterodorsal surface of the
cleithrum.

Cleithrum. — Greatly expanded both laterally and
medially along all the length of its anterior edge, except
for a short distance ventraily, so that a large thin vertical
plate is formed at right angles to the axis of the body;
also greatly expanded posteriorly in the ventral two-
thirds of its length; articulates dorsally by fibrous tissue
on its lateral surface with the broadly overlying supra-
cleithrum and on its medial surface with the anterior end
of the dorsal postcleithrum; along its posterior edge it ar-
ticulates by fibrous tissue and interdigitation with the
anterior edge of the scapula, which it somewhat overlies
ventraily, while more ventraily along its posterior edge it
articulates through cartilage and slight interdigitation
with the coracoid. Ventromedially the cleithrum ar-
ticulates by fibrous tissue with its opposite member,
while the dorsal edge of its medially expanded platelike
portion is held by fibrous tissue to the ventral surface of
Baudelot's ligament.

Postcleithra. — The postcleithra are greatly
expanded into an extremely wide, thin plate whose
lateral surface is closely held by fibrous tissue to the
cuirass. There are distinct dorsal and ventral segments,
the ventral segment much larger than the dorsal. The
anterior end of the dorsal segment articulates by fibrous
tissue with the medial surface of the dorsal end of the
cleithrum, while posteriorly it interdigitates with the
ventral postcleithrum. The ventral edges of both
postcleithra are thicker than the dorsal region, and the
anterior end of the dorsal postcleithrum is thickened and
shaftlike in the area of its articulation with the
cleithrum.

Coracoid. — Somewhat wider ventraily than
dorsally; its posterior edge with a laterally directed
flange throughout its length, the flange ending dorsally
as a posteriorly directed prong below the lowermost ac-
tinost; cartilage filled at its dorsal and anterior edges; an
upraised flange present on its lower lateral surface, run-
ning, with increasing height of the flange, from postero-

dorsally to anteroventrally where it contacts the pos-
teroventral end of the cleithrum; articulates dorsally
through cartilage with the scapula, while posterodor-
sally it articulates through cartilage (and interdigitation
in one of the two study specimens) with the third actinost
and through cartilage and interdigitation with the fourth
actinost. It is probable that in large specimens the third
actinost would also interdigitate with the coracoid.
Anteriorly the coracoid articulates through cartilage and
in some places slight interdigitation with the posterior
edge of the cleithrum.

Scapula. — Completely encloses the scapular
foramen; cartilage filled at its posteroventral and
anterior edges; articulates by fibrous tissue and inter-
digitation anteriorly and anteroventrally with the
cleithrum, while posteroventrally it articulates through
cartilage with the coracoid. Posterodorsally the scapula
articulates through fibrous tissue and interdigitation
with the bases of the first two actinosts and, in one of the
two study specimens, with the anterior end of the base of
the third actinost. Just anterior to its articulation with
the first actinost the scapula bears a projection with a
concave surface for support of the short uppermost pec-
toral fin ray, to which it is held by fibrous tissue.

Actinosts. — Four elements; cartilage filled at their
ventral edges; first actinost the smallest, the others of
slightly increasing size posteriorly in the series; all four
elements held to one another by interdigitation; first and
second actinosts articulated ventrally by interdigitation
with the posterodorsal edge of the scapula; third actinost
articulated ventrally either through cartilage or by inter-
digitation with the dorsal edge of the coracoid; fourth ac-
tinost interdigitated with the posterodorsal edge of the
coracoid. Distally the actinosts support by fibrous tissue
all of the fin rays, except for the first, which is supported
by the scapula, and perhaps the second, which may be
supported in part by the scapula as well as by the first
actinost.

Fin rays. — Twelve to thirteen fin rays in most
specimens, with the first ray only about one-fourth or
one-fifth the length of the second ray and articulated
directly with the scapula instead of with the actinosts, as
are all the other rays with the possible exception of the
second, which articulates more or less in the region of in-
terdigitation between the scapula and first actinost; first
ray with its two halves distinct and of about equal size.
The first two rays unbranched, the others branched. The
first ray has cross-striations at its extreme distal end and
the other rays are more fully cross-striated.

VERTEBRAL COLUMN. —All vertebrae with
biconcave centra, except the last, which ends posteriorly
in the fused hypural plate, and the highly modified first
two vertebrae. However, the third and fourth vertebrae
particularly are much compressed anteroposteriorly and
the biconcaveness of the centra is only slightly indicated.

Abdominal Vertebrae.

First two vertebrae. — In the two adult study
specimens, the first two abdominal vertebrae are
rudimentary and are not only fused to one another and to
the skull but are much smaller than any of the subse-
quent vertebrae. The neural arches of the first two
vertebrae are slender and the arches from each side do
not meet their opposite members in the midline above
the spinal cord. They are distinct from one another in the
dorsal halves of their lengths and are fused to one
another and to the exoccipitals and basioccipital only
basally. The fused centra of the first two vertebrae are so
compressed anteroposteriorly and shortened dorsoven-
trally that they form only a thin plate fused to the dorsal
region of the posterior end of the basioccipital. The
neural arches of both vertebrae bear small neural
foramens. The structure of the first two vertebrae and of
the vertebral column in general has been described by
Tyler (1963a:164-171).

Other abdominal vertebrae. — As used here the term
abdominal vertebrae refers to all of those vertebrae
anterior to the vertebra to which is attached the first anal
fin basal pterygiophore. In the case of ostracioids, with
their highly unusual anal fin basal pterygiophore ar-
rangement, this definition is not applicable. The more
anterior of the anal fin basal pterygiophores of os-
tracioids do not remain in the midvertical plane of the
body, but, rather, are laterally divergent from their
medially placed bases and thus no longer connected with
the ventral surface of the vertebral column. Only the last
three or four basal pterygiophores are invariably in the
midline of the body, and the proximal ends of even these
pterygiophores in ostracioids do not make as close a con-
tact with the vertebrae as in other plectognaths. Under
these circumstances (discussed more fully by Tyler
1963a: 169) it is necessary to arbitrarily state that the first
caudal vertebra is that vertebra to which the proximal
end of the first basal pterygiophore that consistently lies
entirely in the midline of the body is most closely
associated. Although this divides the vertebral column
into abdominal and caudal segments that are not com-
parable to those of other plectognaths, it has consistency
within the superfamily, for all specimens of ostracioids
examined have, by this definition, 9 abdominal
vertebrae, regardless of how many of the anterior
elements are fused together, and either 9 (usually) or 10
caudal vertebrae. In Kentrocapros aculeatus the first
basal pterygiophore of the anal fin that lies entirely in
the midline of the body is supported or most closely as-
.sociated with the sequentially 10th vertebra, which ar-
bitrarily is thus defined as the first caudal vertebra. The
third to ninth abdominal vertebrae have normal, un-
fused, centra, and complete neural arches with well-de-
veloped neural spines. These abdominal vertebrae differ
from the more anterior of the caudal vertebrae only in
that the first few vertebrae following the two rudimen-
tary vertebrae (i.e., the third to about the sixth

vertebrae), have their centra anteroposteriorly com-
pressed, the degree of which decreases posteriorly in the
series, and by the fact that the third to eighth abdominal
vertebrae have transverse processes laterally from the
centra. The posterolaterally directed transverse process
of the third abdominal vertebra arises from the upper
part of the centrum and the lower part of the neural arch.
The processes of the fourth and fifth abdominal
vertebrae are placed successively lower on the centra and
the processes of the sixth to eighth vertebrae are low on
the ventral edge of the centra. None of these abdominal
vertebrae, up to and including the eighth, have haemal
arches or spines. The ninth and arbitrarily designated
last abdominal vertebra has a short haemal arch and
spine differing from those in the caudal series only by its
smaller size. The neural spines of the third to seventh ab-
dominal vertebrae lie anterior to the first basal pterygio-
phore of the dorsal fin. The neural spines of the third,
fourth, and sometimes fifth abdominal vertebrae are
held distally in a narrow concavity along the ventral edge
of the ventral keel of the supraneural element or carina,
while the neural spines of the fifth to seventh vertebrae
simply articulate by fibrous tissue with the ventral edge
of the keel. The neural spines of the eighth and ninth ab-
dominal vertebrae articulate by fibrous tissue between
the basal pterygiophores of the anterior region of the dor-
sal fin. The haemal arch and spine of the ninth abdomi-
nal vertebra is symmetrically placed in the midline
below the centrum and there is no indication here or on
any of the more posterior vertebrae that the haemal
canal lies other than in the midline, in contrast to the
condition in the Ostraciidae. The neural spine and arch
of the seventh abdominal vertebra bears a low ridge
along its length, while a similar ridge is present on the
eighth and ninth abdominal vertebrae, but much more
prominently developed, ending ventrally on the anterior
end of the centrum as a flange.

Caudal Vertebrae. — The caudal vertebrae numbered
nine in two specimens. All of the caudal vertebrae pos-
sess complete neural and haemal arches and spines. The
haemal spine of the first caudal vertebra is better de-
veloped than that of the last abdominal vertebra, but
still shorter than that of the second caudal vertebra,
which is the largest in the series. Posterior to the second
caudal vertebra the haemal spines decrease gradually in
size in the series to the seventh caudal vertebra. The
haemal spine of the eighth caudal vertebra is larger than
that preceding it and it is autogenous to the centrum.
The neural spines decrease in length gradually in the
series from the first to the seventh caudal vertebrae,
while the neural spine of the eighth caudal vertebra is
somewhat larger than that preceding it. The haemal
spines of the first to fourth caudal vertebrae support the
distal ends of the last four basal pterygiophores of the
anal fin, these being the only four which consistently lie
in the midline of the body, rather than being divergent
from it as are the more anterior pterygiophores. The
neural and haemal canals lie in the midline of the body
and are not divergent from it. Each of the caudal verte-

brae, as with the abdominal vertebrae, possesses a neural
foramen low on the neural arch region.

Caudal Skeleton. — The caudal complex consists of a
large rectangular plate, with a rounded expansion in the
middle region of its anterior edge representing the cen-
trum of the last caudal vertebra, and a deep cleft in the
middle of its posterior edge representing the division be-
tween what in more generalized plectognaths such as
triacanthodids would be the second and third hypurals.
However, the hypurals are fully fused to the centrum and
to themselves and no real distinction can be made be-
tween them. The anteroventral region of the rectangular
plate represents the parhypural which is fully fused to
the centrum and hypurals. That this is the region of the
parhypural is indicated by the haemal canal penetrating
the anterior region of the rectangular plate just below the
centrum and exiting at a foramen a short distance pos-
terior to where it first enters the plate. This foramen
represents the region of fusion between the parhypural
and the hypurals. The anterodorsal region of the rec-
tangular plate represents the completely fused epural,
for the neural canal penetrates the plate, again just
above the centrum, and courses through it to exit about
midway along the dorsal edge of the plate.

Caudal fin rays. — Eleven fin rays are present in
both specimens and generally throughout the Aracani-
dae; the uppermost ray and the lowermost ray un-
branched, the intervening nine rays branched; all rays
with cross-striations. The fin is vaguely divided into an
upper and lower lobe, the upper lobe containing five rays
articulated to the fused hypural plate above the indenta-
tion in its posterior edge and the lower six rays articu-
lated to the lower half of the plate. The bifid bases of the
rays do not overlap the hypural plate, to which they are
articulated through fibrous tissue. The branched fin rays
become increasingly branched toward the center of the
fin, where the rays are branched in up to incomplete
quadruple dichotomies.

DORSAL AND ANAL FINS.

Dorsal Fin.

Fin rays and pterygiophores. —Eleven fin rays are
present in both study specimens; the first ray un-
branched, the other rays branched in single to triple
dichotomies. Distal pterygiophores are either absent or
unossified. The bifid bases of the fin rays are supported
through fibrous tissue by 9 or 10 basal pterygiophores.
The specimen with 9 basal pterygiophores (the il-
lustrated specimen) has the last pterygiophore a com-
pound element obviously representing the fusion product
of 2 pterygiophores, for in the other study specimen there
are 2 separate basal pterygiophores in this region for a
total of 10. With the exception of the first and last
pterygiophores the basal pterygiophores are more or less
slender rods, cartilage filled at both ends, and of slightly

decreasing length posteriorly in the series. At their dorsal
ends the pterygiophores are slightly expanded into ar-
ticular knobs. The pterygiophores articulate with one
another and with the neural spines supporting them by
fibrous tissue with little or no interdigitation between the
pterygiophores. Just below their expanded articular faces
the pterygiophores are slightly constricted so that there
is a definite gap between their apposed surfaces. The first
pterygiophore is unlike the others only in that it is much
larger and more expanded anteriorly and posteriorly in-
to a thin plate filling the space between the distal regions
of the neural spines of the seventh and eighth abdominal
vertebrae. Anterodorsally the first basal pterygiophore
articulates by fibrous tissue and interdigitation with the
posterior end of the supraneural element or carina. The
supraneural is a long strut reaching nearly to the pos-
terior end of the skull. Its dorsal surface is laterally ex-
panded and its bears a deep ventral keel. Anteriorly the
ventral edge of the keel is concave to enclose the distal
tips of the neural spines of the third, fourth, and, in one
specimen, fifth abdominal vertebrae, articulating with
them by fibrous tissue. The distal tips of the neural
spines of the sixth and seventh vertebrae end at the ven-
tral edge of the keel and do not penetrate it. The dorsal
surface of the supraneural lies just below the cuirass. The
basal pterygiophores of the dorsal fin articulate between
the neural spines of the seventh abdominal to third cau-
dal vertebrae.

Fin rays and pterygiophores. — Eleven fin rays are
present; the first ray unbranched, the others branched in
up to incomplete triple dichotomies. Distal pterygio-
phores are either absent or unossified. The bifid bases of
the fin rays articulate by fibrous tissue to nine basal
pterygiophores. These pterygiophores are basically
similar to those of the dorsal fin except that they tend to
be longer, more anteroposteriorly expanded and much
more firmly held to one another, usually by extensive in-
terdigitation. However, while all of the dorsal fin basal
pterygiophores lie in the midvertical plane of the body,
only the last four anal fin basal pterygiophores con-
sistently lie entirely in this medial plane, the five
pterygiophores anterior to them diverging to the right
and to the left from their ventral ends, which are in the
midvertical plane. The sixth to ninth basal pterygio-
phores have their dorsal ends held by fibrous tissue at
some distance from the ventral ends of the haemal spines
of the first to fourth caudal vertebrae. The second to fifth
pterygiophores have their slightly expanded ventral ar-
ticular ends in the midvertical plane, but their long more
or less rodlike anterodorsally directed portions lie to the
right or to the left of the midline in the anterior portion of
the large muscle mass connected to the anal fin. In the
two study specimens three of these pterygiophores
diverge to the right and two to the left. The first basal
pterygiophore is expanded just above its knoblike distal
end into a thin plate which lies only slightly to one side of
the midline. It has the largest keellike expansion of any

of the pterygiophores. It is the second to the fifth
pterygiophores that diverge widely from the midline, two
to the left and two to the right. The second to fifth
pterygiophores remain rodlike throughout their lengths,
while the sixth to ninth pterygiophores, that lie in the
midline of the body throughout their lengths, are
variously expanded into thin plates for parts of their
lengths. The knoblike distal ends of these more posterior
basal pterygiophores are closely apposed to one another
and articulate by fibrous tissue but they are not fused or
sutured. The shafts of the first to fifth pterygiophores are
tightly held to one another in the midline before they di-
verge to the right or left, and the surfaces of apposition
seem to be at least extensively interdigitated, if not, in
some cases, perhaps even fused. The concave region on
the posteroventral edge of the last basal pterygiophore
rests against the edge of the carapace, just as the similar-
ly indented region on the last dorsal basal pterygiophore
supports the carapace in that region.

Anatomical diversity. — Very little anatomical
diversity is present in this small family of deepwater box
fishes comprising about 10 species from the Indo-Pacific
(Hawaii to South Africa), but mainly from Australia.
They have been relatively poorly collected and the
genera as presently recognized (McCulloch and Waite
1915; Fraser-Brunner 1935b, 1941c) seem extremely fine-
ly split, the distinctions based exclusively on carapace
characteristics, many of which are rather trivial (num-
ber of spiny processes and degree of development of iso-
lated caudal peduncular scale plates). Three of the
genera {Kentrocapros,^ Capropygia, Caprichthys) are
monotypic, while Aracana and Strophiurichthys each
have two species, and Anoplocapros two, perhaps three
(if grayi is valid).

The carapace in all genera except Kentrocapros and
Aracana has six major ridges or angles: single dorsal
and ventral crests and two paired ridges along the side of
the body, dorsolateral and ventrolateral in position. In
Kentrocapros and Aracana there is no dorsal crest, the
back being relatively flat or gently convex rather than
high crested, so that there is a total of five ridges. In the
single fossil species of the family, Proaracana dubia
(Blainville 1818) from the Eocene of Monte Bolca, Italy,
the dorsal crest is well -developed and terminates at its
greatest height in a large spiny process similar to that of
the ventral crest, spines in these positions being unique
in the family.

In addition to the major ridges, a mediolateral ridge or
slightly convex region may be present about midway
between the dorsolateral and ventrolateral ridges, but it
is usually less distinct than the other ridges.

The carapace extends further posteriorly and the cau-
dal fin is less strongly developed in Caprichthys and
Capropygia than in the other genera.

Posterior to the carapace there are always isolated

Aracanostracion Smith (1949b) is tentatively considered a synonym
of Kentrocapros, until the single known specimen, the type of rosapinto,
can be examined.

Figure \3S.—Kentrocapros aculeatus: lateral
of head, 90.0 mm SL, Japan.

Figure \37 .—Kentrocapros aculeatug:

ventral (left) and dorsal (right) views of

skull, 90.7 mm SL, Japan.

ring around the rear of the peduncle, and another set of
plates more anteriorly just behind the bases of the dorsal
and anal fins form broad saddles above and below the
peduncle which do not meet one another mediolaterally.
In Kentrocapros, Aracana, and Strophiurichthys the two
anterior saddles are similar to those of Capropygia and
Anoplocapros, except that the individual scales making
up the anterior saddle in Kentrocapros and Aracana are
less well-consolidated into a single unit than they are in
Strophiurichthys, Capropygia, and Anoplocapros.
Posteriorly, instead of a complete ring around the pedun-
cle as in Capropygia and Anoplocapros, there are similar
saddles over the dorsal and ventral regions of the pe-
duncle that fail to meet mediolaterally, although in large
specimens the edges of these two posterior saddles are al-
most in contact. In the Eocene Proaracana the dorsal and

Figure 139.— Kentrocapros aculeatus:

dorsal view of branchial arches

(extended on lower side); lateral

view of hyoid arch and urohyal;

90.7 mm SL, Japan.

lU^i

1111=1

scale plates on the caudal peduncle. These plates are
least developed in Caprichthys, in which there are only
one to three relatively small scutes dorsally and ventrally
on the anterior region of the caudal peduncle just behind
the bases of the dorsal and anal fins. The caudal plates
are best developed in Capropygia and Anoplocapros, in
which a posterior set of plates forms a completely closed

ventral regions of the caudal peduncle are covered with
numerous semi-isolated small scale plates that do not
form compact saddles, and the mediolateral region is
scaleless.

The center of each carapace scale plate in most spe-
cies usually bears a spinule larger than those of the rest of
the plate, while large prominent spiny processes are vari-
ously developed. In Kentrocapros there is a small
supraorbital spine, a larger one on the dorsolateral ridge
and several on the mediolateral and ventrolateral ridges.
Aracana is similar to Kentrocapros, but with the
supraorbital spine larger and with two or more spines on
the dorsolateral ridge. In Capropygia and Caprichthys
there is a single large spine on the dorsolateral and ven-
trolateral ridges, with Caprichthys additionally having a
small supraorbital spine in the young which is resorbed
in the adult. In Anoplocapros and Strophiurichthys
prominent spines of the magnitude found in the other
genera are absent, although in large adults of one of the
species of the latter genus, S. robustus, there is a small
supraoccipital spine and the central spinule of many of
the individual scale plates of the carapace is relatively
much larger than the others, forming spines interme-
diate in size between those on the ridges of such genera as
Kentrocapros, Aracana, Capropygia, and Caprichthys,
and those of the center of the plates in all other species.
Young specimens of all species probably have the

basibranchiaU
I rT~_3 — hypobranchials

pharyngobranchials

mterhyal "^^

^1 ,/^ ^ dorsal hypohyal

epihyal U ^^

anchiostegal rays

spinules of the individual scale plates, including the
variously enlarged central spinule, better developed than
in adults, i.e., the plates in the young are more highly
sculptured and ornamented than in adults.

Aracanids generally have between 10 and 13 dorsal and
anal fin rays, with Capropygia and Caprichthys tending
to have closer to 13 dorsal and 12 anal and the other
genera tending to have closer to 10 or 11 in these fins, the
anal with the same number, or one less, than the dorsal.
All of the species tend to have either 10 or 11 pectoral fin
rays in addition to the small uppermost ray.

There are usually 8 to 10 teeth in the upper jaw and 8
in the lower. The numbers of teeth found in the jaws of
the specimens examined (including alcohol preserved
material) are as follows, with the total in the upper jaw
followed by that of the lower jaw: Strophiurichthys
robustus, 9 and 8 in four specimens, 8 and 8 in one, 10
and 7 in one; S. inermis, 10 and 8 in one; Kentrocapros
aculeatus, 8 and 8 in one, 7 and 8 in one; Aracana aurita,
8 and 8 in one; A. ornata, 8 and 8 in two; Caprichthys
gymnura, 10 and 8 in two; Capropygia unistriata, 10
and 8 in two.

In most osteological characteristics the aracanids are
remarkably similar, closely following the same general
plan in genus after genus.

In Capropygia and Caprichthys, with relatively high
numbers of dorsal and anal fin rays, there are usually 12
dorsal fin basal pterygiophores and 11 anal fin basal
pterygiophores, while in the other genera these range in
number from 8 to 10. There are always seven predorsal
vertebrae, the more numerous rays and basal pterygio-
phores in Capropygia and Caprichthys being accommo-
dated toward the rear of the fin; e.g., the last pterygio-
phore of the dorsal fin in these two genera being placed
between the neural spines of the 13th and 14th verte-
brae, while in the other genera it is placed between the
12th and 13th.

The branchiostegal rays are usually 2-1-4, except that
the single specimen of Caprichthys gymnura examined
had 2 -I- 4 on one side and 1 -(- 4 on the other, and that of
the two species of Aracana examined, the single speci-
men of A. aurita had 1 -(- 4 on both sides, while of the two
specimens of A. ornata examined one had 1 -I- 4 on both
sides and the other had 1 -(- 4 on one side and 2 -I- 3 on the
other. It would appear that in Aracana the branchios-
tegals are usually reduced to five, while the normal sit-
uation in Caprichthys awaits the examination of further
specimens, as it does also in Anoplocapros, no species
of which have been examined internally for this work.

In Kentrocapros, Aracana, and Strophiurichthys the
second pharyngobranchial bears minute teeth, while in
Caprichthys and Capropygia it is entirely toothless. The
number of large teeth respectively on the third and
fourth pharyngobranchials are as follows for single speci-
mens of each of the species closely examined for this fea-
ture: Strophiurichthys robustus 6 and 6 (plus a few
smaller teeth on the third element in addition to the larg-
er ones); Aracana ornata 6 and 3; Aracana aurita 4 and 2;
Kentrocapros aculeatus 4 and 3; Caprichthys gymnura 3
and 2; Capropygia unistriata 3 and 2. Thus, Strophi-
urichthys has the best developed pharyngobranchial
dentition and Caprichthys and Capropygia the least,
with Aracana and Kentrocapros more or less interme-
diate.

The ventral flange of the carina or supraneural is
deeper for a greater proportion of its length in the high
crested Strophiurichthys, Capropygia, and Caprichthys
than it is in the relatively flat-backed Kentrocapros and
Aracana.

In Kentrocapros the autogenous haemal spine of the
penultimate vertebra is smaller than in the other genera.
In Kentrocapros, Aracana, and Strophiurichthys the
haemal canal enters the fused centrum-epural-hypural-
parhypural plate at the end of the vertebral column and
exits shortly behind at a prominent foramen low on the
plate, the foramen marking the region of fusion between
what would be the parhypural and lowermost hypural in
more generalized forms. In Capropygia and Caprichthys
this foramen is either absent or minute (Tyler 1970b: 18)
and the haemal canal which pierces the front edge of the
plate either stops within the plate or has only a minute
foramen that could not be detected in either of the single
specimens examined of the two species involved.

In one of the species of Aracana, A. ornata, the snout
becomes convex and laterally expanded in adults, this

*

200

^v^x^V^UUUuy

Figure 143 —Lateral views of

heads of tvpical aracanids
left, Strophturichthye robiutiu,
150 mm SL, Australia; right, Amcana
.3 mm SL, Tasmania.

from a great enlargement of the anterior prolongation of
the frontals from the level of the eye forward. This swell-
ing of the snout is more prominent in males {ornata) than
in females (the synonymous flavigaster) and occurs in
the family only in this species, even the closely related A.
aurita showing no signs in large specimens of any snout
or other enlargements.

The structure of the parasphenoid presents variation
of special interest. In Kentrocapros and Strophiurichthys
the ventral edge of the vertical platelike portion of the
parasphenoid anterior to its articulation with the
anteromedial edges of the prootic shelves is relatively un-
expended laterally throughout its length, being only
slightly wider anteriorly where it sutures with the vomer
than more posteriorly. In Aracana the ventral edge of the

Figure 144.— Ventral views of the two types
of configuration of the ventral surface of the
vertically expanded anterior portion of the
parasphenoid: A, Caprichthys gymnura,
74.1 mm SL, Australia, representative of the

shape in Caprichthya and Capropygia;

B, Aracana aurita, 87.3 mm SL, Tasmania,

representative of the shape in the

two species of Aracana.

203

,.r

If^

parasphenoid is relatively as narrow as in Kentrocapros
and Strophiurichthys in the anterior half of its length
anterior to the prootic shelves, but it is much expanded
laterally just in front of the shelves, forming a partial
hard palate over the rear of the oral cavity. In
Caprichthys and Capropygia the ventral edge of the
parasphenoid is relatively as narrow as in Kentrocapros
and Strophiurichthys throughout most of its length
{interior to the prootic shelves, but it becomes laterally
expanded anteriorly behind the vomer, forming a partial
hard palate over the front of the oral cavity. The possible
phylogenetic implications of these differences and of the
others mentioned above stfe discussed in the section on
generic relationships that follows.

Generic relationships. — Probably having been deriv-
ed from the same line of fishes that connects the tria-
canthids and balistids, the primitive aracanids also
probably had a laterally compressed body whose cross-
sectional outline would be a dorsoventrally elongate oval.
The body was probably fully covered with enlarged, per-
haps slightly overlapping, scale plates. While the
balistids retained a more flexible body and flexible scale
covering, the aracanids specialized in a thicker and less

Figure H5.—Proaracana dubia: lateral view

of entire specimen, composite based on three

specimens, including the holotype, 31.4-54.5 mm SL,

all specimens from the Eocene of Monte

Bolca, Italy (Tyler 1973a:fig. 7).

flexible scale covering with a concomitant reduction in
swimming ability but greatly increased protection from
the exoskeleton. The relatively complete covering of
somewhat flexible scales would have become con-
solidated into a relatively inflexible carapace of hex-
agonal plates with marginal interdigitations, except that
many of the scale plates on the check remained small
and free from one another to allow for respiratory move-
ments and that the scales around and behind the dorsal
and anal fins remained similarly small and free from one
another to allow for lateral flexion of the caudal peduncle
when the caudal fin was used in rapid swimming. The
tendency would be for the scales that originally entirely
covered the caudal peduncle to become reduced in size
and in coverage of the peduncle, within the limits of also
providing some protection from the bites of predators.
The Eocene Proaracana dubia provides a firm example
of an early aracanid, the carapace over most of the body
being fully consolidated and the caudal peduncle bearing

204

a large number of small, at least semi-isolated, scale plates
in a long patch both dorsally and ventrally but not
mediolaterally. The carapace is high crested both dor-
sally and ventrally so that in cross-sectional outline it
would be in the form of a laterally compressed oval. The
carapace does not extend posteriorly beyond the level of
the rear of the bases of the dorsal and anal fins. This can
probably be taken as the generalized aracanid carapace
condition, and the one on which the Recent species have
modified mainly only in a greater specialization of the
caudal peduncular scale plates. Unfortunately, the in-
ternal anatomy of P. dubia is mostly unknown, being
largely hidden from view by the carapace in the few
specimens presently available. However, the more pos-
terior portion of the vertebral column is relatively easily
seen in some of the specimens, and the indications are.
that the caudal plate is as fully fused together as in Re-
cent species and that the anterior anal fin basal pteryg-
iophores diverged from the midline. The teeth were dis-
crete and low in number, similar to those of Recent
species. Proaracana has only one known specialized
feature relative to the Recent species of the family, this
being that it has only 10 caudal fin rays rather than the
11 found in all Recent species.

In triacanthids and balistids the ventral edge of the
parasphenoid anterior to the orbit is narrow, and the
pharyngobranchials consist of a toothless suspensory ele-
ment followed by three (triacanthids) or two (balistids)
elements bearing large teeth. One would expect that the
more generalized of the aracanids would have been
derived from these conditions. Strophiurichthys and
Kentrocapros have the ventral edge of the parasphenoid
the least expanded in the family, and Strophiurichthys
has the best developed dentition on the two pharyngo-
branchials (third and fourth) bearing large teeth, as well
as minute teeth on the second pharyngobranchial. Ken-
trocapros has minute teeth on the second pharyngo-
branchial but fewer large teeth on the third and fourth
pharyngobranchials than in Strophiurichthys.
Strophiurichthys has a high dorsal crest, similar in mag-
nitude to that of Proaracana if the high dorsal spiny
process in the latter is ignored.

In short, of the genera examined, Strophiurichthys ap-
pears to be the most generalized of the Recent forms. A
form like it probably gave rise on the one hand to a line
leading to Kentrocapros and hence Aracana and on the
other hand to a line leading to Anoplocapros (not studied
internally) and Capropygia-Caprichthys, the latter two
monotypic genera being so similar that they scarcely
merit distinction. In the line leading to Kentrocapros the
high crested dorsal ridge was lost, with the back becom-
ing flattened or only very gently convex, while the ven-
tral carapace ridge was also reduced in size. The pharyn-
geal dentition was slightly decreased and the degree of
development of large spiny processes increased. Aracana
is obviously closely related to Kentrocapros, sharing with
it a flattened back and a more or less comparable
pharyngeal dentition, and differing from it externally
only by a continuing slight increase in the number of
large spiny carapace processes and the probable reten-

tion from a slightly more generalized ancestor than the
Recent Kentrocapros of a deeper ventral carapace ridge.
Internally Aracana is further specialized by the ap-
parently usual loss of one of the two anterior branchios-
tegal rays, and, more importantly, by the development of
a wide lateral expansion of the ventral edge of the para-
sphenoid just in front of its articulation with the antero-
medial edges of the prootic shelves. In one respect Ken-
trocapros and Aracana have remained slightly more
generalized than the ancestral Strophiurichthys-like
form in that the individual scale plates of the two an-
terior caudal peduncular saddles are less fully con-
solidated into functionally single pieces.

Probably a close derivative of the ancestral Strophi-
urichthys-like form is Anoplocapros, whose internal
structure has not been studied but which is very similar
to Strophiurichthys externally in the features of its
relatively spineless high crested carapace. Anoplocapros
differs from Strophiurichthys mainly in having a com-
plete ring of scales around the posterior region of the
caudal peduncle instead of two saddles. However, this is
surely a minor distinction, and McCulloch and Waite
(1915:479) pointed out that the ring "may be incomplete
in the young." In short, the two only narrowly separated
posterior peduncular saddles as found in Strophiurich-
thys have simply met mediolaterally in adult Anoplo-
capros. The Strophiurichthys-Anoplocapros line
probably also gave rise to Capropygia and Caprichthys,
both of which are specialized by the loss of even minute
teeth on the second pharyngobranchial and the reduc-
tion in number of large teeth on the third and fourth
pharyngobranchials, while the carapace extends back
further posteriorly than in any of the other genera.

The most distinctive internal feature linking Capro-
pygia and Caprichthys is the development of a moderate
lateral expansion of the ventral edge of the para-
sphenoid just behind its articulation with the vomer, in
contrast to the more posteriorly placed lateral expansion

Aracana
Kentocapros

Caprichthys

Capropygia

Anoplocapros

Strophiurichthys

Figure 146.— Hypothesized phylogenetic
relationshipg of the genera of Aracanidae.

205

just in front of the prootics found in Aracana. Capro-
pygia and Caprichthys differ almost exclusively on the
basis of the caudal peduncular scales, although Caprich-
thys sometimes develops a supraorbital spine whereas
Capropygia never does. In both Caprichthys and Capro-
pygia the small scale plates that make up what are the
anterior peduncular saddles in other genera are less con-
solidated and do not form distinctive saddles that are
functionally single pieces. This can be considered a
reduction from the condition seen in Strophiurichthys
and Anoplocapros. The posterior series of peduncular
plates forms a complete ring around the peduncle in
Capropygia, just as in Anoplocapros, while the posterior
plates are completely lost in Caprichthys.

It seems evident that Capropygia is a derivative of an
Anoplocapros -like ancestral stock specialized by the
reduction of the anterior peduncular plates, a posterior
elongation of the carapace, a reduction in the size of the
caudal fin, the development of prominent spiny
processes on the carapace, the reduction in the pharyn-
geal dentition and the development of a partial hard
palate over the front of the oral cavity. Caprichthys is
simply a closely related derivitive of Capropygia in which
the anterior peduncular plates are further reduced in size
and the posterior plates completely lost, and with a ten-
dency for one more carapace spine to develop.

Capropygia and Caprichthys have a slightly higher
number of dorsal and anal fin rays than do the other
genera, and a relatively large number of fin rays might be
expected to be a primitive character in a group derived
from a posttriacanthid and prebalistid ancestral stock,
both of which families have long-based and many-rayed
dorsal and anal fins. However, Proaracana has only 10 or
11 dorsal and anal fin rays, like all of the Recent genera
except Capropygia and Caprichthys. I suspect that the
number of dorsal and anal fin rays in Capropygia and
Caprichthys is a de novo increase associated with the
greater dependence on these two fins for locomotion
demanded by the longer posterior extension of the
carapace and concomitant reduction in lateral flexibility
of the caudal peduncle, at the end of which is a caudal fin
less well developed than ii; all of the other genera.

Relationship to the Ostraciidae. — The deepwater
aracanids are so obviously closely related and ancestral
to the shallow water, more speciose (about 20 species),
ostraciids, that it is needless to belabor the point. The
changes from the aracanid to ostraciid levels of or-
ganization mostly are associated with the increased
length of the carapace and its more complete enclosure
posterior to the dorsal and anal fins of the more elongate
body concomitant with a reduction in the size and
strength of the orbital and postorbital regions of the skull
and an extensively sutured, inflexible, generally weaker
vertebral column within the carapace, with four or five
rather than only two vertebrae involved in the variously
fused and sutured complex at the rear of the skull, and a
general reduction in the size of most of the neural and
haemal arches. The caudal fin supporting apparatus in

the more generalized ostraciids is similar to that of
aracanids, except that there are only 10 fin rays rather
than 11, while the apparatus becomes highly specialized
in most other ostraciids.

Differences in the type of food available in deep versus
shallow water that the aracanids and ostraciids are
specialized in feeding on may account for the slight dif-
ferences in dentition between the two families, ostraciids
having a much reduced pharyngobranchial dentition but
usually a few more teeth in the jaws, especially in the up-
per jaw, relative to aracanids. The differences in the
structure of the ethmoid supporting the upper jaw and of
the completeness of the parasphenoid forming a hard
palate over the roof of the oral cavity between the two
families are also undoubtedly related to differences in
diet, the shallow-water ostraciids perhaps making
greater use of plant material in their omnivorous diet
than do the deepwater aracanids.

Other evolutionary trends from the aracanids to the os-
traciids are the reduction in the size of the postcleithral
apparatus, perhaps associated with a lesser need for but-
tressing the side of the carapace in the more completely
and solidly enclosed ostraciid body, the reduction of the
size of the uppermost pectoral fin ray, the development
variously by the basal pterygiophores and neural and
haemal spines of a firmer system of support for the
carapace around the dorsal and anal fins, the diversion of
the haemal canal away from the midline and its close ap-
position to the ventral surface of the centra, perhaps a
space saving device in a crowded abdominal region from
which even muscles are largely excluded laterally, the
loss of the ventral keel of the carapace and the reduction
in the number of isolated scale plates on the caudal
peduncle posterior to the carapace.

With the aracanids obviously ancestral to the os-
traciids, a critical question is whether the latter can be
shown to have evolved from one or the other of the two
main evolutionary lines postulated here as diverging
from a Strophiurichthys-\ike ancestral stock, or whether
it was from such a generalized form as Strophiurichthys
or Proaracana. This is discussed under the section on
generic relationships of the Ostraciidae, with the con-
clusion being that the ostraciids, and more particularly
the lactophrysins, the more generalized of the two sub-
families, probably evolved from one of the two lines
possessing a laterally expanded parasphenoid shelf over
the roof of the oral cavity, but whether from the line with
the shelf placed posteriorly (Aracana) or anteriorly
(Capropygia and Caprichthys) cannot be determined.

It is of interest that in both the triacanthoids, in which
the deepwater Triacanthodidae are clearly ancestral to
the derived shallow-water Triacanthidae, and in the os-
tracioids, in which the deepwater Aracanidae are clearly
ancestral to the derived shallow-water Ostraciidae, all of
the families occur together in the same strata of the up-
per portion of the lower Eocene of Monte Bolca, Italy.
The Triacanthodidae are represented there by
Spinacanthus cuneiformis, Protobalistum imperiale, and
Eoplectus bloti, all so different from the Recent species

of the family that they are placed in different sub-
families than the Recent species, while the
Triacanthidae are represented by Protacanthodes om-
bonii, a somewhat intermediate form between the
Triacanthodidae and Triacanthidae and different
enough from the Recent triacanthids and those known
from the Oligocene and Miocene to be placed in a
separate subfamily.

By contrast, of what little is known of them, the single
species of Eocene aracanid and ostraciid are relatively
modem enough in general countenance to be easily ac-
commodated in the same higher categories as the Recent
species. The diversifications of the families apparently
was rapid and in both cases with the derived families
specialized for a shallow-water existence coextant with
the ancestral families successful continuance in deeper
waters.

Relationship to the Balistidae. — The derivation of
the ostracioids presents a slight enigma. Both families of
ostracioids appeared in the Eocene with genera not
markedly different from Recent genera, and paleon-
tology offers no clue as to what plectognath group the os-
tracioids are most closely related. As pointed out in the
historical section of the Introduction, the ostracioids
have been considered variously as a third major sub-
ordinal group of plectognaths, or as members of both the
Sclerodermi and Gymnodontes. Regardless of how they
have been classified, their relationships to the
triacanthoids and balistoids, with which they are usually
placed, have never been analyzed much beyond the sim-
ple statement that ostracioids have individual discrete
teeth protruding from the jaws and hence show an af-
finity with the scleroderms and that since they lack the
spiny dorsal and pelvic fins, they are probably related to
the balistoids, which have the spiny dorsal and pelvic
fins at least reduced in size or in number of elements in
relation to the triacanthoids.

This is not to belittle the clue that the presence of dis-
crete protruding relatively normal teeth and lack of spiny
dorsal and pelvic fins in ostracioids affords, for I agree
that they are important indications of their relationship
with the balistoid scleroderms. Any such clues are impor-
tant, for ostracioids are so highly modified for life within
a shell that many of their features are unique.

The divergence of many of the anal fin basal pterygio-
phores away from the midline is not found in fishes other
than ostracioids, while the fusion of several abdominal
vertebrae to the skull and the heavy ossification of
Baudelot's ligament are at least highly unusual features
among fishes. Other features of the ostracioids that are
less distinctive, but at least unique among the plectog-
naths, are the heavy lateral expansions of the ventral
edge of the parasphenoid forming a partial hard palate,
the articulation of the preoperculum with a groove on the
lateral surface of the hyomandibular, the complete fu-
sion of all epural, hypural, and parhypural elements into
a single largo plate, the displacement in ostraciids of the
haemal canal to one side or the other of the midline, and

the great expansion in aracanids of the postcleithrum.
Beset with a multitude of such specializations one is for-
tunate to find even a few structures in ostracioids that
give evidence concerning the origin of the group.

The forward extension of the prootics as a shelf under
the orbit is unusual among fishes, and in the plectog-
naths it is found only in the balistoids and ostracioids.
The interoperculum is a short and simple rod not ex-
tending posteriorly past the level of the epihyal only in
balistoids and ostracioids among the plectognaths, and
the operculum and suboperculum are about equally
reduced in the two groups. Only in balistoids and os-
tracioids is the hyomandibular supported dorsally by the
prootic and pterotic alone, without contacting the
sphenotic. The palatine in balistoids and ostracioids is
reduced in size in comparison to other plectognaths, al-
though it is articulated differently in these two super-
families. The shape and size of the inflexibly articulated
premaxillary and maxillary are very similar in balistoids
and ostracioids, as is the rotation of the upper jaw around
a laterally expanded and buttressed ethmoid-vomerine
region. The shape of the teeth of Recent ostracioids is
somewhat intermediate between that of triacanthodids
and that of triacanthids and balistids, but there is an in-
dication that the teeth of the Eocene ostracioids were
more balistidlike than at present (see description of
Eolactoria).

The ostracioids do not show any such similarities with
any of the gymnodonts, and it is obvious from the above
that they have their closest anatomical affinity with the
balistoids. Since the ostracioids have completely lost the
spiny dorsal and pelvic fins, and the pelvis, one must
consider the possibility that they are, among the
balistoids, more closely related to the derivative mona-
canthids than to the ancestral balistids, for it is among
the monacanthids that the reduction in size and number
of dorsal fin spines (to the presence of a single and
sometimes weakly developed spine in a few genera) and
in the size of the pelvic fin (completely lost in many
genera) and pelvis (to a relatively slender shaft in a few
genera) has reached its extreme for the superfamily.
However, the monacanthids possess a number of
specialized features in comparison to balistids that show,
as discussed below, that the ostracioids are more closely
related to balistids than to monacanthids.

The T-shaped palatine of balistids, with the foot of the
T articulated against the ectopterygoid and the crossbar
articulated between the maxillary and ethmoid-vomer-
ine region, becomes a simple rod in monacanthids repre-
senting only the crossbar of the balistid palatine, and it is
usually well removed from the ectopterygoid, even though
connected to it by a strong ligament. The columnar
palatine of ostracioids, with its ventral end sutured to the
ectopterygoid and its dorsal end firmly held to the eth-
moid-vomerine (especially to the latter) region is dif-
ficult to derive from the highly specialized and greatly
reduced in size monacanthid condition but easy to derive
from the balistid condition. The ostracioid palatine
represents basically the shaftlike foot of the T-shaped

balistid palatine, the foot now sutured to the ectoptery-
goid (and mesopterygoid) rather than simply ar-
ticulated closely to it through a short ligament, with the
top of the ostracioid palatine probably representing a
combination of the top part of the shaft and the pos-
terior part of the crossbar held by ligament to the eth-
moid-vomerine (especially the former) region in
balistids.

The specialized nibbling incisors of monacanthids are
far less like the dentition of ostracioids than is that of
balistids. Balistids have four teeth in an outer series in
each half of the upper and lower jaws, while the derived
monacanthids have reduced the number to three in the
upper jaw and to three or, in some cases, only two in the
lower jaw. In ostracioids there are usually four or five
teeth in each half of both the upper and lower jaws. It is
unlikely that the reduced in number and relatively broad
incisors of monacanthids could be ancestral to the more
elongate and basically conical teeth in greater numbers
as found in ostracioids.

While the number of teeth in the outer series in balis-
tids is similar to that in the single series in ostracioids,
their shapes, at least at present, are not. The teeth in
balistids are somewhat wider, thicker, and more notched
than in Recent ostracioids. However, the teeth of the two
species of Eocene ostracioids probably were wider,
thicker, and more notched than in the Recent species,
approximating the balistid condition. It is suggested here
simply that the Eocene balistids (for which there are
as yet no fossils from that period) and ostracioids could
easily have had rather similar numbers, sizes, and shapes
of teeth, those of balistids somewhat smaller and less
notched than at present and those of ostracioids some-
what larger and more notched.

There are marked similarities between the balistid and
ostracioid parasphenoid, ethmoid, and prefrontal that do
not exist in the monacanthids. In balistids and os-
tracioids the prefrontal is relatively well developed and
extends ventrally to articulate with a thickened region of
the parasphenoid, while in monacanthids the prefrontal
is greatly reduced in size and is far removed from any
contact with the parasphenoid. In balistids and os-
tracioids the ethmoid has only a relatively shallow ven-
tral keel, if present at all, while in monacanthids the ven-
tral keel is always extremely well developed. In balistids
and ostracioids the parasphenoid is expanded dorsally
into a thick plate in front of the orbit, but no such ex-
pansion is present in monacanthids. The large, thick,
usually rhomboidal scale plates of balistids are far closer
to the larger, even thicker, usually hexagonal scale plates
of ostracioids than are the small, thin, usually more or
less rounded to rectilinear scale plates of monacanthids,
and in some balistids the scale plates are hexagonal.

In all ostracioids there are 18 vertebrae, with the ex-
ception of one specialized genus with a secondary in-
crease to 19, and all balistids normally have 18 vertebrae
also, while monacanthids always have 19 or more verte-
brae. Balistids and ostracioids have mostly branched
rays in the soft dorsal, anal, and pectoral fins, while these

rays are all unbranched in monacanthids. A supraneural

is present in both balistids and ostracioids, although it is
possible that their origins are different, while a supra-
neural is absent in monacanthids. The lesser reduction in
number of pharyngobranchial elements in balistids ver-
sus monacanthids is more similar to the condition in
most ostracioids. In balistids and ostracioids the para-
sphenoid is only slightly, if at all, expanded laterally just
behind the orbit, while in monacanthids it is moderately
to greatly expanded there, except in one highly specializ-
ed genus. In balistids and ostracioids the postcleithrum
is sometimes composed of two pieces, while in mona-
canthids it is always a single piece.

By contrast, there are only a few ways in which mona-
canthids are more similar to ostracioids than £U"e
balistids. In monacanthids and ostracioids the medial
edges of the pterotics on the ventral surface of the skull
are widely separated, while they are only narrowly
separated in balistids. However, balistids and mona-
canthids are similar in that the separation is by the para-
sphenoid and basioccipital, while in ostracioids it is by
the prootic and exoccipital, a rather different arrange-
ment. The posttemporal of monacanthids and os-
tracioids is more superficially held to the skull than is
that of balistids, which is always placed in a deep groove
on the skull. However, the posttemporal in ostracioids is
usually much larger and more extensively sutured to the
pterotic than is that of monacanthids, and the slight
resemblance between the posttemporal of monacanthids
and ostracioids is probably coincidental and of no phylo-
genetic significance.

In monacanthids and ostracioids the supracleithrum is
not posteriorly expanded and the postcleithrum does not
have a dorsal prong, while in balistids a posterior expan-
sion and dorsal prong are present. However, these struc-
tures in balistids probably are involved with support of
the specialized tympanal region above the pectoral fin
base only found in balistids and not to be expected in
groups lacking a tympanum. In monacanthids and os-
tracioids the fifth ceratobranchial is toothless but has a
series of gill rakers along its anterior edge, while in
balistids the fifth ceratobranchial bears teeth but no gill
rakers. These differences are surely correlated with the
usually coarser diet of balistids versus monacanthids and
ostracioids, and the loss of teeth from and the gain of gill
rakers along the anterior edge of the fifth cerato-
branchial probably occurred independently in mona-
canthids and ostracioids. It is perhaps instructive that in
triacanthoids, from which the balistids are derived, the
fifth ceratobranchial also is toothed but lacks gill rakers.

Thus, the evidence strongly indicates that it is the
balistids among the balistoids to which the ostracioids
are most closely related, and not to the monacanthids
that are derivatives of the balistids. Moreover, since the
ostraciids are clearly derived from the aracanids, and the
aracanids have remained far more generalized, the
closest relationship of the balistids to the ostracioids is to
the aracanids and not to the ostraciids.

There are only a few ways in which balistids are more

specialized than aracanids, and these are mostly con-
cerned with dentition and correlated with the coarser
diet of balistids. In balistids there are only three pharyn-
gobranchials, a toothless suspensory element followed by
two elements bearing prominent teeth, while in
aracanids the toothless suspensory element is followed by
three elements, the first of which is either toothless or
with minute teeth but the third and fourth elements with
teeth almost as well developed as in balistids but usually
of lesser number. The relatively conical teeth of
aracanids are more generalized than the heavier notched
incisors of balistids in the sense that they are closer in
structure to those of the more generalized
triacanthodids, the basal plectognaths, while those of
balistids are more similar to those of the triacanthids
derived from the triacanthodids. However, as indicated
previously, there is evidence that the structure of the
teeth in balistids and aracanids might not have been
much different in the Eocene. By the retention of an in-
ner series of teeth in the upper jaw, balistids are more
generalized than aracanids, while the variable presence
in aracanids of a fifth tooth in what is the outer series of
balistids, which always have only four in each half of
both jaws, is slightly more generalized than in balistids.

The uppermost pectoral fin ray of aracanids is relative-
ly well developed and the two halves are of about equal
length, while in balistids this ray is greatly reduced in
size and the medial half is larger than the splintlike nub-
bin that represents the lateral half of the ray. Here again
the condition in aracanids is similar to that in
triacanthodids while that of balistids is similar to that in
triacanthids, although the lateral half of the ray in
balistids is even more reduced than in triacanthids. It is
presumed here that these few more specialized features
of balistids were acquired after the divergence of the
common triacanthoid phyletic line leading on the one
hand to balistids and on the other to aracanids, as dis-
cussed subsequently.

The conversion of a balistidlike fish into an aracanid
requires mainly the following: 1) the complete loss of
the pelvis along with its rudimentary pelvic fin-ray ele-
ment and encasing scales; 2) the complete loss of the
three dorsal fin spines and of their three pterygial and
supraneural supports, with the possible exception of one
of them; 3) a rearrangement of the posterodorsal part of
the head that no longer supports the spiny dorsal fin, in-
cluding the development of a posteriorly projecting
supraoccipital crest, subsequently to be lost by the os-
traciids; 4) a flattening of the dorsal surface of the
supraoccipital; 5) the elimination of the articulation
foramen of the first basal pterygiophore of the spiny dor-
sal fin between the epiotic and supraoccipital, and a
reduction in the posteromedial region of the exoccipitals
so that they no longer are closely apposed to one another
in the midline on the posterior wall of the skull; 6) the
development of a dorsal flange on the parasphenoid into
the orbital septum to meet an anteroventral prolonga-
tion of the pterosphenoids; 7) an enlargement of the
vomer with a broader surface of suturing to the ethmoid

and parasphenoid; 8) a reduction of the anterior cross-
bar of the T-shaped palatine and a firm suturing of the
foot of the T to the ectopterygoid and mesopterygoid; 9)
the loss of the inner series of teeth but the retention of a
slightly more generalized form and number in the outer
series than in Recent balistids; 10) a foreshortening and
deepening of many of the hyoid arch elements along with
the reduction in width of the first branchiostegal and the
development of a constant articulation between the last
branchiostegal and the suboperculum; 11) the retention
of a third toothed pharyngobranchial and the reduction
in size of the teeth on the first of the toothed elements;
12) loss of teeth on the fifth ceratobranchial and develop-
ment of gill rakers along its anterior edge; 13) great
reduction in the size of the urohyal; 14) great increase in
the size of the hyomandibular, posttemporal, postcleith-
rum, cleithrum, coracoid, epiotic, and pterotic; 15)
reduction in the size of the myodome and the elimination
of the posterior opening into it; 16) the ossification of
Baudelot's ligament; 17) the elaboration of the prootic
shelf including a recurved lateral wing; 18) a wider
separation of the pterotics on the ventral surface of the
skull; 19) the reduction in size and the fusion of the first
two vertebrae to the skull; 20) the loss of epipleurals; 21)
complete fusion of all the caudal fin supporting elements
of the last vertebral centrum; 22) an increase in the size
of the neural spines of the predorsal vertebrae and a great
reduction in length but a broadening of the haemal
spines of most of the caudal vertebrae; 23) a great reduc-
tion in the number of dorsal and anal fin rays and of their
basal pterygiophores and the development of divergent
anal fin basal pterygiophores; 24) concomitant with the
reduction in the dorsal and anal fins, an increase in the
number of vertebrae with neural or haemal spines an-
terior to the first basal pterygiophores of both fins (i.e.,
an increase in predorsal and abdominal vertebrae); 25)
development of lateral flanges on many of the more cen-
trally located vertebrae; 26) loss of one branched caudal
fin ray; 27) the enlargement and thickening of the scales
into heavy, more or less hexagonal, plates over the head
and body anterior to about the level of the ends of the
dorsal and anal fin bases; 28) the enlargement of a supra-
neural element into a long strut projecting anteriorly
from the dorsal fin base; 29) the suturing of the actinosts
to one another and to the scapula and coracoid.

Some of these above changes from the balistid to
aracanid type of organization are not shared by the os-
traciids. For example, ostraciids do not have: 1) the
posteriorly prolonged supraoccipital crest; 2) the es-
pecially elongate supraneural in front of the dorsal fin; 3)
the immensely enlarged postcleithrum; 4) the pos-
teriorly prolonged epiotic and generally more elongate
postorbital region of the head; 5) the elongate neural
spines of most of the predorsal vertebrae. This indicates
either that the ostraciids diverged from the aracanids at
a time before the ancestral line had developed these
latter differences with balistids or that the ostraciids at
any early stage had similar differences with balistids
which they have subsequently modified. I suspect, with-

out proof, that it is mainly a case of the latter, but it
could well be some combination of both possibilities.

As in the evolution of the triacanthoids and balistoids,
the major features in the diversification of the aracanids
from a prebalistidlike group and of ostraciids from
aracanids are largely reductive.

It is not suggested here that the aracanids evolved
from a group with a level of organization like that of the
Recent or few known fossil balistids, but rather that
aracanids and balistids share a common ancestry at a
lower level of organization, around that of the
triacanthids. However, this joint ancestral line of the
aracanids and balistids must have been more generalized
than any of the Recent triacanthids, and probably split
off from the triacanthids at a level of organization
somewhat like that exemplified by Protacanthodes, the
Eocene ancestral triacanthid that probably evolved from
hoUardiinlike triacanthodids. That is, the triacanthid
line leading to the balistids and aracanids probably did
not yet have such specialized features as found in the Re-
cent triacanthids as: 1) the especially large notched in-
cisors; 2) the forward extension of the prefrontal suturing
to the vomer; 3) a fully elongate soft dorsal fin with many
more rays than the anal fin; 4) the uppermost pectoral
fin ray much reduced in size and one half smaller than
the other; 5) the second and third basal pterygiophores of
the spiny dorsal fin greatly reduced in size; 6) the ab-
sence of a free parhypural and autogenous haemal spine
of the penultimate vertebra; 7) an elongate tapering
caudal peduncle with a deeply forked caudal fin. The
triacanthid line leading to the balistids and aracanids
would have had to retain from its triacanthodid ancestry
more generalized conditions of these features, much as
did Protacanthodes, the basal triacanthid. Whether the
balistid-aracanid ancestral line diverged from the early
triacanthids at a level of organization slightly more
generalized than that represented by Protacanthodes or
at a level between that oi Protacanthodes and the Recent
triacanthids is difficult to ascertain, for too many critical
features of Protacanthodes remain unknown, but it was
probably the former, as discussed under the Balistidae.

This ancestral pretriacanthid line is seen as diverging
into two radiations, one line leading with little change to
the Eocene Protacanthodes and the other triacanthids on
the one hand, and the other line leading through greater
changes to the balistids and aracanids, with the balistids
(as discussed under that family) remaining more
generalized and thus anatomically closer to their pre-
triacanthid ancestors than did the aracanids, whose
specializations are built around the defensive shell that
encases their bodies. While the balistid line continued to
lengthen the soft dorsal and anal fin bases and to develop
a locking mechanism between the first two dorsal spines
and an elaborate flexing mechanism of rudimentary rays
at the end of the pelvis, the aracanid line lost the spiny
dorsal fin and pelvic apparatus completely while
shortening the bases and decreasing the number of rays
in both the soft dorsal and anal fins.

When the aracanid line split off from that of the pre-
balistids it rapidly lost, still in the Eocene, the spiny dor-

Figure 147.— Typical body form in the Recent
Ostraciidae: Ostracion lentiginoaum.

sal and pelvic fins as the developing carapace took on the
major line of defensive protection. The aracanids became
less flexible and less swift in sustained swimming as the
soft dorsal and anal fins were reduced in number of rays
and basal pterygiophores. While the pelvis was com-
pletely lost, there remains some trace of either the now
absent spiny dorsal fin supports or of those of the now ab-
sent anterior part of the originally longer based soft dor-
sal fin. The long supraneural element of aracanids that
extends out from the anterior end of the dorsal fin is un-
doubtedly a modified basal pterygiophore, but whether
from the spiny dorsal fin or the anterior region of the soft
dorsal fin is impossible to say with assurity, because as
the spiny dorsal was being lost so was some of the soft
dorsal fin, probably from anteriorly to posteriorly judging
from the position of that which remains.

In balistoids the tendency has been for the spiny dor-
sal fin and its pterygiophores to migrate anteriorly as the
fin is reduced in size and number of elements from
balistids to monacanthids, and the latter retain no rem-
nants of the third spine, second pterygiophore, and
supraneural element of balistids. If the spiny dorsal fin of
aracanids migrated anteriorly as it became rudimen-
tary, it may have left free one of the more posterior pte-
rygiophores (corresponding in balistids to either the
second pterygiophore or to the supraneural strut, the lat-
ter itself being a modified third pterygiophore) which
then became enlarged to support the carapace. Con-
versely, in the only group of plectognaths other than
aracanids to lose the spiny dorsal fin, the gymnodonts,
the loss of the fin, on the evidence of Triodon, has been
by the posterior migration of the spines and pterygio-
phores as they became rudimentary. If the loss in
aracanids was by posterior migration, then the supra-
neural still represents an enlarged pterygiophore, or an
enlarged consolidation of two or more of them. Equally
possible is that the aracanid spiny dorsal fin, regardless
of whether it migrated anteriorly or posteriorly, lost not
only the spines but also all the pterygiophores, and that
the supraneural is actually an enlarged pterygiophore, or
an enlarged consolidation of two or more of them, from
the soft dorsal fin that became available for carapace
support as the soft dorsal was shortened from anteriorly
to posteriorly.

Family Ostraciidae

Comparative diagnosis (contrast with that of the
Aracanidae). — Supraorbital region of the skull rela-
tively weak, the frontal extremely thin in this region and
tightly held to the undersurface of the carapace; frontal
without a long forward extension anterior to the orbit,
ending anteriorly over or only slightly in front of the
prefrontal and not overlying the ethmoid; supraoccipital
crest absent; epiotic little, if any, prolonged postero-
laterally, scarcely if at all beyond the level of the base of
the skull; prootic shelf without a recurved wing;
myodome either small and shallow or essentially absent;
usually three pharyngobranchials, occasionally only two,
the first a toothless suspensory element and sometimes
absent, the second with a series of small to minute teeth,
and the third either toothless or with minute rudi-
mentary teeth; branchiostegals 2 -I- 4; ethmoid a more or
less rounded shaft posteriorly and much expanded
laterally in the anterior one-third or more of its length;
pterosphenoid and parasphenoid with only a very limited
region of suturing in the interorbital septum; para-
sphenoid much expanded laterally throughout its length
anterior to its region of suturing with the anteromedial
edges of the prootic shelves, forming a complete hard
palate over the roof of the oral cavity; posterodorsal
region of coracoid without a posterior prong; four or five
vertebrae, represented by both neural arch and centrum
material, at the front of the vertebral column variously
fused and sutured to one another and to the exoccipitals
and basioccipital, the fusion region with a prominent
ventral process below the level of the centra; vertebrae
immediately following the fusion region with relatively
short neural spines and with centra of about normal
length without transverse processes; the centra of two or
more of the vertebrae immediately preceding the caudal
plate much compressed anteroposteriorly; haemal arch
and spine of the penultimate vertebra relatively small
and either autogenous or fused to the centrum, in which
latter case it may be rudimentary; caudal fin rays, i, 8, i;
postcleithrum of moderate size, formed of one or two
pieces, and not extending posteriorly much beyond the
level of the posterior edge of the pectoral girdle; carina
(supraneural) relatively short and either without a ven-
tral flange or with only a weak shallow one, not extending
anteriorly as far as the level of the anteroventral end of
the first dorsal fin basal pterygiophore and not in contact
with any of the neural spines of the predorsal vertebrae;

haemal spines poorly developed and absent on many of
the abdominal and caudal vertebrae, except for the
heavy haemal spine(s) which supports by close apposition
and sometimes interdigitation the last anal fin basal
pterygiophore, the other anal fin basal pterygiophores
that lie in the midline of the body also making relatively
close contact along their distal ends with the vertebral
column; first anal fin basal pterygiophore expanded
anterolaterally at its distal end into wings supporting the
carapace; haemal canal not entirely straight, diverted
under the centra of the more posterior abdominal
vertebrae (usually between the seventh abdominal to
first caudal vertebrae) either to the left or to the right of
the midline, or alternatively to the left and right, and
relatively poorly enclosed under most vertebrae by weak-
ly developed haemal arches and spines; most of the
vertebrae posterior to the anteriormost four or five fused
vertebrae and anterior to the postanal vertebrae sutured
to one another (suturing between at least the fifth and
sixth abdominal vertebrae to the third and fourth caudal
vertebrae); uppermost pectoral fin ray short, composed
of a single piece without cross-striations and bearing a
foramen representing the region of otherwise complete
fusion between the original halves of the ray; body (ex-
clusive of vertical carapace spines) relatively less deep,
the distance between the distal ends of the first dorsal
and anal fin basal pterygiophores being contained more
than 3 times in SL, and the distance between the top of
the rear of the cranium and the ventralmost edge of the
pectoral girdle being contained about 2.3 to 2.5 times in
SL; carapace closed behind the anal fin and, with the ex-
ception of one species, behind the dorsal fin as well;
carapace never with a ventral keel; caudal peduncle
usually without scale plates isolated from the carapace
proper.

Detailed description o( Acanthostracion quadricor-

Material examined: — Thirteen cleared and stained
specimens, 8.2-350 mm; three dry skeletons, 130-163 mm.

Occipital Region.

Basioccipital. — A short column, expanded antero-
laterally; cartilage filled along its anterior and dorsal
edges; articulates by interdigitation dorsally with the
exoccipitals, anterolaterally with the prootics and
anteriorly with the overlying posterior end of the para-

Figure 148. — Acanthostracion quadricornis:

upper left, nasal region as seen externally

(olfactory epithelium relatively smooth in

this and all other species of ostraciids);

lower left, outline of cross section of middle

of body; pattern of scale plates shown

only behind pectoral fin.

sphenoid. The posterior half of the basioccipital is
hidden from view by the forward displacement of the
highly modified first five abdominal vertebrae, as ex-
plained in the section on the vertebral column.

Exoccipital. — Cartilage filled along all of its edges
of articulation with the other cranial bones; articulates
by interdigitation posterodorsally with the epiotic, ven-
trolaterally with the pterotic, ventromedially with the
basioccipital, and anteroventrally with the prootic.
Posteriorly the exoccipitals form the lateral walls and
most of the dorsal walls of the foramen magnum, while
ventrally the foramen is closed by the dorsal surface of
the basioccipital. Posterodorsally the medial edges of the
exoccipitals closely approach, but do not meet, one
another and are held together by a thin sheet of fibrous
tissue. More anteriorly, however, the dorsomedial edges
of the exoccipitals are more distantly separated and the
space between them is filled by a thin sheet of cartilage.
The lateral surface of the extreme posterior end of the
exoccipital is overlain by and interdigitated with the
more anterior of the highly modified first five abdominal
vertebrae, as explained in the section on the vertebral
column.

Supraoccipital. — Laterally expanded; no supraoc-
cipital crest; cartilage filled along all of its edges of ar-
ticulation with the other cranial bones; articulates by
interdigitation anteriorly and anterolaterally with the
frontals and posterolaterally with the epiotics.

Otic Region.

Pterotic. — Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates by
interdigitation posterodorsally with the epiotic, antero-
dorsally with the sphenotic, anteroventrally with the
prootic and posteroventrally with the exoccipital. For a
short distance medially along its anteroventral edge the
pterotic supports by fibrous tissue the posterodorsal edge
of the hyomandibular. The anterolateral region of the
pterotic is prolonged ventrally into a stout shaft that is
mostly obscured from lateral view by the broadly over-
lying posttemporal, to which it is interdigitated. At the
ventromedial end of this shaft the pterotic helps to sup-
port by fibrous tissue the laterally expanded end of the
ossified Baudelot's ligament and medially the postero-
lateral edge of the hyomandibular.

Sphenotic. —Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates by
interdigitation posterodorsally with the frontal, postero-
ventrally with the pterotic and for a short distance with
the anterodorsal end of the posttemporal, dorsomedially
with the pterosphenoid, and ventromedially with the
prootic. The dorsolateral surface of the sphenotic is
broadly overlain by the frontal.

Epiotic. — More or less rounded, but with a short
ventrally directed process from its ventrolateral edge;

cartilage filled along all of its edges of articulation with
the other cranial bones; articulates by interdigitation
dorsally with the frontal, anteroventrally along its lateral
edge with the sphenotic and pterotic, posteroventrally
with the exoccipital, and medially with the supraoc-
cipital. For a short distance the medial edge of the
epiotic is in contact with the cartilaginous plate that
separates the medial edges of the exoccipitals.

Prootic. — Cartilage filled along all of its edges of
articulation with the other cranial bones, except at its
anteriormost region of articulation with the para-
sphenoid; articulates by interdigitation ventromedially
with the laterally compressed keellike dorsal region of
the parasphenoid, ventrolaterally with the pterotic, pos-
teroventrally with the basioccipital and exoccipital, dor-
somedially with the pterosphenoid and dorsolaterally
with the sphenotic. Along most of the anterior edge of its
laterally expanded posterior portion the prootic possesses
a slight concavity with which the dorsal head of the hyo-
mandibular articulates by fibrous tissue. The antero-
medial region of the prootic possesses a long forward
extension under the orbit which makes contact by inter-
digitation with the parasphenoid at the level of the
prefrontals. A bony myodome is essentially absent.

Orbital Region.

Frontal. — Wide posterolaterally; extremely thin
and delicate anterolaterally, where it is in close contact
with the ventral surface of the cuirass; articulates by in-
terdigitation posteriorly with the supraoccipital and
posterolaterally with the epiotic and sphenotic, broadly
overlying the latter. Anteroventrally, in the orbital
region, the frontal interdigitates with the ptero-
sphenoid. Anteriorly the frontal is prolonged into a very
thin projection that overlies the dorsomedial edge of the
prefrontal and the dorsal surface of the ethmoid cartilage
and reaches almost to the posterior end of the ethmoid.
Since the thin anterolateral expansion of the frontal is
closely held by fibrous tissue to the ventral surface of the
cuirass over the orbital region, a large portion of the fron-
tal is very easily torn away when the cuirass is removed.
Thus, the few figures of the skulls of trunkfishes that
have been previously published usually show the frontal
incompletely and unrealistically with ragged edges.

Prefrontal. — Large and wedge-shaped; cartilage
filled along its medial edge where it is continuous with
the ethmoid cartilage; articulates by fibrous tissue dor-
somedially with the frontal, while ventrally it articu-
lates by fibrous tissue and, in large specimens, by slight
interdigitation with the parasphenoid. Its major surface
of articulation, however, remains with the ethmoid car-
tilage that broadly intervenes between the prefrontal and
the ethmoid.

Parasphenoid. — Elongate, running almost the
entire length of the skull; expanded ventrally along near-
ly all of its length into a thin keel, which itself is laterally

expanded along the anterior third of its ventral edge to
form a slightly concave horizontal plate over the roof of
the oral cavity. The parasphenoid articulates by inter-
digitation posterolaterally with the prootics, while at its
extreme posterior end it broadly overlies and interdigi-
tates with the basioccipital. At its blunt anterior end the
parasphenoid is deeply and firmly interdigitated with
both the ethmoid and vomer, and in large specimens
(over approximately 200 mm) these articulations become
truly fused. Anterodorsally the parasphenoid interdigi-
tates with the ethmoid, while dorsally in about the mid-
dle of its length the parasphenoid becomes slightly ex-
panded laterally and articulates by fibrous tissue and, in
large specimens, by interdigitation with the bases of the
prefrontals. In the anterior half of the orbital cavity the
parasphenoid is present as a thin flat plate, the extreme
posterodorsal edge of which makes contact, in large
specimens at least, with the extreme anterior end of the
pterosphenoid. In large specimens the articulation of the
parasphenoid and pterosphenoid is by interdigitation,
whereas in smaller specimens the articulation is through
the interorbital sheet of fibrous tissue. In about the mid-
dle of its length, where it is laterally expanded, the para-
sphenoid makes contact by interdigitation with the
anterior prolongation of the prootics.

Pterosphenoid. — Cartilage filled along all of its
edges of articulation with the other cranial bones, except
anteriorly; articulates by interdigitation posterodorsally
with the somewhat overlying frontal, posteriorly with the
sphenotic and posteroventrally with the prootic. The
anterior edge of the pterosphenoid is prolonged anteriorly
into a flattened process which makes contact with the
posterior edge of the platelike orbital expansion of the
parasphenoid, as explained above.

Ethmoid Region.

Ethmoid. — Elongate; a rounded shaft posteriorly,
but becoming laterally expanded anteriorly; cartilage
filled at its posterior edge, where it is continuous with the
ethmoid cartilage; articulates by interdigitation ventral-
ly along the posterior two-thirds of its length with the
parasphenoid, while ventrally in the anterior one-third of
its length it firmly articulates by extensive interdigi-
tation with the vomer. In large specimens the articu-
lation between the ethmoid and vomer becomes fused. At
its anterior edge the ethmoid helps support, through
fibrous tissue, the upper jaw.

Mandibular Region.

Hyomandibular. — Much expanded posteriorly,
tapering gradually to a blunt end anteriorly; cartilage
filled at its posterior and anterior edges; articulates by
fibrous tissue posteriorly with the slight concavity on the
anterior edge of the laterally expanded posterior portion
of the prootic and, for a short distance, with the con-
tinuation of that concavity on the anteromedial edge of
the pterotic, as well as along the medial surface of the
ventral shaft of the pterotic. The dorsomedial edge of the
hyomandibular is also supported through fibrous tissue
by the lateral surface of the ventral flange of the prootic
shelf. The hyomandibular possesses a slight concavity on
its ventral edge in which the rounded dorsal end of the
operculum is held by fibrous tissue. Just anterior to this
articular area the hyomandibular bears an elongate
groove on its lateral surface into which the slender pos-
terodorsal end of the preoperculum fits and is tightly
held by fibrous tissue. Anteriorly the hyomandibular is
attached to the fibrous tissue sheet between the
metapterygoid, symplectic, and preoperculum.

Quadrate. — Wide posteriorly, tapering anteriorly
to a knob for articulation with the articular in the lower
jaw, while from its posteroventral edge it possesses a
short posteriorly directed process; cartilage filled at its
posterior edge; articulates by interdigitation dorsally
with the ectopterygoid, posterodorsally with the
mesopterygoid and posteroventrally with the symplectic.
Along the posterior two-thirds of its ventral edge the
quadrate articulates by fibrous tissue with the preoper-
culum.

Metapterygoid. — Elongate; wide anteriorly but
becoming narrower posteriorly; cartilage filled at its
anterior edge; articulates by interdigitation anterodor-
sally with the mesopterygoid and anteroventrally with
the symplectic, which it somewhat overlies. Posteriorly
the metapterygoid articulates by fibrous tissue with the
hyomandibular and interhyal.

Symplectic. — Slender and elongate; no cartilage
filled edges evident; articulates by interdigitation an-
teriorly with the quadrate, posteriorly with the
metapterygoid and by fibrous tissue posteroventrally
with the preoperculum and hyomandibular.

Palato-Pterygoid Region.

Vomer. — Short, slightly expanded laterally in
about the middle of its length; articulates by interdigi-
tation dorsally with the ethmoid and posteriorly with the
parasphenoid, but in large specimens both of these ar-
ticular surfaces become fused. Anterolaterally, just in
front of its laterally expanded region, the vomer supports
through fibrous tissue the palatines, while at its extreme
anterior end it supports the upper jaw. A slight concavity
in the lateral surface of the vomer marks the place of ar-
ticulation between the vomer and palatine.

Palatine. — A small wedge of bone; articulates by
interdigitation ventrally with the ectopterygoid and, to a
lesser extent, with the mesopterygoid. Even though the
palatine has often been said to be absent in ostraciids, it
is distinct from the ectopterygoid even in large specimens
and the suture between the two can be seen upon close
examination. Dorsomedially the palatine articulates by
fibrous tissue from a slightly rounded prominence on its
upper medial surface with the slight concavity on the
lateral surface of the vomer.

Ectoptcrygoid. —Elongate; articulates by inter-
digitation dorsally with the palatine, posteriorly with the
mesopterygoid, which it somewhat overlies, and ven-
trally with the quadrate, which it also overlies.

Mesopterygoid. —Large; articulates by inter-
digitation posteroventrally with the metapterygoid,
while anteriorly it interdigitates with and is somewhat
overlain by the palatine, ectopterygoid, and quadrate.

Opercular Region.

Operculum. —Relatively short and thin; more or
less rounded in outline, with a protuberance dorso-
medially for articulation by fibrous tissue with a slight
concavity on the ventral edge of the hyomandibular.
Ventrally the operculum broadly overlies and ar-
ticulates by fibrous tissue with the suboperculum.

Suboperculum. —Very thin; widest in the middle,
broadly rounded anteriorly, tapering to a point pos-
teriorly; articulates by fibrous tissue dorsally with the
broadly overlying operculum. At the anterior edge of the
region where the operculum overlies the suboperculum, a
strong ligament coming from the interoperculum makes
contact with the suboperculum and, to a lesser extent,
with the operculum.

Interoperculum. — A short rod; extends from the
level of the anterior end of the preoperculum to the level
of the junction of the ceratohyal and epihyal; connects by
a strong ligament anteriorly to the angular in the lower
jaw, while posteriorly it connects by a short ligament to
the epihyal and by a longer more diffuse ligament to the
anterior edge of the suboperculum and, to a lesser ex-
tent, of the operculum.

Preoperculum. — Relatively short; not much ex-
panded in its middle region, and tapering to narrow ends
anteriorly and posteriorly; articulates by fibrous tissue
anteriorly along its dorsal edge with the quadrate and, to
a slight extent, with the symplectic; posteriorly the nar-
row shaft of the preoperculum is held by fibrous tissue in
an elongate concavity on the lower lateral surface of the
hyomandibular.

Upper Jaw.

Premaxillary. — A slightly curved plate, wider dor-
sally than ventrally; its posterodorsal region somewhat
concave for articulation by fibrous tissue with the an-
terior edges of the vomer and ethmoid; the anterior edge
of the upper jaw formed by the premaxillary, except for a
short distance ventrally where it is formed by the maxil-
lary; the dorsomedial edges of the two premaxillaries
held in close apposition by fibrous tissue. Each pre-
maxillary in adults usually bears five or six teeth in a
single row. The teeth are borne in relatively deep and
elongate grooves on the outer surface of the premaxil-
lary. The lowermost teeth are shaftlike and taper to

sharp points, but more dorsally and medially they are
flattened and blunter at their distal ends. At the pos-
terior end of each tooth-bearing groove there is a deep
socket in which new teeth develop before moving for-
ward to replace the old ones. Most of the interior of the
premaxillary contains the dental pulp from which the
new teeth develop. This pulp cavity communicates with
the exterior not only by the deep sockets in which the
new teeth develop but also by a large hole in the postero-
dorsal surface of the premaxillary. The premaxillary ar-
ticulates by extensive interdigitation with the maxillary
along all of its posterior edge, except for a short distance
dorsally.

Maxillary. — Widest ventrally, constricted in the
middle of its length; articulates by extensive interdigita-
tion anteriorly along all of its length with the pre-
maxillary, which it somewhat overlies. The maxillary ar-
ticulates by fibrous tissue posterodorsally with the
anterior edges of the vomer and ethmoid and ventro-
medially with the dorsolateral surface of the dentary.

Lower Jaw.

Dentary. — Wider posteriorly than anteriorly; its
posteromedial surface concave dorsally to accommodate
the articular, with which it interdigitates. Just below the
articular, the dentary interdigitates with the angular.
The ventromedial edges of the two dentaries are held
closely together by fibrous tissue. Laterally from its
posterodorsal region the dentary articulates by fibrous
tissue with the medial surface of the maxillary. Each
dentary in adults usually bears four or five teeth, like
those of the upper jaw, in deep grooves on its outer sur-
face, each of the tooth-bearing grooves ending pos-
teriorly in a deep socket in which new teeth develop. The
sockets are in communication with the large pulp cavity
that fills most of the hollow interior of the dentary. The
pulp cavity communicates with the exterior not only at
the tooth sockets but also at its posterior concave region
of articulation with the articular.

Articular. — Small; its posterior edge with a con-
cavity for articulation by fibrous tissue with the anterior
knoblike process of the quadrate. Anteriorly the articular
interdigitates with the concave upper half of the postero-
medial surface of the dentary. On the medial side of its
ventral edge the articular interdigitates with the angular,
but the lateral surfaces of these two bones are not in con-
tact; rather, they are slightly separated by the dentary.
The sesamoid articular is a roundish nubbin of bone
closely held to the medial surface of the articular just
behind the region where the upper medial edge of the ar-
ticular meets the medial surface of the dentary.

Angular. —Small, slightly elongate; articulates by
interdigitation dorsomedially with the articular, while
dorsolaterally, anteriorly, and ventrally it interdigitates
with the dentary. Posteriorly the angular connects by
ligament with the anterior end of the interoperculum.

BRANCHIAL APPARATUS.

Hyoid Arch, Branchiostegal Rays, and Urohyal.

Hypohyals. — Both hypohyal elements well de-
veloped; ventral element larger than the dorsal ele-
ment; dorsal element cartilage filled along its posterior
and ventral edges, the ventral element cartilage filled
along its posterior and dorsal edges. The two elements ar-
ticulate with one another and with the ceratohyal
through cartilage; the anteromedial edges of both
elements articulate by fibrous tissue with their opposite
members. The dorsomedial edge of the dorsal hypohyal
articulates by fibrous tissue posteriorly with the first
basibranchial and anteriorly with the reduced urohyal,
while the dorsomedial edge of the ventral hypohyal also
articulates by fibrous tissue with the urohyal.

Ceratohyal. — A wide flat plate; shortened antero-
posteriorly and expanded dorsoventrally to such an ex-
tent that the former dimension is shorter than the latter;
cartilage filled along all of its edges except at the in-
dented regions posterodorsally and anteroventrally; ar-
ticulates through cartilage anteriorly with the ventral
hypohyal, anterodorsally with the dorsal hypohyal and
posteriorly with the epihyal. The first two branchios-
tegal rays articulate by fibrous tissue with slight depres-
sions on the ventral edge of the ceratohyal at the an-
terior edge of its deeply indented anteroventral region.
The next three branchiostegal rays articulate by fibrous
tissue with the posteroventral edge of the ceratohyal, and
the last ray with the ventral edge of the area of ar-
ticulation between the epihyal and ceratohyal.

Epihyal. — Large, elongate dorsoventrally; cartilage
filled at its anterior, anterodorsal and ventral edges; ar-
ticulates through cartilage anteriorly with the cerato-
hyal, while posterodorsally it supports the interhyal by
fibrous tissue. Just anterior to its articulation with the
interhyal, the lateral surface of the epihyal articulates by
fibrous tissue with the posterior end of the interoper-
culum.

Interhyal. —A short, slender rod; cartilage filled at
both ends; articulates by fibrous tissue ventrally with the
epihyal and dorsally with the slight concavity on the ven-
tral edge of the metapterygoid immediately behind the
posterodorsal end of the symplectic.

Branchiostegal rays. —Six in number; the first ray
shorter than the second; the third to fifth rays of about
the same length as the first ray; the sixth ray by far the
longest. The rays articulate by fibrous tissue to the
ceratohyal and, to a lesser extent, the epihyal, as ex-
plained under those bones.

Urohyal. -—Reduced to a short depressed plate
without a ventral keel; articulates by fibrous tissue
anterodorsally with the anteromedial edge of the dorsal
hypohyal and dorsomedial edge of the ventral hypohyal;

articulates by fibrous tissue posteriorly with the ventral
surfaces of the first two basibranchial elements.

Branchial Arches. — All the elements are cartilage
filled at their edges of articulation with the other
elements in the series, and the articulations are usually
through cartilage and fibrous tissue. The branchial
arches are composed of three basibranchials, three pairs
of hypobranchials, five pairs of ceratobranchials, four
pairs of epibranchials, and two (sometimes three) pairs
of pharyngobranchials. Four gills are present, with a
small slit between the fourth arch and the lower pharyn-

First arch. — Basi-, hypo-, cerato-, epi-, and,
sometimes, pharyngobranchial elements present. First
basibranchial short, slightly wider posteriorly than an-
teriorly; displaced forward so that it articulates pos-
teriorly with the second basibranchial and anteriorly
with the hypohyals and urohyal, but with no direct con-
nection with the first hypobranchials. First hypo-
branchial wide dorsally but very narrow ventrally; the
largest of the hypobranchial elements, which decrease in
size posteriorly in the series; articulates ventrally with
the lateral surface of the middle region of the second
basibranchial and dorsally with the first cerato-
branchial. First ceratobranchial a long sturdy com-
pressed rod; the longest of the ceratobranchial elements,
which decrease in length posteriorly in the series; from
the first to the fourth ceratobranchial the ventral regions
become increasingly enlarged (the degree of compression
of the rods, and hence their width, appears to increase
somewhat with increasing specimen size); no ventrally
directed flange present on any of the ceratobranchials;
articulates ventrally with the first hypobranchial and
dorsally with the first epibranchial. First epibranchial a
short rod, somewhat narrowed in the middle; articulates
dorsally with the anterolateral surface of the second
pharyngobranchial and with the base of the first or
suspensory pharyngobranchial when such is present; the
shortest of the epibranchial elements, which increase
slightly in size posteriorly in the series. First
pharyngobranchial absent, at least as an ossification, in
the majority of the specimens examined, but present in 3
out of 13 specimens as a short slender toothless rod whose
distal end is held by fibrous tissue to the medial surface
of the extreme anterior end of the ventral flange of the
prootic shelf just lateral to the region where the prootic
interdigitates with the laterally expanded portion of the
parasphenoid in the anterior region of the orbit.

Second arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Second basi-
branchial the longest of the basibranchials, the first and
third being of about the same length, but of different
widths; articulates anteriorly with the first basi-
branchial, laterally with the first hypobranchials, and
posteriorly with the third basibranchial. Second hypo-
branchial somewhat wider dorsally than ventrally; ar-
ticulates ventrally with the area of articulation between

215

the first and second basibranchials, and dorsally with the
second ceratobranchial. Second ceratobranchial only
slightly shorter than the first ceratobranchial; slightly
expanded anteroventrally; articulates dorsally with the
second epibranchial. Second epibranchial a short rod; ar-
ticulates dorsally with the posterolateral surface of the
second pharyngobranchial. Second pharyngobranchial
the first and larger of the two toothed pharyngo-
branchial elements; even in the largest specimen studied
(350 mm) it remains partially unossified, because a large
cartilaginous region is present anterodorsally in its mass;
more or less squarish in outline, except for its concave
and vertically oriented posterior surface; articulates
along its lateral surface posteriorly with the second epi-
branchial and anteriorly with the first epibranchial; ar-
ticulates along its concave posterior surface by fibrous
tissue to the rounded anterior surface of the third
pharyngobranchial. The posterior surface of the second
pharyngobranchial is more deeply concave ventrally
than dorsally, so that the ventral surface of the element
is U-shaped. It bears an irregular band of up to about 25
minute teeth on its ventral surface along the bend of the
U. The teeth are so small that it is doubtful if they are
functional or even if they protrude to the surface through
the skin of the oral cavity.

Third arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Third basi-
branchial the widest of the basibranchial elements; rec-
tangular in shape; articulates anteriorly with the second
basibranchial, anterolaterally with the second hypo-
branchials, posterolaterally with the third hypo-
branchials and third ceratobrtmchials, and posteriorly
with the fourth ceratobranchials. Third hypobranchial a
relatively straight and slender rod; articulates postero-
laterally with the third ceratobranchial and postero-
medially with the third basibranchial, while its ven-
trally directed anterior end articulates by fibrous tissue
with the under surface of the more anterior basi-
branchial elements. Third ceratobranchial articulated
ventrally with the posterior ends of the third hypo-
branchial and third basibranchial, and dorsally with the
third epibranchial. Third epibranchial rodlike; slightly
expanded anteriorly in the middle of its length, at which
place it makes fibrous tissue contact with the posterior
surface of the second epibranchial; articulates dorsally
with the lateral surface of the third pharyngobranchial.
Third pharyngobranchial smaller than the second
pharyngobranchial; more or less columnar; its ventral
surface concave; better ossifed than the second pharyn-
gobranchial, but nevertheless with several cartilaginous
areas present; articulates anteriorly with the concave
posterior surface of the second pharyngobranchial and
laterally with the dorsal ends of the third and fourth epi-
branchials. It usually bears one to six minute and non-
protruding teeth in a row along the anterior edge of its
ventral surface, but no such even rudimentary teeth
could be found in some of the specimens examined.

Fourth arch. — Cerato- and epibranchial elements

only. Fourth ceratobranchial much expanded ventrally;
articulates ventrally with the posterior end of the third
basibranchial and dorsally with the fourth epibranchial.
Fourth epibranchial the longest of the epibranchial
elements; much wider ventrally than dorsally; ar-
ticulates dorsally with the lateral surface of the third
pharyngobranchial.

Fifth arch. — Ceratobranchial (lower pharyngeal)
element only. Fifth ceratobranchial the shortest of the
ceratobranchial elements; much expanded ventrally
throughout the middle region of its length; articulates
ventrally with the ventral end of the fourth cerato-
branchial; toothless.

PAIRED FIN GIRDLES.

Pectoral Fin.

Posttemporal. — A large long shaft broadly over-
lying the lower half of the lateral surface of the pterotic,
to which it is firmly interdigitated; anterodorsally it also
interdigitates, to a much lesser extent, with the
sphenotic, while its ventral head is firmly held by fi-
brous tissue to the supracleithrum. On its ventromedial
surface the posttemporal helps, along with the pterotic
and, to a far lesser extent, the cleithrum, to support by fi-
brous tissue the expanded lateral end of the ossified
Baudelot's ligament.

Baudelot's ligament. — The ligament fully ossified
as a large stout shelf giving great support to the pectoral
girdle. It articulates firmly by fibrous tissue at its medial
end with the extreme posterolateral edge of the pttra-
sphenoid immediately anterior to the posteriormost
point of contact between the parasphenoid and basioc-
cipital. Its expanded lateral end is firmly held by fibrous
tissue mainly to the ventromedial surface of the ventral
flange of the pterotic and to the ventral end of the post-
temporal, and, to a much lesser extent, to the extreme
dorsolateral end of the cleithrum. The dorsal edge of the
medially expanded platelike portion of the cleithrum is
firmly held by a sheet of fibrous tissue to the ventral sur-
face of Baudelot's ligament.

Supracleithrum. — Located slightly obliquely pos-
terodorsally to anteroventrally in relation to the axis of
the body; relatively short and broadly overlain by the
posttemporal; articulates firmly by fibrous tissue, and
slight interdigitation in large specimens, dorsally with
the overlying posttemporal and ventrally with the cleith-
rum, which it broadly overlies.

Cleithrum. —Greatly expanded both laterally and
medially along all the length of its anterior edge, except
for a short distance ventrally, so that a large thin ver-
tical plate is formed at right angles to the axis of the
body; also greatly expanded posteriorly in the ventral
half of its length; articulates dorsally by fibrous tissue

216

and slight interdigitation with the broadly overlying
supracleithrum and by fibrous tissue with the anterior
edge of the dorsal postcleithrum; posteriorly in about the
middle of its length it articulates by fibrous tissue and
slight interdigitation with the scapula and coracoid. Ven-
tromedially the cleithrum articulates by fibrous tissue
with its opposite member, while the dorsal edge of its
medially expanded platelike portion is held by fibrous
tissue to the ventral surface of Baudelot's ligament.

Postcleithra. —The postcleithra form a thin wide
plate, closely held by fibrous tissue to the cuirass, from
the upper end of the cleithrum to about the level of the
posteriormost point of the pectoral girdle. The dorsal
postcleithrum is widest and thickest at its anterior edge
where it articulates by fibrous tissue with the cleithrum.
Posteriorly the dorsal postcleithrum articulates by fi-
brous tissue, and in large specimens by slight inter-
digitation, with the ventral postcleithrum. The ventral
postcleithrum is an extremely thin plate of variable
shape, but it is always much smaller than the dorsal
postcleithrum. The two postcleithra overlie one another
in a variable manner at their region of articulation.

Coracoid. — Wider ventrally than dorsally; its pos-
terior edge with a laterally directed flange throughout its
length, except at the extreme dorsal and ventral ends; a
short dorsal projection present from its posterodorsal
edge which makes contact with the posterior edge of the
last actinost; cartilage filled at its dorsal edge; an up-
raised flange present on its lateral surface, running, with
increasing height of the flange, from posterodorsally to
anteroventrally; articulates dorsally through cartilage
with the scapula, while posterodorsally its small dor-
sally directed flange interdigitates with the posterior
edge of the last actinost. Dorsally along its anterior end
the coracoid articulates by fibrous tissue and slight in-
terdigitation with the cleithrum, which broadly overlies
it.

Scapula. — Completely encloses the scapular
foramen; cartilage filled at its ventral and posterior
edges; articulates by fibrous tissue and slight inter-
digitation anteriorly and anteroventrally with the cleith-
rum, and through cartilage ventrally with the coracoid;
posterodorsally the scapula interdigitates with the bases
of the first two actinosts, while more anteriorly on its dor-
sal edge it bears a concavity to which the reduced first
pectoral fin ray articulates by fibrous tissue.

Actinosts. — Four elements; all cartilage filled at
their dorsal edges; first actinost the smallest, the others
of slightly increasing size posteriorly in the series; all four
elements held to one another by interdigitation; first and
second actinosts articulated ventrally by interdigitation
with the posterodorsal edge of the scapula; third actinost
articulated ventrally by fibrous tissue with the scapula

and coracoid; fourth actinost articulated by fibrous tis-
sue ventrally with the coracoid, as well as by inter-
digitation posteriorly with the dorsally directed process
of the coracoid. Distally the actinosts support by fibrous
tissue all of the fin rays, except for the first, which is sup-
ported by the scapula, and the second, which is sup-
ported by both the scapula and first actinost.

Fin rays. — Twelve fin rays in most specimens, with
the first ray about one-sixth the length of the second ray
and articulated directly with the scapula instead of with
the actinosts, as are all the other rays, except for the sec-
ond, which articulates with the region of interdigitation
between the scapula and first actinost; first ray with its
two halves fused together, the division between the two
halves being indicated only by an oblong cavity running
horizontally through the base of the ray; the medial half
of the ray larger than the lateral half. The first two rays
unbranched, the others branched. The first ray without
cross-striations, the other rays cross-striated.

VERTEBRAL COLUMN. —All vertebrae with bicon-
cave centra, except the last, which ends posteriorly in the
fused hypural plate, and the highly modified first five
vertebrae.

Abdominal Vertebrae.

First five vertebrae. — What has usually been called
the first vertebra is in fact a compound element resulting
from the fusion of the first five abdominal vertebrae to
themselves and to the posterior end of the basioccipital
(as described in detail by Tyler 1963a). In adult
specimens there is slight evidence of this fusion of the
first five vertebrae into a single piece, for there are five
neural foramina in the upper half of the lateral surface of
the fusion product and its neural arch shows, except in
extremely large specimens, a division or lobation dor-
sally into five successive regions separated from one
another by clefts which extend downward about one-
third to one-half the distance from the top of the neural
arch to the region of the centrum. In the 8.2 mm
specimen each of the last three of the first five vertebrae
possesses a very thin centrum and neural arch in a single
continuous piece separate from that succeeding and
preceding it. The first and second vertebrae are even
thinner than the others and their centra are in close ap-
position and may even be fused at this early develop-
mental stage. These two centra, in turn, appear to be in
the process of fusing with the basioccipital, but their
neural arches are still separate from one another. In the
15.3 mm specimen the first two vertebrae have their cen-
tra and the lower one-fourth of their neural arches indis-
tinguishably fused, while the centra are similarly fused
to the basioccipital. The third and fourth vertebrae have
their centra and the lower three-fourths of their neural
arches indistinguishably fused, but separate from the
fused first two vertebrae and the fifth vertebra. In the

217

study specimens of 25.0 mm and larger the centra of the
first five vertebrae are all fully fused to one another and
to the basioccipital, while their neural arches become
progressively more fused together until only the lobation
dorsally remains from their original distinctness (see
Tyler 1963a: 158- 160 for details). After the centra have
fused with the basioccipital I am unable to distinguish
where the one ends and the other begins. In the adult, the
posterior end of the fused basioccipital and first five
vertebral centra is a concave rounded surface which ar-
ticulates by fibrous tissue with the sixth abdominal
vertebra. One can suppose that this concave articular
surface is formed from material contributed by the cen-
trum of the fifth vertebra. At least the first two vertebrae
seem to lie principally lateral to the posterior region of
the basioccipital, for it appears that they have fused not
to the extreme posterior end of the basioccipital but
rather to its posterolateral surface. The neural arch of the
first vertebra overlies and interdigitates with the postero-
dorsal surface of the exoccipitals. Ventrolaterally from
each side of the fused first five vertebral centra there is a
wide, but delicate, ventral process which attaches by fi-
brous tissue posteriorly with a similar pair of projections
from the sixth abdominal vertebra. The small muscles
which attach to these ventral projections have a com-
plex relationship with the air bladder. The dorsomedial
edges of the neural arches of the first five vertebrae meet
in the midline and are held together at first by fibrous
tissue, but in large specimens this articulation becomes
interdigitated to form a complete bony roof over the
neural canal. No neural spines are present. The lower
half of the posterior edge of the fused neural arches of the
first five vertebrae strongly interdigitates with the an-
terior edge of the lower portion of the neural arch of the
sixth abdominal vertebra.

Other abdominal vertebrae. —As explained in the
preceding section on the Aracanidae, the first caudal
vertebra is arbitrarily designated as the anteriormost
vertebra whose ventral surface is most closely associated
with the proximal end of the first anal fin basal pteryg-
iophore which consistently lies in the midvertical plane
of the body. In 13 specimens there are nine abdominal
vertebrae; which is to say that the first five vertebrae are
fused into a single piece, and that this element is fol-
lowed by four separate and distinct vertebrae. The sixth
abdominal vertebra articulates by fibrous tissue at its
concave anterior end with the posterior concave surface
of the fused first five vertebrae and basioccipital, while
laterally the lower half of the anterior edge of the neural
arch of the sixth vertebra is extensively interdigitated
with the lower posterior edge of the neural arch of the fus-
ed first five vertebrae. The sixth abdominal vertebra
possesses a short and thin neural spine along most of the
length of the dorsal surface of its neural arch. The
haemal processes from the ventrolateral edges of the an-
terior end of the sixth vertebra are anteriorly expanded
at their distal ends. The distal ends of these processes do
not fuse to one another in the midline of the body; rather,
they are held to one another by fibrous tissue. This in-

complete haemal arch is held by fibrous tissue an-
teriorly to the similar structure from the fused first five
vertebrae and, like the latter, it is a point of attachment
for a complex of muscles associated with the air bladder.
The seventh to ninth abdominal vertebrae have large
neural spines and complete, although much reduced,
haemal arches without haemal spines. The sixth to ninth
abdominal vertebrae articulate with one another not only
by fibrous tissue between their concave articular faces
but also by deep and firm interdigitation of the edges of
the neural arches. The neural spine of the seventh ab-
dominal vertebra is long, thin, and delicate; the antero-
posterior depth of its basal region is about equal to that
of the vertebra, but the spine tapers to a point distally.
The posterior edge of the neural spine of the seventh ab-
dominal vertebra articulates by fibrous tissue with the
first dorsal fin basal pterygiophore. The neural spines of
the eighth and ninth abdominal vertebrae are narrow
rods arising from the posterodorsal surfaces of their
neural arches and directed posterodorsally between the
dorsal fin basal pterygiophores almost to the distal ends
of the latter elements. The neural spine of the eighth ab-
dominal vertebra lies between the first and second
pterygiophores, and the neural spine of the ninth
between the third and fourth pterygiophores. The neural
spines and pterygiophores are held to one another by a
sheet of fibrous tissue. The haemal arches of the seventh
to ninth abdominal vertebrae are complete, although
thin and very delicate and of somewhat variable shape
from specimen to specimen. No haemal spines are pres-
ent. The haemal arches arise from the anterior end of the
ventrolateral surfaces of the centra and do not project
very far ventrally, thus enclosing only a shallow cavity
through which the haemal canal runs. The arches of the
eighth and ninth vertebrae are distinctly displaced to one
side of the midline of the body, usually to the right. The
longitudinal concavity on the ventral surface of the
seventh to ninth vertebrae, which marks the course of the
haemal canal, is likewise displaced to a greater or lesser
degree to one side of the midline of the body. The
longitudinal concavity on the ventral surface of the sixth
vertebra is in the midline of the body, as it is on the an-
terior end of the seventh vertebra. As the concavity
passes from the anterior to the posterior end of the
seventh vertebra, it is displaced usually to the right of
the midline, and along the eighth and ninth vertebrae it
is similarly usually displaced to the right of the midline.
The ventrolateral edges of the seventh to ninth ab-
dominal vertebrae are interdigitated with one another.
The eighth and ninth abdominal vertebrae possess
slightly upraised ridges diagonally across the lateral sur-
faces of their centra from the bases of the neural spines to
the anterolateral edges of the centra. The neural foramen
of each of the abdominal vertebrae from the sixth to
ninth emerges relatively low on the lateral surface of the
vertebra.

Caudal Vertebrae. — The caudal vertebrae numbered
10 in 13 specimens, although all other species of os-
traciids, including the other three species of Acantho-

stracion, have only 9 caudal vertebrae (see Tyler
1965b:268). The first and second caudal vertebrae are
like the eighth and ninth abdominal vertebrae, except
that their neural spines are shorter and that their antero-
ventral surfaces support by fibrous tissue the first two
anal fin basal pterygiophores which lie entirely in the
midvertical plane of the body. The neural spine of the
first caudal vertebra lies between the fifth and sixth dor-
sal fin basal pterygiophores. The haemal arches of the
first, second, and third caudal vertebrae are delicate and
complete, like those of the seventh to ninth abdominal
vertebrae. The haemal arches of the second and third
caudal vertebrae are more or less symmetrically placed
in the midline of the body, but that of the first caudal
vertebra is displaced somewhat to the right. The longi-
tudinal concavity present on the ventral surfaces of the
first to fourth caudal vertebrae, which marks the course
of the haemal canal, is more or less in the midline, ex-
cept in the case of the first caudal vertebra, where the
concavity is displaced to the right at the anterior end of
the vertebra. The fourth caudal vertebra usually
possesses only the longitudinal concavity on its ventral
surface, and no haemal arch. The third and fourth anal
fin basal pterygiophores that lie in the midvertical plane
of the body articulate by fibrous tissue with the antero-
ventral surfaces of, respectively, the third and fourth
caudal vertebrae. The neural spine of the third caudal
vertebra is anteroposteriorly expanded into a flattened
plate which along the dorsal half of its anterior edge is
firmly articulated by fibrous tissue with the posterior
edge of the last dorsal fin basal pterygiophore, while
along its posterior edge it is deeply interdigitated with
the anterior edge of the neural spine of the fourth caudal
vertebra. The neural spine of the fourth caudal vertebra
is anteroposteriorly expanded into a flat plate which is
firmly interdigitated at its anterior and posterior edges
with, respectively, the posterior edge of the neural spine
of the third caudal vertebra and the anterior edge of the
neural spine of the fifth caudal vertebra. Whereas the
third caudal vertebra is the last caudal vertebra to make
contact with the dorsal fin basal pterygiophores, the fifth
caudal vertebra is the last to make contact with the anal
fin basal pterygiophores. The fifth caudal vertebra has
its neural spine anterodorsally expanded, with its an-
terior edge interdigitating with the posterior edge of the
neural spine of the fourth caudal vertebra, while its pos-
terior edge is well separated fi-om the neural spine of the
sixth caudal vertebra and is held to the latter only by fi-
brous tissue. The fifth caudal vertebra possesses a well-
developed haemal arch and spine, with the haemal arch
running the length of the vertebra. The haemal spine of
the fifth caudal vertebra is a thick compressed plate
which is concave along the lower half of its anterior edge
and articulates by fibrous tissue with the middle region
of the posterior surface of the last anal fin basal pteryg-
iophore. The sixth to ninth caudal vertebrae are basically
similar to one another. They possess sturdy neural arches
and spines and haemal arches and spines, all of which
decrease slightly in size posteriorly in the series. The cen-
tra of these four vertebrae are much shortened antero-

posteriorly and are held to one another by fibrous tissue.
These vertebrae, and the last vertebra, are thus the only
vertebrae which can be laterally flexed to any ap-
preciable extent, since all the vertebrae anterior to the
sixth caudal vertebra are interdigitated with one another
over large areas of their surfaces of contact. Thus, the
third to sixth caudal vertebrae are interdigitated with
one another along the surfaces of contact of their neural
arches and spines; the seventh abdominal to the second
caudal vertebrae are interdigitated with one another not
only along the surfaces of contact of their neural arches
but also at their ventrolateral edges of contact; the sixth
abdominal vertebra is interdigitated along the anterior
and posterior edges of its neural arch with the vertebrae
anterior and posterior to it; the first five abdominal
vertebrae are completely fused to one another.

Caudal Skeleton. — The caudal complex consists of a
large rectangular plate, with a rounded expansion in the
middle region of its anterior edge representing the cen-
trum of the last caudal vertebra, and a deep cleft in the
middle of its posterior edge representing the division
between what in more generalized plectognaths would be
the second and third hypurals. However, the hypurals
are fully fused to the centrum and to themselves and no
real distinction can be made between them. Examina-
tion of young specimens shows that the anterior portion
of the caudal skeleton develops from a centrum which
has a complete neural arch and spine and haemal arch
and spine. The neural (epural) and haemal (parhypural)
apparatus in the 8.2 mm specimen already appear to be
continuous with the centrum, although a deep indenta-
tion in the dorsal edge of the plate partially separates the
epural from the fused hypurals. In all of the small
specimens a urostylar thickening can be seen projecting a
short distance posterodorsally from the posterior end of
the last vertebra, just in front of the anterodorsal edge of
the hypural region. This complex that represents the last
caudal vertebra corresponds to about the anterior one-
seventh of the adult caudal skeleton. Posterior to the
region of the last caudal vertebra in the 8.2 mm
specimen, and continuous with it, is the hypural plate,
representing the hypural elements fully fused already to
one another and to the last vertebra, except in the region
of the deep cleft in the posterior edge, the cleft being
much deeper in young specimens than in adults. In
adults all of these elements are simply more extensively
and indistinguishably fused into the single rectangular
plate already apparent at 8.2 mm. A small foramen is
present in the anteroventral region of the plate and this is
the last opening into the haemal canal, which forms a
short tube in the substance of the plate anterior to the
foramen and is continuous with the haemal canal of the
more anterior vertebrae. This foramen thus marks the
region of fusion of the lower hypural region with the
parhypural. The neural canal enters the anterior edge of
the plate just above the centrum and courses through the
plate to exit at a foramen on its posterodorsal edge. The
region of the plate above the neural canal represents the
fully fused epural.

219

Caudal fin rays. —Ten in number, the uppermost
ray and the lowermost ray unbranched, the others
becoming increasingly branched toward the middle rays,
which are branched in triple dichotomies. The five upper
rays articulate by fibrous tissue at their bifid bases with
the upper lobe of the fused hypural plate and the lower
five rays with the lower lobe.

DORSAL AND ANAL FINS.

Fin rays and pterygiophores. — Ten fin rays are
present in most specimens; the first ray unbranched, the
other rays branched in single or double dichotomies. Dis-
tal pterygiophores are either absent or unossified. The
bifid bases of the fin rays are supported through fibrous
tissue by nine basal pterygiophores. With the exception
of the first pterygiophore, all of them are slender rods,
cartilage filled at both ends and of decreasing length pos-
teriorly in the series. At their dorsal ends these rodlike
pterygiophores are somewhat compressed and are closely
held to one another by fibrous tissue and slight inter-
digitation, with the degree of interdigitation between
them increasing with increased specimen size. From the
second to the ninth pterygiophore, slight concavities are
present in the surfaces of contact of the pterygiophores
with each other, so that gaps are present in the other-
wise interdigitated surfaces. The first pterygiophore is
unlike the others only in that it is expanded anteriorly
throughout most of its length into a thin flange. The
anterodorsal edge of the first pterygiophore inter-
digitates with the posterior edge of the supraneural. The
supraneural is a short, laterally expanded, plate whose
dorsal surface is convex and whose ventral surface is con-
cave. The dorsal surface of the supraneural is in close fi-
brous tissue contact with the cuirass. The dorsal fin basal
pterygiophores articulate by fibrous tissue between the
neural spines of the seventh abdominal to third caudal
vertebrae. The articulation of the ninth pterygiophore is
somewhat different from the others in that it makes a
much more intimate contact with the adjacent neural
spine than do the others, for it lies in close apposition
with the concave anterior edge of the neural spine of the
third caudal vertebra.

Fin rays and pterygiophores. — Ten fin rays are
present in most specimens; the first ray unbranched, the
others branched in single or double dichotomies. Distal
pterygiophores are either absent or unossified. The bifid
bases of the fin rays are supported through fibrous tissue
by 9 (occasionally 10) basal pterygiophores. These
pterygiophores are basically similar to those of the dor-
sal fin in that they are slender rods, compressed at their
ventral ends into flattened narrow plates held to one
another by fibrous tissue and slight interdigitation. Con-
cavities are present in their surfaces of contact with each
other so that gaps are present in their otherwise inter-

digitated surfaces. Whereas all of the dorsal fin pteryg-
iophores lie in the midvertical plane of the body, only the
last four anal fin pterygiophores lie entirely in this
medial plane, the five pterygiophores anterior to them
diverging to the right and to the left from their ventral
ends, which are in the midvertical plane. The sixth to
ninth pterygiophores have their dorsal ends held by fi-
brous tissue to the anteroventral surfaces of, respec-
tively, the first to fourth caudal vertebrae. The middle
region of the posterior edge of the ninth pterygiophore is
also held by fibrous tissue to the concavity on the an-
terior edge of the haemal spine of the fifth caudal verte-
bra. The second to fifth pterygiophores have their flat-
tened ventral regions in the midvertical plane, but their
long rodlike anterodorsally directed portions lie to the
right or to the left of the midline in the anterior portion of
the large muscle mass connected to the anal fin (only the
two pterygiophores that diverge to the left are shown in
the figure of the lateral view of the entire skeleton). The
anterodorsal rodlike portion of the first anal fin basal
pterygiophore is placed only very slightly to one side of
the midvertical plane, either to the right or to the left
depending on the individual. The ventral edge of its rod-
like portion is expanded into a low and thin keel through-
out most of its length. The first pterygiophore has by far
the largest distal end of any of the pterygiophores. Its
base is expanded anterolaterally to either side to form a
U-shaped projection which lies just under the cuirass. In
large specimens the usually rodlike dorsal portions of the
second to fifth pterygiophores may become laterally ex-
panded into thin flanges in the region where these
pterygiophores converge toward the midvertical plane of
the body.

Anatomical diversity. — The carapace extends well
behind the level of the posterior ends of the dorsal and
anal fins and completely encloses the fin bases in all
species of the family except Lactophrys trigonus, the lat-
ter having the carapace somewhat extended beyond the
dorsal and anal fin bases but only completely closed
behind the anal fin. Posterior to the dorsal fin in L.
trigonus the two posterodorsal extensions of the carapace
often closely approach one another in the midline, es-
pecially in large specimens, and are separated from one
another and from the anterior half of the large fi-ee plate
that lies just behind the dorsal fin only by a very narrow
scaleless space. In one especially large specimen of L.
trigonus examined (ANSP 102757, 366 mm) externally,
the posterodorsal extensions of the carapace and the
usually free plate are in intimate contact and are as fully
interdigitated with one another as are the other scale
plates of the carapace, the carapace in this specimen
forming a complete wide bridge over the caudal peduncle
behind the dorsal fin.

In most genera (Acanthostracion, Rhinesomus, Lac-
tophrys, Tetrosomus) the carapace has a basically tri-
angular shape, the horizontal distance between the
ventrolateral ridges being greater than between the
dorsolateral ridges, with the carapace extended above
into a dorsal ridge. A mediolateral ridge is scarcely devel-

oped, if at all, and there is no ventral ridge, the flattened
bottom of the carapace being either gently convex or slightly
concave. At the opposite extreme to the forms with a
triangular carapace is Ostracion, with a rectangular
carapace in which the distance between the dorso-
lateral ridges is about equal to that between the ventro-
lateral ridges, and with both the back and belly flattened
or only gently convex, the dorsal and ventral ridges

Fi({ure laO.—Acanthostracion quadricornis:
lateral view of head, composite based on
several specimens, 58.2-350 mm SL,
Gulf of Mexico and Caribbean.

pterosphenoid

frontal

prefrontal

maxillary
premaxillaryXp"'*"'"*

dentary

articular
anguia

quadrate
-interoperculom
ectopterygoid

parosphenoid

cartilage '

ethmoid \ V#f^^

vomer

supraoccipital

sphenotic

symplectic

basioccipital

prootic \ parasphenoid

metapterygoid , operculum

■^ preoperculum \ i» i

"^^^'y^J^ suboperculum hyomandlbular PO^«emporal

essentially being absent. Two other genera are in various
ways intermediate between the two extremes of carapace
shape. In Lactoria the carapace is basically pentagonal,
with the distance between the ventrolateral ridges much
greater than that between the dorsolateral ridges, but
with only a low dorsal ridge, extended in two species (for-
nasinii and diaphana) into a prominent spine. In
Rhynchostracion the carapace in one species
(rhinorhynchus) is as rectangular as in Ostracion, while
in the other species (nasus) there is a prominent and
relatively high dorsal ridge so that the carapace is pen-
tagonal. In the pentagonal carapace of R. nasus the dis-
tance between the ventrolateral ridges is only very slight-
ly greater than that between the dorsolateral ridges and
this shape may be most closely related to the rec-
tangular form, while the pentagonal carapace of Lac-
toria, with the distance between the ventrolateral ridges
much greater than that between the dorsolateral ridges,
may be most closely related to the triangular form.

Isolated scale plates on the caudal peduncle behind
the carapace are found only in Acanthostracion quad-
ricornis (commonly) and in A. polygonius (rarely, see
Tyler 1965a:8), in addition, of course, to the single large
plate found in Lactophrys trigonus, discussed above.

In the only known fossil specimen of ostraciid, Eolac-
toria sorbinii from the Eocene of Monte Bolca, Italy, it is
impossible to say with assurity whether the carapace was
triangular, pentangular, or rectangular. However, there
is a slight indication (Tyler 1973a: 108) that the scale
plates along the region that would be the dorsolateral
ridge had larger central tubercules or spines than the sur-
rounding plates, suggesting that the dorsolateral ridge

was prominent and that there was a dorsal crest, the
carapace perhaps being pentangular as in the Recent
Lactoria and Rhynchostracion rhinorhynchus.

Spiny processes are present on the carapace except in
Ostracion, Rhynchostracion, and Rhinesomus triqueter.
In Acanthostracion, Lactoria, and the Eocene Eolac-
toria there are paired preorbital and postanal spines,
with Eolactoria having in addition an unpaired an-
teriorly directed spine between the eyes and Lactoria
having a well-developed spine on the dorsal ridge in some
species. In Lactophrys and Rhinesomus bicaudalis there
are postanal spines, while in Tetrosomus there are three
to five spines along the ventrolateral ridge, a supraor-
bital spine and one or two spines along the dorsal ridge.

In general, the carapace of ostraciids is slightly more
fully consolidated than in aracanids, the only region with
much flexibility and unsutured scale plates being the
cheek, whereas in aracanids many of the scale plates
around the mouth and along the ventral region and
around the anus are often unsutured and slightly flexi-
ble.

The number of teeth is slightly variable in ostraciids,
which tend to have a few more teeth in the jaws, es-
pecially the upper, than in aracanids. Most ostraciids
have 10 to 12 teeth in the upper jaw and 8 to 10 in the
lower jaw. There is only a slight tendency for an increase
in the number of teeth with increasing specimen size, ex-
cept in Lactoria diaphana, a relatively small species
which at sizes below about 30 mm tends to have 10 to 12
teeth in the upper jaw and 6 to 7 in the lower jaw, but at
larger sizes to have about 13 to 16 in the upper and 7 to 9
in the lower. The numbers of teeth in the upper and lower

222

jaws of those species for which the most counts were
made are shown in Table 3 and Figures 165-166 (for Lac-
toria cornuta, L. fornasinii, and L. diaphana; Ostracion
tuberculatus and 0. lentiginosum; Acanthostracion
quadricomis and A. polygonius; Lactophrys trigonus;
and Rhinesomus triqueter and R. bicaudalis). There is
nothing out of the ordinary in the tooth counts recorded
for the fewer specimens of the other species of ostraciids
examined but not listed above. In the Eocene Eolactoria
the total number of teeth in the jaws is not known, but
they seem to be slightly wider basally and more decided-
ly constricted distally or cusped than in the Recent
species, somewhat reminiscent of the teeth of balistids
(Tyler 1973a).

In all species of the Indo-Pacific genera (Lactoria,
Tetrosomus, Ostracion, and Rhynchostracion) the dor-
sal and anal fin rays are modally 9 and the pectoral fin
rays modally 10, excluding the uppermost rudimentary
pectoral ray. In all of the species of the Atlantic genera
{Acanthostracion, Rhinesomus, and Lactophrys) the
dorsal and anal fin rays are modally 10 and the pectoral
fin rays modally 12 in all but Acanthostracion quad-

Figure 151. — Acanthostracion quadricornis:
dorsal (left) and ventral views of skull, composite
based on several specimens, 58.2-350 mm SL,
Gulf of Mexico and Caribbean.

ricornis, which has only 11 (see Tyler 1965a, b). Corre-
lated with the differences in the number of dorsal and
anal fin rays are similar differences in the number of
basal pterygiophores. In the Indo-Pacific genera there are
modally eight dorsal and anal fin basal pterygiophores
and in the Atlantic genera nine. In both cases the an-
terior divergent anal fin basal pterygiophores number
five. When one more than the modal number of rays is
present, there is also often one additional basal pteryg-
iophore.

In all ostraciids there are two nostrils on each side,
nearly always in the form of two short upright tubes.
However, in Rhinesomus bicaudalis the anterior tube is
extremely bulbous (see Bohlke and Chaplin 1968:678, fig.
212; and Fig. 159 here) at all specimen sizes, while in
Lactophrys trigonus it becomes bulbous, to about half
the extent of that in R. bicaudalis, in large specimens.

Figxire 152.— Acanthost
quadricomis: posterior views
skull (left) and of orbit (right) (
section of skull: dashed lines represent
cut surfaces of frontals, pterosph(
prootics, and parasphenoid). composite
based on several specimens,
58.2-350 mm SL, Gulf of Mexico
and Caribbean.

Figure l53.—Acantho»tracion

quadricomia: lateral (to left) and

medial views of palatopteroquadrate

region, composite based on several

specimens, 58.2-350 mm SL,

Gulf of Mexico and Caribbean.

Figure \5i.—Acantho8tracion
quadricornig: (left) lateral view of
occipital region of skull and the
fused five abdominal verte-
brae; (right) posterior view of
fused first five abdominal vertebrae;
77.7 mm SL, Florida.

1st 2nd 3rd

Figure \55.—Acantho8tracion

quadricomis: medial (left) and

lateral (right) views of pectoral

girdle, composite based on several

specimens, 58.2-350 mm SL, Gulf of

Mexico and Caribbean.

Figure 156.— /Icanf/iostrocion
quadricomis: above, ventral view

of vertebral column from the
fused first five abdominal vertebrae

to the fourth caudal vertebra,

showing the haemal spines and the

displacement of the haemal canal;

below, ventral view of the basal

pterygiophores of the anal fin,

showing the displacement of the

more anterior pterygiophores to

the left and to the right of the

midline; composite based on several

specimens, 58.2-350 mm SL,

Gulf of Mexico and Caribbean.

ceratobranchials

epibranchials rS^'

pharyngobranchials

::;= basibranchials
hypobranchials

epihyal

' }j ventral hypohyal
ceratohyol

Figure 159. (opp. page)— External features

of other representative ostraciid genera:

A, Rhinesomus bicaudalis—

upper left, nasal region as seen

externally and, lower left,

outline of cross section of

middle of body; B, R. triqueter—

to show the course of the

inconspicuous lateral line

canals and their major pores.

as decifered by placing drops

of ink on each pore found by

microscopic search of a partially

drying specimen; C, R. triqueter—

upper left, nasal region as

seen externally and. lower

left, outline of cross section

of middle of body, with pattern

of scale plates shown only

behind pectoral fin on main

body illustration.

branchlostegal rays

Figure 157.— 4cant/io»tracion lateral and dorsal views of urohyal;

quadricomig: dorsal view of composite based on several specimens,

branchial arches (extended on lower 58.2-350 mm SL, Gulf of

side); lateral view of hyoid arch; Mexico and Caribbean.

Figure 158.— 4conf/io»trocion quadri-
comis: lateral view of caudal fin
supporting structures, composite

based on several specimens,

58.2-350 mm SL, Gulf of Mexico

and Caribbean.

l;^

I u~^i

Figure 162.— External features of other representative

ostraciid genera: Tetrosomus gibboeus— upper left,

nasal region as seen externally; lower left, outline

of cross section of middle of body; pattern of

scale plates shown only behind pectoral fin.

Figure 163.— External features of other representative

ostraciid genera: Ostracion lentiginosum— upper

left, nasal region as seen externally; lower left,

outline of cross section of middle of body; pattern

of scale plates shown only behind pectoral fin.

Figure 160. — External features of other representative

ostraciid genera: Lactophrys trigonus — upper left,

nasal region as seen externally; lower left, outline

of cross section of middle of body; pattern of

scale plates shown only behind pectoral fin.

Figure 161.— External features of other representative

ostraciid genera: Lactoria forntuinii— upper left,

nasal region as seen externally; lower left, outline

of cross section of middle of body; pattern of

scale plates shown only behind pectoral fin.

Figure 164.— External features of other representative

ostraciid genera: Rhynchostracion nasus (above) and

R. rhinorhynchua— in both cases, upper left, nasal

region as seen externally; lower left, outline of

cross section of middle of body; pattern of scale

plates shown only behind pectoral fin.

227

L. TRIGONUS • O
R. TRIQUETER ▼ V
R. BICAUDALIS ■ □

I f

The fundamental plan of the skull is even more con-
servative in the ostraciids than in the aracanids, and
nearly all of the differences between the species to be
pointed out involve the vertebral column.

In all of the species of the Indo-Pacific genera (Lac-
toria, Tetrosomus, Ostracion, and Rhynchostracion) a
small but distinct myodome is present, while in all of the
species of the Atlantic genera (Acanthostracion, Rhine-
somus, and Lactophrys) the myodome is essentially ab-
sent. In the Atlantic species there is no anterodorsal roof
to the prootic at the appropriate place over the posterior
eye muscles, although in the rear of the orbit where the
prootics are in contact medially the two bones are curved
upward to form a shallow concavity bounded posteriorly
by a low crest, which region can be considered as the pos-
terior and posterodorsal walls of a rudimentary
myodome. If these surfaces of the prootics in the Atlan-
tic species were a little more concave and the posterior
myodome region more uplifted and anterodorsally
oriented, a small myodome could be said to be present,
but such is not quite the case. In the Indo-Pacific species
a small and shallow myodome is more distinctly present.
In all of the species of the more generalized family
Aracanidae, a small but distinct myodome is present.

In all of the Indo-Pacific species at least two of the
vertebrae posterior to the last which helps support the
last dorsal fin basal pterygiophore have trifid neural
spines, while none of the neural spines are trifid or even
bifid in the Atlantic species. The greatest number of
trifid neural spines is found in Lactoria cornuta, in which
the 14th to 17th vertebrae have them. In L. fomasinii
there is one less, the trifid spines being present on the
14th to 16th vertebrae. In Tetrosomus concatenatus and
T. gibbosus well-developed trifid neural spines are pres-
ent on the 15th and 16th vertebrae, and a less strongly
developed one may be present or absent on the 17th
vertebra. In Ostracion lentiginosum they are trifid on the
15th to 17th vertebrae, as they also are in 0. tuber-
culatus and Rhynchostracion rhinorhynchus, except that
the spine of the 17th vertebrae is sometimes single rather
than trifid. The trifid neural spines apparently afford a
broader surface of support for the carapace in this region
than does the single neural spine.

In all of the Atlantic species and in the Indo-Pacific
Lactoria fornasinii the first five vertebrae are involved in
the fusion complex with the base of the skull, while all of
the Indo-Pacific species except L. fornasinii have only
the first four vertebrae involved in the fusion complex. Of
the two species of Ostracion examined, at least one has
slightly less fusion between the first four vertebrae than
in the other Indo-Pacific species. In Ostracion tuber-
culatus the first two vertebrae are fully fused to one

i »

7 8 9 10

NUMBER OF TEETH

Figure 165.— Number of teeth in upper (solid
symboU) and lower (open symbols) jaws in relation
to sUndard length, to show the slightly greater
number of teeth in the upper versus the lower jaw,
and the relatively negligible increase in number of
teeth with increasing standard length at sizes
greater than about 40 mm SL; Lactophry» trigonut
Rhinesomiu triqueter, R. bicaudalit.

O. TUBERCULATUS • O

a LENTIGINOSUM ■ D

A. OUADRICORNIS ▼ V

A. POLYGONIUS • O

I I

8 9 10 11 12 13

JUMBER OF TEETH

another and to the basioccipital and exoccipital basally,
while distally there is only partial fusion of their neural
spines. The third and fourth vertebrae are similarly fully
fused with one another basally, while their neural spines
are only partially fused. The degree of fusion of the
neural spines increases with increased specimen size,
much as described by Tyler (1963a) (or Acanthostracion
quadricomis. But even in the two largest specimens of 0.
tuberculatus examined (96.7-122 mm) the fused first-
second vertebrae and the fused third-fourth vertebrae re-
main distinct from one another. In 0. lentiginosum the
two smallest specimens examined (24.4-33.2 mm) have
the first and second vertebrae fully fused to themselves
and to the basioccipital and exoccipital basally, with the
neural spines only partially fused, and the third and
fourth vertebrae are similarly fused to themselves, with
the second and third remaining separate. But in all of the
larger specimens (68.2-115 mm) the fused first-second
vertebrae and the fused third-fourth vertebrae are at
least partially fused together between the second and
third, and in the largest specimens almost completely
fused. However, this variously partial to almost com-
plete fusion between the two fusion complexes in 0. len-
tiginosum is less complete and occurs at a larger
specimen size than in the other Indo-Pacific genera.

The divergence of the haemal canal away from the
midline under most of the unfused abdominal vertebrae
is highly variable. There follows a listing of the course of
the divergence in the specimens closely examined for
it: Tetrosomus concatenatus, to the right of the mid-
line in 1 specimen; T. gibbosus, left in 1; Lactoria cor-
nuta, right in 4, left in 1, alternately right and left in 2,
almost straight in 1; L. fornasinii, right in 1, essentially
straight in 2; Ostracion lentiginosum, right in 4, left in 1;
0. tuberculatus, right in 5; Rhynchostracion rhinorhyn-
chus, right in 1; Acanthostracion quadricornis, right in 6,
left in 2; A. polygonius, right in 1, alternately right and
left in 1; A. guineensis, alternately right and left in 1; A.
notacanthus, right in 1; Lactophrys trigonus, right in 2,
left in 1, alternately right and left in 2; Rhinesomus
bicaudalis, left in 2; R. triqueter, left in 2, alternately
right and left in 2, almost straight in 1. Thus, in both the
Indo-Pacific species and in the Atlantic species, the
canal can be diverted to either the right or left, or al-
ternately to the right and left. If enough specimens were
examined it might be possible to show that there was a
greater frequency of right deviation in the Indo-Pacific
species and of left or alternately right and left deviation
in the Atlantic species, but such cannot yet be done.

The structure of the last few vertebrae shows interest-
ing variation, as described by Tyler (1970b) and briefly

Figure 166.— Number of teeth in upper (solid
symbols) and lower (open symbols) jaws in relation
to standard length, to show the slightly greater
number of teeth in the upper versus the lower jaw,
and the relatively negligible increase in number
of teeth with increasing standard length at sizes
greater than about 40 mm SL: Ostracion
tuberculatus, O. lentiginosum, Acanthostracion
quadricornis, A. polygonius.

229

%

/4^

Figure i~ I .—Rhinesomus triqueter: lateral
view of head, 134 mm SL, Colombia.

pterosphenoid \

porasphenoid
proolic \

\^

\^

'%f

ethmoid

tj^

porasphenoid X/

%^

palatine \

A

premaxillary (^--^T'^ A ^^

-■-:^' /

J

iupraocclpital
/ epiotic

suboperculum

hyomandibulo
preoperculum

metaplerygold
sympiectic
mesopterygoid

Figure 172.— Ostrocion tuberculatum:

lateral view of head, 54.5 mm SL,

Aldabra Atoll.

angular

meta pterygoid
symplectic
ectoDterygoid
quadrate
interoperculum

parasphenoid
posttemporol
pterotic
hyomandibular

Figure 173.— Tetrosomu*
concatenatua: lateral view of
head, 43.4 mm SL, Thailand.

Figure m.—Lactoria comuta: ventral
view of skull, 88.2 mm SL, Philippines.

Figure 175.— Lactoria fomaainii: posterior view of

orbit (cross section of skull; dashed lines
represent cut surfaces of frontals. pterosphenoids,
prootics, and parasphenoid), 65.2 mm SL, Hawaii.

summarized here. The most generalized condition of the
caudal skeleton is found in Lactophrys trigonus and the
two species of Rhine somus, the haemal spine of the
penultimate vertebra being relatively large and autog-
enous and the haemal canal passing through it and into
the caudal plate to exit at a foramen just below the
region of the centrum, the foramen marking the region of
fusion between what in more generalized plectognaths
are the parhypural and lowermost hypural. All three
species have three postanal vertebrae. The moderately
specialized condition of the caudal skeleton is found in
Lactoria cornuta and the four species of Acantho-
stracion, the haemal spine of the penultimate vertebra
being reduced in size and fused to its centrum, while the
haemal canal pierces and exits in the caudal plate just as
in Lactophrys and Rhinesomus. There are three post-
anal vertebrae in Lactoria cornuta, four in A. polygonius,
A. guineensis, and A. notacanthus, and five in A. quad-
ricornis. Since all ostracioids with the exception of A.
quadricornis have 18 vertebrae, and quadricornis 19, it is
assumed that 18 is the generalized number and that the
addition of a vertebra in quadricornis has taken place in
the postanal series.

The most specialized condition of the caudal skeleton
is found in Lactoria fomasinii, the two species of Ostra-
cion, the two species of Tetrosomus, and in Rhyncho-
stracion rhinorhynchus. In these species the haemal
arches and spines of all the postanal vertebrae are usual-
ly much reduced in size, solid, fused to the centra, and
not pierced by the haemal canal, which also does not
enter the foramenless caudal plate. Of these species, the
solid haemal spines are best developed in Ostracion
tuberculatus (illustrated here), the haemal spines being

shorter in 0. lentiginosum (Tyler 1963a:fig. 13, which in-
correctly shows nontrifid neural spines) and just as
rudimentary if not more so in Tetrosomus and Rhyncho-
stracion. There are four postanal vertebrae in Ostracion
and Rhynchostracion, and three in Lactoria and
Tetrosomus.

All species of ostraciids have 9 abdominal and 9 caudal
(10 caudal in A. quadricornis) vertebrae, and all have 7
predorsal vertebrae, with the postanal, flexibly ar-
ticulated, vertebrae varying from 5 to 3.

The postcleithrum is nearly always composed of two
pieces, both of which, but especially the ventral piece,
are relatively thin and delicate, only the anterior end of
the dorsal piece around its region of articulation with the
cleithrum being of much strength. Only in the single
specimens examined of the two species of Tetrosomus
was the division into two pieces not seen, and it may
have simply escaped notice in the posterior half of the
bone which tends to break away from the front portion
during dissection and hold fast to the carapace.

In all ostraciids the distal end of the first anal fin basal
pterygiophore is expanded into a pair of anterolateral
wings more or less surrounding the rear half of the anus
and supporting the carapace in this region. The carapace
has additional specialized supporting structures around
both the dorsal and anal fins in the Indo-Pacific species,
but not in those of the Atlantic. In addition to the trifid
neural spines of two or more of the postdorsal vertebrae
that broaden the base of support for the carapace in the
Indo-Pacific species, the last dorsal and anal fin basal
pterygiophores are also variously modified to help sup-
port the carapace.

Just above its distal end the last anal fin basal pteryg-
iophore is always laterally expanded into a shelf upon
which the carapace rests. In Lactoria cornuta the shelf is
mainly a simple lateral expansion, but in L. fomasinii
and all of the other Indo-Pacific species it is expanded
anteriorly as well as laterally into a pair of broad antero-
laterally directed prongs supporting the carapace.

The last dorsal fin basal pterygiophore, just below the
distal end, is variously expanded laterally and posterior-
ly into a supporting flange. In Ostracion, Rhynchostra-
cion, and Lactoria cornuta this basal portion of the last
dorsal fin basal pterygiophore is much expanded antero-
posteriorly and closely held between the stout neural
spines of the 11th and 12th vertebrae, while the distal
portion of the pterygiophore is expanded both laterally
and posteriorly into a shelf upon which the carapace
rests. The posterior expansion of the distal end of the
pterygiophore usually makes contact with the neural
spine of the 13th vertebra. In Lactoria fornasinii the dis-
tal end of the last dorsal fin basal pterygiophore is only
slightly expanded laterally and posteriorly into a sup-
porting shelf and the basal portion is rodlike, not ex-
panded anteroposteriorly. Moreover, the trifid neural
spines of the 14th to 16th vertebrae are not as massive in
L. fornasinii as in L. cornuta, and the neural spine of the
13th vertebra in L. fornasinii is long and slender and not
in contact with the last pterygiophore rather than mas-
sive and in contact with it, while that of the 12th verte-

bra is also much smaller than in L. comuta and only con-
tacts the basal half of the rodlike portion of the last
pterygiophore.

The last dorsal fin basal pterygiophore in the two
species of Tetrosomus is similar to that of L. fornasinii,
but the neural spines of the vertebrae below it are even
more reduced in size. In Tetrosomus the distal end of the
pterygiophore is only slightly expanded laterally and
posteriorly and the basal portion is rodlike, its extreme
basal end being supported variously by the long neural
spine of the Uth vertebra. The rest of the length of the
pterygiophore is not in contact with any other vertebrae,
mainly because the posterior extension of the distal end
is short and the neural spines of the 13th and 14th verte-
brae are extremely low. As in L. fornasinii, the trifid
neural spines of the 15th and 16th vertebrae in
Tetrosomus are not massive.

The degree of development of lateral flanges extend-
ing down from the neural spines across the lateral sur-
faces of the centra to end in broad transverse processes
extending out anterolaterally from the lower surface of
the centra is difficult to compare between species other
than relatively subjectively. The flanges are best
developed in Lactoria and Ostracion, moderately
developed in Tetrosomus and Rhynchostracion and only
slightly, if at all, less so in Rhinesomus, and poorly
developed in Lactophrys and, especially, Acantho-

stracion. Thus, there is a tendency for the Indo-Pacific
species to have the lateral flanges better developed than
in at least most of the Atlantic species.

Probably associated with the more bulbous snout of
Rhynchostracion is the fact that the ethmoid in this ge-
nus is wider and more massive than in all other genera.

Variation in the branchial arches of ostraciids is main-
ly confined to the presence or absence of minute and per-
haps nearly functionless teeth on the third pharyngo-
branchial. The second pharyngobranchial in all species
bears equally minute or only slightly larger teeth than
those of the third pharyngobranchial. The first pharyn-
gobranchial is a toothless suspensory element that is

Figure 176.— Rhynchostracion
rhinorhynchus: ventral (left) and dorsal
(right) views of skull, 88.2 mm SL, Java.

tyO CO basibranchials
l_J' —hypobranchials

branchiostegal

ventral hypohyal
ceratohyal

Figure ill .—Lactoria comuta: dorsal view of branch-
ial arches (extended on lower side); lateral view
of hyoid arch and urohyal; 88.2 mm SL, Philippines.

present in all species, with the exception that only a
minority of the specimens of Acanthostracion quad-
ricornis examined had a first pharyngobranchial present,
at least as an ossification, and that the first pharyngo-
branchial may also have been absent in the single
specimen of A. notacanthus examined, the specimen
having been poorly cleared and crudely dissected in this
region and the presence or absence of that pharyngo-
branchial now being impossible to decipher. In most of
the species examined the third pharyngobranchial is
toothless, but minute teeth are present on it in Lac-
tophrys trigonus, Tetrosomus concatenatus, and in most
but not all of the specimens examined of Acantho-
stracion quadricornis and Lactoria cornuta. Such varia-
tion is to be expected in what appear to be highly
rudimentary structures, and if more specimens of the
species in which no third pharyngobranchial teeth were
found had been examined, the suspicion is that at least
some of these species could be added to the list above of
those known to at least sometimes have teeth on this ele-
ment.

Generic relationships and comparative diagnoses
of subfamilies (Ostraciinae, Lactophrysinae).— Fra-

ser-Brunner (1941c) divided the ostraciids into two sub-
families, the Indo-Pacific Ostraciinae and Atlantic Lac-
tophrysinae, on the basis of the lower number of dorsal,
anal, and pectoral fin rays in the former and of supposed
vertebral differences between the two groups. Tyler
(1963a:185) showed that the vertebral differences men-
tioned by Fraser-Brunner did not stand close examina-
tion of a larger number of species than Fraser-Brunner
apparently had examined, but Tyler only hinted at a
few other vertebral characters that might eventually be
used to separate the two groups. These and additional
characters that distinguish the two subfamilies are dis-
cussed more fully in the preceding section on anatomical
diversity in the family.

The separation of the ostraciids into two phyletic lines
of subfamilial rank seems justified to me on the basis of
the following differences:

OSTRACIINAE

dorsal and anal fin rays modally 9
dorsal and anal fin basal pterygio-

phores modally 8
pectoral fin rays modally 10

four vertebrae involved in the fu-
sion complex at the rear of the
skull, except for one species with
five

two or more postdorsal vertebrae
with trifid neural spines

last anal fin basal pterygiophore
laterally or anterolaterally ex-
panded into a prominent fiange
for carapace support

last dorsal fin basal pterygiophore
moderately or greatly expanded
laterally and posteriorly for cara-
pace support

haemal spine of penultimate verte-
bra fused to the centrum, either
with a foramen or solid

myodome small and shallow, but
present

Indo-Pacific distribution.

LACTOPHRYSINAE

dorsal and anal fin rays modally 10

dorsal and anal fin basal pterygio-
phores modally 9

pectoral fin rays modally 12, ex-
cept one species with 11

five vertebrae involved in the fu-
sion complex at the rear of the
sliull

all vertebrae with unbranched
neural spines

last tinal fin basal pterygiophore
not expanded for carapace sup-
port

last dorsal fin basal pterygiophore
not expanded for carapace sup-
port

haemal spine of penultimate verte-
bra autogenous or fused to the
centrum, always with a foramen

myodome essentially absent

Atlantic distribution (to South
Africa).

If, as suggested, the two subfamilies represent separate
phyletic lines, it is assumed that the involvement of the
fifth vertebra in the anterior fusion complex of the Indo-
Pacific Lactoria fornasinii has taken place indepen-
dently by convergence from that in the Atlantic species,
and that the fusion of the haemal spine of the
penultimate vertebra to its centrum in the four species of
Atlantic Acanthostracion has been similarly indepen-
dent from that in the Indo-Pacific species. This, it seems
to me, is far easier to believe than that the entirely con-
stant differences: 1) the numbers of dorsal, anal, and
pectoral fin rays; 2) the structure of the neural spines of
two or more postdorsal vertebrae; 3) the structure of the
last basal pterygiophore of the dorsal and anal fins; and
4) the different distributions and myodome develop-
ment, between the two groups have no phylogenetic
meaning.

Generalized conditions in ostraciids can be considered
those which deviate the least from the structure of the
ancestral aracanids. Only two anterior vertebrae are fus-
ed in aracanids, so the involvement of only four vertebrae
in the fusion complex of all but one species of ostraciins is
less specialized than the involvement of five in the lacto-
phrysins. Aracanids have single unbranched neural
spines, as do lactophrysins, and the trifid neural spines of
some of the vertebrae in ostraciins can be considered as
specialization. Aracanids have no lateral or other expan-
sions toward the distal ends of the last dorsal and anal fin
basal pterygiophores for support of the carapace, nor do
the lactophrysins, and the expansions of these pterygio-
phores in ostraciins can be considered a specialization.
The higher numbers of dorsal, anal, and pectoral fin
rays, and of the dorsal and anal fin basal pterygio-
phores, in lactophrysins is more similar to that of
aracanids than are the lower numbers found in ostra-
ciins, the latter again more specialized. In aracanids the
haemal spine of the penultimate vertebra is autogenous,
while only among the lactophrysin ostraciids are there
species with the penultimate haemal spine autogenous,
the ostraciins again having the more specialized condi-
tion. Moreover, it is only among the species of ostraciins
that the penultimate and the preceding one or two
haemal spines are reduced in size, solid and without a
foramen for the haemal canal, the ostraciins again at
least tending to have a more specialized caudal fin sup-
porting apparatus than in the lactophrysins. The
moderate degree of lateral flange development across the
surface of the centra in aracanids is about intermediate
between that of the moderate to well-developed flanges
found in ostraciins and that of the moderate to poorly
developed flanges of lactophrysins.

In only two ways can ostraciins be considered more
generalized than lactophrysins. In aracanids and os-
traciins a small myodome is present, but this has been
essentially lost by lactophrysins. Aracanids and ostra-

Ostracion

Rhynchostracion

Acanthostracion

Eolactoria
(Eocene)

Figure 178.— Hypothesized phylogenetic relation-
ships of the genera of Ostraciidae.

ciins both have an Indo-Pacific distribution, while that
of lactophrysins is Atlantic.

In short, in most of the anatomical characters discuss-
ed above, with the notable exception of the anterior
vertebral fusion and mydome development, the lacto-
phrysins are seen as the more generalized of the two sub-
families. It seems best to assume that the lactophrysins
have remained more generalized than the ostraciins in
nearly all respects, except that the myodome became
reduced and the fifth vertebra became involved in the
anterior fusion complex in all species, while the generally
more specialized ostraciins retained from the same
generalized stock that gave rise to both subfamilies a
slightly better developed myodome and a less extensive
anterior vertebral fusion complex. In this view it is likely
that the Ostraciidae diverged into two lines in the Indo-
Pacific, with the more specialized ostraciins eventually
becoming dominant there to the exclusion of the lacto-
phrysins, which remained anatomically closer to the
ancestral stock, and were only permanently successful in
the Atlantic, in what is perhaps at least a partially relict
distribution.

Within the Lactophrysinae three genera were recogniz-
ed by Fraser-Brunner (1935b:317, 1941c:307), exclu-
sively on the basis of carapace characters, Acantho-
stracion for the species (quadricornis, polygonius,
guineensis, and notacanthus) with the carapace closed
behind the dorsal fin and with preorbital as well as
postanal carapace spines, Rhinesomus for the two species
(bicaudalis and triqueter) with the carapace closed
behind the dorsal fin, no preorbital carapace spines but
with the postanal spines present or absent, and Lacto-
phrys for trigonus, with the carapace open behind the
dorsal fin, no preorbital carapace spines but with
postanal spines. Since the presence of preorbital spines
seems to have no more magical phylogenetic quality than
the presence of postanal spines, one could with equal
simplicity group together those species with postanal
spines (Acanthostracion, Lactophrys, and Rhinesomus
bicaudalis) as distinct from R. triqueter, and within the
former group split off R. bicaudalis and L. trigonus from
Acanthostracion on the basis of the presence or absence
of preorbital spines, and further separate R. bicaudalis
and L. trigonus on the basis of whether the carapace is
closed behind the dorsal fin. A great many such com-
binations are as logically valid on the basis of carapace
characteristics alone. With this in mind, most workers
have not recognized all three genera as valid, variously
lumping them all together in one genus (Lactophrys) or
recognizing Lactophrys for the three species without
preorbital spines and Acanthostracion for the four
species with preorbital spines.

Several skeletal characters indicate the advisability of
recognizing two genera rather than one or three for the
Lactophrysinae. In Lactophrys and Rhinesomus the
penultimate haemal spine is autogenous while in
Acanthostracion it is fused to the centrum. In Lacto-
phrys and Rhinesomus there are three postanal (pos-
terior to the last vertebra supporting the anal fin)
vertebrae while in Acanthostracion there are four in three

species and five in one (quadricornis) , i.e., the 15th
vertebra helps to support the last anal fin basal pteryg-
iophore in Lactophrys and Rhlnesomus but not in Acan-
thostracion. In Lactophrys and Rhlnesomus suturing be-
tween adjacent vertebrae extends from the 5th and 6th
back to between the 12th and 13th, while in Acantho-
stracion it extends back to between the 13th and 14th
(erroneously stated by Tyler 1963a:171, to be between
the 12th and 13th in A. quadricornis). Lateral flanges
across the centra are slightly better developed in Lacto-
phrys and Rhlnesomus than in Acanthostracion.

On the basis of the above, only two genera should be
recognized in the Lactophrysinae, with Lactophrys =
Rhlnesomus being more generalized in all of the charac-
teristics listed above than Acanthostracion. Whether the
condition of having the carapace slightly open behind the
dorsal fin in L. trlgonus is a hold over from the ancestral
aracanids that gave rise to the ostraciids or is a case of
the subsequent reduction of the carapace in this region
from a more immediate ancestor with a complete bridge
across the back behind the dorsal fin is impossible to say
with assurance. In the only fossil ostraciid, the Eocene
Eolactorla, the carapace is complete behind the dorsal
fin. I suspect, especially in light of the fact that the
carapace is complete behind the dorsal fin at least oc-
casionally in large specimens of L. trlgonus just as it is in
Eolactorla and all other ostraciids, that the condition in
L. trlgonus is a secondary reduction of the posterior end
of the carapace and not the retention of a primitive con-
dition. It seems possible that L. trlgonus and R.
blcaudalis are slightly more closely related to one
another than to R. trlqueter, despite the behind the dor-
sal fin carapace similarity between blcaudalis and trl-
queter. In addition to blcaudalis and trlgonus both hav-
ing postanal carapace spines, both have the anterior
nostril bulbous, more so in blcaudalis and at all sizes,
and less so and only in large specimens of trlgonus. It
would be intriguing to know whether the anterior nostril
in trlqueter, which is a smaller species than the other
two, would become bulbous were it to reach sizes com-
parable to trlgonus.

Acanthostracion is probably derived from a Lacto-
phrys = Rhlnesomus-Vike stock, undoubtedly from a
form with the carapace closed behind the dorsal fin and
with postanal spines. Acanthostracion is specialized in-
ternally in the ways mentioned above, while the presence
of preorbital spines could indicate either that they are a
new development from the Lactophrys = Rhlnesomus-
like line which give rise to it or that the immediate
ancestry of the Lactophrys = Rhlnesomus line had preor-
bital spines which were lost by the Recent species of that
group but retained by the line leading to Acantho-
stracion. Within Acanthostracion, quadricornis is ob-
viously the most specialized form (the addition of 1
vertebra to the postanal series for a total of 5 in the series
and 19 for the entire column, both unique in the family,
and the total number of vertebrae unique in the super-
family; the modal reduction by one of the number of pec-
toral fin rays, unique in the subfamily; the frequent
presence of isolated scale plates above and below on the

caudal peduncle, unique in the family), and notacan-
thus, polygonlus, and gulneensls are more closely related
to one another than to quadricornis, with the rela-
tionship between notacanthus and polygonlus being es-
pecially close on the basis of the coloration, carapace,
and caudal fin shape characteristics discussed by Tyler
(1965b:275).

In the Ostraciinae the generic relationships, and which
genera are generalized versus specialized, are less clear
than in the lactophrysins. The rectangular carapace of
Ostraclon and Rhynchostraclon rhlnorhynchus can be
considered a specialization, being further removed from
the laterally compressed oval of aracanids and the high
crested triangle of lactophrysins and of Tetrosomus, with
the high crested but otherwise rectangular form of R.
nasus being a variant of the Ostraclon and R. rhlno-
rhynchus form and the pentangular carapace of Lactorla
a low crested variant of the triangular form.

In the caudal skeleton the presence of a haemal canal
piercing the more posterior haemal spines and entering
the caudal plate, as found only in Lactorla cornuta
among the ostraciins, is clearly a relatively generalized
condition in comparison to the highly specialized solid
postanal haemal spines, usually reduced in size and
never pierced by the haemal canal, which also does not
enter the caudal plate, found in L. fornaslnll, Tetro-
somus, Ostraclon, and Rhynchostraclon. The reduction
in size of several of the neural spines below the last dor-
sal fin basal pterygiophore, so that the pterygiophore is
supported only basally by a neural spine, as found in L.
fornaslnll, and, especially, Tetrosomus, seems special-
ized in comparison to the more normal neural spines in
this position found in L. cornuta, Ostraclon, and Rhyn-
chostraclon. However, the relatively small size and poor
development of shelves supporting the carapace on this
pterygiophore in L. fornaslnll and Tetrosomus are much
closer to the condition found in the lactophrysins and
aracanids, and must be considered either more general-
ized than or a secondary reduction in size from the far
more massive last dorsal fin basal pterygiophore with
well-developed specialized shelves found in L. cornuta,
Ostraclon, and Rhynchostraclon.

The only moderately developed lateral flanges across
the centra in Tetrosomus and Rhynchostraclon would
seem slightly less specialized than the larger flanges of
Lactorla and Ostraclon. The relatively small lateral shelf
for carapace support on the last anal fin basal pterygio-
phore in L. cornuta seems the least removed from the
condition in lactophrysins and aracanids and thus less
specialized than the much larger anterolateral shelf
found in L. fornaslnll, Tetrosomus, Ostraclon, and
Rhynchostraclon.

The number of postanal vertebrae, three in Lactorla
and Tetrosomus and four in Ostraclon and Rhyncho-
straclon, can be compared only to the condition in lacto-
phrysins. In aracanids there is a far less close association
of the last (and other) anal fin basal pterygiophore with
the haemal spines of the caudal vertebrae (with only the
proximal end of the element at all associated with a par-
ticular haemal spine instead of most of its posterodorsal

edge closely in contact with several vertebrae) which
leads to incomparably high postanal counts. In lacto-
phrysins the more generalized group, Lactophrys =
Rhinesomus, have three postanal vertebrae and the more
specialized group, Acanthostracion, four, while Ostra-
cion and Rhynchostracion would thus be considered to
have a more specialized number of postanal vertebrae.
Trifid neural spines on some of the postdorsal
vertebrae being a specialization found in all ostraciins, it
seems reasonable to assume that the greater the number
of vertebrae with trifid neurals, the greater the degree of
specialization in this character. The greatest number,
four, is found in L. cornuta, with L. fornasinii having
three, and Tetrosomus, Ostracion, and Rhynchostracion
either two or three.

While the number of vertebrae involved in the fusion
complex with the rear of the skull in all species of the
subfamily is four, except five in L. fornasinii, the lack of
fusion between the fused first-second vertebrae and the
fused third-fourth vertebrae in 0. tuberculatus even in
large adults and of the sometimes only partial fusion in
this region in 0. lentiginosum seem more generalized
than having at least the basal regions of the four
elements fully fused, as in Rhynchostracion, Lactoria,
and Tetrosomus.

As seen above, no one genus has a relatively full com-
plement of either the specialized or generalized charac-
teristics discussed, all of the genera having a difficult to
interpret mixture of both. What follows is a highly specu-
lative interpretation of the evidence.

The probably triangularly or pentagonally carapaced
ancestral group of the Ostraciidae had no more than four
vertebrae involved in the anterior fusion complex, the
haemal spine of the penultimate vertebra autogenous,
the last dorsal and anal fin basal pterygiophores without
shelves for carapace support, and the neural spines of the
more posterior vertebrae undivided. This ancestral group
gave rise on the one hand to the Lactophrysinae, which
remained relatively unchanged from the ancestral condi-
tion except for subsequently involving the fifth vertebra
in the fusion complex and having the penultimate
haemal spine become fused to the centrum in the most
specialized genus, in which the number of postanal
vertebrae was also increased. On the other hand the
ancestral group gave rise to the Ostraciinae, which
retained the ancestral condition of the anterior vertebral
fusion complex but which specialized in a number of
ways mostly involving the simplification of the postanal
haemal spines and of the development of carapace sup-
porting structures on the last dorsal and anal fin basal
pterygiophores and on some of the neural spines of the
postdorsal vertebrae.

The line immediately ancestral to the Recent ostra-
ciins is envisioned as having the last dorsal and anal fin
basal pterygiophores moderately well developed (i.e.,
larger than in the lactophrysins), with moderate shelves
(probably less well developed than in their greatest
development in the Recent species), and with only two or
three of the postdorsal vertebrae with trifid neural
spines. This hypothetical ancestral line leading from the

immediate ancestry of the lactophrysins would probably
have had a triangular or pentangular carapace, and it is
postulated to have diverged into two lines of Recent
species, one line containing basically triangular forms
with many carapace spines {Lactoria and Tetrosomus)
and the other basically rectangular forms with a rela-
tively spineless carapace (Ostracion and Rhyncho-
stracion).

In the Lactoria- Tetrosomus line of ostraciins, a form
close to L. cornuta, but with a still only moderately
enlarged last dorsal fin basal pterygiophore and only two
or three trifid neural spines, was probably ancestral, L.
cornuta being the only species to have the haemal canal
pierce the fused haemal spine of the penultimate
vertebra and enter into and exit from the caudal plate
and to have a relatively moderate lateral shelf on the last
anal fin basal pterygiophore. The L. cornuta-like (with
the exceptions noted) ancestral line probably gave rise to
L. cornuta on the one hand by the further enlargement of
the last dorsal fin basal pterygiophore and its shelf and
by the trifid division and enlargement of additional post-
dorsal neural spines.

On the other hand the pre-L. cornuta ancestor is seen
as having given rise to L. fornasinii with no increase in
the number and size of trifid neural spines, the reduction
of the size of the haemal arches of the postanal vertebrae,
and the elimination of their foramina so that they are
solid and not pierced by the haemal canal which also
does not pierce the caudal plate, an enlargement of the
moderate shelf on the last anal fin basal pterygiophore
into a more substantial anterolateral shelf, a reduction in
the size of the neural spine of the vertebra (13th) between
that which supports the base of the last dorsal fin basal
pterygiophore and the most anterior of the three
vertebrae (14th to 16th) with trifid neural spines, and a
slight reduction in the size of the last dorsal fin basal
pterygiophore and its posterior carapace shelf so that the
basal portion of the pterygiophore is rodlike rather than
anteroposteriorly expanded while the posterior end of the
distal shelf does not make contact with any neural
spines.

Probably a close derivative of the immediate ancestor
of L. fornasinii is Tetrosomus. Tetrosomus shows the
same tendencies, but more strongly, as L. fornasinii does
in the reduction of the size of the neural spines below the
last dorsal fin basal pterygiophore and in the size of the
latter. In Tetrosomus the size of the neural spines of the
12th to 14th vertebrae is greatly reduced, and that of the
14th is no longer trifid. The trifid neural spines of the
15th and 16th vertebrae are of about the same moderate
size as in L. fornasinii, the anterolateral shelf on the last
anal fin basal pterygiophore just as massive, and the
solid haemal spines of the postanal vertebrae just as
reduced in size and not pierced by the haemal canal,
which likewise does not enter the caudal plate. Thus, L.
fornasinii shares many peculiarities with Tetrosomus,
and a form like it is very likely to have been ancestral to
Tetrosomus. The higher crested carapace in Tetrosomus
could be either an enlargement of the lower crest in Lac-
toria or a hold over from the line leading from the pre-

fornasinii-like ancestor, which may have been higher
crested than the Recent Lactoria.

In the Ostracion-Rhynchostracion line of ostraciins
derived from a probably pentangular form, there seems
to have been a tendency for the distance between the
ventrolateral ridges to be decreased to about the same as
that between the dorsolateral ridges, while the dorsal
carapace crest tended to be reduced in height, leading to
a basically rectangular form with overtones of the
ancestral pentangular form depending on the degree of
development, if any, of the dorsal crest. Ostracion and
Rhynchostracion probably evolved from the same pre-
Lactoria comuta-\ike line that gave rise to Lactoria and
Tetrosomus. The Ostracion-Rhynchostracion line is en-
visioned as diverging from them in the shape of the
carapace and in the loss of spines from the probably
spiny carapaced ancestor, while developing large last
basal pterygiophores with large carapace supporting
shelves in both the dorsal and anal fins and reducing the
size of the haemal spines of the four postanal vertebrae,
with the haemal spines losing the foramen and being
solid.

Ostracion shows the extreme in dorsal carapace crest
reduction and increased distance between the dorso-
lateral ridges, the carapace being essentially rectangular,
while it retains the most generalized ostraciid condition
of the anterior vertebral fusion complex. In Rhyncho-
stracion, however, one species {nasus) retains a rela-
tively well-developed dorsal carapace crest while in the
other (rhinorhynchus) it is obsolete, the carapace being
almost as rectangular as in Ostracion, with both species
of Rhynchostracion having the anterior four vertebrae
fully fused to one another. Rhynchostracion has the addi-
tional slight specialization of a bulbous or protruding

snout. Ostracion and Rhynchostracion can be viewed as
closely related sibling genera from a common ancestral
group, with Rhynchostracion slightly more specialized in
the greater anterior vertebral fusion and expanded
ethmoid region while retaining a slightly more general-
ized carapace shape.

The Eocene Eolactoria is impossible to place with con-
fidence in either one or the other of the two subfamilies
recognized here for the Recent species, simply because its
dorsal, anal, and pectoral fin rays are not preserved and
none of the internal features equally diagnostic between
the two subfamilies can be seen (Tyler 1973a). The
presence of enormous preorbital and postanal carapace
spines in Eolactoria would seem to relate it to either the
Atlantic Acanthostracion (Lactophrysinae) or the Indo-
Pacific Lactoria (Ostraciinae), from both of which it
differs in the immensity of the elongation of the spines
(equal to the standard length of the fish) and in the
presence of a much shorter unpaired spine directed
anteriorly from between the front of the eyes, the latter
unique in the family. Without knowing the appropriate
critical diagnostic characteristics, Eolactoria is simply
placed here questionably in the Lactophrysinae, the
more generalized of the two subfamilies, on the theory
that an Eocene ostraciid is more likely to be generalized
in most respects than specialized. It is possible that

FifTure i79.— Eolactoria aorbinii: lateral view

of holotype, 15.5 mm SL, Eocene of Monte Boica,

ItalyCIVlerigrSaifig. 1).

242

Eolactoria is a representative of the line which gave rise
to Acanthostracion on the one hand and to Lactophrys =
Rhinesomus on the other, with the latter variously losing
the preorbital and postanal carapace spines, and that
another more specialized derivative of the Eolactoria to
Acanthostracion line gave rise to that of Lactoria and
Tetrosomus on the one hand and to that of Ostracion and
Rhynchostracion on the other, with the latter two genera
also losing the preorbital and postanal carapace spines.
Since all ostraciids have the ventral edge of the para-
sphenoid laterally expanded into a roof over the oral
cavity anterior to the level of the prefrontals, they can be
expected to have evolved from one of the two lines of
aracanids with a similar but less extensive expansion of
the ventral edge of the parasphenoid, either the Aracana
line with the expansion limited to a posterior region or
the Capropygia-Caprichthys line with the expansion
limited to an anterior region. However, it is impossible to
say with any confidence which one of these two lines was
more likely ancestral to the ostraciids (see discussion un-
der Aracanidae).

SUBORDER TETRAODONTOIDEI

(GYMNODONTES)

Comparative diagnosis (contrast with that of the
Balistoidei). —Teeth either small rounded units or long
rodlike structures, but always nonprotruding and fully
incorporated into the matrix of the jaw bones, in one
family indistinguishably so, the otherwise discrete teeth
in molids no longer distinguishable from the bony
matrix, at least at 30 magnifications; dentaries and/or
premaxillaries often fused to their opposite members;
the medial articulation of the premaxillaries to one
another, if unfused, strengthened by alternating emar-
ginations and indentations, least developed in Triodon,
the most generalized member of the suborder; post-
temporal absent; urohyal absent, except in Triodon;
pelvis absent, except in Triodon; pelvic fin always ab-
sent; palatine relatively large and massive, always firmly
sutured or otherwise closely and immovably held to both
the ethmoid-vomerine region and the pterygoid arch;
myodome with a complete dorsal roof absent, except in
Triodon; prootic shelf under the orbit never present;
supracleithrum always placed distinctly obliquely to the
axis of the skull; scapular foramen incomplete, except in
Triodon; scapula without any special knob or crest for ar-
ticulation with the uppermost pectoral fin ray, except in
Triodon; distal pterygiophores of the soft dorsal and anal
fins always unossified; spiny dorsal fin absent, except
present as a rudiment placed far behind the skull in
Triodon; first I ^ranchiostegal ray modified, with a slight-
ly (Triodon) to enormously enlarged and inturned dorso-
medial edge, except unmodified in molids.

Figure 180.— Body form in the only known Recent
species of Triodontidae: Triodon macroptenis.

Infraorder Triodontoideo

Comparative diagnosis (contrast with that of the
Tetraodontoideo), which is also that of its only con-
tained Superfamily, the Triodontoidea, and family, the
Triodontidae. — A rudimentary spiny dorsal fin of one or,
more usually, two small spines borne on two basal pte-
rygiophores present in most specimens of one of the
populations (Indonesia to Japan) of the single included
Recent species, the second basal pterygiophore succeed-
ed by two supraneural elements; ribs and epipleurals pre-
sent; caudal fin with 12 principal rays and numerous
procurrent rays; caudal fin deeply forked; caudal
skeleton with four separate hypurals, two uroneurals,
and a hypurapophysis; haemal spine of antipenultimate
vertebra autogenous; neural and haemal spines of penul-
timate vertebra long rounded shafts oriented highly
obliquely and directly supporting caudal fin rays; caudal
peduncle distinctly tapered to narrow transversely in-
dented regions above and below just in front of the
caudal fin, the least depth of the peduncle being about
3% SL and the least width at this region always greater
than the least depth; many of the caudal vertebrae with
large anterolateral processes from the region of the neural
arch and base of the neural spine; pelvis present; a huge
expansible dewlap of skin between the end of the pelvis
and anus, but no inflatability of the abdominal region;
cleithrum greatly elongate anteriorly, reaching forward
to between the lower jaw; basisphenoid a small rod plac-
ed far posteriorly in the interorbital septum and articu-
lated with the anterior edge of the dorsal roof of the
myodome; a large myodome present with a complete dor-
sal roof; a shallow channel present between the apposed
surfaces of the parasphenoid and basioccipital leading
into the myodome; scapular foramen complete, entirely
enclosed by the scapula; scapula with a distinct knob for
articulation with uppermost pectoral fin ray; four ac-
tinosts, none of which are sutured to one another or to the
scapula or coracoid; urohyal present; four pharyngo-
branchials present; fifth ceratobranchial well-covered
with prominent teeth; dentaries and premaxillaries
totaling three separate pieces; premaxillaries articu-
lated to one another by interdigitation or by only minute

243

more regular interlocking emarginations; sphenotic rela-
tively small and confined to the posterior wall of the or-
bit, not reaching the lateral or dorsal surface of the skull;
first branchiostegal ray with its dorsal edge slightly in-
tumed, articulated with the ventrolateral surface of the
ceratohyal; interoperculum with a ventral flange and a
short posterior shaft extending only slightly behind the
ventral flange and level of the epihyal; pterotic prolonged
posteroventrally as a stout laterally compressed shaft
broadly articulating with the hyomandibular and supra-
cleithrum; epiotic confined to the lateral region of the top
of the skull, separated from the supraoccipital by the
exoccipital; exoccipital in contact with the frontal; fron-
tal in contact posteriorly with the pterotic in the rear of
the orbit; olfactory epithelium in the form of a rosette,
the lamellae radiating out from a horseshoe-shaped base,
the olfactory sac below the surface of the body and the
two nostrils more or less flush with the surface, the
anterior nostril with an upraised flap posteriorly and the
posterior nostril with an upraised flap anteriorly.

Detailed description of Triodon macropterus.

Material examined. — Two cleared and stained
specimens, 391-463 mm. It will be advantageous in the
future to compare the amount of interdigitation in the
skull structure of these two large adult specimens with
that in a young specimen, but the species as yet is known
only from several dozen relatively large (about 285-480
mm, see Tyler 1967:90) adults. This species has often
been called T. bursarius, but Boeseman (1962) has thor-
oughly reviewed the literature £md shown T. macropterus
to have priority.

SKULL.

Occipital Region.

Basioccipital. — A very short column, slightly
expanded anterodorsally; cartilage filled at its anterior
ledge; articulates by extensive interdigitation anteriorly
with the overlying parasphenoid, anterolaterally with the
prootics, and posterolaterally with the exoccipitals. A
slight depression is present medially on the ventral sur-
face of the anterior end of the basioccipital just behind
the region where it is overlain by the parasphenoid. This
depression continues anteriorly as a narrow canal
between the otherwise interdigitated surfaces of the
basioccipital and parasphenoid, and opens into the ex-
treme posterior end of the myodome. The extreme pos-
teroventral portion of the wall of the myodome is formed
by the anterodorsal end of the basioccipital. The rim of
the round concave posterior end of the basioccipital ar-
ticulates by fibrous tissue with the rim of the concave
anterior face of the centrum of the first vertebra.

Exoccipital. —Extremely large; cartilage filled
along its dorsal and dorsolateral edges; articulates by ex-
tensive interdigitation dorsomedially with the supra-
occipital, anterodorsally with the frontal, dorsolaterally

Figure 181. — Triodon macropterus: showing the course

of the lateral line; lower left, the nasal region as

seen externally (above) and the olfactory lamellae

as seen with the top of the nasal sac removed; lower

right, scales from upper middle region of body,

including three lateral line canal bearing scales.

with the epiotic and pterotic, ventrolaterally with the
prootic, and ventromedially with the basioccipital. The
posteromedial edges of the two exoccipitals form the ven-
tral and lateral walls of the foramen magnum, while dor-
sally the wall of the foramen is formed mostly by the
exoccipitals, but also for a short distance medially by the
ventral edge of the supraoccipital. Strong condyles pro-
ject from the posteromedial portions of the exoccipitals
and make fibrous tissue contact over the anterior half of
the lateral surfaces of the centrum of the first vertebra.

Supraoccipital. —Somewhat expanded anteriorly,
but drawn out into a short, stout, laterally compressed
spine posteriorly; cartilage filled along all of its edges of
articulation with the other cranial bones, except pos-
teroventrally; articulates by interdigitation anteriorly
and anterolaterally with the frontals and postero-
laterally with the exoccipitals.

Otic Region.

Pterotic. — Very large; cartilage filled along its
medial edge; articulates by interdigitation anterodor-
sally with the epiotic, anteroventrally with the
sphenotic, ventromedially with the prootic, and postero-
medially with the exoccipital. Posteroventrally the
pterotic is prolonged into a stout, laterally compressed
shaft which articulates by fibrous tissue along its ante-
rior edge with the hyomandibular and along its posterior
edge with the supracleithrum. The articulation with the
supracleithrum appears to also involve some interdigi-
tation in the smaller specimen, but not in the larger; in
neither case is the articulation very flexible. The more
normal articulation of the pterotic with the supracleith-
rum is through fibrous tissue along the slightly concave
ventral surface of the extreme posterodorsal region of the
pterotic. Just anterior to the base of its stout shaft the
pterotic is deeply concave to receive and articulate by fi-
brous tissue with most of the length of the dorsal end of
the hyomandibular.

Sphenotic. — Relatively small and confined to the
posterior wall of the orbit; cartilage filled along all of its
edges of articulation with the other cranial bones; ar-
ticulates by interdigitation anterodorsally with the fron-
tal, posterodorsally with the pterotic, anteroventrally
with the pterosphenoid, and ventrally with the prootic.
The ventral edge of the sphenotic is deeply concave for
articulation by fibrous tissue with the anterolateral
region of the dorsal head of the hyomandibular. This con-
cavity of the sphenotic is continuous with that along the
lateral edge of the prootic, which supports the antero-
medial region of the head of the hyomandibular, and with
that of the pterotic forming the main support for that
head.

Epiotic. — Relatively large, but thin; cartilage filled
along its medial edge; articulates by interdigitation
anteriorly with the frontal, medially with the exoccipital
and posteriorly with the pterotic.

Prootic. — Cartilage filled along all of its edges of
articulation with the other cranial bones, except anteri-
orly; forms all of the walls of the myodome, except for the
medial portion of the ventral wall and the short section of
the anterior wall formed by the parasphenoid and for the
extreme posterior portion of the wall formed by the
basioccipital; articulates by interdigitation anteriorly
with the dorsolateral wing of the parasphenoid, dorso-
medially with the pterosphenoid, dorsolaterally with the
sphenotic, ventrolaterally with the pterotic, ventro-
medially with the parasphenoid and basioccipital, and
posteriorly with the exoccipital. From its oblique medial
surface the prootic gives rise to the horizontal plate of
bone which meets its opposite member in the midline to
form the dorsal roof of the myodome. These two projec-
tions interdigitate with one another in the midline and
each possesses a small foramen laterally near its dorsal
edge. Another pair of foramina are present in the ventral
wall of the myodome to either side of the midline, where
there is a slight gap in the interdigitated surfaces of the
prootics and the parasphenoid. Anteromedially from the
dorsal roof of the myodome the prootics support by fi-
brous tissue the posterior end of the basisphenoid. At the
concavity along its lateral edge the prootic articulates by
fibrous tissue with the anteromedial region of the dorsal
head of the hyomandibular.

Orbital Region.

Frontal. — Large and massive; laterally expanded
posteriorly but tapering to a bluntly rounded end ante-
riorly. Throughout its entire length the medial edge of
the frontal is broadly, but not deeply, concave and covers
a core of cartilage that is continuous anteriorly with the
cartilage of the ethmoid region and posteriorly with that
of the occipital region. The frontal articulates by inter-
digitation dorsomedially in the posterior half of its length
with its opposite member, while directly postero-
medially it interdigitates with the supraoccipital.
Posterolaterally on its dorsal surface it interdigitates

with the exoccipital and epiotic. Posteroventrally it in-
terdigitates with, from anterior to posterior, the ptero-
sphenoid, sphenotic, and pterotic. The ventromedial
edges of the two frontals are in close fibrous tissue con-
tact for most of their lengths, diverging from one another
only posteroventrally just in front of the pterosphenoids
and anteriorly just behind the prefrontals. Anteriorly the
frontal broadly overlies and slightly interdigitates with
the ethmoid, while anterolaterally it interdigitates with
the dorsomedial edge of the prefrontal.

Prefrontal. — Relatively large, both dorsally and
ventrally; cartilage filled along all of its ventromedial
edges where it is continuous with the ethmoid cartilage;
articulates dorsally by interdigitation with the frontal and
the posterolateral edges of the ethmoid, while postero-
ventrally it articulates through cartilage and fibrous tis-
sue with the anterodorsal surface of the parasphenoid.
Along most of its ventral edge the prefrontal articulates
through cartilage and interdigitation with the long pos-
terior portion of the palatine. In about the middle of its
medial surface the prefrontal is deeply concave, so that a
horizontal canal is formed between the prefrontal and
the ethmoid cartilage, with which the prefrontal is
otherwise continuous. The nerves and blood vessels that
course through this canal run not only to the olfactory
organ but also to the anteriormost regions of the head by
way of a similar canal between the palatine and ethmoid.

Parasphenoid. — Elongate, with a thin ventral keel
throughout most of its length. The anterior end of the
parasphenoid is deeply concave to accommodate the
posterior end of the vomer, with which it is interdigi-
tated. The parasphenoid also interdigitates antero-
laterally with the posterior portion of the palatine, while
anterodorsally it articulates through cartilage and fi-
brous tissue with the prefrontal. Posteriorly the para-
sphenoid somewhat overlies and strongly interdigitates
with the basioccipital, while at the same time leaving
open the small canal in the midline between their
otherwise interdigitated surfaces. About five-sixths of
the way back in its length, the parasphenoid gives rise to
its paired dorsolateral wings which interdigitate with the
anteromedial edges of the prootics and form a section of
the anterior wall of the myodome. On its ventral surface
between the level of its dorsolateral wings and its
meeting with the basioccipital, the lateral edges of the
parasphenoid interdigitate with the ventromedial edges
of the prootics, and in so doing the parasphenoid forms
the medial portion of the ventral wall of the myodome.
At the posterior end of its ventral keel the parasphenoid
becomes thickened and posterolaterally expanded into a
bony knob to which the distal end of the first (suspen-
sory) pharyngobranchial and associated muscles are at-
tached.

Pterosphenoid. — A thick plate; broadly cartilage
filled along its dorsal and lateral edges; articulates by in-
terdigitation dorsally with the frontal, posterolaterally
with the sphenotic, and posteroventrally with the prootic.

Basisphenoid. — Laterally compressed throughout
its length, except posteriorly where it is laterally expand-
ed into a rounded process which articulates by fibrous
tissue with the anterior edge of the dorsal roof of the
myodome in the smaller specimen, while in the larger
specimen the posterior end is attached more dorsally to
the membranous wall between the two pterosphenoids.
From its posterior place of attachment to the prootics,
the basisphenoid is directed anteroventrally and ends in
the fibrous tissue sheet of the interorbital septum. This
anterior edge, which is slightly cartilage filled, is some
distance removed from the dorsal surface of the para-
sphenoid below it.

culum and anteriorly with the symplectic and interhyal,
as well as with the slightly overlying posterior edge of the
metapterygoid.

Quadrate. — Widest posteriorly, tapering to a knob
anteriorly for articulation by fibrous tissue with the ar-
ticular in the lower jaw; cartilage filled at its posterior
edge; prolonged from its posteroventral edge into a thin
process which articulates by fibrous tissue with the
symplectic; articulates by fibrous tissue anterodorsally
with the ectopterygoid, ventrally with the preoperculum,
posteriorly with the metapterygoid, and posterodorsally
with the mesopterygoid.

Ethmoid Region.

Ethmoid. —Large and wide; relatively thin in the
posterior half of its length, but becoming much deeper
anteriorly; the posterior half of its ventral surface
irregularly concave and continuous with the ethmoid car-
tilage; articulates posterodorsally by interdigitation with
the overlying frontals and posterolaterally through car-
tilage and interdigitation with the prefrontals. Antero-
laterally the ethmoid is extensively interdigitated with
the medial edges of the palatines. The medial region of
the anteroventral surface of the ethmoid is extensively
interdigitated with the dorsal surface of the vomer in the
smaller specimen, but only lightly interdigitated to it in
the larger. The more or less vertical anterior face of the
ethmoid supports the posteromedial ends of the premax-
illaries and maxillaries. The lateral surface of the anteri-
or one-third of the ethmoid is concave, so that when it
articulates laterally with the equally concave medial sur-
face of the palatine a spacious canal is left between the
two bones. Through this canal run the nerves and blood
vessels from the similar canal between the medial surface
of the prefrontal and lateral surface of the ethmoid car-
tilage.

Vomer. — A narrow shaft throughout its length,
except anteriorly where it becomes somewhat expanded
dorsally to articulate by fibrous tissue with the postero-
medial arms of the premaxillary and maxillary; articu-
lates by interdigitation posteriorly with the concave
anterior end of the parasphenoid, laterally with the
palatines, and anterodorsally with the ethmoid. Postero-
dorsally the ethmoid cartilage intervenes between the
dorsal surface of the vomer and the ventral surface of the
ethmoid.

Mandibular Region.

Hyomandibular. — Very wide for most of its length,
only tapering to a stout shaft for a short distance antero-
ventrally; cartilage filled at its dorsal and anteroventral
edges; articulates by fibrous tissue dorsally with the
elongate concavity formed anterolaterally by the
sphenotic, anteromedially by the prootic, and pwsteri-
orly by the pterotic. The hyomandibular articulates by
fibrous tissue along its posterior edge with the preoper-

Metapterygoid. — Large, thick, more or less square;
cartilage filled at its anterior edge; articulates by fibrous
tissue anterodorsally with the mesopterygoid, anteriorly
with the quadrate, and posteriorly with the hyomandib-
ular. Ventrally the metapterygoid slightly overlies and
interdigitates with the symplectic and also articulates by
fibrous tissue with the interhyal.

Symplectic. — Large and elongate; cartilage filled at
its posterior edge; articulates dorsally by slight inter-
digitation with the overlying metapterygoid, while it ar-
ticulates by fibrous tissue posteroventrally with the
preoperculum and interhyal, and anteroventrally with
the overlying quadrate.

Palato-Pterygoid Region.

Palatine. — Expanded into a thick plate postero-
ventrally, but becoming laterally expanded anterodor-
sally; articulates by firm interdigitation along the ante-
rior half of its dorsomedial edge with the ethmoid and
along the anterior half of its ventromedial edge with the
vomer. The palatine is equally firmly interdigitated pos-
terodorsally with the prefrontal and posteroventrally
with the parasphenoid. The medial surface of the upper
region of the anterior half of the palatine is concave, so
that a canal is formed between it and the ethmoid, as de-
scribed previously. The medial and anterior surface of
the anterodorsal end of the palatine forms the articular
facet for support through fibrous tissue of the lateral sur-
face of the posteromedial wing of the maxillary.

Ectopterygoid. — Expanded posteriorly; only
slightly curved along its anterior edge; articulates by fi-
brous tissue dorsally with the palatine and ventrally with
the quadrate, while posteriorly it articulates by fibrous
tissue and slight interdigitation with the mesopterygoid.
The palatine and quadrate slightly overlie the ec-
topterygoid, while the ectopterygoid slightly overlies the
mesopterygoid.

Mesopterygoid. — Thin; irregular in its posterior
and ventral outline; articulates by slight interdigitation
anteroventrally with the ectopterygoid and by fibrous
tissue anteriorly with the palatine, ventrally with the
quadrate and posteroventrally with the metapterygoid.

246

Opercular Region.

Operculum. — Thin and expanded ventrally; a
dorsally directed process for muscle attachment present
above its articular region with the preoperculum; articu-
lates ventrally by fibrous tissue with the suboperculum,
which it broadly overlies, while along its upper anterior
edge the operculum is slightly expanded laterally into
the articular facet whose concave face attaches by fi-
brous tissue to the preoperculum.

Suboperculum. — Thin and delicate in its regions
that lie behind and below the operculum, but slightly
thicker at its anterodorsal end; articulates by fibrous tis-
sue dorsally with the overlying operculum, while at the
end of its emterodorsal process it connects by a ligament
with the upper prong at the posterior end of the in-
teroperculum.

Interoperculum. — A long sturdy rod for all of its
length, except posteriorly where it bifurcates into two
thin prongs. The more ventral of these two prongs lies in
the fibrous tissue sheet between the branchiostegal rays
and the region of the operculum and suboperculum. The
dorsal prong connects by ligament with the anterodorsal
process of the suboperculum. Anteriorly the interoper-
culum connects by ligament with the angular in the
lower jaw. Just in front of its ventral prong the medial
surface of the interoperculum is held by fibrous tissue to
the area of articulation between the epihyal and in-
terhyal.

Preoperculum. — Relatively thin and narrow
throughout its length, being only slightly expanded ven-
trally in its middle region in the smaller specimen, but
about one-third wider in the larger specimen; articu-
lates by fibrous tissue along the posterior half of its dor-
sal edge with the hyomandibular and along the anterior
half of its dorsal edge with the quadrate and the region of
the symplectic, metapterygoid, and interhyal. Along the
upper portion of its posterior edge the preoperculum
possesses a slight convexity for articulation through fi-
brous tissue with the operculum.

Upper Jaw.

Premaxillary. — Posteromedial arm thick and
sturdy; together with the fused teeth forms a massive
crushing plate; its anterior edge forming the anterior
border of the upper jaw, except for a short distance ven-
trally where the maxillary forms the border. The dorso-
medial surfaces of the two premaxillaries are firmly held
to one another by fibrous tissue and extensive interdigi-
tation, which, especially in the larger specimen, takes
the form of numerous small projections from the
otherwise flattened medial surfaces of apposition of each
premaxillary alternating with one another. The postero-
lateral surface of the premaxillary is strongly interdigi-
tated with the overlying maxillary. The premaxillary

contains an elongate internal cavity which houses the
dental pulp and which communicates with the exterior
by two openings on the posterior edge of the interdigi-
tated surfaces of the premaxillary and maxillary, as well
as at numerous smaller openings along the lateral surface
of the tooth bearing region, especially dorsally. The pos-
teromedial arm of the premaxillary articulates by fi-
brous tissue posteriorly with the vertical anterior face of
the ethmoid. The fused teeth of the jaws retain much of
their individual identity and are particularly clearly seen
at the edge of the jaw. About 25 to 30 of these small den-
tal units, with the rounded faces oriented toward the
biting edge of the jaw and the apparently slightly con-
cave edges oriented toward the pulp cavity, can be seen
packed closely together at the edge of the jaw, being held
immovably in place by an extremely dense bony matrix.
How much fusion actually takes place between the tooth
elements themselves and between the tooth elements
and the bony matrix of the premaxillary could only be
told by a histological examination not attempted here.
The tooth elements become less and less distinctly seen
at the surface further away from the edge of the jaw, ex-
cept at the extreme dorsal edge of the tooth bearing
lateral surface, where the primordia of the individual
teeth can be seen forming in the small pockets in the
matrix that open by pores to the exterior on the lateral
surface. The primordia obviously move gradually toward
the distal crushing edge of the jaw and become more fully
impacted with the matrix of the premaxillary as new
primordia form behind it. There are about 10 to 15 such
small teeth in various stages of incorporation with the
matrix between the generative region at the inner edge of
the tooth bearing surface and the crushing outer edge of
the jaw. In the medial region of its ventral surface the
premaxillary bears a massive trituration plate formed of
fused teeth. The process by which this plate is formed on
the undersurface of the premaxillary can be seen at the
posterior edge of the plate. Four or five anteroposterior^
compressed teeth are formed in deep sockets at the pos-
terior edge of the plate and are clearly evident as entirely
individual and separate units held relatively loosely in
their sockets. Immediately in front of this row of newly
formed teeth in their individual sockets, the surface of
the trituration plate no longer shows individual tooth
elements, and only the ridges and grooves on the surface
give evidence of what are evidently the now fused teeth
that had moved up from the formation area. The medial
edges of the trituration plates of either premaxillary do
not meet in the midline but, rather, are separated by a
slight gap.

Maxillary. — Somewhat curved forward at both
ends; forms the anterior border of the upper jaw at its
ventral end. Posteromedially the maxillary possesses a
relatively long, but thin, process which overlies the
lateral and most of the ventral edges of the postero-
medial arm of the premaxillary. The lateral surface of
this posteromedial arm of the maxillary is slightly con-
cave, forming the articular surface which makes fibrous
tissue contact with the medial surface of the anterodor-

sal region of the palatine. The posterior edge of the pos-
teromedial arm of the maxillary is supported against the
anterior vertical face of the ethmoid. The maxillary is
firmly interdigitated along most of its medial surface
with the premaxillary, except at the extreme ventral end
where it no longer overlies the premaxillary and articu-
lates by fibrous tissue with the lateral surface of the pos-
terodorsal end of the dentary. In the larger of the two
specimens the posteromedial arm of the premaxillary is
narrower and that of the maxillary wider than that illus-
trated for the smaller specimen, and the maxillary,
rather than the premaxillary, forms the major articu-
lation with the ethmoid and vomer.

Lower Jaw.

Dentary. — Dentaries indistinguishably fused
together at what would normally be their medial edges of
contact, so that the combined dentaries form a massive
U-shaped bone. No medial line of fusion between the two
halves can be observed externally or internally, but the
two canals leading into the pulp cavity occur side by side
to either side of the midline. The pulp cavity of the com-
bined dentaries is smaller than that of the premaxil-
laries. The teeth of the crushing jaw are exactly like those
described for the premaxillary, as are those of the tri-
turation plate, and are formed from primordia in the
same way. The trituration plate of the dentaries,
however, is a single, almost squarish, massive block
which shows no evidence of the fusion line between what
may be assumed to have been at one time the separate
right and left halves. The posterior end of the dentary is
extremely concave to accommodate the articular, to
which it attaches by interdigitation. Posteroventrally the
dentary articulates by fibrous tissue with the angular.
The dorsolateral surface of the dentary articulates by fi-
brous tissue with the ventromedial surface of the max-
illary.

Articular. — More or less triangular in shape; its
posterior edge somewhat laterally expanded; cartilage
filled at its anterior edge where it is continuous with the
remains of Meckel's cartilage; articulates by interdigi-
tation over most of its lateral surface and over the ante-
rior portions of its medial surface with the dentary.
Posteroventrally it interdigitates with the angular. At
the concave surface of the lateral expansion on its lower
posterior edge the articulfir is supported through fibrous
tissue by the anterior knob of the quadrate. The sesa-
moid articular is a slightly elongate nubbin of bone held
alongside the anteromedial edge of the articular just
above the remains of Meckel's cartilage and just below
the posteromedial edge of the dentary, and is relatively
much larger in the larger specimen than in the small-

Angular. — Small; articulates dorsally by inter-
digitation with the articular and anteriorly by fibrous tis-
sue with the dentary. Posteriorly it connects by ligament
with the anterior end of the interoperculum.

BRANCHIAL APPARATUS.

Hyoid Arch, Branchiostegal Rays, and Urohyal.
Hypohyals. — Both hypohyal elements present, the
dorsal the larger of the two; dorsal hypohyal cartilage fill-
ed at its posterior and ventral edges, ventral hypohyal
cartilage filled at its posterior and dorsal edges; both
hypohyals articulate through cartilage and, in the larger
of the two specimens, by slight interdigitation of their
medial edges with one another and with the ceratohyal;
dorsomedially the dorsal hypohyal articulates by fibrous
tissue with its opposite member.

Ceratohyal. — Elongate; dorsoventrally expanded
at both its anterior and posterior ends; cartilage filled at
its anterior and posterior edges; articulates through car-
tilage anteriorly with the dorsal and ventral hypohyals
and posteriorly with the epihyal, strengthened in the
case of the larger of the two specimens by slight inter-
digitation. Of the six branchiostegal rays, the first three
are held directly to the ceratohyal and the fourth is held
near the posterior edge of the ceratohyal on the cartilage
separating the latter from the epihyal. The somewhat
enlarged first branchiostegal ray is held to the lower
lateral surface of the ceratohyal about one-third the way
back its length. The second branchiostegal ray is held to
a very slight concavity in about the middle of the ventral
edge of the ceratohyal or slightly dorsomedial to it, while
the third branchiostegal attaches to the lower postero-
lateral surface of the ceratohyal.

Epihyal. — Rounded posteroventrally but slightly
prolonged anterodorsally; cartilage filled at its ventral
and anterior edges; articulates through cartilage and, in
the larger specimen, by interdigitation anteriorly with
the ceratohyal, while at a slight prominence on its pos-
terodorsal edge it articulates by fibrous tissue with the
interhyal. Just below its articulation with the interhyal,
the epihyal articulates by fibrous tissue laterally with the
interoperculum.

Interhyal. — An elongate column; cartilage filled at
its dorsal and ventral edges; articulates by fibrous tissue
ventrally with the epihyal and dorsally with the fibrous
tissue sheet between the hyomandibular, metaptery-
goid, symplectic, and preoperculum.

Branchiostegal rays. — Six in number; the first
branchiostegal ray shorter but, in the anterior two-thirds
of its length, wider than the other rays; its dorsal edge
distinctly curved medially while its ventral platelike ex-
pansion lies in the vertical plane; articulates by fibrous
tissue with the lower lateral surface of the ceratohyal
about one-third the way back its length. The second
branchiostegal ray is a long thin shaft held to about the
middle of the length of the ventral edge of the ceratohyal
in the smaller specimen (illustrated), but distinctly on
the medial surface just above the middle of the ventral
edge in the larger specimen. The third and fourth
branchiostegal rays are slightly shorter, and much
stouter anteriorly, than the second ray. The third ray is

held to the lower posterolateral surface of the ceratohyal,
while the fourth is held to the posterior edge of the
ceratohyal and to the cartilage between the epihyal and
ceratohyal. The fifth and sixth branchiostegal rays are
held to the lateral surface of the epihyal; the sixth
branchiostegal is the longest of the branchiostegal rays.
All of the branchiostegal rays articulate with the
ceratohyal or epihyal by fibrous tissue.

Urohyal. — Thin and shallow; laterally expanded
and wider than long; articulates by fibrous tissue
posterodorsally with the anterior edge of the first basi-
branchial and anterodorsally with the dorsomedial sur-
face of the dorsal hypohyal; in the larger specimen the
urohyal is even more laterally expanded than as illus-
trated for the smaller specimen.

Branchial Arches. — All the elements are cartilage
filled at their edges of articulation with the other
elements of the series, and the articulations are usually
through cartilage and fibrous tissue. The branchial
arches are composed of three basibranchials, three pairs
of hypobranchials, five pairs of ceratobranchials, four
pairs of epibranchials, and four pairs of pharyngo-
branchials. Four gills are present, with a slit between the
fourth arch and the lower pharyngeal.

First arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. First basi-
branchial elongate, slightly compressed in the middle;
displaced forward so that it articulates posteriorly with
the second basibranchial and only posterolaterally with
the first hypobranchials. First hypobranchial an elongate
rod; articulates with the anterior end of the second basi-
branchial and the posterior end of the first basi-
branchial. First ceratobranchial a narrow rod; slightly
shorter than the second ceratobranchial but of about the
same length as the third and fourth ceratobranchials; ar-
ticulates ventrally with the first hypobranchial and dor-
sally with the first epibranchial. First epibranchial
rodlike ventrally but becoming wider and flatter dorsal-
ly; the largest of the epibranchials; its dorsal end with
two articular prominences, the anterior of which con-
nects with the first pharyngobranchial, while the poste-
rior process connects with the base of the second
pharyngobranchial. First pharyngobranchial (suspensory
pharyngeal) a curved rod; toothless; articulates ventrally
with the first epibranchial and dorsally with the lateral
surface of the base of the large ventral process of the
posteroventral surface of the parasphenoid.

posteromedially with the second basibranchial and dor-
sally with the second ceratobranchial. Second cerato-
branchial the longest of the ceratobranchial elements;
articulates dorsally with the second epibranchial. Second
epibranchial columnar, with a slight expansion postero-
ventrally; articulates dorsally with the second pharyngo-
branchial. Second pharyngobranchial roughly L-shaped,
with an irregular series of about 25 well-developed and
sharp-pointed teeth borne on the slightly longer of its two
arms. The toothless arm of the pharyngobranchial ar-
ticulates ventrally with the posterodorsal process of the
first epibranchial and with the dorsal end of the second
epibranchial. The teeth are borne in individual sockets
and are replaced by new teeth developing in sockets
irregularly scattered between the sockets of the old teeth.
The second pharyngobranchial is closely held to the
other two toothed pharyngobranchials by fibrous tissue.

Third arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Third basi-
branchial the longest of the basibranchials; laterally ex-
panded in the anterior two-thirds of its length; articu-
lates anteriorly with the second basibranchial, postero-
laterally with the third hypobranchials, and posteriorly
with the fourth ceratobranchials. Third hypobranchial
produced anteroventrally, at which end it articulates by
fibrous tissue with the ventral surface of the first basi-
branchial; articulates posteromedially with the third
basibranchial and posterolaterally with the third cerato-
branchial. Third ceratobranchial like the others; ar-
ticulates dorsally with the third epibranchial. Third epi-
branchial triangular in shape; articulates anterodorsally
with the third pharyngobranchial and posterodorsally
with the middle of the anterior edge of the fourth epi-
branchial. Third pharyngobranchial rounded ventrally,
only slightly expanded dorsally; toothed in the same
manner as the second pharyngobranchial, but with
somewhat fewer teeth on its slightly shorter and thicker
tooth-bearing edge.

Fourth arch. — Cerato-, epi-, and pharyngo-
branchial elements present. Fourth ceratobranchial like
the others; articulates ventrally with the third basi-
branchial and dorsally with the fourth epibranchial.
Fourth epibranchial a column expanded posteriorly in
the middle of its length; articulates dorsally with the
fourth pharyngobranchial. Fourth pharyngobranchial
the smallest of the toothed pharyngobranchials, but with
the widest tooth-bearing surface; teeth like those of the
other pharyngobranchials, except slightly smaller.

Second arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Second basi-
branchial only slightly longer than the first basi-
branchial, but much wider; articulates anteriorly with
the first basibranchial, anterolaterally with the first
hypobranchials, posterolaterally with the second hypo-
branchials, and posteriorly with the third basibranchial.
Second hypobranchial the largest of the hypo-
branchials; much expanded anterolaterally; articulates

Fifth arch. — Ceratobranchial (lower pharyngeal)
element only. Fifth ceratobranchial a stout shaft; round-
ed anteroventrally but tapering to a narrow blunt end
posterodorsally; articulates anteroventrally with the
base of the fourth ceratobranchial. The rounded anterior
region of the fifth ceratobranchial bears teeth like those
of the pharyngobranchials, except slightly smaller, but
more posteriorly on the dorsal surface of the shaftlike
portion the teeth are extremely small.

249

PAIRED FIN GIRDLES.

Pectoral Fin.

Supracleithrum. — In position at about a 45° angle
to the axis of the body; a massive straight shaft, very
thick dorsally but thinner ventrally. At its wide dorsal
edge the supracleithrum articulates by fibrous tissue
with the ventral surface of the posterolateral portion of
the pterotic, while anterodorsally the supracleithrum
abuts against the posterior edge of the posteroventral
flange of the pterotic and articulates with it by fibrous
tissue and, in the smaller specimen, by interdigitation,
the supracleithrum being immovably held to the
pterotic. Ventrally the supracleithrum broadly overlies
the cleithrum and anterior end of the dorsal post-
cleithrum, to both of which bones it is firmly held by fi-
brous tissue. There is no evidence in either of the two
study specimens of a posttemporal bone intervening
between the supracleithrum and pterotic, even as a small
bone extensively interdigitated or partially fused to one
of them. It would be of interest to know whether young
specimens of this species, the most generalized of the Re-
cent gymnodonts, give evidence to the fate of the post-
temporal in gymnodonts.

Cleithrum. — Extremely long, reaching anteriorly to
between the dentaries in the lower jaw; laterally expand-
ed along most of its length but expanded posteriorly only
to a very slight extent in the area of the scapula; articu-
lates by fibrous tissue dorsolaterally with the overlying
supracleithrum and posterodorsally with the anterior
edge of the dorsal postcleithrum. Along its upper medial
edge the cleithrum articulates by fibrous tissue with the
anterior edge of the dorsal postcleithrum. Along its upper
medial edge the cleithrum articulates by fibrous tissue
with the anterior edge of the scapula and the anterodor-
sal edge of the coracoid. The ventral end of the coracoid
also articulates by fibrous tissue with the medial edge of
the cleithrum about one-third the way back the length of
the cleithrum. At their extreme anterior ends the medial
surfaces of the two cleithra are firmly bound to one
another by fibrous tissue, while just posterior to this
region the medial surfaces of the two cleithra support the
anterior end of the pelvis.

Postcleithra. —The postcleithra form a long strut
from the ventral end of the supracleithrum along the ab-
dominal wall musculature to about one-third the way
back the length of the abdominal cavity. The dorsal post-
cleithrum articulates by fibrous tissue anteriorly with
the posterodorsal edge of the cleithrum and antero-
laterally with the posteromedial surface of the overlying
supracleithrum. Along nearly all of its ventral edge the
dorsal postcleithrum articulates by fibrous tissue with
the dorsal edge of the anterior half of the ventral post-
cleithrum, which is a somewhat slimmer shaft than the
dorsal postcleithrum.

Coracoid. — Rounded anterodorsally, but tapering
to a slender shaft in the ventral half of ito length, a large

posterodorsally directed process present from the poste-
rior edge of the rounded portion of the coracoid; cartilage
filled along the anterior edge of its rounded portion; ar-
ticulates by fibrous tissue anteriorly with the cleithrum
and posterodorsally with the third and fourth actinosta;
anterodorsally it articulates through cartilage with the
scapula.

Scapula. — Scapular foramen entirely enclosed by
the scapula; cartilage filled at its anterior and ventral
edges; articulates by fibrous tissue anteriorly with the
cleithrum and ventrally through cartilage with the cor-
acoid. Along the upper half of its posterior edge the
scapula possesses two slightly protruding articular
facets, the more dorsal for articulation with the first fin
ray and the other for articulation with the ventral edge of
the first actinost. At the bottom of its posterior edge the
scapula supports the second actinost.

Actinosts. — Four elements; all cartilage filled at
both ends; first actinost by far the smallest of the
elements and articulated by fibrous tissue with the up-
raised area on the posterodorsal edge of the scapula; sec-
ond, third, and fourth actinosts much larger than the
first and articulated by fibrous tissue with the postero-
ventral edge of the scapula in the case of the second ac-
tinost and to the dorsal edge of the coracoid in the case of
the third and fourth actinosts. Distally the actinosts sup-
port through fibrous tissue all of the pectoral fin rays, ex-
cept for the small first fin ray, which articulates with the
scapula.

Fin rays. — Fifteen fin rays present on both sides of
the smaller specimen and 16 on both sides of the larger;
the first ray short, composed of two distinct halves
throughout its length, the medial half being by far the
stouter of the two; first ray articulated with the scapula,
but the other rays articulated with the actinosts. The
first two rays and the last ray unbranched, the other rays
branched; first ray without cross-striations, the other
rays cross-striated.

Pelvic Fin.

Pelvis.— An extremely long and stout shaft; its
posterior half tapering to a thin rod, while its anterior
half becomes ventrolaterally expanded and then tapers
to a bluntly rounded anterior end; the two halves of the
pelvis firmly interdigitated with one another, except at
the narrow posterior end where the interdigitation
becomes so extensive that distinct halves can no longer
be recognized and where fusion of the medial surfaces of
each half may have taken place. The ventral surface of
the anterior half of the pelvis is deeply concave and
houses the muscle mass whose contraction causes the
downward and forward rotation of the pelvis around its
anterior articulation with the cleithra in the process of
expansion of the huge ventral flap or dewlap. A small gap
is present in the otherwise interdigitated medial surfaces
of the two halves at a point just behind the middle of the

260

length of the pelvis. Anteriorly the pelvis articulates by
tough fibrous tissue with the medial edges of the two
cleithra in the region below the quadrate. There is no
trace of pelvic fins or of any of the rudimentary fin ray
elements such as are found in balistoids.

VERTEBRAL COLUMN. —All vertebrae with bicon-
cave centra, except for the last, which ends posteriorly in
the urostyle.

Abdominal Vertebae.

First vertebra. — Lateral wall of neural arch from
each side of the centrum meeting its opposite member
dorsally above the neural canal but not continuous or in-
terdigitated with it; rather, the medial edges are held to
one another by fibrous tissue; the dorsal surface of this
articular area is slightly upraised medially to give the
suggestion of a neural spine. The upper anterior portion
of the centrum and lower portion of the neural arch are
deeply concave and form a facet for articulation by fi-
brous tissue with the exoccipital condyle.

Other abdominal vertebrae. — Nine abdominal
vertebrae in two specimens, all of which, except for the
first and ninth in the smaller specimen, bear true ribs;
from the first to the ninth vertebrae the neural spines
become increasingly well developed; the two halves of
the neural spine of the second vertebra, like that of the
first, are not fused together above the neural canal, but
are held to one another by fibrous tissue, while the neural
spines of the more posterior vertebrae are single pieces.
The first to the sixth abdominal vertebrae do not possess
haemal arches, but the seventh to ninth have complete
haemal arches and progressively better developed
haemal spines which project slightly posterior to the
haemal arch, that of the seventh with only a short pos-
terior tapering to the haemal arch. Haemal prezyga-
pophyses are not developed and haemal postzyga-
pophyses only begin to develop from about the seventh
vertebra and never extend posteriorly even to the level of
the end of the centrum above them. Neural postzyga-
pophyses are present on the abdominal vertebrae, but
the greatest development of the zygapophyses involves
the neural prezygapophyses. The neural pre-
zygapophyses, and particularly the portions of the neural
arches and spines above them, become enlarged from the
second to about the seventh and eighth abdominal verte-
brae. At its full development the region anterior to the
thickened shaftlike part of the neural spine becomes en-
larged into a relatively flat plate which projects ante-
riorly slightly past the anterior edge of its vertebral cen-
trum to fit below the posteriorly projecting neural spine
of the preceding vertebra. These anterior expansions are
held by fibrous tissue to the posterior edge of the neural
spine region of the preceding vertebra and increase the
rigidity of the vertebral column. The tip of the neural
spine of the fourth vertebra supports the anteroventral
edge of the first supraneural ( = first basal pterygiophore

of the spiny dorsal fin when the latter is present), while
the neural spines of the fifth and sixth abdominal verte-
brae articulate with the ventral edge of this element fur-
ther back along the length of the latter. The neural spine
of the seventh abdominal vertebra articulates dorsally
with the ventral edges of the second and third supra-
neurals (terminology when no spiny dorsal fin is present)
and posteriorly with the anterior edge of the first soft dor-
sal fin basal pterygiophore. The neural spines of the
eighth and ninth abdominal vertebrae articulate be-
tween the more anterior dorsal fin basal pterygiophores.

Ribs and epipleurals. — True ribs are present on all
of the abdominal vertebrae in the large specimen, but are
absent on the first and last abdominal vertebrae in the
smaller (illustrated) specimen. The point of articulation
of the rib with the vertebra is at first with the lower
anterolateral surface of the centrum, but it becomes
progressively lower on the centrum until finally the ribs
on the last few abdominal vertebrae articulate with the
lateral surfaces of the haemal arches. The ribs become
progressively longer from the first to the fourth or fifth,
and then become slightly shorter again. The ribs border
the internal surface of the abdominal cavity and there
can be no doubt that they are pleural ribs. The first two
ribs (attached to the first and second vertebrae) of the
larger specimen and the first rib (attached to the second
vertebra) of the smaller specimen do not bear epi-
pleurals (intermuscular bones), but all of the other ribs
have well-developed epipleurals attached to their dorsal
surfaces. In the smaller of the two study specimens the
epipleurals attached to the second to fifth abdominal
vertebrae are composed of two pieces which are variously
branched. The two pieces of these intermuscular bones
are held together by fibrous tissue. The epipleurals at-
tached to the sixth and seventh ribs are composed of a
single unbranched piece, and while that of the sixth
vertebra is held by fibrous tissue to the pleural rib, that
of the seventh vertebra is fused to the rib. The ninth ab-
dominal vertebra has an unbranched epipleural attached
by fibrous tissue to the lateral surface of its haemal arch.
Short, unbranched epipleurals are attached by fibrous
tissue to the lateral surft e of the haemal arches of the
first five caudal vertebrae. In the larger study specimen
the epipleurals are more extensively branched, and only
one, that attached to the fifth abdominal vertebra, is
composed of two pieces. The epipleurals of the second
abdominal vertebra are both unbranched, while those of
the third are branched on one side and unbranched on
the other. The epipleurals of the fourth to seventh ab-
dominal vertebrae are extensively branched, while those
on the eighth and more posterior vertebrae are un-
branched. Whereas in the smaller specimen the
epipleurals extend to the fifth caudal vertebra, in the
larger there are epipleurals on both sides of the sixth
caudal vertebra as well.

Caudal Vertebrae. — Eleven caudal vertebrae in two
specimens; all, except the last, with well-developed
haemal arches and spines; neural spines well developed

on all but the last of the caudal vertebrae; neural spines
decreasing in length from the first to the eighth, but in-
creasing in length from the 9th to 10th. As the neural
spines of the first to sixth caudal vertebrae decrease in
height, they become laterally compressed into antero-
posteriorly extended plates. The neural spines of the first
to third caudal vertebrae are placed between, and ar-
ticulate by fibrous tissue with, the dorsal fin basal ptery-
giophores. The anterior edge of the neural spine of the
fourth caudal vertebra articulates by fibrous tissue with
the posterior edge of the 11th, or last, basal pterygio-
phores. The neural spine of the 9th caudal vertebra
elongates almost directly posteriorly to form a roof over
the 10th caudal vertebra for almost the entire length of
the centrum of the latter. The neural spine of the 10th
caudal vertebra is an elongate shaft which distally sup-
ports the first 4 or 5 procurrent caudal fin rays. The an-
terior ends of the neural arches and bases of the neural
spines of all of the caudal vertebrae, except for the last
two or three, are prolonged anterolaterally into thick
spinelike processes to which are attached large muscle
masses. Just above the centrum on the lower anterior
edge of the neural arch there is a shorter anterior process
which articulates by fibrous tissue with the postero-
lateral surface of the neural arch and base of the neural
spine of the preceding vertebra. This anterior process,
representing the neural prezygapophysis, is increasingly
well developed from the first to about the fifth and sixth
caudal vertebrae, posterior to which it becomes
progressively smaller again. The haemal arches and
spines of the first to fifth or sixth caudal vertebrae bear
epipleurals laterally and the first to fifth support through
fibrous tissue ventrally the basal pterygiophores of the
anal fin. Along most of its posterior edge the haemal
spine of the first caudal vertebra supports the upper an-
terior edge of the enlarged first basal pterygiophore of the
anal fin. The haemal spines of the second to fourth
caudal vertebrae support most of the anal fin basal
pterygiophores. The posterolaterally expanded and
slightly concave anterior surface of the large haemal
spine of the fifth caudal vertebra supports the small
ninth, or last, pterygiophore. The ventrolateral surface of
the posterior half of each centrum from the first to fifth
caudal vertebrae becomes increasingly expanded ven-
trally to form the side walls of the trough through which
the haemal canal courses. At the sixth caudal vertebra
these ventral expansions articulate by fibrous tissue with
the dorsal edge of the posteriorly prolonged flange of the
haemal spine. There are thus, in effect, two arches over
the haemal canal — the haemal arch proper and the arch
formed by the meeting of the haemal spine with the ven-
tral flange from either side of the posteroventral surface
of the centrum. The seventh and eighth caudal verte-
brae have the same arrangement, except that the sur-
face of contact between the haemal spine and the ven-
tral flanges is fused, leaving only a foramen to indicate
what was formerly the wide space between the dorsal
edge of the posteriorly directed haemal spine and the
ventrolateral surface of the centrum. The haemal spine
of the 9th caudal vertebra, like its neural spine, is

prolonged posteriorly to form a roof over the haemal arch
and base of the haemal spine of the 10th caudal verte-
bra. The haemal spine of the 10th caudal vertebra, like
its neural spine, forms a long shaft that distally supports
most of the procurrent rays of the lower lobe of the caudal
fin. The haemal arches and spines of the 9th and 10th
caudal vertebrae are autogenous.

Caudal Skeleton. — The caudal fin supporting struc-
tures are composed of the Uth caudal vertebra and its
urostylar projection, an epural, two pairs of uroneurals,
four hypurals, and a parhypural, as well as the shaftlike
neural spine and haemal spine of the 10th caudal verte-
bra. The centrum of the 11th caudal vertebra is prolong-
ed posterodorsally into a urostylar process whose pos-
terior edge is concave and supports the upper three
hypurals, while its anterodorsal edge is similarly con-
cave and helps support the uroneurals. Because of its
concave anterodorsal and posterior surfaces, the upper
half of the urostyle is forked into right and left halves,
since there is no bony material in the midline of the uro-
style in this region. From the anterolateral region of its
dorsal surface the centrum possesses an anteriorly
directed prong or neural prezygapophysis which ar-
ticulates by fibrous tissue with the posterolateral sur-
face of the neural spine of the 10th caudal vertebra. The
dorsomedial edges of the two sides of the neural arch of
the last centrum are not in close contact in the midline
over the neural canal, only partially roofing over the lat-
ter. The epural is a large column of bone which is con-
cave along the upper half of its posterior edge. It ar-
ticulates by fibrous tissue anteriorly with the posterior
edge of the neural spine of the 10th caudal vertebra, pos-
teriorly with the first uroneural and ventrally with the
dorsal edges of the incomplete neural arch of the 11th
caudal vertebra. The second uroneural is a long rounded
shaft, with a medial groove along its length representing
the region of complete fusion of its two halves, whose
anteroventral edge rests against the concave dorsal sur-
face of the urostyle mainly in the region of its pwsterior
bifurcation. The posterodorsal end of the second
uroneural supports the deeply forked base of the first
principal ray of the upper lobe of the caudal fin. Along
the ventral two-thirds of its round anterior surface the
second uroneural has closely applied to it a large, but
very thin, pair of bones, the first uroneurals, which ar-
ticulate with it, and with each other medially, by fibrous
tissue. Along their anteroventral edges the first uro-
neurals articulate by fibrous tissue with the dorsal sur-
face of the neural arch and urostyle of the last centrum.
The fourth hypural, or uppermost, is more or less
triangular in shape, with the lower two-thirds of its an-
terior edge resting against the concave posterior surface
of the urostyle. Along its ventral edge the fourth hypural
is closely held by fibrous tissue to the dorsal edge of the
third hypural. The third hypural is widest posteriorly,
narrow anteriorly to a slightly convex and laterally ex-
panded articular facet which fits against a concavity on
the anterior one-third of the dorsal edge of the second
hypural. The extreme anterior end of the second hypural

makes only slight contact with the posterior edge of the
urostyle. The anterior end of the second hypural rests
against the lower posterior edge of the urostyle, while its
dorsal edge is slightly expanded laterally along the an-
terior one-third of its length into a concave articular facet
for contact with the anterior end of the third hypural, as
explained above. The ventral edge of the second hypural
articulates with the dorsal edge of the first hypural by fi-
brous tissue posteriorly but more anteriorly by slight in-
terdigitation. The first hypural articulates dorsally with
the second hypural as just described, while anteriorly it
rests against, and possibly slightly interdigitates with,
the posteroventral edge of the last centrum. The first
hypural articulates ventrally with the dorsal edge of the
parhypural by fibrous tissue posteriorly but more an-
teriorly by slight interdigitation. The parhypural is ob-
viously the haemal arch and spine of the last, or 11th,
caudal vertebra, but, like the haemal arches and spines
of the 9th and 10th caudal vertebrae, it is autogenous.
Dorsally the parhypural articulates with the first hypural
through fibrous tissue and interdigitation, as described
above. The portion of the dorsolateral surface of the
parhypural that lies below the anterior one-third of the
first hypural is laterally expanded into a flange to which
caudal fin-ray muscles are attached. A similar, but much
smaller, lateral flange is present from the anterior edge of
the first hypural. The anterior end of the flange from the
first hypural overlies the anterodorsal surface of the
flange from the parhypural. The combined flanges from
the parhypural and first hypural compose the hypura-
pophysis. The anterior half of the dorsal articular sur-
face of the parhypural is concave, so that where it inter-
digitates with the last centrum and with the anteroven-
tral end of the first hypural, a canal is formed in the mid-
line of the otherwise interdigitated surfaces between the
three bones. The canal opens to the exterior posteriorly
at a slight gap in the articulated surfaces of the parhy-
pural and first hypural just behind the lateral flange of
the parhypural. The canal is continuous anteriorly with
the haemal canal of the preceding vertebrae.

Caudal fin rays. — Twelve principal (branched rays
plus the unbranched ray directly above and directly
below the series of branched rays) fin rays preceded by
eight (smaller specimen) or nine procurrent rays above
and six (smaller specimen) or seven procurrent rays
below. The 10 branched rays become increasingly
branched toward the middle 2 rays, which are branched
in triple dichotomies. The uppermost branched ray and
the lowermost branched ray are the longest of the caudal
fin rays, while the two middle rays are the shortest, being
about one-half the length of the longest branched rays.
The two unbranched principal caudal rays are some-
what shorter than the longest branched rays, and the
procurrent caudal rays rapidly decrease in length an-
teriorly in the series. The first two or three procurrent
rays are without cross-striations, while all the other
procurrent rays and principal rays are cross-striated. The
upper unbranched principal caudal ray articulates with
the second uroneural; the lower unbranched principal

caudal ray with the parhypural. The upper three and
lower two branched rays of the dorsal lobe articulate,
respectively, with the fourth hypural and with the third
hypural. The upper two and lower three branched rays of
the ventral lobe articulate, respectively, with the second
hypural and with the first hypural. All the rays ar-
ticulate by fibrous tissue with the supporting elements.

DORSAL AND ANAL FINS.

Spines and pterygiophores. — The presence or ab-
sence of a spiny dorsal fin in T. macropterus has been
discussed by Tyler (1962a:794-796; 1967:92-93), it being
shown that most specimens from Japan to Indonesia
have a minute spiny dorsal fin, usually of two spines,
while the specimens now known from the Indian Ocean
lack the spiny dorsal fin. The smaller of the two
specimens studied for the present work, SU 13747, from
the Philippines, had been skinned, then cleared and
stained, and then dissected by the author prior to his dis-
covery of a spiny dorsal fin in other whole specimens, and
it is impossible to say, upon reexamination of the cleared
and stained SU 13747 specimen, whether spines had
been present or not. Being from the Philippines, it
probably did have spines which were lost during its
processing as an osteological preparation. The larger
study specimen, ANSP 98917, from the Volcano Islands
south of Japan, was processed with the knowledge of its
possession of a spiny dorsal fin. The lateral view illus-
tration of the entire skeleton presented here is based on
SU 13747, except that the dorsal spines are based on
ANSP 98917, as are the detailed lateral and dorsal views
of the spiny dorsal fin.

When dorsal spines are present, the first is longer than
the second, and both can be laid back in a shallow groove
in the skin in the midline of the body. The bases of the
two spines are relatively close together, the first ar-
ticulating with the posterodorsal surface of the long first
basal pterygiophore ( = first supraneural when no spiny
dorsal fin is present), and the second spine articulating
in the middle of the much shorter second basal pterygio-
phore. From the anterolateral surface on either side of
the base of the first spine a tendon runs anteroventrally
along a groove in the dorsal and lateral surfaces of the
first basal pterygiophore to connect to a small muscle
whose contraction causes the erection of the spine. The
second spine does not have such muscles and tendons as-
sociated with it, but a low membrane connects the first
and second spines so that the erection of the first spine
causes the erection of the second as well. The ventral
ends of both spines articulate by fibrous tissue to the
relatively flat surfaces of their basal pterygiophores.

There are rudiments of what is probably another dor-
sal spine in the form of two nubbins of bone, one to either
side of the midline, lying beneath the skin on the dorsal
surface of the first basal pterygiophore immediately in
front of the base of the first relatively well-developed dor-

sal spine. How constant this pair of nubbins are in their
occurrence cannot be determined until more specimens
of this species are available for clearing and staining, but
I suspect that the nubbins represent the right and left
basal regions of a rudimentary spine. In all other plec-
tognaths with two or more dorsal spines, the first two
spines are always borne on the first basal pterygiophore,
while the spines posterior to them are each borne on in-
dividual basal pterygiophores, which additionally leads
me to believe that the paired nubbins in T. macropterus
are homologous to the first dorsal spine in other plec-
tognaths and that the two relatively well -developed
spines in T. macropterus are homologous to the second
and third spines of other plectognaths. U that is true,
Triodon exhibits a feature not seen in other plectog-
naths with a spiny dorsal fin, for when the size and
number of the dorsal spines decreases in the
triacanthoids and balistoids it is always by reduction and
loss of elements from posteriorly to anteriorly in the
series, and the first spine is always much larger than the
others.

Two other bony elements without intemeural processes
are placed above the dorsal ends of the neural spines be-
tween, and in series with, the two basal pterygiophores
supporting the dorsal spines and the first basal pterygio-
phore of the soft dorsal fin. When a spiny dorsal fin is not
present, these four elements anterior to the anterodorsal
end of the first basal pterygiophore of the soft dorsal fin
would be called supraneurals. But where the two more
anterior elements in the series support dorsal spines,
they must be referred to as basal pterygiophores, even
while the two more posterior elements technically remain
supraneurals. Thus, the distinction between supra-
neurals and basal pterygiophores in Triodon is ar-
bitrary, and indicates that the supraneural element that
occurs in ostracioids and gymnodonts is probably a
simplified basal pterygiophore that no longer supports
dorsal fin spines.

Of the four elements that are supraneural in position
in Triodon, the first (the first basal pterygiophore when a
spiny dorsal fin is present) is by far the largest and ar-
ticulates by fibrous tissue anteriorly with the distal end
of the neural spine of the fourth abdominal vertebra,
ventrally with the distal ends of the neural spines of the
fifth and sixth abdominal vertebrae, and posteriorly with
the anterior edge of the second element. The second ele-
ment is much shorter than the first and articulates by fi-
brous tissue anteriorly with the first, ventrally with the
neural spine of the seventh abdominal vertebra and pos-
teriorly with the anterior edge of the third element. The
third element is slightly shorter than the second and ar-
ticulates anteriorly by fibrous tissue with the second and
with the neural spine of the seventh abdominal verte-
bra. Posteriorly the third element slightly interdigitates
with the anterior edge of the fourth, while posteroven-
trally it articulates by fibrous tissue with the first basal
pterygiophore of the soft dorsal fin. The small fourth ele-
ment articulates by interdigitation ventrally with the
first basal pterygiophore of the soft dorsal fin and an-
teriorly with the third element.

Fin rays and ptery^ophores. — Eleven fin rays pres-
ent; the first ray unbranched, the others branched in
single, double, or triple dichotomies; third and fourth
rays longest; first ray about equal in length to the next to
last ray. Each fin ray has a well-developed pair of distal
pterygiophores as two distinct halves between the bifur-
cate base of the ray. In the smaller specimen the two
halves are not in close contact with one another, while in
the larger specimen the two halves have their medial
edges in close apposition and are held firmly together by
fibrous tissue and in some cases apparently by slight in-
terdigitation. The fin rays are supported basally through
fibrous tissue by ii basal pterygiophores, which ar-
ticulate with one another and with the neural spines of
the vertebrae by fibrous tissue. The dorsal ends of the
basal pterygiophores are slightly enlarged and are in
close contact with one another. The upper lateral sur-
faces of the first three or four basal pterygiophores
possess shallow lateral flanges, but these flanges de-
crease in size posteriorly in the series and are effectively
absent on the last few pterygiophores. The pterygio-
phores are cartilage filled at both ends. The intemeural
portions of the pterygiophores articulate between the
neural spines of the seventh abdominal to the fourth
caudal vertebrae.

Anal Fin.

Fin rays and pterygiophores. — Ten fin rays pres-
ent; the first ray unbranched, the others branched in
single, double, or triple dichotomies; third and fourth
rays longest; first ray about as long as the last ray. Each
fin ray has a well-developed pair of distal pterygiophores
as two distinct halves between the bifurcate base of the
ray, with the two halves in close apposition medially only
in the larger specimen, just as with the soft dorsal fin.
The fin rays are supported basally through fibrous tissue
by nine basal pterygiophores, which are cartilage filled at
both ends. At their distal ends the pterygiophores are
slightly expanded anteroposteriorly, but they do not
make as close a fibrous tissue contact with one another as
do the soft dorsal fin basal pterygiophores. Other than at
their distal ends, the anal fin pterygiophores, with the
exception of the first, are slender shafts without any
lateral projections or flanges from their surfaces. The
first pterygiophore is a much stouter shaft and ventrally
it possesses a large lateral flange to either side of its an-
terior edge. The second pterygiophore is extremely thin,
and is held between the first and third pterygiophores
without reaching dorsally to make contact with a haemal
spine. The other pterygiophores articulate by fibrous tis-
sue dorsally with the haemal spines of the first to fifth
caudal vertebrae.

Anatomical diversity.— The single Recent species of
this family apparently consists of two at least partially
reproductively isolated populations (Tyler 1967) differ-
ing grossly only in that the population from Japan to In-
donesia usually retains a rudimentary spiny dorsal fin

254

while the Indian Ocean population usually completely
lacks even rudimentary dorsal fin spines.

Fragments of premaxillaries and fused dentaries with
small rounded teeth incorporated into the matrix of the
biting edge, found as early as the Eocene of Europe (see
Material Examined), have been assigned to the Triodon-
tidae as Triodon antiquus Leriche (1905, 1906, 1919,
1920), entirely on the rationale that the jaws are in three
major pieces rather than the two of diodontids or four of
tetraodontids.

Tetraodontids would be eliminated from considera-
tion for the antiquus jaw fragments because in antiquus
the individual dental units are small and rounded as in
triodontids and diodontids rather than long rods as in
tetraodontids. Diodontids can be eliminated from con-
sideration only because the premaxillaries in antiquus
remain separate from their opposite members, while they

parasphenoid
prefrontal frontal

Fiffure 183. — Triodon macropterus:

lateral view of tiead, 391 mm SL,

Philippines.

premaxillcry

bosisphenoid epiotic

angular
interopercolum quadrate sympleetic

\ mesopterygoid metapterygoid

Flsure liS.—Triodon iruicroptenu: poitcrior
view of skull (left), with, below, lateral and
posterior views of Hrst abdominal vertebra;
posterior view of orbit (right) (cross section of

skull; dashed lines represent cut surfaces of

frontals and parasphenoid), with, below, lateral

view of basisphenoid (anterior to left);

391 mm SL, Philippines.

ceratobranchials

'^C^ C^C=^ basibranchials

epibranchials // jT //^ <??/ "hypobranchials

pharyngobranchials

Figure 184. (opp. page)— TViodon

macropterus: dorsal (left) and

lateral (right) views of skull,

391 mm SL, Philippines.

.^

T4

interhyal

pihyal

Figure 186. (right)— TViodon

macropterus: dorsal view of branchial

arches (extended on lower side);

lateral view of hyoid arch; dorsal

and lateral views of urohyal;

391 mm SL, Philippines.

dorsal
lateral

urohyal

dorsal hypohyal
ventral hypohyal

branchiostegol rays

dorsal postcleithrum

ventral postcleithrum

Figure 187.— Triodon macropterus: lateral

view of pectoral girdle, with inset showing

full extent of scapula in lateral view,

391 mm SL, Philippines.

Parahollardia lineala Balistapus undulatus Acanthostracion quadricomis

Triodon macropterus Lagocephalus laevigatus Carrthigaster rostrata

Dicxion holocanthus

Figure 188.— Lateral views of scapula and

first (uppermost) actinost in representatives

of all superfamilies, showing the completely

enclosed scapular foramen of Triodon macropterus,

unique among the gymnodonts and similar to

the scleroderms.

Figure lS9.—Triodon macroplenu: lateral views
of alternate pleural ribs and their epipleurals
(left); dorsal view of lower jaw (right) to show
the trituration plate; 391 mm SL, Philippines.

Figure 190.— Triodon macropterua: lateral

(above) and dorsal views of rudimentary spiny

dorsal fln and its basal pterygiophores,

463 mm SL, Japan.

Figure 191.— TWodon macroptenu: ventral, dorsal,
and lateral views of pelvis, 391 mm SL, Philippines.

2nd uroneural

epural

neural spines

haemal spines

caudal vertebrae

procurrent fin rays

are fully fused together in diodontids. Even accepting
this reasoning, the presence of Triodon-like jaw frag-
ments tells us nothing about what the fish behind the
jaws was like, there being no guarantee that the general
form of antiquus was at all like that of the Recent
Triodon macropterus . Moreover, apper jaws without fus-
ed premaxillaries but with small rounded dental units in
the matrix are what one would expect to find in the
Eocene ancestors of the early to late Eocene diodontids
such as Prodiodon and Progymnodon.

Relationships to the Balistoidei and to the other
Tetraodontoidei. — The relationship of the Triodon-
tidae as an intermediary between the eoplectin
Triacanthodidae and the other more specialized
Tetraodontoidei is discussed under the subfamilial rela-
tionships of the Triacanthodidae, it being shown, in es-
sence, that Triodon, which is by far the most generalized
of the gymnodonts, clearly is derived from the eoplectins,
which retain many generalized triacanthodid features
while having highly specialized gymnodontlike jaws with
small rounded dental units incorporated into the matrix
of the premaxillaries and dentaries. Since it is highly im-
probable that such complex dentitional changes as found
in the gymnodonts have arisen independently in various
lines, the eoplectins are obviously ancestral to the gym-
nodonts, with their closest relatives among the gym-
nodonts being the triodontids.

While the evolution of small rounded dental units in-
corporated into the matrix of the jaw bones is a highly
complex matter that has arisen only once among plec-
tognaths (although similar dentition has been in-
dependently developed in some of the Scaridae among
the perciform fishes), the same is not true of the fusion of
the dentaries and premaxillaries to their opposite mem-

Figure \92.— Triodon macropterus: lateral

view of caudal fin supporting structures

(see T>ler 1970b:figs. 42a and 42b,

for details), 391 mm SL, Philippines.

bers, which, relative to the dentitional and general struc-
tural changes involved in developing a crushing or biting
beak, is a simple matter. The relative superficiality of
the jaw bone fusion is attested to by the fact that the
dentary and premaxillary of very large specimens of
tetraodontids occasionally fuse to their opposite mem-
bers and that Reuvens (1894:130) described a young
molid in which the dentaries were fused but the pre-
maxillaries were separate. Thus, the form of the dental
units in the biting edge of the jaws is a far more
anatomically complex and phylogenetically important
indicator than is the fusion or lack of fusion of the den-
taries and premaxillaries.

The fusion of these bones to their opposite members
undoubtedly at least slightly increases the rigidity and
strength of the beak beyond that which it would possess
if the premaxillaries and dentaries remained separate,
but I doubt that the difference is functionally of great
magnitude, for the two halves of the upper and lower
jaws of tetraodontids are firmly and relatively inflexibly
held together medially by a combination of interlocking
emarginations and thick bands of tough fibrous tissue.
The two halves of the upper jaw in Triodon, for example,
are so extensively interdigitated that they are ex-
ceedingly difficult to force apart, and I doubt that much
more pressure would be needed to break apart the fused
dentaries than to separate the interdigitated premaxil-
laries.

The most generalized dentitional condition in gym-
nodonts is small rounded units incorporated into the

matrix of the biting edge of the premaxillaries and den-
taries, as found in the triodontids, the most generalized
gymnodonts, and in their ancestral eoplectin triacan-
thodids (as well as in the moderately specialized diodon-
tids). Of the two known eoplectins, the biting edges of
the jaws are exposed only in Eoplectus, which has small
rounded dental units, but this can also be expected to be
the condition in the related Zignoichthys. The eoplec-
tins undoubtedly evolved from a group of early
triacanthodids with a generalized dentition of numerous
conical teeth in an outer series, internal to which there
were a smaller number of conical inner series teeth.

Conversion of this generalized dentition into that as
found in eoplectins and triodontids probably involved a
great increase in the number of outer series teeth and
their placement in more than one row concomitant with
a great reduction in size of the individual teeth, which
eventually protruded less and less to the exterior beyond
their basal sockets in the premaxillaries and dentaries
and finally became entirely nonprotrudant, being fully
surrounded by the matrix of the bone. It is not known
whether eoplectins had trituration plates internal to the
biting edges of the jaws, but it is probable that they did,
for trituration plates are found in all triodontids, diodon-
tids, molids, and in many tetraodontids, and the jaws of
eoplectins are just as massive as in most gymnodonts and
the eoplectin diet probably consisted of hard shelled in-
vertebrates for which trituration plates would be useful
in crushing and grinding. These trituration plates are ob-
viously formed from modified internal series teeth as
found in the more generalized triacanthodids, the teeth
becoming greatly increased in number, and sometimes in
size, to form a series of rows which variously retained
much of their individual identity or became incor-
porated into a plate in which some of the teeth lost their
individual identity, especially anteriorly in the plate
away from the posterior region of tooth replacement.
However, the development of trituration teeth, whether
as plates or not, is probably a highly labile feature in
various lines of gymnodonts closely correlated with
dietary changes as species evolved in differing habitats.

The generalized gymnodont dentition of small rounded
units in the biting edge has been retained by triodontids
and diodontids, but tetraodontids and molids have great-
ly modified the dentition of the biting edge. In molids
discrete dental units are no longer present, at least as ob-
served at 30 magnifications, the ancestral small rounded
units apparently indistinguishably fused to the bony
matrix. In tetraodontids the ancestral small rounded un-
its have been modified into less numerous long rods lying
in series approximately parallel to the biting edge. The
tetraodontid dental units of the biting edge obviously did
not become specialized long rods until after the division
of the joint tetraodontid-diodontid ancestral line, for
diodontids retained small rounded units from this line.

It seems reasonable to assume that the tetraodont jaw
type (but not tooth type), with both the premaxillaries
and dentaries closely articulated but not fused to their
opposite members, was the primitive condition of early
gymnodonts, and that both the triodont jaw type, with

only the dentaries fused to their opposite members, and
the diodont jaw type, with both the dentaries and pre-
maxillaries fused to their opposite members, are derived
from a tetraodont jaw type ancestry, either in-
dependently or with the diodont type perhaps having
been preceded by a triodont type ancestral group.

It is not known whether the premaxillaries and den-
taries were fused or articulated to their opposite mem-
bers in the Eocene triacanthodid Eoplectus, but in the
apparently closely related Zignoichthys from the same
strata, at least one of the jaws, probably the lower, had
the two halves (presumedly dentaries) fully fused
together.

If these two Eocene triacanthodids, representing the
subfamily Eoplectinae, are indeed representatives of the
ancestral line leading to Triodon and the other gym-
nodonts, then at least Zignoichthys is already too
specialized in its jaw structure (which must be either
triodont or diodont type) to be considered as ancestral to
the tetraodontids. It seems to me extremely unlikely that
once the full fusion of the dentaries and/or premaxil-
laries was established in a line of gymnodont evolution
that consistent reversal to an unfused condition could
take place, especially in light of the intricate inter-
locking interdigitations that are almost always present
when the premaxillaries or dentaries are articulated to
one another rather than fused.

It is assumed here that the immediate ancestry of the
few known eoplectins and thus of the triodontids had the
premaxillaries and dentaries separate from their op-
posite members, and that the tetraodontids diverged
from this ancestral line at a time in the Eocene when at
least some members of that line retained separate pre-
maxillaries and dentaries. This line was probably
TViodon-like except in the retention of separate den-
taries and also probably before the elongate tapered
caudal peduncle as seen in the Recent Triodon had been
fully established. Since it is here considered that the
diodontids are more closely related to tetraodontids than
to molids, an additional implication of the above reason-
ing is that the fusion of the premaxillaries and dentaries
in molids and diodontids has taken place in-
dependently, that in diodontids from a tetraodontid an-
cestral group, and that in molids from the same eoplec-
tin-triodontid line retaining separate premaxillaries and
dentaries from which tetraodontids are evolved.

The features in the conversion of an eoplectinlike fish
into a triodontidlike one are discussed under Eoplectus
in the account of the Triacanthodidae. The conversion of
a triodontidlike fish into a fish like that of any of the
other derived families of gymnodonts involves primarily
the loss of most of its more generalized or triacanthodid-
like features.

The features of Triodon that are lost by all other gym-
nodonts are: 1) the rudimentary spiny dorsal fin and all
but one of the basal pterygiophores and supraneural
elements supporting it; 2) the ribs and epipleurals; 3) the
pelvis; 4) uroneurals and hypurapophysis, and the con-
solidation of the hypurals so that only one free element is
present and the conversion from an autogenous to fused

haemal spine on the antipenultimate vertebra; 5) the
procurrent caudal fin rays and at least one of the prin-
cipal rays (from the upper lobe of the fin); 6) a complete
dorsal roof to the myodome and the channel leading into
it posteriorly; 7) the knob on the scapula for articulation
with the uppermost pectoral fin ray, and the complete
scapular foramen; 8) urohyal; 9) one of the pharyngo-
branchials and well-developed teeth on the fifth cerato-
branchial; 10) the deep olfactory sac and normal olfac-
tory rosette; 11) four actinosts, none of which are sutured
to one another or to the scapula or coracoid.

Triodon, as represented by the single Recent species
upon which our entire knowledge of the family is based
except for the fossil jaw fragments possibly related to it,
has certain features found in neither the triacanthodids
nor in other gymnodonts. These undoubtedly are
specializations which were acquired by Triodon after its
ancestral stock had given rise to the lines leading to the
molids on the one hand and to the tetraodontids and
diodontids on the other hand. These specialized features
of Triodon are: 1) the long caudal peduncle tapering to
a transversely indented region just in front of the base of
the deeply forked caudal fin; 2) the prominent antero-
lateral processes from the neural arch region of many of
the caudal vertebrae and the long rodlike neural and
haemal spines of the penultimate vertebrae, all as-
sociated with the musculature and support of the caudal
peduncle and caudal fin used for apparently sustained
rapid swimming; 3) the rotatability of the pelvis and the
expansible dewlap of skin between the end of the pelvis
and the anus; 4) the great anterior elongation of the
cleithrum; 5) the small sphenotic entirely confined to the
rear of the orbit; 6) the articulation of the first branchi-
ostegal ray on the ventrolateral surface of the cerato-
hyal; 7) the short posterior shaft of the interoperculum
behind the level of the epihyal; 8) the posteroventrally
prolonged shaft of the pterotic articulating broadly with
the hyomandibular; and 9) the exoccipital in contact
with the frontal and excluding the epiotics from contact
with the supraoccipital.

Since neither the triacanthodids nor any of the other
gymnodonts possess anything similar to these specializa-
tions of Triodon, the ancestral line leading to the Recent
Triodon can be expected to have had more triacantho-
didlike conditions in these features and that they
remained relatively generalized in them until after the
other gymnodont linages split off from the early Triodon-
like fishes.

The ancestral line connecting the eoplectin tri-
acanthodids and the triodontids is hypothesized as hav-
ing given rise to the line leading to the great diversifica-
tion of the gymnodonts, this line diversifying in two basic
directions, one leading to the molids and one to the
tetraodontids and diodontids, at a pre-Recent Triodon
level of organization, i.e., at a level of organization that
retained all of the generalized features of Triodon but
that did not yet have the specialized features of the Re-
cent Triodon noted above, as well as with both the pre-
maxillaries and dentaries unfused and the first branchi-
ostegal ray entirely unmodified.

The molid line of radiation from the pre-Recent
Triodon level of organization is represented by only a few
species, at least surviving today, which became vastly
modified for a slow swimming oceanic and mostly pelagic
existence and relatively huge size within a protective
wall of thickened skin, while retaining from its triodon-
tid ancestry several generalized features lost in the
tetraodontid-diodontid line of radiation. Molids have
retained: 1) the basisphenoid; 2) fourth gill and gill slit
between the fourth and fifth arches; 3) unmodified first
branchiostegal ray; 4) uninflatable gut; and 5) un-
sutured ceratohyal and epihyal.

Moreover, the configuration of the bones of the snout
in molids is in many ways remarkably similar to that in
Triodon, far more so than is the case in the tetraodontid-
diodontid line. In Triodon and molids the ethmoid tends
to be a large squarish block of bone bordered on either
side by the large palatines and prefrontals, with the
palatine broadly held medially to the ethmoid, vomer
(unossified in Ranzania), parasphenoid and prefrontal.
The shape and size of the frontals in Triodon and molids
is also remarkably similar. An anteriorly directed prong
in the suboperculum, as found in Triodon and diodon-
tids, is retained in molids, even though the molid oper-
culum and suboperculum are greatly reduced in size.

While these similarities between Triodon and molids
indicate the relationship between the two groups, molids
possess an array of specializations beyond the generaliz-
ed Triodon level of organization, many of which are cen-
tered around the aborted rear end of the body and the
support of the continuous dorsal, anal, and pseudo-
caudal fins. The major ways in which molids differ from
Triodon, other than those already mentioned above that
distinguish Triodon from all other gymnodonts, whether
they be the generalized triacanthodidlike features of
Triodon that are lost or the specialized features of
Triodon developed after the radiation of the other gym-
nodonts from a triodontid ancestral group, are: 1) the
apparently indistinguishable incorporation of the dental
material into the matrix of the jaws, along with the fu-
sion of the premaxillaries, even though well-developed
individual trituration teeth are retained in both the up-
per and lower jaws; 2) the loss of an ossified sesamoid ar-
ticular; 3) the great enlargement of the basisphenoid and
the complete loss of the dorsal roof of the myodome and
of the posterior opening into it; 4) the loss of teeth on the
fifth ceratobranchial and the development of gill rakers
along the front of the fifth ceratobranchial and along the
anterior edge of the first gill slit, even though only
Triodon and molids among the gymnodonts retain the
fourth gill, well-developed teeth on all three of the tooth-
ed pharyngobranchials, unsutured ceratohyal and
epihyal, the consistent presence of both dorsal and ven-
tral hypohyals and the interhyal, and a relatively un-
modified first branchiostegal ray; 5) the loss of the
pharyngobranchial of the first arch; 6) the loss of the first
or uppermost actinost; 7) the consolidation of the post-
cleithra into a single piece and the development of an an-
terior spur from it to the region of the actinosts; 8) the
development of a dome or posterior prolongation on the

epiotics; 9) the development of a posterior prolongation
of the pterotics broadly articulating with the supra-
cleithrum; 10) the dorsal prolongation of the basioc-
cipital excluding the exoccipitals from contact with the
first vertebra; 11) the loss of haemal arches and spines on
at least some of the abdominal vertebrae and the reduc-
tion in size of their neural spines; 12) the loss of all of the
supraneural elements and the vast rearrangements of the
vertebral column and basal pterygiophores of the dorsal
and anal fins in the formation of the continuous fin
around the posterior end of the body with the abortion of
the true caudal fin and its supporting structures and the
development of a pseudocaudal fin of posteriorly
migrated dorsal and anal fin rays and their basal ptery-
giophores; 13) reduction in the number of vertebrae from
20 to between 16 and 18; 14) reduction in the size of the
olfactory apparatus; 15) development of thickened or
hardened skin; 16) reduction in the size of the gill open-
ing; and 17) the loss of the air bladder.

The tetraodontid-diodontid line of radiation from the
pre-Recent Triodon level of organization diversified
greatly, with well over a hundred species alive today, and
have become in many ways just as morphologically dif-
ferent if not more so from triodontids as have molids,
while retaining none of the few generalized characters
shared by triodontids and molids, even though in general
body form the tetraodontid-diodontid line is more
similar to that of Triodon (exclusive of the expansible
dewlap) than is that of molids.

The major ways in which tetraodontids and diodon-
tids differ from Triodon, other than those previously
mentioned that distinguish Triodon from all other gym-
nodonts, whether they be the generalized triacanthodid-
like features of Triodon that are lost or the specialized
features of Triodon developed after the radiation of the
other gymnodonts from a triodontid ancestral group,
are: 1) the loss of the basisphenoid; 2) the loss of the
fourth gill and of the gill slit between the fourth and fifth
arches; 3) the loss of at least well-developed teeth on the
fifth ceratobranchial and on one of the three pharyngo-
branchials; 4) the loss of the pharyngobranchial of the
fourth arch; 5) the loss of gill rakers behind the anterior
edge of the fourth arch; 6) the great expansion of the first
branchiostegal ray into a horizontal pumping plate ar-
ticulated to the medial surface of the ceratohyal; 7) the
development of an inflatable diverticulum of the gut; 8)
the suturing of the ceratohyal with the epihyal; 9) the
frequent loss of the interhyal and dorsal hypohyal; 10)
the loss of at least most of the dorsal roof of the myo-
dome and of the posterior opening into it; 11) the loss of
all, or all but one, of the supraneural elements; 12) the
development in one of the two families of long rodlike
dental units and of the frequent reduction in size or loss
of the trituration plates or teeth; 13) the development of
a more limited area of articulation of the palatine with
the cranium either at a notch and flange between the
palatine and ethmoid-vomerine region or by extensive
suturing between the palatine and frontal; 14) a reduc-
tion in the massiveness of the prefrontal and supra-
cleithrum; 15) the usual better development of the

mesopterygcid (except absent in one species) and its firm
suturing to the metapterygoid as well as to the palatine;
and 16) the development in three or more of the ab-
dominal vertebrae of bifid neural spines, always in-
cluding the first three abdominal vertebrae.

Thus, even though Triodon is by far the most
generalized gymnodont and retains many primitive
features of its triacanthodid ancestry, the structure of the
only species of which we have knowledge other than jaw
fragments, the Recent Triodon macropterus, is too
specialized for an Eocene fish closely similar to it to have
been the ancestral line from which the molid and tetrao-
dontid-diodontid lines radiated. This ancestral triodon-
tid line had a more generalized organization than that of
T. macropterus, not yet possessing such specializations
as the small size and entirely orbital placement of the
sphenotics allowing the frontals and exoccipitals to meet,
and the elongation and tapering of the caudal peduncle
and associated vertebral modifications. Nevertheless, an
early Eocene Triodon-like fish minus the specializations
as seen in the single Recent species is undoubtedly an-
cestral to the other two major subgroups of gymnodonts
and itself evolved from the eoplectin triacanthodids in
the early Eocene.

Infraorder Tetraodontoideo

Comparative diagnosis (contrast with that of the
Triodontoideo).— Spiny dorsal fin completely absent
and no more than one supraneural element present; ribs
and epipleurals absent; caudal fin with 11 or fewer prin-
cipal rays and no procurrent rays; caudal fin varying
from moderately forked to rounded; caudal skeleton with
no more than one separate hypural, and no uroneurals
and hypurapophysis; haemal spine of antipenultimate
vertebra fused to its centrum; neural and haemal spines
of penultimate vertebra various but not long rounded
shafts directly supporting caudal fin rays; caudal pedun-
cle not distinctly tapered to narrow transversely in-
dented regions above and below just in front of the
caudal fin, the least depth of the peduncle being about
6% SL or greater, and the least width at this region al-
ways less than the least depth; none of the caudal verte-
brae with anterolateral processes above the centra; no
pelvis; no expansible dewlap of skin in front of the anus,
but inflatability of the abdominal region present in all
families except molids; cleithrum not elongate an
teriorly, reaching forward no further than about the level
of the middle of the ceratohyal; basisphenoid either ab
sent or present as a large plate in the interorbital sep
tum; myodome either absent or present only as a rudi
ment of the dorsal roof represented by medial wings ol
the prootics, the roof always highly incomplete; no chan
nel present between the apposed surfaces of the para
sphenoid and basioccipital; scapular foramen incom-
plete, closed anteriorly by the cleithrum; scapula
without a distinct knob for articulation with the upper-
most pectoral fin ray; three or four actinosts, if four pre-
sent at least some sutured either to one another or to the

scapula or coracoid, the uppermost always sutured to the
scapula; urohyal absent; no more than three pharyngo-
branchials present; fifth ceratobranchial usually tooth-
less, rarely with even a small patch of minute teeth; den-
taries and premaxillaries totaling either two or four
separate pieces; premaxillaries, if separate, articulated
to one another by prominent regular interlocking
emarginations; sphenotic relatively large and not con-
fined to the posterior wall of the orbit, but present on
both the lateral and dorsal surfaces of the skull as well;
first branchiostegal ray with its dorsal edge either not in-
turned at all and articulated to the ventral edge of the
ceratohyal or greatly inturned into a huge plate and ar-
ticulated to the medial surface of the ceratohyal; inter-
operculum with a ventral flange and a long posterior
shaft extending well behind the ventral flange and level
of the epihyal, except in molids in which the interoper-
culum is a short to long simple slender rod without a ven-
tral flange; pterotic not prominently prolonged postero-
ventrally as a stout shaft; epiotic placed more medially
on the top of the skull, always in contact with the
supraoccipital; exoccipital never in contact with the
frontal; frontal never in contact posteriorly with the
pterotic in the rear of the orbit, separated from it by the
sphenotic, while in some tetraodontins posterolateral
wings from the dorsal surface of the frontal make con-
tact with the pterotic on the top of the rear of the skull;
olfactory epithelium smooth, pitted or in parallel folds,
never in a rosette, the olfactory sac either above the sur-
face in a tube or tentacle with one or two nostrils, or at or
below the surface and rudimentary.

SUPERFAMILY TETRAODONTOIDEA

Comparative diagnosis (contrast with that of the
Moloidea). — Inflatable diverticulum of the gut present;
first branchiostegal ray with the dorsal edge inturned
and enormously enlarged into a more or less horizontal
pumping plate articulated to the medial surface of the
ceratohyal; air bladder well developed; three gills, not
greatly expanded dorsally; no gill slit between the gill-
less fourth and fifth arches and no gill rakers along the
posterior edge of the fourth arch and the anterior edge of
the fifth arch; no gill rakers along the anterior edge of the
first gill slit; pharyngobranchials with moderate to
minute teeth, and one sometimes toothless, the three
elements being those of the first to third arches (that of
the third sometimes absent); ceratobranchial and epi-
branchial sutured to one another; interhyal and dorsal
hypohyal often absent; teeth in biting edge of jaws
retaining much of their individual identity; basi-
sphenoid absent; caudal fin relatively normal, with 9 to
11 rays supported by variously consolidated and fused
but normal vertebral elements; most dorsal, anal,
caudal, and pectoral fin rays less extensively branched
but with a normal amount of cross-striations, these not
confined to only the distal ends of the rays; sesamoid ar-
ticular usually present; postcleithrum without an an-
teriorly directed process; four actinosts; supracleithrum

less elongate, only its extreme proximal end articulated
directly to the pterotic; coracoid less long and slender,
with a posterodorsal prong below the lower actinost al-
ways developed to some degree; operculum and sub-
operculum of much greater complexity of structure and
lateral surface area; interoperculum with a ventral flange
and posterior shaftlike portion extending posteriorly well
beyond the level of the epihyal; basioccipital not
prolonged dorsally behind the exoccipitals, the exoc-
cipitals bordering all but the bottom edge of the foramen
magnum; exoccipitals with condyles and in contact with
the first vertebra, which articulates anteriorly with both
the basioccipital and exoccipitals; epiotic without any
kind of posterodorsal prolongation; pterotic not prolong-
ed posteriorly, not reaching posteriorly to the level of the
end of the basioccipital; bony canal, when present, for
the nerves and blood vessels running from the orbit to the
nasal region nearly always incomplete, surrounded by
the prefrontal laterally, above and below, but not
medially; palatine receiving its main support dorsally
either by a complex interlocking with the vomer and eth-
moid or by extensive suturing to the frontal; at least the
first three abdominal vertebrae with bifid neural spines
projecting dorsally or dorsolaterally on each side of the
neural arch; centra of at least some of the abdominal
vertebrae always with ventral or ventrolateral processes
of some sort, whether or not forming complete haemal
arches or zygapophyses; dorsal and anal fin rays not
widely separated from their basal pterygial supports by a
large block of cartilage; scales, when present, small to
large, the basal plates not forming a completely con-
tinuous covering over the entire body, and some of the
scales always forming prominent prickles or spines; one
or two nostrils in a prominent upraised sac or tentacle, or
a single minute nostril in a low tube; lateral line present,
whether conspicuous or inconspicuous, but nearly al-
ways clearly seen within 30 magnifications.

Family Tetraodontidae

Comparative diagnosis (contrast with that of the
Diodontidae). — Teeth incorporated into the matrix of
the biting edge of the jaws as long slender rods; pre-
maxillaries and dentaries not fused to their opposite
members in the midline, the articulation strengthened
by interlocking emarginations; lateral surface of maxil-
lary neither deeply indented nor laterally flanged, the
surface relatively even; the jaws usually less massive;
small trituration teeth often present in the upper and,
less frequently, lower jaw as well, but seldom with a large
trituration plate in both jaws; first pharyngobranchial
with small or, usually, minute teeth, or sometimes tooth-
less, but second and third pharyngobranchials with small
teeth; dorsal hypohyal often absent; interhyal some-
times present; anterior edge of ectopterygoid distinctly
concave; ethmoid and vomer relatively well developed
and sturdy, although sometimes at least partially fused
together; palatine with a notched region posterodorsally
for firm articulation with a crested region of the vomer
and/or ethmoid, and never making contact with the fron-
tal; prefrontal usually well developed, absent only in two
closely related specialized genera, Xenopterus and
Chonerhinos; frontals less wide and massive, except in
Xenopterus; anterior end of parasphenoid less wide and
less deeply concave, the concavity filled by the posterior
articulating shaft of the vomer; rear margin of the orbit
formed by the frontal and sphenotic; sphenotic often
laterally expanded anteriorly, but never as a long,
slender lateral prong; frontal not in contact posteroven-
trally in the rear of the orbit with the prootic, separated
from it by the sphenotic and pterosphenoid; suboper-
culum without an anteriorly directed prong, and the in-
teroperculum articulated posteriorly by a short ligament
to the anterior edge of the operculum; supracleithrum
usually positioned at about a 45° angle to the axis of the
body, but with great variation between species on the
degree of obliqueness to the body; postcleithrum in two
pieces, and much longer than the distance along the
scapula to the lowest actinost; supraoccipital spine
laterally compressed and mainly in a vertical plane, al-
though its dorsal edge may be thicker than the ventral
plate; exoccipital condyles well developed; at least the
first two vertebrae anterior to the first basal pterygio-
phore of the dorsal fin without bifid neural spines, only
the first three to five abdominal vertebrae with bifid
divergent neural spines; none of the vertebrae with
prominent lateral flanges from the centra; neural spines
of the vertebrae supporting the basal pterygiophores of
the dorsal fin relatively normal long slender shafts pene-
trating relatively deeply the interspaces between the

Figure 193.— Range of diversity in

body form in the Recent Tetraodontidae:

A. Lagocephalus spadiceus,

B, Xenopterus naritua,
C, AmblyrhynchoteB pioaae,
D, Canthigaater rogtrata.

pterygiophores; a supraneural element often present; at
least some of the basal pterygiophores of the dorsal and
anal fins usually, but not always, interdigitated with one
another; one or more of the more posterior abdominal
vertebrae usually, but not always, with a complete
haemal arch; none of the vertebrae posterior to the bases
of the last basal pterygiophores of the dorsal and anal
fins anteroposteriorly compressed, of about the same
centrum length as those more anteriorly; abdominal
vertebrae usually fewer in number than the caudal verte-
brae, sometimes of equal number but never of greater
number; dorsal and anal fins more anterior in position;
at least four vertebrae fully posterior to that whose hae-
mal spine is the last support of the last anal fin basal
pterygiophore; caudal fin supporting skeleton with a free
epural, one free uppermost hypural that in a few species
may be partially fused to the centrum, a free parhypural
and an autogenous haemal spine of the penultimate
vertebra; no prominent lateral flange present on the fus-
ed hypural-centrum plate; haemal canal penetrating the
last vertebral complex to exit between the parhypural
and fused hypural-centrum plate; caudal fin rays 11, the
two lowermost rays unbranched; scales always relatively
smaller, even when best developed, as in the moderate
quills of Torquigener piosae or the enlarged basal plates
of adult Ephippion guttifer.

Comparative diagnoses of Subfamilies (Tetraodon-

tinae and Canthigasterine). — Subfamily Tetraodonti-
nae: ethmoid varying from long to short, and of a vari-
ety of shapes, but never distinctly T-shaped in cross
section; supraoccipital with an only moderate to low
crest, usually arising from a flattened anterior region
of the bone from which it is prolonged posteriorly, the
posterior end of the crest not visible externally through
the skin; posterolateral region of frontal variously lateral-
ly expanded or not, but never as two flanges whose distal
ends closely approach or are in contact with one another
to form a bony well around the muscle mass leading to
the operculum; sphenotic a prominent component of the
dorsal surface of the skull, and usually extended antero-
dorsally at least a short distance forward of the rear edge
of the orbit; base of prefrontal always either in contact
with or in very close proximity to the vomer; trituration
teeth often present; one or two nostrils in a moderately
low to high sac or tube, or a single or bifid tentacle or
flap, but always conspicuous; lateral line conspicuous
(with the possible exception of the long-spined Tor-
quigener piosae); gill opening usually extending ven-
trally below the level of the middle of the pectoral fin
base; vertebrae modally 17 to 29; vertebral column
usually not highly arched anteriorly, the axis of the an-
terior portion of the column usually not especially ob-
lique to that of the skull; haemal arches and spines
usually not especially well developed on most of the ab-
dominal vertebrae.

Subfamily Canthigasterinae: ethmoid long and T-
shaped in cross section, its upper surface a laterally ex-

panded more or less flat plate of decreasing width an-
teriorly and its lower surface a similarly flat vertical
plate of decreasing depth anteriorly; supraoccipital with
a high well-developed crest throughout most of its
length, the dorsal edge of the posterior end of the crest of-
ten indicated externally by a break in the contour of the
dorsal profile; posterolateral region of frontal prolonged
laterally as two flanges whose distal ends closely ap-
proach or are in contact with one another to form a par-
tial to complete bony well around the muscle mass
leading to the operculum; sphenotic mostly excluded
from the dorsal surface of the skull, mostly confined to
the ventrolateral surface of the skull behind the orbit;
base of prefrontal well removed from the vomer, broadly
separated from it by the ethmoid and parasphenoid;
trituration teeth never present; a single nostril on each
side, set in a very low inconspicuous tube; lateral line in-
conspicuous, apparent only with magnification; gill
opening restricted, extending ventrally no further than
about the level of the middle of the pectoral fin base;
vertebrae always modally 17; vertebral column highly
arched anteriorly, the axis of the arched portion highly
oblique to that of the skull; haemal arches and spines es-
pecially well developed on most of the abdominal verte-
brae.

Detailed description of Lagocephalus laevigatus.

Material examined— Three cleared and stained
specimens, 61.4-166 mm; one wet completely disar-

ticulated specimen, approximately 290 mm, prepared by

maceration; one dry skeleton, approximately 290 mm.

Occipital Region.

Basioccipital. —A short column, slightly expanded
anteriorly; cartilage filled at its anterior and antero-
lateral edges; articulates by interdigitation anteriorly
with the slightly overlying parasphenoid, anterolaterally
with the prootic and laterally with the exoccipitals. The
anterodorsal end of the basioccipital reaches to and
forms the lower posterior wall of the vestigial myodome
(see description of the prootic). The rim of the round con-
cave posterior end of the basioccipital articulates by fi-
brous tissue with the rim of the concave anterior face of
the centrum of the first vertebra. The posterodorsal sur-
face of the basioccipital forms the ventral wall of the
foramen magnum.

Exoccipital. — Cartilage filled at its anterior and
ventromedial edges; articulates by interdigitation ven-
tromedially with the basioccipital, posterolaterally with
the pterotic, anterodorsally with the epiotic, and antero-
ventrally with the prootic. Dorsomedially the exoc-
cipital forms the lateral wall of the foramen magnum,
while the dorsal wall of the foramen is formed by the
close fibrous tissue articulation of the extreme dorso-
medial edges of the two exoccipitals. Ventrally the
foramen magnum is closed not by medial projections of

Figure 194.— Lagocephalus laevigatus (above),

with L. lunaris (center) and

L. scleratus (below) for comparison: in all

three examples, upper left, nasal region as

seen externally (far left) and the olfactory

lamellae as seen with the top of the nasal

sac removed (center), and the outline of an

anteroposterior cross section of the sac and

lamellae; center left, scales from upper

middle region of body, including, in L.

lunaris and L. scleratus, the small scales

associated with the lateral line.

the exoccipitals, but, rather, by the posterodorsal sur-
face of the basioccipital. From the posterior end of its
ventromedial edge the exoccipital possesses a modified
condyle in the form of a posteroventral process that ar-
ticulates by fibrous tissue with the lateral surface of the
neural arch and centrum of the first vertebra.

Supraoccipital. — A rounded plate anteriorly, but
drawn out posteriorly into a long, laterally compressed
spine; cartilage filled along all of the edges of its rounded
anterior portion; articulates by interdigitation postero-
laterally with the epiotics and anteriorly and antero-
laterally with the frontals. Along the middle of its an-
terior end, as seen dorsally, the supraoccipital is broadly
overlain by the posteromedial edges of the frontals. The
supraoccipital spine is drawn out posteriorly between the
bifid neural spine of the first vertebra and sometimes
reaches posteriorly to above the second vertebra. The an-
terior half of the ventral edge of the supraoccipital spine
articulates by tough fibrous tissue with the dorsomedial
edges of the epiotics and exoccipitals.

Otic Region.

Pterotic. —Cartilage filled along its anteromedial
edge; articulates by interdigitation posteromedially with
the exoccipital, dorsomedially with the epiotic, antero-
dorsally and along the lateral part of its anteroventral
edge with the sphenotic, and anteroventrally with the
prootic. At the anterodorsal edge of its most laterally ex-
panded portion the pterotic articulates by fibrous tissue
with the posterolaterally projecting arm of the frontal.
Along the middle of its posterior surface the pterotic has
a concavity for articulation by fibrous tissue with the
supracleithrum. In the middle of its ventral surface the
pterotic articulates by fibrous tissue with the hyoman-
dibular.

Sphenotic. — Broadly cartilage filled at its medial
edge; articulates by interdigitation anterodorsally and
dorsomedially with the frontal, posteromedially on its
dorsal surface with the epiotic, posterodorsally with the
pterotic, posteromedially on its ventral surface with the
pterosphenoid, and anteroventrally with the frontal.
Posteromedially on its ventral surface the sphenotic ar-
ticulates with the hyomandibular by fibrous tissue.

Epiotic. —Flat, more or less rounded; broadly
cartilage filled along all of its edges of articulation with
the other cranial bones; articulates by interdigitation an-
teriorly with the frontals, anteromedially with the round-
ed anterior portion of the supraoccipital, anterolaterally
with the sphenotic, posterolaterally with the pterotic,
and posteromedially with the exoccipital. The medial
edges of the two epiotics articulate with one another and
with the ventral edge of the supraoccipital spine by fi-
brous tissue.

teriorly; articulates by interdigitation anteroventrally
with the dorsolateral wing of the parasphenoid, antero-
dorsally mostly with the sphenotic but also with the
pterosphenoid, laterally with the sphenotic, postero-
laterally with the pterotic, and anteromedially with the
parasphenoid. From about the middle of its vertical
anteromedial edge the prootic possesses a thin splint of
bone projecting medially almost to the midline of the
skull, where it articulates by fibrous tissue with the
splintlike projection from the other prootic. These
delicate spines are evidently all that remains of the dor-
sal roofing of the myodome. The small myodome is now
roofed over only by a sheet of dense fibrous tissue which
begins anteriorly where it binds together the two medial
projections of the prootics and continues on posteroven-
trally to the floor of the cranial cavity, attaching pos-
teriorly to the region where the parasphenoid and basioc-
cipital interdigitate with the prootics. The myodome is
thus enclosed dorsally mainly by a fibrous tissue sheet,
and only at its upper anterior edge is the roof enclosed by
the medial projections of the prootics. Laterally the
myodome is bordered by the medial surfaces of the
prootics, ventrally by the dorsal surface of the para-
sphenoid, and posteriorly by the anterior end of the
basioccipital.

Orbital Region.

Frontal. — Laterally expanded in the middle of its
length; with a long posterolaterally directed arm. The
bone is generally dense and hard, but along the approxi-
mately anterior two-thirds of its medial portion a layer of
spongy connective tissue is present between the medial
surfaces of the two frontals. This spongy connective tis-
sue layer extends laterally into the frontal for about one-
fourth or one-fifth the width of the bone and anteriorly it
becomes cartilaginous and continuous with the ethmoid
cartilage. The frontal articulates by interdigitation dor-
somedially in the midline of the skull with its opposite
member, anteriorly with the ethmoid, which it broadly
overlies, anterolaterally with the prefrontal, postero-
laterally with the sphenotic, posteriorly on its dorsal sur-
face with the epiotic, and posteromedially on its dorsal
surface with the supraoccipital. From above the region
where it articulates with the sphenotic, the frontal
possesses a long posterolaterally directed process that ar-
ticulates by fibrous tissue with the anterodorsal edge of
the most laterally expanded portion of the pterotic. The
lateral fossa thus enclosed is bordered above by the fron-
tal and below by the sphenotic and pterotic. On its ven-
tral surface the frontal sends a ventromedially directed
process into the deeply concave dorsal surface of the ver-
tical interorbital projection of the parasphenoid, with ex-
tensive interdigitation occurring between the two bones.
Just lateral to its ventromedial projection, the frontal is
broadly overlain by the pterosphenoid, with slight inter-
digitation occurring between the two surfaces.

Prootic. — Cartilage filled along all of its edges of
articulation with the other cranial bones, except an-

Prefrontal. — In the form of a dorsolaterally ex-
panded column; cartilage filled at its rounded ventral

267

edge; articulates by fibrous tissue ventrally with the
region of articulation between the vomer and para-
sphenoid; articulates at its dorsal surface by interdigi-
tation anteromedially with the ethmoid and postero-
medially and posteriorly with the frontal. Along most of
its medial surface it is attached to the ethmoid cartilage,
with which it is continuous.

Parasphenoid. — Elongate, with a well-developed
ventral keel in the suborbital region. The anterior end of
the parasphenoid is deeply concave and into this con-
cavity the shaftlike posterior end of the vomer interdigi-
tates. The parasphenoid articulates laterally by fibrous
tissue with the base of the prefrontal in this region. Pos-
teriorly the parasphenoid slightly overlies the anterior
end of the basioccipital and strongly interdigitates with
it. About three-fourths the way back its length the para-
sphenoid gives rise to its paired dorsolateral wings that
interdigitate with the anteromedial edges of the prootics.
Posterior to this area of interdigitation with the antero-
medial edges of the prootics, the lateral edges of the
parasphenoid interdigitate with the ventromedial edges
of the prootics. In about the middle of its length the para-
sphenoid gives rise to an interorbital projection into
whose deeply concave dorsal surface the ventromedial
projections of the frontals interdigitate.

Pterosphenoid. — A thin plate; cartilage filled along
its dorsal edge; articulates by interdigitation postero-
ventrally with the prootic and posterolaterally with the
sphenotic. For most of its length the pterosphenoid over-
lies the posteromedial portion of the ventral surface of
the frontal, and slightly interdigitates with it.

ncave anterior end of the parasphenoid. Along the dor-
solateral surface of the vomer there is a laterally projec-
ting bony shelf above and below which fits the deeply
notched posterodorsal region of the palatine, the articu-
lation being by fibrous tissue.

Mandibular Region.

Hyomandibular. — Expanded dorsally, tapering to a
stout shaft anteroventrally; cartilage filled at its antero-
ventral and dorsal edges; articulates by fibrous tissue
dorsolaterally with the sphenotic and pterotic and dorso-
medially with the prootic and pterotic. Along the ventral
three-fourths of its posterior edge it articulates by fibrous
tissue with the preoperculum, while just above the dorsal
end of the latter the hyomandibular supports the articu-
lar face of the operculum. Anteriorly the hyomandibular
articulates by fibrous tissue principally with the pos-
terior region of the metapterygoid and symplectic.

Quadrate. — Wide posteriorly, but tapering to a
knob anteriorly for articulation with the articular in the
lower jaw; a short, posteriorly directed process present
from its posteroventral edge; cartilage filled at its pos-
terior edge; articulates by fibrous tissue anteriorly with
the articular, anterodorsally by interdigitation with the
slightly overlying ectopterygoid, dorsally by interdigi-
tation with the broadly overlying mesopterygoid, and
ventrally by fibrous tissue with the preoperculum.
Posteriorly the quadrate articulates through cartilage
with the metapterygoid and symplectic, the anterior part
of the latter being overlain by the short posterior process
of the quadrate.

Ethmoid Region.

Ethmoid. — Large, more or less rectangular; articu-
lates by interdigitation posterodorsally with the overly-
ing frontals and posterolaterally with the prefrontals.
Along the anterior half of its ventral surface the ethmoid
articulates with the vomer, the posterior part of this ar-
ticular surface showing interdigitation, while more
anteriorly the two bones are completely fused. In small
specimens (under approximately 100 mm) it can be seen
that the anterior edge of the ethmoid region is actually
formed by the upturned anterior edge of the vomer, even
though in large specimens no such distinction can be
made. Anterolaterally the ethmoid articulates by fibrous
tissue with the palatine, although the main articulation
of the latter is with the vomer. In large specimens the
ethmoid tends to become swollen and porous, or hy-
perostotic. Just behind its articulation with the vomer,
the posteroventral surface of the ethmoid is continuous
with the remains of the ethmoid cartilage.

Vomer. — Broad anteriorly, but tapering to a stout
shaft posteriorly; articulates with the ethmoid by fusion
anterodorsally but by interdigitation posterodorsally.
The posterior shaft of the vomer interdigitates with the

Metapterygoid. — Broad anteriorly; cartilage filled
at its anterior edge; articulates through cartilage
anteriorly with the quadrate and ventrally with the
symplectic, the latter also slightly overlying and inter-
digitating with the metapterygoid. The metapterygoid
articulates by fibrous tissue posteroventrally with the
hyomandibular, preoperculum, and interhyal, while
anterodorsally it broadly overlies and slightly interdigi-
tates with the mesopterygoid.

Symplectic. — Small, somewhat irregular in shape
from specimen to specimen; cartilage filled at its pos-
terior edge; articulates through cartilage and slight inter-
digitation anteriorly with the overlying process of the
quadrate and dorsally with the metapterygoid; articu-
lates by fibrous tissue posteroventrally with the preoper-
culum and interhyal.

Palato-Pterygoid Region.

Palatine. — Expanded posteriorly; a deep, ante-
riorly directed cleft present in its upper posterior edge
which fits tightly above and below the thin lateral
shelf of the vomer, articulating with it by fibrous tissue.
Anteroventrally the palatine interdigitates with the ec-
topterygoid, while posteriorly it overlies and interdigi-

tates with the mesopterygoid. Anterodorsally the pala-
tine is expanded laterally to form the articular facet for
support of the maxillary through fibrous tissue.

Ectopterygoid. — Somewhat V-shaped, with the
apex directed posteriorly; articulates by interdigitation
dorsally with the palatine, posteriorly with the
mesopterygoid, which it broadly overlies, and ventrally
with the quadrate, which slightly overlies it.

Mesopterygoid. — Variable in shape, but usually
thin anteriorly and thickened at its posterior edge. It ap-
pears to be relatively small as seen laterally, because it is
overlain by each of the four bones with which it articu-
lates by interdigitation; anterodorsally with the pala-
tine, anteriorly with the ectopterygoid, anteroventrally
with the quadrate, and posteroventrally with the
metapterygoid.

Opercular Region.

Operculum. — Thin and expanded posteroven-
trally; a dorsally directed process from its dorsal sur-
face present for muscle attachment; articulates by
fibrous tissue dorsally at its flattened articular surface
with the upper posterior edge of the hyomandibular,
while ventrally it articulates with the deeply cleft region
of the suboperculum, which it slightly overlies. In about
the middle of its anterior edge the operculum articulates
by a tough ligament with the rodlike interoperculum.

Suboperculum. — Thin; deeply and broadly cleft in
its upper region; the portion of the bone anterior to the
cleft somewhat prolonged dorsally; articulates by fibrous
tissue along the edges of its cleft region with the slightly
overlying operculum.

Interoperculum. — A long rod, with a small ventral
flange from about the middle of its ventral edge; articu-
lates by strong ligaments posteriorly with the operculum
and anteriorly with the angular in the lower jaw. The
ventral flange articulates by fibrous tissue with the
lateral surface of the epihyal.

Preoperculum. — Large; expanded posteroventral-
ly; slightly convex laterally; the anterior half of its dorsal
edge somewhat laterally expanded to present a broad
surface for fibrous tissue articulation with the ventral
edge of the quadrate; articulates by fibrous tissue along
the posterior half of its dorsal edge with the quadrate,
symplectic, interhyal, and metapterygoid.

Upper Jaw.

Premaxillary. — Posteromedial arm short; together
with the fused teeth forms a massive crushing plate; its
anterior edge forming the anterior border of the upper
jaw, except for a short distance ventrally where the max-
illary forms the border. The premaxillaries articulate
dorsomedially with one another by fibrous tissue, with

the articulation strengthened by a single row of a dozen
or more well-developed medial projections from the dor-
somedial surface of each premaxillary. These projec-
tions alternate with one another and decrease in size
anteriorly. The short posteromedial arm of the premaxil-
lary articulates by fibrous tissue with the anteroventral
surface of the vomer. Posteriorly and laterally the
premaxillary is firmly articulated by extensive interdigi-
tation with the broadly overlying maxillary. The
premaxillary contains an internal cavity which commu-
nicates with the exterior through a small opening placed
just lateral to the posteromedial arm that articulates
with the vomer. This cavity in the premaxillary contains
the dental pulp, which continually gives rise to the long,
thin, rodlike, highly modified teeth. These teeth lie
parallel to the anterior edge of the premaxillary and are
constantly being moved forward to replace those dental
lamellae worn down through use. There are usually 5 to
10 of these dental lamellae present in each premaxillary,
of which only the most anterior one or two are actually
exposed at the edge of the jaw. However, the lamellae
posterior to them can be seen easily through the very thin
portion of the premaxillary covering them, giving the im-
pression that many more of the dental lamellae are ex-
posed than is actually the case. On its ventral surface the
premaxillary bears a variable number (two to six) of
more or less anteroposteriorly compressed blunt teeth,
set in shallow sockets, in a longitudinal row just lateral to
its medial edge. These teeth, like the dental lamellae, are
replaced from the posterior end of the series by the ac-
tivity of the pulp tissue.

Maxillar>'. — Curved forward and slightly expanded
dorsally and ventrally; broadly overlies and firmly inter-
digitates with the premaxillary along all of its length, ex-
cept for a short distance ventrally where it is free of the
premaxillary and forms the anterior border of the upper
jaw. Posterodorsally at a slight groove on its surface the
maxillary is supported through fibrous tissue by the
anterodorsal edge of the palatine. The ventromedial sur-
face of the maxillary articulates by fibrous tissue with
the dorsolateral surface of the dentary.

Lower Jaw.

Dentary. — Somewhat squarish; articulates poste-
riorly by fibrous tissue and interdigitation with the artic-
ular, which it broadly overlies laterally but only slightly
overlies medially. Posteroventrally the dentary articu-
lates by fibrous tissue, or in large specimens by inter-
digitation, with the angular. The dorsolateral surface of
the dentary articulates by fibrous tissue with the ventro-
medial surface of the maxillary. The dentary is hollow in-
ternally, although not as extensively so as the premaxil-
lary. The cavity contains the dental lamellae producing
pulp material, which functions in the same way as de-
scribed for the premaxillary. The pulp cavity opens to
the exterior at the region of articulation of the anterior
end of the articular with the dentary. Only long, thin,
rodlike, dental lamellae are produced, there being no

stubby trituration teeth as there are on the ventral sur-
face of the premaxillary. Ventromedially the dentary ar-
ticulates by fibrous tissue with its opposite member, with
about 15 stubby projections strengthening the articu-
lation, as in the case of the premaxillary.

Articular. — More or less triangular in shape, with
the apex pointing anteriorly into the hollowed out pos-
terior part of the dentary; cartilage filled at its anterior
edge where it is continuous with the remains of Meckel's
cartilage; articulates by fibrous tissue and interdigi-
tation anteriorly with the dentary, posteriorly by fibrous
tissue at a groove on its surface with the knoblike
anterior end of the quadrate, posteroventrally by fibrous
tissue, and interdigitation in large specimens, with the
angular. The sesamoid articular is a rod of bone lying
alongside the anteromedial surface of the articular just
posterior to the region where the latter interdigitates
with the dentary.

Angular. — Small; articulates by fibrous tissue with
the dentary and articular, or in large specimens by inter-
digitation with both of these bones. Posteriorly the angu-
lar makes ligamentous connection with the anterior end
of the interoperculum.

BRANCHIAL APPARATUS.

Hyoid Arch and Branchiostegal Rays.

Hypohyals. — Both hypohyal elements present but
not large, the dorsal hypohyal being particulfirly small;
dorsal hypohyal cartilage filled at its ventral edge and
ventral hypohyal cartilage filed at its posterior edge. The
hypohyals articulate through cartilage with one another
and with the ceratohyal, while they articulate by fibrous
tissue anteromedially with their opposite members.

Ceratohyal. — Elongate, somewhat expanded pos-
teriorly; cartilage filled at its anterior and posterior
edges; articulates through cartilage anteriorly with the
hypohyals and through cartilage and interdigitation
posterodorsally with the epihyal. Of the six branchios-
tegal rays, only five usually articulate by direct contact
through fibrous tissue with the ceratohyal.

Epihyal. — Rounded posteriorly; cartilage filled at
its anterior edge; articulates anteriorly through cartilage
and interdigitation with the ceratohyal and by fibrous
tissue posterodorsally with the interhyal and laterally
with the ventral flange of the interoperculum.

Interhyal. — Short and columnar; cartilage filled at
its dorsal and ventral edges; articulates by fibrous tissue
ventrally with the epihyal and dorsally with the fibrous
tissue sheet between the symplectic, metapterygoid, and
preoperculum.

Branchiostegal rays. — Six in number; the first
branchiostegal ray a large, relatively flat horizontal plate

with a down -turned lateral edge which anteriorly be-
comes a thick vertical flange for articulation by fibrous
tissue with a vertical groove in the middle of the medial
surface of the ceratohyal. The second branchiostegal ray
is the longest of the six elements and it is a normal unex-
panded shaft articulating by fibrous tissue with the ven-
tral edge of the ceratohyal just posterior to the region of
articulation of the first branchiostegal ray. The third,
fourth, and fifth branchiostegal rays are also normal in
shape, increasing slightly in length from the third to fifth
and articulating by fibrous tissue with the lateral sur-
face of the posteroventral region of the ceratohyal. The
sixth branchiostegal ray is sometimes broad posteriorly,
but always tapers to a narrow shaft anteriorly where it
articulates by fibrous tissue with the dorsal edge of the
fifth branchiostegal ray, rather than with the cera-
tohyal.

Branchial Arches. — All the elements are cartilage
filled at their edges of articulation with the other ele-
ments of the series, and the articulations are usually
mediated through cartilage and fibrous tissue. The bran-
chial arches are composed of three basibranchials, three
pairs of hypobranchials, five pairs of ceratobranchials,
four pairs of epibranchials, and three pairs of pharyngo-
branchials. Three gills are present; the fourth arch pos-
sesses no gill and there is no slit between it and the lower
pharyngeal.

First arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. First basi-
branchial elongate and rodlike; displaced forward so that
it articulates posteriorly with the second basibranchial
and posterolaterally with the first hypobranchials. First
hypobranchial slightly expanded dorsally; the largest of
the hypobranchial elements, which decrease in size pos-
teriorly in the series; articulates ventrally with the re-
gion of articulation of the first basibranchial with the
second basibranchial and dorsally with the first cera-
tobranchial. First ceratobranchial the shortest of the
ceratobranchial elements, which, except for the fifth
ceratobranchial, increase in size posteriorly in the series;
possesses a ventrally directed flange along most of its
ventral surface, this flange on the succeeding cera-
tobranchials decreases in size posteriorly until it is al-
most absent on the last ceratobranchial; articulates dor-
sally with the first epibranchial. First epibranchial a nar-
row rod; the shortest of the epibranchial elements, which
increase in length posteriorly in the series; articulates
dorsally with the first pharyngobranchial. First pharyn-
gobranchial (suspensory pharyngeal) a rounded plate
with a short ventral process for articulation with the first
epibranchial; toothless; placed in line with the two tooth
bearing pharyngobranchials rather than being oriented
toward, and having its distal end attached by fibrous
tissue to, the ventral surface of the skull. The dorsal re-
gions of the branchial arches are held to the ventral sur-
faces of the parasphenoid and prootics by fibrous tissue
attaching to the dorsolateral surfaces of the epibran-
chials and pharyngobranchials.

Second arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Second basi-
branchial short, but much expanded laterally both
anteriorly and posteriorly; articulates anteriorly with the
first basibranchial, posteriorly with the third basi-
branchial, anterolaterally with the first hypobranchials,
and posterolaterally with the second hypobranchials.
Second hypobranchial a short rod; articulates ventrally
with the posterolateral surface of the second basi-
branchial and dorsally with the second ceratobranchial,
which in turn articulates dorsally with the second
epibranchial. Second epibranchial a slender rod; articu-
lates dorsally with the ventral arm of the second
pharyngobranchial. Second pharyngobranchial roughly
L-shaped; bearing a single row of 12 to 17 well-de-
veloped teeth, which are relatively sharp pointed and
curved slightly posteriorly, along the larger of the two
wings of the L. The toothless ventral wing articulates
principally with the second epibranchial, but also, to a
lesser extent, with the third epibranchial. The teeth are
set in shallow sockets and are replaced by new teeth de-
veloping in sockets just anterior to and below the bases of
the old teeth. The second pharyngobranchial is held to
the other two pharyngobranchial elements by fibrous tis-

Third arch. — Basi-, hypo-, cerato-, epi-, and pha-
ryngobranchial elements present. Third basibranchial
like second basibranchial, but larger; articulates
anteriorly with the second basibranchial, posteriorly
with the fourth ceratobranchials and posterolaterally
with the third hypobranchials. Third hypobranchial with
an anteroventral process that articulates by fibrous
tissue with the ventral surface of the first basibranchial;
posteriorly it articulates with the posterolateral surface
of the third basibranchial and with the ventral edge of
the third ceratobranchial. Third ceratobranchial articu-
lated ventrally with the third hypobranchial and dorsally
with the third epibranchial. Third epibranchial a stout
rod, with a stubby projection from its posterolateral sur-
face articulating by fibrous tissue with a similar process
from the fourth epibranchial; articulates dorsally with
the second and third pharyngobranchials. Third pharyn-
gobranchial roughly L-shaped, with the larger of the two
arms bearing about 10 to 12 teeth in a single row; teeth
similar to those of the second pharyngobranchial and
replaced in a similar manner; articulates by the smaller
of its two arms ventrally with the third and fourth
epibranchials.

Fourth arch. — Cerato- and epibranchial elements
only. Fourth ceratobranchial the longest of the cerato-
branchial elements; articulates ventrally with the third
basibranchial and dorsally with the fourth epibranchial.
Fourth epibranchial an elongate rod, the longest of the
epibranchial elements; possesses a stubby projection
from its anterolateral edge which articulates through
fibrous tissue with the similar process on the third
epibranchial; articulates dorsally with the third
pharyngobranchial.

Fifth arch. —Ceratobranchial (lower pharyngeal)
element only. Fifth ceratobranchial a stout rod, slightly
expanded medially in about the first one-third of its
length; toothless; articulates ventrally with the base of
the fourth ceratobranchial.

PAIRED FIN GIRDLES.

Pectoral Fin.

Supracleithrum. —In position at about a 45° angle
to the axis of the body; a stout shaft with a lateral flange
along most of its length; articulates by fibrous tissue dor-
sally with a slight concavity on the posterior surface of
the pterotic and ventrally with the cleithrum, which it
overlies.

Cleithrum. — Laterally expanded along most of its
length; articulates by fibrous tissue dorsally with the
overlying supracleithrum and with the dorsal post-
cleithrum, which it overlies. Along its posterior surface it
articulates by fibrous tissue with the scapula and cora-
coid. Ventromedially the cleithrum articulates by tough
fibrous tissue with its opposite member in the midline of
the body between the medial edges of the horizontal
platelike portions of the first branchiostegal rays.

Postcleithra. — The postcleithra form a long strut,
widened posteriorly, along the abdominal wall muscula-
ture from the supracleithrum to about halfway back the
length of the abdominal cavity. The dorsal postcleithrum
articulates by fibrous tissue anterolaterally with the
overlying cleithrum and posterolaterally with the ventral
postcleithrum, which is much deeper and more laterally
compressed than the dorsal postcleithrum.

Coracoid. — Rounded dorsally, but tapering to a
point anteroventrally; a posterodorsally directed spine-
like process present from its posteroventral edge; carti-
lage filled at its dorsal edge; articulates anteriorly by
fibrous tissue with the cleithrum and dorsally through
cartilage with the scapula and the second to fourth ac-
tinosts.

Scapula. — Scapular foramen not entirely enclosed
by the scapula, but, rather, with its anterior edge closed
by the cleithrum; cartilage filled at its dorsal, ventral,
and anterior edges; articulates by fibrous tissue anterior-
ly with the cleithrum, posteriorly by slight interdigi-
tation dorsally with the small first actinost and ventrally
with the base of the second actinost, and through carti-
lage ventrally with the coracoid.

Actinosts. — Four elements; all cartilage filled at
both ends, except for the first, which is only cartilage fill-
ed at its dorsal edge. The first actinost is reduced to a
small wedge between the upper regions of the scapula
and the second actinost, articulating with the former by
interdigitation and with the latter through cartilage. The
second and third actinosts are constricted in the middle,

271

with the third being slightly larger than the second. The
fourth actinost is only concave on the side toward the
third actinost. The actinosts articulate with one another
dorsally through cartilage and ventrally, except for the
small first actinost, with one another by slight interdigi-
tation. The second actinost interdigitates antero-
ventrally with the base of the scapula, while the second
to fourth actinosts articulate ventrally through cartilage
with the coracoid. Distally the actinosts support all of
the fin rays, except for the small first fin ray, which ar-
ticulates with the scapula.

Fin rays. — Usually 17 to 19 fin rays present; the
first ray small, its medial half thicker than its lateral
half; first ray articulated with the scapula, but the other
rays articulated to the actinosts. The first two rays and
last ray unbranched, the other rays branched. First ray
without cross-striations, the other rays cross-striated.

VERTEBRAL COLUMN. —All vertebrae with bicon-
cave centra, except the last, which ends posteriorly in the
modified urostyle and fusion with some of the hypurals.

Abdominal Vertebrae.

First vertebra. — Neural spine befid and relatively
short; neural arch with a complete bony roof over the
neural canal, as do all the other vertebrae; articulates by
fibrous tissue laterally at a shallow but broad concavity
over the upper anterior half of its centrum with the
posteroventrally projecting spinelike exoccipital con-
dyles. Along the lizn of its concave anterior end the
centrum articulates by fibrous tissue with the rim of the
round posterior end of the basioccipital. The first
vertebra articulates by fibrous tissue poaterodorsally at a
shallow concavity on its lateral surface with the neural
prezygapophysis of the second vertebra, while
posteroventrally from its lateral edge it sends a large
process posteriorly to fit under a shallow groove on the
anterior half of the ventrolateral surface of the centrum
of the second vertebra.

Other abdominal vertebrae. — In 12 specimens, the
abdominal vertebrae numbered eight. The second and
third abdominal vertebrae have bifid neural spines very
similar to that of the first vertebra. The fourth abdomi-
nal vertebra, however, has the neural spine bifid
anteriorly but single posteriorly. All the vertebrae, both
abdominal and caudal, posterior to the fourth abdominal
vertebra have undivided neural spines. The neural spines
of the fifth to eighth abdominal vertebrae become in-
creasingly thiimer and longer. From the second to eighth
abdominal vertebrae the neural and haemal prezy-
gapophyses gradually become larger, but the haemal
postzygapophyses remain about the same size so that
each of the last few abdominal vertebrae articulates ven-
trally by the apposition of its posteroventrally projecting
haemal postzygapophyses with the anteroventrally pro-
jecting haemal prezygapophyses of the vertebra pos-

terior to it. No haemal arches or spines are present. The
neural spine of the seventh abdominal vertebra articu-
lates by fibrous tissue along its posterior edge with the
first basal pterygiophore of the dorsal fin, while the
neural spine of the eighth abdominal vertebra articu-
lates between the first and second basal pterygiophores
of the dorsal fin.

Caudal Vertebrae. — In 12 specimens, the caudal
vertebrae numbered 11 in 10 specimens and 10 in 2 speci-
mens. All the caudal vertebrae, and only the caudal
vertebrae, possess haemal arches, while only those verte-
brae just below and posterior to the last basal pterygio-
phore of the anal fin possess large haemal spines. The
haemal arches of the first four caudal vertebrae (those
without conspicuous haemal spines) are modified for ar-
ticulation by fibrous tissue with the dorsal ends of the
basal pterygiophores of the anal fin. From the haemal
prezygapophyses of each of these vertebrae there is one
projection directed anteriorly and another directed pos-
teromedially. The anterior projections from the haemal
prezygapophyses on each side of the vertebra do not meet
in the midline under the centrum, while the postero-
medial projections from each side do meet and fuse in the
midline under the centrum at the open area between the
anterior projections from the haemal prezygapophyses of
the vertebra behind it. Thus, the posteromedial projec-
tions from the haemal prezygapophyses form the haemal
arch, which is here displaced posterodorsally from the
position it would more normally assume. The anterior
projections from the haemal prezygapophyses are much
shorter than the posteromedial projections, except in the
case of the first caudal vertebra, in which the anterior
projection is much enlarged and somewhat ventrally di-
rected for support along its ventral surface of the dorsal
end of the enlarged first basal pterygiophore of the anal
fin. The anterior projections from the haemal prezyga-
pophyses of the second and third caudal vertebrae sup-
port through fibrous tissue, respectively, the second and
third anal fin basal pterygiophores. The fourth caudal
vertebra supports the fourth and fifth anal fin basal
pterygiophores from, respectively, its anterior haemal
prezygapophyseal projections and from its haemal arch.
The fifth caudal vertebra supports the sixth anal fin
basal pterygiophore from its haemal arch. The fifth cau-
dal vertebra, in all but one of the study and radio-
graphed specimens, possesses a haemal spine. The pos-
teromedial projections of the haemal prezygapophyses of
the fifth caudal vertebra not only fuse to each other in
the midline, but also fuse to the ventrolateral edges of
the centrum, leaving open a medial canal beneath the
centrum and enclosing a large foramen laterally. From
the area of fusion with the centrum, a long haemal spine
is given off ventrally from the fifth caudal vertebra to
support by fibrous tissue the last anal fin basal pterygio-
phore. The haemal apparatus of the sixth to ninth verte-
brae is similar to that of the fifth vertebra in that the
haemal arch fuses posterolaterally to the centrum and
completely encloses the haemal canal, except laterally at
the foramen. The haemal spines of the sixth and seventh

caudal vertebrae are the largest of those of the caudal
vertebrae. In large specimens (over 200 mm) these two
haemal spines become enlarged and swollen, taking on
the hyperostotic consistency that is also characteristic of
the ethmoid in large specimens. The haemal structures
of the last two vertebrae are autogenous.

The neural spines of the first five caudal vertebrae are
slender rods which support most of the basal pterygio-
phores of the dorsal fin. They arise from the posterodor-
sal surface of the neural arches at the region of the neural
postzygapophyses. The neural spine of the first caudal
vertebra is placed between the third and fourth basal
pterygiophores and that of the fourth caudal vertebra
behind the ninth and last basal pterygiophore. The
neural spine of the fifth caudal vertebra is larger than
those of the caudal series anterior to it, but it is rela-
tively small in comparison to the neural spine of the sixth
caudal vertebra. Posterior to the sixth caudal vertebra
the neural spines become progressively smaller. It is of
interest to note that whereas the haemal spines of the
sixth and seventh caudal vertebrae are swollen and
spongy in the two largest study specimens, the neural
spines of these same vertebrae are normal thin hard
plates.

Caudal Skeleton. — The supporting structure of the
caudal fin appears to be a single large block of bone,
square in its lateral outline, but it is a compound struc-
ture formed of four separate bony elements firmly inter-
digitated with one another. The ventral portion of the
caudal skeleton is formed by the parhypural. A slight
concavity is present in about the middle of its dorsal sur-
face just below a similar concavity on the ventral surface
of the bony element that lies above it, which is the fused
centrum-lower hypurals. The parhypural and the fused
centrum-lower hypurals are extensively interdigitated
with each other, except in the region where the concavi-
ties occur in their apposed surfaces so that a foramen is
enclosed between them. This foramen is the last opening
into the haemal canal, the latter being connected to the
canal in the preceding vertebrae by a small longitudinal
groove along the articular surface between the
parhypural and the centrum-lower hypural element
anterior to the foramen. Anteriorly the parhypural is
laterally expanded to form haemal prezygapophyses.
The epural is much like the parhypural in shape, but it is
shorter and only forms the dorsal margin of the anterior
two-thirds of the caudal skeleton. Along most of its ven-
tral surface it firmly interdigitates with the centrum but
posteriorly it interdigitates with the autogenous upper
hypural plate. Concavities exist on the ventral and dorsal
surfaces, respectively, of the epural and centrum so that
a foramen is enclosed by the otherwise apposed surfaces
of the two bones. A deep longitudinal groove is present
along the ventral surface of the epural and the dorsal sur-
face of the centrum which permits the neural canal to be
present in the midline of the otherwise interdigitated sur-
faces of these two elements. The neural canal opens
anteriorly into the neural canal of the preceding verte-
brae, while posteriorly it opens to the exterior between

the posterior end of the epural and the dorsal surface of
the autogenous upper hypural plate. The anterior end of
the epural is expanded laterally to form the neural
prezygapophyses. The fused centrum-lower hypurals
form the largest bone of the caudal skeleton, its anterior
portion corresponding to the last vertebral centrum and
abbreviated urostyle, and its posterior portion corres-
ponding to what in generalized plectognaths would be
the first and second hypurals. The anterior portion inter-
digitates dorsally with the epural, posterodorsally with
the upper hypural plate and ventrally with the
parhypural. The posterodorsal region of the anterior por-
tion, which represents the urostyle, is extensively inter-
digitated with the anteroventral edge of the upper
hypural plate. The posteroventrally expanded portion
representing the lower hypurals articulates by interdigi-
tation dorsally with the upper hypural plate and ventral-
ly with the parhypural. There is a short concavity, or
blindly ending tube, proceeding posteriorly and slightly
dorsally into the lower hypural region from the area of
the haemal foramen between the centrum and parhy-
pural. This tube probably encloses the end of the haemal
canal. The lower hypural region supports the lower six
caudal fin rays, with the support of the lowermost ray
shared with the parhypural. The autogenous upper
hypural plate forms the posterodorsal region of the
caudal skeleton and represents what in generalized plec-
tognaths would be the third to fifth (or, in Triodon, third
and fourth) hypurals. It interdigitates strongly with the
centrum-lower hypural plate and with the posterior end
of the epural. The upper hypural plate supports the up-
per five caudal fin rays.

Caudal fin rays. — Eleven in number, the upper-
most ray and lowermost two rays unbranched, the inter-
vening rays becoming increasingly branched toward the
middle rays, which are branched in at least triple dicho-
tomies. The upper five rays articulate by fibrous tissue at
their bifid bases to the upper hypural plate and the lower
six rays mainly to the lower hypural region, as described
above.

DORSAL AND ANAL FINS.

Fin rays and pterygiophores.— Fi/tecn fin rays pres-
ent in most specimens, the first two rays and the last one
or two rays unbranched, the others branched in single or
double dichotomies. The first ray is small and rudimen-
tary, barely protruding to the surface, formed of two
separate halves and without cross-striations, while all
the other rays are cross-striated. Distal pterygiophores
are either absent or unossified. The fin rays are sup-
ported below by nine basal pterygiophores which are
strongly interdigitated with one another dorsally. The in-
temeural processes of these pterygiophores are large
anteriorly in the series but decrease in size posteriorly.
The first and second basal pterygiophores are placed well

in advance of the base of the dorsal fin. The intemeural
process of the first basal pterygiophore lies between the
neural spines of the seventh and eighth abdominal verte-
brae and those of the last two basal pterygiophores
between those of the third and fourth caudal vertebrae.
On the upper end of each of the basal pterygiophores
there is a short columnar projection whose deeply con-
cave surface is cartilage filled, except for the last
pterygiophore, which has two of these facets with which
the fin rays articulate by fibrous tissue. The pterygio-
phores are also cartilage filled at their ventral edges.
From its anterodorsal edge the first basal pterygiophore
has an anteriorly directed process, lying in the midline of
the body just beneath the skin, which reaches to about
the level of the tip of the neural spine of the sixth ab-
dominal vertebra. At its anterior end it is attached by
fibrous tissue to a rodlike supraneural. The supraneural
lies just below the skin in the midline of the body from
the level of the tip of the neural spine of the sixth ab-
dominal vertebra to about the region above the fourth
abdominal vertebra. The supraneural is surrounded by a
longtitudinal band of muscles which lies between the
dorsomedial edges of the general epaxial muscle masses.
This longtitudinal band of muscles continues anteriorly
and partially fills the space between the bifid neural
spines of the first three abdominal vertebrae and ends by
connecting firmly with the supraoccipital spine. The
ninth, or last, basal pterygiophore has a posteriorly di-
rected spine from its posterodorsal edge that ends over
the neural spine of the fifth caudal vertebra. The fact
that the ninth basal pterygiophore bears two columnar
articular facets may indicate that it is a fusion product of
two originally separate basal pterygial elements. None of
the basal pterygiophores possess lateral projections along
its length for muscle attachment.

the sixth basal pterygiophore, which usually has three of
them. From these columnar facets the fin rays articulate
by fibrous tissue. The pterygiophores are also cartilage
filled at their dorsal ends. As is the case with the last
basal pterygiophore of the dorsal fin, the presence of
more than one columnar articular process may indicate
that the first and the last anal fin basal pterygiophores
are the fusion products of two or more originally sep-
arate elements. The sixth basal pterygiophore bears from
its posteroventral edge a posterior projection that ex-
tends back under the haemal spine of the sixth caudal
vertebra. None of the pterygiophores possess lateral pro-
jections along their lengths for muscle attachment.

Detailed description of Canthigaster rostrata.

Material examined. — Five cleared and stained speci-
mens, 10.0-59.6 mm, and one dry, partially disarticu-
lated skeleton, approximately 96.5 mm.

SKULL.

Occipital Region.

Basioccipital. — A long column, laterally expanded
anteriorly; cartilage filled at its anterior and postero-
lateral edges; articulates anteriorly by interdigitation
with the slightly overlying parasphenoid, anterolaterally
with the prootics, and laterally with the exoccipitals. The
posterior concave end of the basioccipital is somewhat
depressed, its rim articulating by fibrous tissue with the
rim of the anterior concave face of the first vertebra. The
dorsal edge of the basioccipital forms the ventral wall of
the foramen magnum.

Anal Fin.

Fin rays and pterygiophores. — Fourteen fin rays
present in most specimens, the first two rays and the last
one or two rays unbranched, the others branched in
single or double dichotomies. The first ray is small and
rudimentary, like that of the dorsal fin. Distal pterygio-
phores are either absent or unossified. The fin rays are
supported below by six basal pterygiophores which are
strongly interdigitated with one another ventrally. The
"interhaemal" processes of these pterygiophores are
large anteriorly in the series but decrease in size pos-
teriorly, with the first basal pterygiophore being by far
the largest of the series. The first, second, third, and
fourth basal pterygiophores articulate by fibrous tissue
dorsally with the haemal prezygapophyses of, respec-
tively, the first, second, third, and fourth caudal verte-
brae. The fifth and sixth basal pterygiophores articulate
similarly with about the middle of the haemal arches of,
respectively, the fourth and fifth caudal vertebrae. On
the ventral surface of each pterygiophore there is a short
columnar projection whose ventral edge is deeply con-
cave and cartilage filled, except for the first basal
pterygiophore, which has two of these projections, and

Exoccipital. — Cartilage filled at its anterior and
ventromedial edges; articulates by interdigitation ven-
tromedially with the basioccipital, ventrolaterally with
the pterotic, anterodorsally with the epiotic, and dorso-
medially with the supraoccipital. At the extreme pos-
terior end of its dorsomedial edge the exoccipital inter-
digitates with its opposite member to form the dorsal
wall of the foramen magnum, while the posteromedial
edges of the exoccipitals form the lateral walls of that
foramen. From its posteroventral edge the exoccipital
condyle projects posteriorly well past the posterior end of
the basioccipital and articulates by fibrous tissue over
the lateral surface of the anterior half of the centrum of
the first vertebra.

Supraoccipital. — Flattened anteriorly, but drawn
out into a deep laterally compressed spine posteriorly;
cartilage filled along its anterior edge; articulates by in-
terdigitation anteriorly with the frontals, which
somewhat overlie it, laterally with the epiotics and pos-
terolaterally with the exoccipitals. The posterior edge of
the supraoccipital spine is slightly expanded laterally,
particularly in the dorsal region. The spine does not pro-
ject posteriorly far beyond the end of the skull.

Otic Region.

Pterotic. —Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates by
interdigitation anterodorsally with the sphenotic, dorso-
medially with the epiotic, ventromedially with the exoc-
cipital, and anteroventrally with the prootic and
sphenotic. Placed transversely along the middle of its
ventral surface, the pterotic possesses a low ventrally di-
rected flange for muscle attachment. Anterior to this
ventral flange the pterotic has a slight concavity on its
surface for articulation by fibrous tissue with the pos-
terior half of the dorsal edge of the hyomandibular.

Sphenotic. — Cartilage filled at its medial edge; ar-
ticulates by interdigitation anterodorsally with the
frontal, anteromedially with the pterosphenoid, ventro-
medially with the prootic, posteroventrally and postero-
dorsally with the pterotic, and posteromedially on its
dorsal surface with the epiotic. On its ventral surface the
medial edge of the sphenotic articulates with the anterior
end of the dorsal edge of the hyomandibular by fibrous
tissue.

Epiotic. —Somewhat expanded laterally; cartilage
filled along all of its edges of articulation with the other
cranial bones; articulates by interdigitation anteriorly
with the frontal, anterolaterally with the sphenotic,
laterally with the pterotic, posteriorly with the exoccipi-
tal, and medially with the supraoccipital.

Figure l%.—Lagocephalus laevigatua:
lateral view of head, composite

based on several specimens,
61.4-166 mm SL, Gulf of Meitico.

^-^r

W

Figure \97 .—Lagocephaliu laevigaliu: dorsal
(left) and ventral (right) views of skull, with in
at top right showing a ventral view detail of

the articulation of the upper jaw with the
palatines and vomer, composite based on several
61.4-166 mm SL, Gulf of Mexico.

276

Fipire \9S.—Lagocephalus laevigatua:

posterior view of skull (left); above,

lateral and posterior views of first

abdominal vertebra; posterior view of

orbit (right) (cross section of skull;
dashed lines represent cut surfaces of
frontals, pterosphenoids, and parasphenoid);
composite based on several

61.4-166 mm SL, Gulf of Mexic

Prootic. — Cartilage filled along all of its edges of
articulation with the other cranial bones, except
anteriorly; articulates by interdigitation anteroventrally
with the dorsolateral wing of the parasphenoid, antero-
dorsally with the pterosphenoid and sphenotic, ventro-
laterally with the sphenotic, ventromedially with the
parasphenoid, and posteromedially on its ventral sur-
face with the basioccipital. The more or less vertical an-
teromedial edge of the prootic possesses a slight medial
protuberance which probably marks the former dorsal
limit of the now absent bony myodome and corresponds
to the medially directed prongs representing the rudi-
ments of the bony dorsal roof of the myodome in such
tetraodontids as Lagocephalus laevigatas. The func-
tional myodome is roofed over by a fibrous tissue sheet
between the two prootics and the region of articulation
between the parasphenoid and basioccipital. The func-
tional myodome is thus bounded dorsally by a fibrous
tissue sheet, laterally by the prootics, ventrally by the
dorsal surface of the parasphenoid, and posteriorly by
the anterior end of the basioccipital.

Orbital Region.

Frontal. — Thin throughout most of its length, ex-
cept posterolaterally; articulates by slight interdigi-
tation dorsomedially with its opposite member, while its
ventromedial edge makes fibrous tissue contact with the
cartilaginous rod which is continuous anteriorly with the
ethmoid cartilage; articulates by slight interdigitation
anteriorly with the ethmoid, which it slightly overlies,
anterolaterally with the prefrontals, which it broadly
overlies, posteromedially with the supraoccipital, which
it slightly overlies, posteriorly with the epiotic, and pos-
terolaterally with the sphenotic. In the posterior wall of
the orbit the frontal articulates laterally with the spheno-

tic and medially with the pterosphenoid. The most dis-
tinctive feature of the frontal concerns the shape of its
laterally expanded posterior portion. Just behind the
region that forms the posterolateral wall of the orbit, the
dorsal surface of the frontal possesses a thin laterally di-
rected flange. This flange makes contact with the ex-
treme posterolateral edge of the frontal, just above the
area of articulation between the frontal and sphenotic. In
small specimens the articulation between the flange of
the frontal and the posterolateral edge of the portion of
the orbital wall formed by the frontal is by close apposi-
tion through fibrous tissue, but in larger specimens these
two surfaces of the frontal tend to fuse together. The
space between the lateral flange and the posterolateral
region of the orbital wall is filled with a muscle mass
which connects ventrally with the uppermost end of the
operculum (see Winterbottom 1974).

Prefrontal. — Thin and more or less flattened
dorsally, but tapering to a stout column anteroventrally;
cartilage filled at its ventromedial edge, where it is con-
tinuous with the ethmoid cartilage; articulates by slight
interdigitation posterodorsally with the frontal, antero-
dorsally with the ethmoid, and anteroventrally with the
parasphenoid.

Parasphenoid. — Elongate, with a well-developed
ventral keel in the suborbital region. The anterior end of
the parasphenoid is concave and receives the posterior
end of the vomer, with which it interdigitates. The dorsal
edge of the parasphenoid anterior to the orbit is slightly
concave and interdigitates with the posterior half of the
ventral edge of the ethmoid, while just posterior to this
region it interdigitates with the base of the prefrontal.
The flattened, somewhat laterally expanded, posterior
end of the parasphenoid slightly overlies and interdigi-

ceratobranchials

epibranchials

Figure l99.—Lagocephalus taevigatut: dorsal

view of branchial arches (extended on lower side);

lateral (above) and medial (below, upside down)

views of hyoid arch; composite based on several

specimeiu, 61.4-166 mm SL, Gulf of Mexico.

— basibranchials

1st

branchiostegal rays

epihyal

Figure 200.— Lagocephalns laevigatus: lateral

view of pectoral girdle, composite based on

several specimens, 61.4-166 mm SL, Gulf of Mexico.

caudal vertebrae

Figure 201.— Lagocephalus laevigatas: lateral view of
caudal Rn supporting structures, composite based on
several specimens, 61.4-166 mm SL, Gulf of Mexico.

Figure 202.— Canthigaater roatrata (A, above),

with C. valentini (B) and two renditions

of C. amboinensia (C): the lower showing

the course of the lateral line canals

and their major pores as decifered by

placing drops of ink on each pore

found by microscopic search of a

partially drying specimen; in front

of the top three figures, nasal

region as seen externally (above)

and scales from upper middle

region of body.

tales with the anterior end of the basioccipital, while just
anterior to this region the parasphenoid interdigitates
with the ventromedial edges of the prootics. About three-
fourths the way back its length the parasphenoid gives
rise to its dorsolateral wings which interdigitate with the
anteromedial edges of the prootics.

Pterosphenoid. — A small, thin plate; cartilage
filled at its dorsal and lateral edges; articulates by slight
interdigitation dorsally with the frontal, laterally with
the sphenotic and ventrally with the prootic.

Ethmoid Region.

Ethmoid. — More or less T-shaped in cross section;
its dorsal surface laterally expanded, increasingly so pos-
teriorly, while its ventral surface is drawn out into a thin
medial flange in the posterior two-thirds of its length; ar-
ticulates by slight interdigitation pwsteriorly with the
frontals and prefrontals, posteroventrally with the some-
what overlying parasphenoid and anteroventrally with
the vomer.

Vomer. — Thin and laterally compressed pos-
teriorly, but dorsoventrally depressed anteriorly; anterior
end concave; anterolateral edge with a deep indentation
around which the anterodorsal forked region of the pala-
tine fits and is held by fibrous tissue; articulates by slight
interdigitation dorsally with the ethmoid and pos-
teriorly with the concave anterior end of the

Mandibular Region.

Hyomandibular. — Expanded dorsally, tapering
gradually to a stout shaft anteroventrally; cartilage filled
at its dorsal and anteroventral edges; articulates by
fibrous tissue dorsolaterally with the sphenotic and
pterotic and dorsomedially with the prootic and pterotic.
The ventral three-fourths of the posterior edge of the hyo-
mandibular articulates by fibrous tissue with the
preoperculum, while just above the dorsal end of the
latter the hyomandibular supports the articular face of
the operculum. At its anteroventral end the hyo-
mandibular makes fibrous tissue contact mainly with the
preoperculum, and only secondarily with the symplectic
and metapterygoid.

Quadrate. — Somewhat wider posteriorly than
anteriorly; prolonged posteroventrally into a thin taper-
ing piece below the metapterygoid; cartilage filled only
for a short distance along its posterodorsal edge; articu-
lates by fibrous tissue anteriorly with the articulfir in the
lower jaw and ventrally with the preoperculum. It ar-
ticulates by interdigitation dorsally with the overlying
ectopterygoid, while posterodorsally it overlies and inter-
digitates with the metapterygoid.

Metapterygoid. — A more or less rounded plate;
cartilage filled only for a short distance along its antero-

ventral edge; articulates by interdigitation anteriorly
with the ectopterygoid and palatine, dorsally with the
mesopterygoid, anteroventrally with the quadrate, and
posteroventrally with the symplectic.

Symplectic. — A thin delicate rod; cartilage filled at
its posterior edge; articulates by interdigitation dorsally
with the metapterygoid and by fibrous tissue ventrally
with the preoperculum.

Palato-Pterygoid Region.

Palatine. —Expanded posteriorly; a short
anteriorly directed cleft present in the anterior part of its
dorsal edge; articulates anterodorsally by firm fibrous
tissue attachment of its cleft surface in the anterolateral
indentation of the vomer; articulates by interdigitation
ventrally with the ectopterygoid, posteroventrally with
the metapterygoid, and posterodorsally with the
mesopterygoid.

Ectopterygoid. — Roughly triangular, its posterior
end tapering to a blunt point; articulates by interdigi-
tation dorsally with the palatine, posteriorly with the
metapterygoid, and ventrally with the quadrate, which it
broadly overlies.

Mesopterygoid. — Broad anteriorly, but becoming
narrower posteriorly; articulates by interdigitation
anteriorly with the palatine and ventrally with the
metapterygoid, both of which bones overlie the
mesopterygoid to a variable extent.

Opercular Region.

Operculum. — Elongate; wide in the middle of its
length, but tapering gradually to points anteriorly and
posteriorly; its dorsolateral surface produced into a
strong keel which extends dorsally as a stout shaft above
the articular facet for connection with the hyomandibu-
lar. This dorsal process serves as a place of attachment
for the muscles originating on the posterolateral surface
of the frontal. The operculum articulates by fibrous
tissue anterodorsally with the upper posterior edge of the
hyomandibular and ventrally with the suboperculum,
which it overlies. In about the middle of its anterior edge
the operculum makes ligamentous contact with the rod-
like posterior end of the interoperculum.

Suboperculum. — Thin and delicate; rounded
ventrally but produced dorsally into two processes, the
more anterior of which is by far the longer. The posterior
edge of the more anterior of these two dorsal processes ar-
ticulates by fibrous tissue with the anterior edge of the
operculum. The lateral surface of the rounded ventral
portion of the suboperculum is broadly overlain by the
ventral end of the operculum.

Interoperculum. — A long rod, with a ventral flange
along the middle region of its ventral edge; articulates by

ligaments posteriorly with the operculum and anteriorly
with the angular in the lower jaw. The ventral flange ar-
ticulates by fibrous tissue with the lateral surface of the
epihyal.

Preoperculum. — Large, expanded posteroventral-
ly; slightly expanded laterally along the anterior half of
its dorsal edge for articulation by fibrous tissue with the
ventral surface of the quadrate; articulates by fibrous tis-
sue posterodorsally with the hyomandibular and antero-
dorsally with the symplectic, metapterygoid, and
quadrate.

Upper Jaw.

Premaxillary. — Posteromedial arm short; together
with the fused teeth forms a relatively sharp-edged cut-
ting plate; its anterior edge forming the anterior border of
the upper jaw, except for a short distance ventrally where
the maxillary forms the border. The dorsomedial edges of
the two premaxillaries articulate with one another by
fibrous tissue, with the articulation strengthened by 15 to
20 short but firm projections, which decrease in size
anteriorly in the series, from the dorsomedial edge of
each premaxillary. The posteromedial arms of the
premaxillaries articulate by fibrous tissue with the
posteroventral surface of the vomer. Posterolaterally the
premaxillary articulates by interdigitation with the
overlying maxillary. The premaxillary is relatively hol-
low interiorly to accommodate the dental pulp. The long,
thin, rodlike teeth produced by the dental pulp lie paral-
lel to the edge of the premaxillary and are continuously
moved forward to replace those being worn away by use
at the biting edge. About 15 to 20 such rodlike dental
lamellae are present in each premaxillary. Only the most
anterior one or two lamellae are exposed at the cutting
edge of the jaw, but the other lamellae are easily seen
posterior to them because of the thinness of the overlying
anterior region of the premaxillary. The pulp-filled inter-
nal cavity of the premaxillary is open to the exterior
through a small canal just lateral to the base of its
posteromedial arm.

Maxillary. —Not especially curved forward; widest
ventrally where it forms the border of the upper jaw;
broadly overlies and firmly interdigitates with the pre-
maxillary along all of its length except at its expanded
ventral edge, where its medial surface articulates by
fibrous tissue with the dorsolateral surface of the den-
tary. The maxillary articulates posterodorsally by
fibrous tissue at a groove on its surface with the pala-
tine.

Dentary- — Somewhat squarish; its posterior end
concave to accommodate and interdigitate with the
anterior portion of the articular; the dentary overlies the
articular more extensively laterally than medially.
Posteroventrally the dentary articulates by fibrous tissue

or slight interdigitation with the angular, while antero-
medially it articulates by fibrous tissue with its opposite
member. The connection with its opposite member is
strengthened by about 15 stubby medial projections from
each dentary which alternate with one another and de-
crease in size anteriorly, as is the case with the pre-
maxillary. The dorsolateral surface of the dentary at-
taches by fibrous tissue to the ventromedial surface of
the maxillary. About 15 dental lamellae are present in
the dentary, in the same manner as described for those of
the premaxillary. The dentary is filled with the pulp
cavity that produces these rodlike teeth, and the cavity
opens to the exterior at the concave posterior end of the
dentary just in front of the anterior end of the articular.

Articular. — Somewhat triangular in shape; carti-
lage filled at its anterior edge; the anterior portion of the
articular fits into and interdigitates with the concave
posterior end of the dentary. The posterior edge of the ar-
ticular is thickened around the concavity at which it ar-
ticulates by fibrous tissue with the quadrate. Postero-
ventrally the articular articulates by fibrous tissue with
the angular, but in larger specimens this articulation is
by interdigitation. The sesamoid articular is a thin nub-
bin of bone lying alongside the medial surface of the ar-
ticular just behind the articulation of the latter with the
dentary.

Angular. — Small; articulates by fibrous tissue, or
slight interdigitation in larger specimens, with the den-
tary and articular. Posteriorly it articulates by a liga-
ment to the anterior end of the interoperculum.

BRANCHIAL APPARATUS.

Hyoid Arch and Branchiostegal Rays.

Urohyal, basihyal, and interhyal— Absent.

Hypohyal. —Dorsal hypohyal absent; ventral
hypohyal wide posteriorly but tapering to a blunt point
anteriorly; cartilage filled at its posterior edge; articu-
lates through cartilage posteriorly with the ceratohyal
and by fibrous tissue medially with its opposite member.

Ceratohyal. —Elongate and slightly expanded pos-
teriorly; cartilage filled at its anterior and posterior
edges; articulates anteriorly through cartilage with the
hypohyal and posterodorsally by interdigitation with the
epihyal. The ceratohyal supports by fibrous tissue all of
the six branchiostegal rays.

Epihyal. — Cartilage filled at its anterior and
ventral edges; articulates anteriorly by interdigitation
with the ceratohyal. Along its lateral surface the epihyal
articulates by fibrous tissue with the ventral flange of the
interoperculum, while dorsally it connects, without the
intervention of an interhyal, with the symplectic-metap-
terygoid region.

Branchiostegal rays. —Six in number; first
branchiostegal ray a large flat oblique plate with a down-
turned lateral edge which becomes thickened anteriorly
to form the articular facet for contact with the vertical
groove on the ventromedial surface in about the middle
of the length of the ceratohyal; like that described for
Lagocephalus laevigatas, except that it is held in a much
more oblique position. Second branchiostegal slightly
laterally compressed; articulates at its knoblike anterior
end with a slight depression on the ventral surface of the
ceratohyal just behind the articular area of the first
branchiostegal; second branchiostegal ray about the
same length as, or slightly longer than, the third to sixth
rays. Third to sixth branchiostegal rays of about the
same length, but the sixth ray much narrower than the
others; all four rays held by fibrous tissue to the lateral
surface of the posteroventral region of the ceratohyal.

Branchial Arches. — All the elements are cartilage
filled at their edges of articulation with the other ele-
ments in the series, and the articulations are usually
through cartilage and fibrous tissue. The branchial
arches are composed of three basibranchials, three pairs
of hypobranchials, five pairs of ceratobranchials, four
pairs of epibranchials, and three pairs of pharyngobran-
chials. Three gills are present; the fourth arch has no gill
and there is no slit between it and the lower pharyngeal.

First arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. First basi-
branchial wider posteriorly than anteriorly; displaced
forward so that it articulates posteriorly with the second
basibranchial and posterolaterally with the first hypo-
branchials. First hypobranchial a slender rod, slightly
wider dorsally than ventrally; the longest of the hypo-
branchial elements, which decrease in size posteriorly in
the series; articulates ventrally with the region of articu-
lation between the first and second basibranchials and
dorsally with the first ceratobranchial. First cerato-
branchial the shortest of the first four ceratobranchial
elements, which increase in length posteriorly in the
series, except for the fifth ceratobranchial which is
slightly shorter than the first ceratobranchial; first
ceratobranchial with a ventrally directed flange along
most of its ventral surface, while a similar flange is pres-
ent on the second and third ceratobranchials, as well as
on the fourth ceratobranchial in a much reduced state; a
large posterodorsally directed flange is present from the
dorsal edge of the ventral region of the first cerato-
branchial, and a similar, although slightly smaller,
flange is present on the second and third cerato-
branchials; articulates ventrally with the first hypo-
branchial and dorsally with the first epibranchial. First
epibranchial a slender rod, the shortest of the epibran-
chial elements; articulates dorsally with the first
pharyngobranchial. First pharyngobranchial a rounded
plate with a ventral shaftlike process for articulation
with the first epibranchial; bears about 15 to 25 small
sharply pointed teeth in a single but somewhat irregular
row along the edge of the rounded platelike region; the

teeth smaller than those of the second and third pharyn-
gobranchials. The dorsal regions of the branchial arches
are held to the ventral surfaces of the parasphenoid and
prootics by fibrous tissue attaching to the dorsolatertd
surface of the epibranchials and phaiyngobranchials.

Second arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Second basi-
branchial somewhat constricted in the middle of its
length; articulates anteriorly with the first basi-
branchial, anterolaterally with the first hypobranchials,
posterolaterally with the second hypobranchials and pos-
teriorly with the third basibranchial. Second hypo-
branchial a short slender rod; articulates ventrally with
the posterolateral edge of the second basibranchial and
dorsally with the second ceratobranchial, which in turn
articulates dorsally with the second epibranchial. Second
epibranchial slender, rodlike; articulates dorsally with a
stubby projection on the ventral arm of the second
pharyngobranchial. Second pharyngobranchial the larg-
est of the pharyngobranchial elements; more or less L-
shaped, bearing a single row of about 10 to 15 sharply
pointed teeth along the edge of the wider of the two arms
of the L. The teeth are set in shallow sockets and are
curved slightly posteriorly. They are replaced by new
teeth developing in new sockets alongside the sockets of
the old teeth. The teeth of the second pharyngo-
branchial are of about the same size as those of the third.
The second pharyngobranchial is held by fibrous tissue
to the other two pharyngobranchials.

Third arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Third basi-
branchial large, rather squarish, articulates anteriorly
with the second basibranchial, posterolaterally with the
third hypobranchials and posteriorly with the fourth
ceratobranchials. Third hypobranchial a slender L-
shaped bone whose ventral arm articulates by fibrous
tissue with the ventral surface of the first basibranchial;
articulates posterodorsally with the posterolateral edge
of the third basibranchial and with the ventral edge of
the third ceratobranchial. Third ceratobranchial articu-
lated dorsally with the third epibranchial. Third
epibranchial a slender rod, with a stubby projection from
about the middle of its posterior edge which articulates
by fibrous tissue with a similar process from the fourth
epibranchial; articulates dorsally with the third
pharyngobranchial. Third pharyngobranchial like the
second pharyngobranchial, but much smaller and bear-
ing only about 7 to 10 small sharply pointed teeth in a
single row; articulates ventrally with the third and fourth
epibranchials.

Fourth arch. — Cerato- and epibranchial elements
only. Fourth ceratobranchial a slender rod with only a
very low ventral flange along a short portion of its length;
articulates ventrally with the third basibranchial and
dorsally with the fourth epibranchial. Fourth epibran-
chial a slender rod with a low flange along its posterior
edge, and a stubby emargination from about the middle

of its anterior edge which articulates by fibrous tissue
with the third epibranchial; the dorsal end of the fourth
epibranchial articulates with the third pharyngobranchial.

Fifth arch. — Ceratobranchial (lower pharyngeal)
element only. Fifth ceratobranchial a stout shaft with a
small medial expansion in about the middle third of its
length; articulates ventrally with the base of the fourth
ceratobranchial; toothless.

PAIRED FIN GIRDLES.

Pectoral Fin.

Supracleithruin. —In position at about a 45° angle
to the axis of the body; a stout shaft with a lateral flange
along most of its length; articulates by fibrous tissue
anterodorsally with the slight concavity on the lateral ex-
pansion of the pterotic and ventrally with the cleithrum,
which it broadly overlies.

Cleithrum. — Laterally expanded along most of its
length; articulates by fibrous tissue dorsally with the
overlying supracleithrum and with the anterior end of
the dorsal postcleithrum, which it overlies; along its pos-
terior edge it articulates by fibrous tissue with the scapu-
la and coracoid. Ventromedially the cleithrum articu-
lates by fibrous tissue with its opposite member in the
midline of the body between the medial edges of the
platelike portions of the first branchiostegal rays.

Postcleithra. —The postcleithra form a stout strut
along the abdominal wall musculature from the
supracleithrum to halfway back the length of the ab-
dominal cavity. The dorsal postcleithrum articulates by
fibrous tissue anterolaterally with the overlying
cleithrum and posteroventrally with the ventral post-
cleithrum, which tapers to a point posteriorly. The dorsal
and ventral postcleithra are only slightly laterally com-
pressed.

Coracoid. —Rounded dorsally but gradually
tapering to a stout shaft ventrally; along the dorsal third
of its length the posterior edge of the coracoid is in-
turned to form a medial flange whose dorsal edge is pro-
longed into a spinelike process below the fourth actinost;
cartilage filled at the edge of its rounded dorsal portion
and at its extreme ventral edge; articulates by fibrous
tissue anteriorly with the cleithrum and through car-
tilage dorsally with the scapula and last two actinosts.

Scapula. — Scapular foramen not entirely enclosed
by the scapula, but, rather, with its anterior edge closed
by the cleithrum; cartilage filled at its anterior and ven-
tral edges; articulates anteriorly by fibrous tissue with
the cleithrum, ventrally through cartilage with the cora-
coid, posterodorsally by slight interdigitation with the
anterior edge of the first actinost, and posteroventrally
by slight interdigitation with the ventral edge of the sec-
ond actinost.

Actinosts. —Four elements; all cartilage filled at
both ends, except for the first actinost, which is filled
with cartilage only at its dorsal edge. First actinost
reduced to a triangular wedge between the upper parts of
the scapula and second actinost, to both of which ele-
ments it interdigitates. Second actinost rounded dorsal-
ly, constricted in the middle and reduced ventrally to a
narrow splint that interdigitates ventrally with the
scapula and posteriorly with the base of the third acti-
nost; articulates by interdigitation anterodorsally with
the first actinost and posterodorsally with the third acti-
nost. Third and fourth actinosts slightly constricted in
the middle; the third actinost articulating dorsally and
ventrally by interdigitation with the second and third
actinosts, while it articulates through cartilage just be-
low its base with the coracoid; the fourth actinost ar-
ticulating by interdigitation anterodorsally and antero-
ventrally with the third actinost and ventrally through
cartilage with the coracoid. Distally the actinosts sup-
port all of the fin rays.

Fin rays. —Usually 17 fin rays present; the first fin
ray short, about one-third or one-fourth the length of the
second ray, but normal in shape and symmetry, its two
halves of equal size; all rays articulated by fibrous tissue
with the dorsal ends of the actinosts. The first tv/o or
three rays unbranched, the other rays branched. The first
ray without cross-striations, the others cross-striated.

VERTEBRAL COLUMN. —All vertebrae with bicon-
cave centra, except the last, which ends posteriorly in the
urostylar process and fusion with some of the hypurals.

Abdominal Vertebrae.

First vertebra. — Neural spine bifid and relatively
short; neural arch with a complete bony roof over the
neural canal, as do all the other vertebrae; articulates by
fibrous tissue over the anterior half of the lateral surface
of its centrum with the posteriorly projecting exoccipital
condyles, while the rim of the concave anterior face of its
centrum articulates with the rim of the concave pos-
terior end of the basioccipital. Posteroventrally on its
lateral surface the centrum of the first vertebra pos-
sesses a flattened posterior process which fits under the
anterior half of the ventrolateral surface of the centrum
of the second vertebra.

Other abdominal vertebrae. — In 17 specimens the
abdominal vertebrae numbered eight. The second and
third abdominal vertebrae have their neural spines bi-
fid, but somewhat longer than that of the first vertebra.
The fourth vertebra has its neural spine enlarged antero-
posteriorly, and while it is bifid anteriorly, it becomes
single posteriorly. All of the vertebrae, both abdominal
and caudal, posterior to the fourth abdominal vertebra
have undivided neural spines. The neural spines of the
fifth and sixth abdominal vertebrae are of about the
same height as that of the fourth vertebra, but they are

less anteroposteriorly expanded than the latter. The
seventh and eighth abdominal vertebrae have their neu-
ral spines in the form of long stout shafts between which
the first basal pterygiophore of the dorsal fin articulates.
Small neural prezygapophyses are present from about
the fourth to the last abdominal vertebra, and increase in
size posteriorly in the series. From the posterolateral sur-
face of the second vertebral centrum a stout ventrally di-
rected process is present which makes fibrous tissue con-
tact with a similar ventrally directed process from the
anterolateral surface of the centrum of the third verte-
bra. These ventral processes from either side of these two
centra do not meet their opposite members in the mid-
line to completely enclose the haemal canal. However,
the abdominal vertebrae posterior to the third vertebra
have stout ventrally directed processes from the antero-
lateral surfaces on either side of their centra which do
meet and fuse in the midline to enclose completely the
haemal canal. The haemal spines possess posteriorly di-
rected flanges along their posterior edges just below the
haemal canal. The haemal spines increase in size pos-
teriorly in the series and form broad surfaces for muscle
attachment.

Caudal Vertebrae. — In 17 specimens the caudal
vertebrae numbered nine. All of the caudal vertebrae
possess neural arches and spines and haemal arches and
spines. Neural prezygapophyses are present on all of the
caudal vertebrae, although they decrease in size pos-
teriorly so that the last caudal vertebra has them only
weakly developed. The neural spines of the first two cau-
dal vertebrae are slender but strong shafts which dis-
tally support the more posterior of the dorsal fin basal
pterygiophores. These two neural spines are shorter than
the last two neural spines of the abdominal vertebrae,
which support the more anterior of the dorsal fin basal
pterygiophores. The neural spines of the third to sixth
caudal vertebrae are slender but strong shafts which are
slightly longer than those of the two preceding caudal
vertebrae. The neural spine of the seventh caudal verte-
bra is enlarged anteroposteriorly, and that of the eighth
caudal vertebra is even more enlarged. The neural spines
of these two vertebrae make close fibrous tissue contact
with one another and form a large thin plate above their
centra. Each neural arch possesses a small neural fora-
men. The haemal arches and spines of the first three cau-
dal vertebrae are like those of the abdominal vertebrae
just anterior to them, except that the haemal spine of the
first caudal vertebra is even more enlarged posteriorly
than are the others and ventrally supports by fibrous
tissue the much expanded first basal pterygiophore of the
anal fin. The posteriorly expanded platelike portions of
the haemal spines of the second and third caudal verte-
brae ventrally support through fibrous tissue the second
to the fifth (last) anal fin basal pterygiophores. The
haemal spine of the fourth caudal vertebra is less antero-
posteriorly expanded than are those anterior to it, but it
is somewhat longer. The haemal spines of the fifth to
seventh caudal vertebrae become increasingly expanded
anteroposteriorly so that the haemal spine of the seventh

is a rounded flat plate whose posterior edge makes
fibrous tissue contact with the anterior edge of the much
modified haemal spine of the eighth caudal vertebra.
Whereas the centrum and the neural arch and spine of
the eighth caudal vertebra are fused together, the haemal
arch and its spine are autogenous and held to the cen-
trum by fibrous tissue and interdigitation. The haemal
spine of the eighth caudal vertebra is much enlarged, and
possesses a posterior extension which reaches almost to
the base of the lowermost caudal fin ray. The edge of the
haemal apparatus of the eighth caudal vertebra that
makes contact with the centrum is somewhat concave so
that a haemal canal is present throughout the length of
the region of articulation of the two elements. The hae-
mal canal is continuous anteriorly with the haemal canal
of the preceding vertebrae, while posteriorly it is continu-
ous with the canal between the surfaces of contact of the
parhypural and the centrum of the last vertebra.

Caudal Skeleton. —The caudal complex is composed
of four separate bony elements closely held to one
another variously by fibrous tissue and interdigitation.
The ventral portion of the caudal skeleton is formed by
the parhypural, which is broad anteriorly but tapers to a
stout shaft posteriorly where it supports the lowermost
caudal fin ray. Anteriorly it is extensively interdigitated
with the posterodorsal edge of the haemal spine of the
penultimate vertebra. A distinct notch is present in its
dorsal edge a little forward of its middle region. Anterior
to this notch the dorsal edge of the parhypural is exten-
sively interdigitated with the region of the centrum of the
large bony plate above it, which is the fused centrum-
lower hypurals. The notch in the parhypural is the last
opening into the haemal canal, the latter being con-
nected to the canal in the preceding vertebrae by a longi-
tudinal groove medially along the otherwise interdigi-
tated articular surface between the parhypural and the
centrum anterior to the foramen. The dorsal surface of
the parhypural posterior to the foramen is held by fibrous
tissue to the ventral edge of the hypural region of the fus-
ed centrum-lower hypural plate. The epural is a flat-
tened plate closely held by fibrous tissue between the
posterior edge of the neural spine of the penultimate
vertebra and the anterior edge of the urostylar region of
the fused centrum-lower hypural plate and of the antero-
ventral region of the upper edge of the autogenous upper
hypural plate. Ventrally the epural slightly interdigi-
tates with the incomplete neural arch region of the last
vertebra, forming the dorsal roof of the neural arch. The
neural canal thus enclosed is continuous anteriorly with
the canal in the preceding vertebrae, while posteriorly
the canal opens to the exterior just behind the end of the
area of articulation between the epural and the neural
arch of the last vertebra. In large specimens (over 1(X)
mm) the articulation between the epural and the lateral
walls of the neural arch of the last vertebra tends to be-
come more intimately interdigitated and perhaps even
fused. Similarly, the articulation of the epural with the
neural spine of the penultimate vertebra and the uro-
stylar region of the last vertebra tends to become slightly

interdigitated in large specimens. The largest element of
the caudal skeleton is that formed from the fusion of the
centrum of the last vertebra posteroventrally with the
lower hypurals, which in generalized plectognaths would
correspond to the first and second hypurals but which are
here indistinguishably fused with one another and with
the centrum. The concave anterior face of the centrum
portion of this fusion element articulates in a normal
manner with the concave posterior face of the centrum of
the penultimate vertebra, but immediately below this
fibrous tissue articulation the ventral expansions of these
two vertebrae interdigitate with one another, while ven-
trally that of the last centrum interdigitates with the
anterodorsal end of the parhypural, except at the medial
longitudinal groove between their apposed surfaces
enclosing the haemal canal. The urostyle of the last cen-
trum is represented by a tapering thickened shaft ex-
tending posterodorsally from the centrum portion of the
composite plate and is extensively interdigitated along
its posterior edge with the anterior edge of the autogenous
upper hypural plate. The anterior edge of the urostylar
region articulates by fibrous tissue, and by interdigita-
tion in larger specimens, with the epural. The ven-
tral edge of the hypural region of the fused centrum-
lower hypurals articulates by fibrous tissue with the
posterodorsal edge of the parhypural, while its dorsal
edge is extensively interdigitated with the anteroventral
end of the autogenous upper hypural plate. The upper
hypural plate is composed of indistinguishably fused hy-
purals corresponding to what in generalized plectog-
naths would be the third to fifth elements, the plate be-
ing extensively interdigitated with the urostylar and
hypural regions of the fused centrum-lower hypural
plate, and held for a short distance anterodorsally by
fibrous tissue to the epural. In the smallest specimen ex-
amined, 10.0 mm, the lower hypurals are as fully and in-
distinguishably fused to one another and to the centrum
as in adults, while the autogenous upper hypural plate,
the epural and parhypural are, as expected, scarcely if at
all interdigitated with the fused centrum-lower hypurals,
but the upper hypural plate more so than the epural and
parhypural.

Caudal fin rays. — Eleven in number, the upper-
most ray and the lowermost ray unbranched, the others
becoming increasingly branched toward the middle rays,
which are branched in triple dichotomies. The upper-
most five rays are held by fibrous tissue to the auto-
genous upper hypural plate, the lowermost ray to the
parhypural and the upper five rays of the lower lobe to
the hypural region of the fused centrum-lower hypural
plate.

DORSAL AND ANAL FINS.

Dorsal Fin.

Fin rays and pterygiophores. — Ten fin rays are pres-
ent in most specimens, the first one or two rays unbranch-
ed, the others branched in single or double dicho-

tomies; the first ray almost as long as the longest ray.
Distal pterygiophores either absent or unossified. The
fin rays are supported basally by eight basal pterygio-
phores which are broadly interdigitated with one another
and which decrease slightly in length posteriorly in the
series. The first basal pterygiophore is by far the largest
and fills much of the space between the neural spines of
the seventh and eighth abdominal vertebrae. The second
to fourth basal pterygiophores have their ventral or in-
temeural portions between the neural spines of the
eighth abdominal and first caudal vertebrae, while those
of the fifth to seventh basal pterygiophores are between
the neural spines of the first and second caudal verte-
brae. The eighth basal pterygiophore has its ventral
region placed just behind the upper end of the neural
spine of the second caudal vertebra. The pterygiophores
articulate with the neural spines by fibrous tissue. Each
pterygiophore ends dorsally in a short columnar projec-
tion whose deeply concave dorsal edge is cartilage filled.
It is to these columnar projections that the fin rays are
held by fibrous tissue. The ventral ends of the pterygio-
phores are all deeply cartilage filled. The anterodorsal
edge of the first pterygiophore is slightly prolonged
anteriorly and articulates by fibrous tissue with the
rounded shaftlike supraneural. The supraneural extends
from above the posterior region of the neural spine of the
fourth abdominal vertebra to just in front of the dorsal
end of the neural spine of the seventh abdominal verte-
bra. The supraneural is connected with a band of longi-
tudinal muscles in the midline of the body which
anteriorly connects with the supraoccipital spine. None
of the pterygiophores possess lateral flangelike projec-
tions along their lengths for muscle attachment,
although they are slightly thicker in the middle regions
throughout their lengths than they are anteriorly or pos-
teriorly.

Fin rays and pterygiophores. — Nine fin rays are
present in most specimens, the first one or two rays un-
branched, the others branched in single or double
dichotomies; the first ray only slightly shorter than the
longest ray. Distal pterygiophores either absent or un-
ossified. The anal fin rays are supported through fibrous
tissue by five basal pterygiophores which are interdigi-
tated with one another only for a short distance ventral-
ly. Posterior to the enlarged first pterygiophore, the re-
maining pterygiophores decrease only very slightly in
length posteriorly in the series. The dorsal ends of the
pterygiophores articulate by fibrous tissue with the ven-
tral edges of the haemal spines of the caudal vertebrae,
with the first pterygiophore attached to the first caudal
vertebra and the remaining pterygiophores attached to
the second and third caudal vertebrae. Like the pterygio-
phores of the dorsal fin, those of the anal fin are carti-
lage filled proximally, while distally they possess short
columnar projections whose deeply concave edges are
cartilage filled. Whereas the second to fourth pterygio-
phores possess a single columnar projection distally, the

w

first pterygiophore possesses two and the last pterygio-
phore possesses two or three columnar projections. The
presence of more than one such projection may indicate
that the first and last pterygiophores are fusion products
of two or more originally separate basal pterygial ele-
ments. The fin rays are articulated by fibrous tissue to
these columnar distal ends of the pterygiophores. Like
the dorsal fin pterygiophores, the anal fin pterygio-
phores are slightly thicker in the middle regions
throughout their lengths than they are anteriorly or pos-
teriorly.

Anatomical diversity and generic relation-
ships. — The fossil record of tetraodontids is meager. The
only relatively complete specimen of a tetraodontid is
Tetraodon pygmaeus Zigno (1887b) from the Eocene of

prefrontal frontal pterosphenoid supraoccipital
frontal

parasphenoid

maxillary palatine
premaxillary

dentary

epiottc
pterotic

basioccipital

suboperculum

mesopterygoid
ectopterygoid metapterygoid I preoperculum
quadrate interoperculum symplectic

Monte Bolca, Italy. The holotype and only known speci-
men, IGUP 6890-91, 18.2 mm SL, in counterpart, is small
and only moderately well preserved. It is a lateral im-
pression and the medial region of the upper and lower
jaws is not exposed. Thus, it cannot be told at present
how the dentaries and premaxillaries articulate to their
opposite members. However, the body is covered by very
fine small prickles and the abdominal vertebrae (either 9
or 10) are fewer than the caudal vertebrae (12), while the
rodlike postcleithrum is in two pieces, all these features
being more characteristic of tetraodontids than diodon-
tids. There are about 10 or 11 dorsal and anal fin rays,
and the caudal fin seems to have 12 rays, one more than
in any Recent species of tetraodontid, but the same num-
ber of principal rays as in the ancestral gymnodont
triodontids and scleroderm eoplectins. Moreover, there is
a strong indication that there are slender ribs present on
many of the abdominal vertebrae, while ribs are absent
in all Recent gymnodonts except the basal triodontids
(and one of the two species of scleroderm eoplectins also
seems to have ribs). The above are the only features of in-
terest that I can distinguish in the specimen, which is
sufficiently distinct from all Recent species to merit
separate generic recognition, and Eotetraodon is pro-
posed herein for pygmaeus Zigno (type-species by mono-
typy and original designation).

The only other tetraodontid fossil materials are jaw
fragments found from the Miocene onward, and usually
referred to as Tetraodon lecointrae Leriche and T.
lawleyi Carraroli. They indicate only that the individual
dental units incorporated into the biting edge of the jaws
were elongate rods such as in the Recent species.

Figiire 204.— Canthigaster
roBtrata: lateral view
of head, composite based
on several specimens,
55.2-96.5 mm SL, Texas.

The three external characters that have been
prominently used in conjunction with internal features in
subdividing the Recent tetraodontids (the number of
dorsal and anal fin rays, the form and number of the
lateral lines, and the form of the nasal apparatus) will be
discussed here first, followed by an analysis of the osteo-
logical diversity of the family. In the two most recent
revisions of the group (Fraser-Brunner 1943 and Le
Danois 1959, 1961a) five families have been recog-
nized in what is here considered a single family with two
subfamilies, and the term Tetraodontidae as used here is
senso lato, although closely corresponding to that in
general use in the contemporary ichthyological
literature.

Chonerhinos and Xenopterus have the dorsal and anal
fins longer based and with more rays than in other tetra-
odontids. In Chonerhinos there are about 23 to 29 dorsal
rays and in Xenopterus about 32 to 38, with the anal fin
in both cases having several less rays than the dorsal fin.
In other tetraodontids the number of dorsal fin rays
ranges modally from about 7 to 18, with the anal fin hav-
ing either the same number or one to several rays less
than the dorsal fin.

Sanzo (1930) was the first to describe in detail the
lateral line of tetraodontids, and Fraser-Brurmer (1943)
the first to call attention to its systematic importance.

Figure 205. (opp. page)— Canthigaater
roBtrata: dorsal (left) and
ventral (right) views of skull, with
inset at top right showing a ventral
view detail of the articulation of
the upper jaw with the palatines and
vomer, composite based on several
specimens, 55.2-96.5 mm SL, Texas.

ceratobranchials

epibranchials

pharyngobranchials

ClI)Cl3c= — basibranchials
hypobranchials

^>

^::s

Figure 206. (opp. page)— Canfftt^osfer
roatrata: posterior view of skull (left);
below, lateral and posterior views of
first abdominal vertebra; posterior view
of orbit (right) (cross section of
skull; dashed lines represent cut
surfaces of frontals and parasphenoid);
composite based on several specimens,
55.2-96.5 mm SL, Texas.

Figure 207.— Conf/i«<Mtcr rostrata: dorsal

view of branchial arches (extended on lower

side); lateral (above) and medial (below, upside

down) views of hyoid arch; composite based on

several specimens, 55.2-96.5 mm SL, Texas.

Wiedersheim (1887a, b; 1888) was the first to describe in
detail the nasal apparatus of tetraodontids, while its
systematic importance was first stressed by Bibron (see
Dumeril 1855), although an overemphasis on the nasal
apparatus has led to confusion. Dunker and Mohr (1931)
described the changes in the nasal tube in Chelonodon
with increasing body size, and Barnard (1927) showed
the variety of nasal organs in South African puffers,
while Le Danois (1959) illustrated it for an even larger
selection of tetraodontids. Inger (1953) illustrated the
differences in the olfactory epithelium between
Tetraodon kretamensis and Chelonodon fluviatilis.

Chonerhinos and Xenopterus have the lateral line
more elaborately developed than in other tetraodontids,
with three lines on the body rather than only one or two.

Figure 20S.~Canthiga8ter rostrata: lateral
view of pectoral girdle, composite based on
several specimens, 5.5.2-96.5 mm SL, Texas.

while in Canthigaster the lateral line system is so in-
conspicuous that it can only be seen with the aid of
magnification. Nasal specializations also distinguish
these genera from other tetraodontids. In Chonerhinos
and Xenopterus (the Chonerhinidae of Fraser-Brunner)
the nasal apparatus is larger than in other tetra-
odontids, in the form of a large, low open cup with
numerous lamellae or folds, while in Canthigaster (the
Canthigasteridae of Fraser-Brunner) the nasal apparatus
is much smaller than in other tetraodontids, in the form
of a small, low simple tube with a single opening.

In other tetraodontids the nasal apparatus ranges from
an upright sac with two nostrils (in Lagocephalus,
Sphoeroides, Guentheridia, Amblyrhynchotes, Fugu,
and Torquigener, these being the Lagocephalidae of
Fraser-Brunner, except that for purposes of the present
discussion the genera are more finely split than as recog-
nized by Fraser-Brunner, and in Colomesus, the
Colomesidae of Fraser-Brunner) to a tube with a single
nostril or a bifid tentacle or a simple single flap (Ephip-
pion, Arothron, Monotreta, Chelonodon, Carino-
tetraodon, and Tetraodon, the Tetraodontidae of Fraser-
Brunner, except, as above, for purposes of the present
discussion, more finely split). While the lateral line
systems in Chonerhinos-Xenopterus and in Canthi-
gaster are each in their own way just as distinctive as
their nasal apparatuses, the situation is not so neatly
simple in the other tetraodontids.

Among those genera with the relatively normal nasal
apparatus as an upright sac with two nostrils,
Sphoeroides and Guentheridia have a single lateral line
on the body, although short segments of an additional
lower line are sometimes present (see illustrations of S.
lobatus, S. dorsalis, and G. formosa), while in
Lagocephalus, Amblyrhynchotes, Fugu, Torquigener,
and Colomesus two lateral lines usually are present, the

Figure 209.— Canthigaster

roBtrata: lateral view of

caudal fin supporting

structures, composite based

on several specimens.

55.2-9G.5 mm SL, Texas.

lower one always better developed than in the traces of it
sometimes found in Sphoeroides and Guentheridia, even
though the lower line may not run the full length of the
caudal peduncle (see illustration of Torquigener pleuro-
stictus). Possible exceptions are Amblyrhynchotes
piosae, in which the exceptionally long spines mostly ob-
scure the lateral line, so that in the single alcohol-
preserved specimen examined I am unable to find any
lateral line at all on the caudal peduncle, and A. richei,
in which I cannot find a lower lateral line in any of the 10
alcohol preserved specimens examined, although Le
Danois (1959:172) showed it as present. A distinctly up-
raised horizontal ridge of skin is present ventrolaterally
along the caudal peduncle in all species of Logo-
cephalus, in most species of Torquigener (absent in
pleurostictus) and Fugu (absent or only weakly
developed in chrysops), in at least some species of
Amblyrhynchotes (honckenii and hyselogenion) and
Sphoeroides (e.g., see illustrations of lobatus, testudi-

neus, and dorsalis), but it is absent in Guentheridia and
Colomesus. The presence or absence of this ridge is of
very limited systematic interest at the generic level,
while the presence or absence of a second, lower, lateral
line is more variable within certain genera than previous-
ly thought.

In those genera with a single nostril or a nasal tentacle,
there is a single lateral line in Arothron, while in
Chelonodon and Monotreta the upper lateral line nor-
mally curves ventrally to join the lower lateral line in the
region above the anal fin, the lower line continuing pos-
teriorly to the tail as well as anteriorly to the abdominal
region from the point of juncture with the upper lateral
line. Tetraodon has some species with basically the same
arrangement as in Chelonodon and Monotreta (see illus-
trations of T. lineatus and T. mbu), but in at least one
species (see illustration of T. miurus) there seems to be
no lower lateral line, the upper line continuing directly
onto the tail, while in another (see illustration of T.

Figure 210.— External features of other
representative tetraodontid genera:

A, Sphoeroides nephelus, upper left, nasal
region as seen externally (far left) and
olfactory lamella as seen with front of

nasal sac removed, and, lower left, scales
from upper middle region of body;

B, S. testudineua, upper left, nasal region
as seen externally (olfactory lamella, if

present, indistinct in specimen examined),

and, lower left, scales from upper middle

region of body; C, S. dorsalis, upper left, nasal

region as seen externally (far left) and

olfactory lamella as seen with front of

nasal sac removed, and, lower left, scales

from upper middle region of body; D. S. lobatus,

upper left, nasal region as seen externally

(olfactory lamella, if present, indistinct

in specimen examined), and, lower left,

scales from upper middle region of body

(larger markings on rear of body, and one

on dorsum, represent flaps rather than

scales); E. S. pachygaster. lower left, nasal region

as seen externally and, upper left,

olfactory lamellae as seen with front

of nasal sac removed, showing more

lerous lamellae than in other species

of Sphoeroides (scales absent).

291

Figure 211.— Nasal organs and
olfactory lamellae of representative
species of Sphoeroides (top row,
A-C) and Lagocephalus (D-F):
to left in each set, the nasal
region as seen externally, and
to its right the olfactory lamellae
as seen with top or front of nasal

sac removed, and to its right,
in Lagocephalus, the outline of
an anteroposterior cross section
of the sac and lamellae.

A, Sphoeroides trichocephalus;

B, S. spengleri; C, S. maculatus

D, Lagocephalus spadiceus;

E, L. lagocephalus: F, L.

Figure 212.— External features of

other representative tetraodontid

genera: Guentheridia formoaa — upper

left, nasal region as seen externally

(far left) and the pitted olfactory

epithelium as seen with the top of

the nasal sac removed; lower left,

scales from upper middle

region of body.

Figure 213.— External features of
other representative tetraodontid
genera: Colomesus psittacus — upper
left, nasal region as seen externally
(far left) and the olfactory lamella
as seen with the front of the nasal

sac removed; lower left, scales
ft'om upper middle region of body.

Figure 214.— External features of

other representative tetraodontid

genera: A, Amblyrhynchotea pioaae —

upper left, nasal region as seen

externally (far left) and the olfactory

lamellae as seen with the front of the

nasal sac removed, and, lower left,

scales from upper middle region of body

(spines on caudal peduncle surrounded

by fleshy flaps which are better
developed there than more anteriorly);
B, A. honckenii— upper left, nasal
region as seen externally (far left)
and the olfactory lamellae as seen
with the top of the nasal sac removed,

and, lower left, scales from upper

middle region of body; C, A. richei —

upper left, nasal region as seen

externally (two lower figures),

indicating that I out of 10 specimens

has the nostrils confluent and the

nasal tube with a single large opening,

with the upper figure showing the

olfactory lamellae with the top of

the nasal sac removed, and, lower left,

scales from upper middle region of body.

Figure 215.— External features of other

representative tetraodontid genera:

Torquigener pleuroalictus (above) and

T. hamiltoni— in both cases, upper left, nasal region

as seen externally (far left) and the

olfactory lamella(e) as seen with the

front of the nasal sac removed, and, lower left,

scales from upper middle region of body.

Figure 216.— External features of other
representative tetraodontid genera:
Fugu vermicularis — left, bottom to
top, nasal region as seen externally,
the olfactory lamellae as seen with
the front of the nasal sac removed,
and the outline of an anteroposterior
cross section of the sac and lamellae.

Figure 217.— External features of

other representative tetraodontid genera:

Arothron armilla (above) and

A. immaculatus— in both cases, upper

left, nasal region as seen externally

and, lower left, scales from upper

middle region of body (nasal epithelium
of A. immaculatus pitted as shown;
that of A, armilla, not exposed in
the illustration, is only a slightly
irregular medial surface of the
single simple nasal flap).

Figure 218.— External features of

other representative tetraodontid

genera: A, Tetraodon miurug,

B, T. lineatua. C, T. kretamensis, and

D, T. achoutedeni— in all cases, upper

left, nasal region as seen externally

(with inset below for T. schoutedeni

showing the pitted nasal epithelium)

and, lower left, scales from upper

middle region of body.

Figure 219.— External features of

other representative tetraodontid

genera: Ephippion guttifer of

decreasing sizes, from top to bottom

(325, 232, and 101 mm SL) to show

the enlargement of the scale plates

with increasing standard length

until, in large adults, a firm

saddle is formed over the back and

sides (but not belly); other two

insets show nasal region as seen

externally; scales of 101 mm SL

shown for both beUy

Figure 220.— Photographic detail of

interdigitated scale plates from lower

side of body of Ephippion guttifer,

325 mm SL, also shown in

Figure 219 (greatest length of largest

scale plate 23.0 mm).

Figure 221.— External features of

other representative tetraodontid

genera: Chelonodon patoca (above) and

C. fluviatilis— in both cases, upper

left, nasal region as seen externally,

and, lower left, scales from upper

middle region of body.

Figure 222.— External features of other

representative tetraodontid genera:

Monotreta palembangensis (above) and

M. leiurus— in both cases, upper

left, nasal region as seen externally,

and, lower left, scales from upper

middle region of body.

Figure 223.— External features of other

representative tetraodontid genera: Chonerhinos

modestus— upper left, nasal region as seen externally

(far left) and the olfactory lamellae as seen with the

top of the nasal sac removed; lower left, scales

from upper middle region of body.

Figure 224.— External features of other representative tetraodontid
genera: Xenopteriui naritua— upper left, nasal region as seen
externally; lower left, scales from upper middle region of body.

Figure 225.— External features of

other representative tetraodontid genera:

Carinotetraodon torteti— upper left, nasal

region as seen externally; lower left, scales

from upper middle region of body.

kretamensis) the course of the lateral line was found to
be impossible to trace posteriorly in the single alcohol-
preserved specimen examined. In Carinotetraodon a
well -developed lower lateral line is present, but, at least
in the few specimens examined, it does not join with the
upper lateral line, continuing on directly to the tail (see
illustration). In Ephippion a lower lateral line is present,
but its juncture with the upper line is highly variable
within the single species, as is the length of its extension
anterior to the point of juncture (see illustrations). None
of the genera with a single nostril or nasal tentacle have
an upraised ridge of skin ventrolaterally along the caudal
peduncle.

In short, it can be said that, of those genera with two
nostrils, Sphoeroides and Guentheridia have only a
single well-developed lateral line on the body, while in
Lagocephalus, Amblyrhynchotes, Fugu, Torquigener,
and Colomesus two lateral lines usually are present, and
that of those genera with a single nostril or nasal ten-
tacle, Arothron has a single lateral line, while
Chelonodon, Monotreta, and Ephippion have two which
usually merge on the caudal peduncle, Carinotetraodon
usually has two which do not merge, and Tetraodon has
either two which merge or, in at least a few species, a
single lateral line.

There is also intraspecific variation in the form of the
nasal apparatus, although probably less than that
described here subsequently for the Diodontidae. In one
species with two nostrils {Amblyrhynchotes richei), 9
(18.9-47.3 mm, ANSP 109916) out of 10 specimens ex-
amined had the normal arrangement of two nostrils
separated by a broad band of tissue, but one specimen
(49.1 mm, ANSP 109916) had apparently resorbed the
middle region of the band in both the right and left nasal
apparatuses to leave the nostrils confluent, and only in-
completely separated by two flaps of skin (see illus-
tration).

In the genera with a single nostril or nasal tentacle, the
olfactory epithelium is usually relatively smooth, but in
at least a few species (see illustrations of Arothron im-
maculatus and Chelonodon fluviatilis) the inner surfaces
of the bifid tentacles are deeply pitted.

In the genera with two nostrils, the olfactory
epithelium ranges from smooth to highly folded into
numerous successive lamellae, but in the monotypic
Guentheridia formosa the olfactory epithelium is deeply
pitted just as in some of the species with bifid tentacles
(all of the relatively normal species oi Arothron and one
of the species of Chelonodon). The degree of lamellar
development is variable within some genera. For exam-
ple, in Lagocephalus lamellae are exceptionally well

developed in all species as folds that cover the entire in-
ner surface of the nasal sac, with one or two of the folds
on the rear wall enlarged as protruding flaps which them-
selves bear lamellae, except in L. scleratus, in which
folds are present only on the rear wall of the sac and the
two prominent protruding flaps are not themselves ex-
tensively folded. Similarly, in Sphoeroides, most species
have the olfactory epithelium either smooth or with only
one or two prominent folds or horizontal ridges on the
rear wall, but one species, S. pachygaster, has the folds
on the rear wall much better developed and more
numerous, between five and eight in number. While the
habitat of L. scleratus is similar to that of some (e.g., the
equally streamlined and open water L. lagocephalus) of
the other species of Lagocephalus with extensive olfac-
tory flap development, S. pachygaster, in contrast to all
other species of Sphoeroides, is a deepwater species,
which may rely more heavily on olfaction than do the
shallower water species. In Fugu the folds are only slight-
ly less well developed than in most Lagocephalus and oc-
cur on the entire inner surface of the nasal sac, although
the larger folds on the rear wall never become so exten-
sively folded themselves as they do in most species of
Lagocephalus. The numbers and size of the lamellae are
especially great in F. oblongus. In the species of
Colomesus, Torquigener, and Amblyrhynchotes ex-
amined, one to about four folds or ridges are developed
on the rear wall of the sac.

There is variation in the size of the prickly spines and
in their coverage of the body. Among the materials ex-
amined there are four genera in which most of the species
have prickles but in which one or two species are com-
pletely spineless. In the speciose Sphoeroides most
species have spines on both the back and belly but two
species are spineless, S. angusticeps in the eastern
Pacific and S. pachygaster (of which Liosaccus cutaneus
is one on many synonyms, according to Shipp 1974:44-
47) in the Atlantic and Indo-Pacific to Japan. These two
species are not closely related within the genus, as dis-
cussed subsequently, and the loss of prickles undoubted-
ly has occurred independently in the two species. More-
over, Shipp and Yerger (1969a) and Shipp (1974:98)
reported that a minority of specimens of S. nephelus are
completely spineless, and nephelus does not seem closely
related to either pachygaster or angusticeps.

In the equally speciose Fugu, most species have spines
on both the back and belly but two species are spineless,
F. chrysops and F. uermicularis, which on the basis of the
osteological data presented by Kuronuma (1943) do not
seem closely related within the genus and are placed in
different subgenera by Abe (1952). Here again, the loss of

spines in these two species has occurred independently.
T. Abe (pers. commun.) has informed me that, addi-
tionally, F. pardalis is at least superficially spineless.
Lagocephalus and Tetraodon each contain a spineless
species, L. inermis and T. mbu. In T. mbu, and, to a far
lesser extent, in the two scaleless species of Fugu, there
are pits in the skin that appear to be filled with a
secreted matter, which is presently under histological
investigation by the author. In L. inermis, and, to a lesser
extent, in S. pachygaster, the skin of the belly is irregular
in comparison to the smooth skin of the rest of the body,
but no spines are present.

With the exception of these spineless species (and
there are probably others among the species of tetra-
odontids not examined for this work), spines are always
present on the belly of tetraodontids, but there is great
variation both between and within genera in the degree
of covering by prickles dorsally and laterally on the body.
In Lagocephalus, for example, several species (lagoceph-
alus and laevigatus) have spines only on the belly, while
others (lunaris, scleratus, and spadiceus) have prickles
variously developed on the dorsum in front of the dorsal
fin. In Sphoeroides prickles are present (except in the
two naked species) on the belly and dorsum, but are
about as frequently absent as present on the side of the
body (see Shipp 1974). In Torquigener prickles usually
are absent from the sides of the body, but in Ambly-
rhynchotes they are present there. In Ephippion the
prickles on the sides of the body become enormously
enlarged into plates forming a partial carapace (see illus-
tration for form of spines at three specimen sizes), while
one species of Amblyrhynchotes (piosae) has the pro-
jecting portion of the prickle much longer than in any
other tetraodontid. In Chonerhinos and Xenopterus the
prickles of the belly and sides of the body tend to be
larger than in other tetraodontids.

In a few tetraodontids the scales along the course of the
lateral line are modified, being much smaller, more
numerous, and with less development of the protruding
spinule. For example, in Lagocephalus, the three species
(lunaris, scleratus, spadiceus) which have prickles along
the back and upper sides in the region of the lateral line
show great differentiation of the small scales immedi-
ately along the line from those adjacent to them. In the
other tetraodontids examined, there is no differentiation
of the prickles immediately along the line from those ad-
jacent to them, although the course of the lateral line is
sometimes clearly indicated in cleared and stained
specimens by a wider spacing of the prickles leading to a
spineless stripe along the course of the line (especially
clearly marked in the two species of Colomesus, and, to a
lesser extent, in a few species of Monotreta).

The internal anatomical diversity of the tetraodontids
is much greater than the external. The long-based,
many-rayed dorsal fin in Chonerhinos and Xenopterus is
reflected in a greater number of vertebrae than in other
tetraodontids, Chonerhinos has 26 vertebrae modally
(25-27), of which 10 are usually abdominal, while Xenop-
terus has 29 or 30 vertebrae, of which 10 or 11 are ab-
dominal. All other tetraodontids have modal values

ranging from 17 to 23, the only species with modal values
above 20 being numerous species of Fugu and some of
Monotreta and Torquigener. At the opposite extreme, 17
vertebrae modally, are all of the species of Canthigaster,
that of Carinotetraodon and of Guentheridia, and sever-
al species of Lagocephalus, Sphoeroides, and Tetra-
odon, the formula always being eight abdominal and
nine caudal when the total is modally 17. Since both the
Eoplectinae and Triodontidae, ancestral to the tetra-
odontids, have 20 vertebrae (9 -(- 11), as do the other
basal Triacanthodidae (8 -(- 12), the presence of some-
where around 20 vertebrae can be considered general-
ized in tetraodontids, and the great increase in number
in Chonerhinos and Xenopterus as well as the decrease in
number to as low as 17 can both be considered speciali-
zations. Moreover, since the Eoplectinae have only a
moderately long soft dorsal fin base and a moderate
number of rays (14 to 17), with an indication that the
number of rays is in the process of being reduced poste-
riorly in the series, and the Triodontidae a short-based
dorsal fin with a low number of rays (modally 11), the
generalized tetraodontid condition can be considered to
be a relatively short-based dorsal fin and lowered
number of rays (approximately 10 to 12).

Thus, the great increase in the length of the base and
in the number of rays in the dorsal fin of Chonerhinos
and Xenopterus must be considered specializations for
stronger more sustained swimming, perhaps associated
with their invasion of fluviatile fresh waters, although
they are also found in coastal marine waters, with the
possible exception that the problematical C. africanus
may be confined to the Congo. Similarly, the elaborate
open cup nasal apparatus and increased number of
lateral lines in these two genera are specializations.
Chonerhinos and Xenopterus also have specialized skulls
(lacking prefrontals in both genera, and with enor-
mously enlarged frontals in Xenopterus).

Chonerhinos and Xenopterus are highly specialized
genera. Chonerhinos, which has lengthened the base of
the dorsal and anal fins and increased the number of
rays to a lesser extent than Xenopterus, and which has a
less specialized skull than Xenopterus, obviously is the
more generalized of the two. Fraser-Brunner (1943:4)
also considered these two closely related genera as highly
specialized, with the increased numbers of vertebrae and
fin rays "a secondary development." The contention of
Le Danois (1959:248) that these two genera are the most
primitive of the tetraodontids and ancestral to the
diodontids cannot be taken seriously.

While Chonerhinos and Xenopterus are highly
specialized, in my opinion they are not sufficiently differ-
entiated anatomically from the other tetraodontids to be
recognized as even subfamilially distinct (much less as a
separate family). However, one genus of tetraodontids,
Canthigaster, is usually recognized as a distinct family
both in the general ichthyological literature and in that
more specifically devoted to plectognaths, a practice
that I have previously followed. Since about two dozen
species of Canthigaster are usually recognized as a
separate family, and it is being suggested here that this

be changed, a detailed osteological description of C.
rostrata is given here for the sake of comparison with that
of another, less specialized, tetraodontid, Lagocephalus
laevigatas.

Following below is first a discussion of the differences
between Canthigaster and the other tetraodontids, and
then of the anatomical diversity of the Tetraodontinae.

The species of Canthigaster are distinguished from one
another externally primarily by coloration, with slight
aid from the number of dorsal fin rays, one species (am-
boinensis) usually having 11 or 12 rather than 9 or 10,
and perhaps from the caudal peduncle length versus
depth (differences between species not yet well worked
out). The species are as similar internally as externally,
there being an unrelieved sameness in the osteological
configuration: 1) the vertebral column is always highly
arched; 2) the haemal arches of the first three vertebrae
are relatively well developed although they do not usual-
ly completely enclose the haemal canal (first arch some-
times complete); 3) the haemal spines of the other
abdominal vertebrae are well developed and have a char-
acteristic rounded and flattened posterior lobe; 4) the
haemal spines of the caudal vertebrae are also well
developed and of similar shape in all the species; 5) the
neural spines of the first three vertebrae are bifid and
that of the fourth is bifurcate anteriorly; 6) the neural
spine of the seventh abdominal vertebra is always a long
shaft lying along the anterior edge of the first basal
pterygiophore of the dorsal fin, while the neural spines of
the successive vertebrae until the antipenultimate are
also elongate shafts; 7) the neural spine of the
antipenultimate vertebra is slightly expanded antero-
posteriorly and that of the penultimate even more so; 8)
the caudal fin supporting skeleton is essentially similar
in all species; 9) the basal pterygiophores of the dorsal fin
are all placed between the neural spines of the seventh
abdominal to the third caudal vertebrae; 10) the basal
pterygiophores of the anal fin are supported mostly by
the haemal spines of the first to third caudal vertebrae
but sometimes with assistance of the fourth; 11) the first
basal pterygiophores of the dorsal and anal fins are
always much larger than the others; 12) the supraneural
is similarly well developed and usually arched; 13) the
snout is relatively long and narrow; 14) there is always
only one hypohyal; 15) there is never an interhyal; 16)
there is only a slight trace of the dorsal roof of the
myodome; 17) there are never trituration teeth; 18) the
parasphenoid never has a dorsal flange in the orbit to the
frontals; 19) the pharyngobranchial of the first arch
always bears small teeth; 20) the supraoccipital is always
high crested; 21) the ethmoid is extremely long and T-
shaped; 22) the frontal has a pair of posterolateral wings
meeting or closely approaching one another distally; 23)
the prefrontal is flattened dorsally and gently down-
curved anterolaterally and is always placed in about the
middle of the skull; 24) the sphenotics are not much ex-
posed on the dorsal surface of the skull; 25) the vomer is
always well developed and relatively well compressed
laterally and well removed from the prefrontals; and 26)
the parasphenoid always has a deep keel.

'S S ^'

III

f/f

■* •J' i'

.r

A\\\ll M 111

Ill

yi^,

^*/>
/y^^

f^m:K

li

In general, there is such skeletal similarity between the
•pecies that for most practical purposes it can be truly
said that to see the skeleton of one species of Canthi-
gaster is to have seen them all, which is decidedly not the
case with other genera of tetraodontids with similar
numbers of species.

Until recently I had thought that a goodly number of
these monotonously similar features of the species of
Canthigaster were not shared by any tetraodontids and
that Canthigaster could with good justification be recog-
nized as familially distinct from the Tetraodontidae. But
the skeletal structure of one relatively specialized tetra-
odontid examined recently, and, to a lesser extent, that
of a few other species related to it, shows so many
similarities to Canthigaster that I am forced to the con-
clusion that they share a close common ancestry. From
this point of view, Canthigaster seems to me to merit
only subfamilial recognition on the basis of the
anatomical differences between it and the other tetra-
odontids which remain after consideration of the struc-
ture of Carinotetraodon lorteti, and its relatives.

Recently described as Tetraodon somphongsi
(Klausewitz, April 1957a and 1957b) and immediately
thereafter as Carinotetraodon chlupatyi (Benl, July
1957; Benl and Chlupaty 1957), based on specimens from
fresh water in Thailand, Dekkers (1975) has shown that
the proper specific name is lorteti Tirant 1885. Benl
(1959) suggested that the species be retained in the genus
Carinotetraodon since it differed from all other tetra-
odontids {Canthigaster being considered familially dis-
tinct) in being able to raise up a large fold of skin in the
midline both dorsally and ventrally, just as in Canthi-
gaster. It was believed that such behavioral and
anatomical skin changes would probably indicate other,
more internal, differences between somphongsi and the
other tetraodontids. Tyler (1978) has reviewed the grow-
ing aquarium literature on lorteti and the history of its
name.

Because of its moderate tubelike nasal apparatus with
a single rounded nostril, Carinotetraodon lorteti
otherwise would have to be accommodated in Mono-
treta. In fact, on the basis of its coloration and external
morphometries, lorteti seems to be most closely related
to M. caria (Hamilton-Buchanan), but lorteti is suffi-
ciently distinct from the four species of Monotreta ex-
amined (cutcutia, which is the type-species, palem-
bangensis, leiurus, and gularis) to warrant its retention
in Carinotetraodon.

Regardless of its nomenclature, C. lorteti shares with
Canthigaster not only the ability to lift up a ridge of skin
dorsally and ventrally along the midline, but a number of
internal features as well. The vertebral column in
Carinotetraodon is highly arched and almost exactly
similar in configuration to that of Canthigaster. The first
three vertebrae have bifid neural spines and well-
developed haemal spines which do not completely
enclose the haemal canal, while the fourth vertebra has
the neural spine bifurcate anteriorly, and a large haemal
spine, although the flattened region of the spine does not
form a distinctive posterior lobe as in Canthigaster. The

haemal spines of the fifth and subsequent vertebrae are
as well developed as in Canthigaster, but again they do
not have distinctive posterior lobes. The neural spine of
the fourth vertebra is shorter than in Canthigaster, that
of the fifth similar to Canthigaster and that of the sixth
longer. The long neural spine of the seventh vertebra,
supporting the anterior edge of the first basal pterygio-
phore, and the succeeding neural spines until the anti-
penultimate, are also long shafts, just as in Canthi-
gaster, and the basal pterygiophores of the dorsal fin are
accommodated between the neural spines of the 7th to
11th vertebrae in both genera. However, even though
both genera possess a specialized reduced number of 17
vertebrae, there are eight abdominal in Canthigaster and
only seven in Carinotetraodon. In both genera there are
seven vertebrae with neural spines anterior to the first
basal pterygiophore of the dorsal fin and seven with
neural spines posterior to the last basal pterygiophore.
The caudal fin supporting skeleton, and, in general, the
last three vertebrae, are highly similar in both genera.
The anal fin basal pterygiophores are closely supported
basally primarily by the haemal spines of the first three
caudal vertebrae, with secondary assistance from that of
the fourth, in both genera. Although the number of basal
pterygiophores in the dorsal (12) and anal (9) fins of
Carinotetraodon is greater than in Canthigaster (usually
8 or 9 dorsal and 5 anal), the first basal pterygiophores of
both fins in Canthigaster are enormously larger than the
others and could represent fusion products of what are
the first few pterygiophores in these fins in Carinotetra-
odon.

The similarities between Carinotetraodon and Canthi-
gaster in the skull are less extensive than in the vertebral
column, but are of phylogenetic interest. In Carino-
tetraodon the frontal does not have a pair of postero-
lateral processes whose distal ends meet or closely ap-
proach one another to partially enclose the muscle to the
operculum, one of the most characteristic features of
Canthigaster, but a perhaps functionally somewhat
similar arrangement is formed by the sphenotic of
Carinotetraodon. The sphenotic in Carinotetraodon is
broadly present on the dorsal surface of the skull, with an
anterolateral wing excluding the posterolateral edge of
the frontal from the upper rear margin of the orbit, and,
more importantly to the present discussion, a postero-
lateral wing prominently projecting out from the lateral
surface of the skull, a feature found in none of the other
tetraodontids examined. The space between the antero-
lateral and posterolateral wings of the sphenotic of
Carinotetraodon would seem to correspond to that
enclosed by the posterolateral wings of the frontal of
Canthigaster. It is suggested here that the condition such
as found in Carinotetraodon could have given rise to than
in Canthigaster by the gradual posterolateral expansion
of the frontals overlying first the anterolateral wing of the
sphenotic, which would become reduced in size and
eliminated eventually, and then overlying the postero-
lateral wing of the sphenotic, which likewise would be
eliminated eventually as the frontal took over its suppor-
tive function, leaving only a small portion of the

sphenotic visible on the dorsal surface of the skull.
Completion of the process would require only the coming
together in close proximity of the distal ends of the two
posterolateral wings of the frontal (those that had replac-
ed the two wings of the sphenotic) to form a more com-
plete well around the muscle to the operculum.

In Carinotetraodon the supraoccipital crest is not as
high and laterally compressed as in Canthigaster, but it
is higher than in most other tetraodontids, and such a
condition could easily give rise to that of Canthigaster.
The parasphenoid of Carinotetraodon has a well-
developed ventral flange and is arched dbrsally in front of
the orbit much as in Canthigaster. In Carinotetraodon
the prefrontal is relatively posterior in position, being
only slightly forward of about the middle of the skull,
and while it does not have the characteristic smooth flat
dorsal surface gently curved downward anterolaterally as
found in Canthigaster, it is in contact anteriorly with the
vomer only by a thin but wide anterior prolongation. If
this anterior prolongation was reduced in size and then
lost, the prefrontal of Carinotetraodon would have the
Canthigaster condition of being far removed from the
vomer. The ethmoid of Carinotetraodon is not especially
long in comparison to many other tetraodontids, but it is
long enough for one to envision its elongation in the
process of lengthening the snout in Canthigaster and the
conversion into a T-shaped cross section. A few other
features that would be necessary to the conversion of a
Carinotetraodon-\ike fish into a Canthigaster-Vike one

are the development of a high anterodorsal wing on the
suboperculum and a straightening of the dorsal edge of
the operculum.

One of the important distinguishing features of
Canthigaster is the single nostril at the end of a very
small short tube, and it is reasonable to suppose that
Canthigaster was derived from a tetraodontid line with a
single nostril at the end of a better developed nasal tube.
Among tetraodontids a nasal tube with a single nostril is
found only in a few species of Monotreta (a few other
species of the genus have the opening bilobed) and in the
closely related Carinotetraodon.

Only Carinotetraodon and Canthigaster are known to
have the ability to lift up a ridge of skin along the
midline, and the osteology of Carinotetraodon shows
striking similarities to that of Canthigaster, more so than
does that of any other tetraodontid. It seems highly likely
that Carinotetraodon and Canthigaster share a close
common ancestry, from which Carinotetraodon has
diverged little but from which Canthigaster has become
highly enough modified to warrant subfamilial recog-
nition. It would be of interest to know if the common
ancestral group was marine, fresh water, or euryhaline.

Monotreta and Chelonodon, closely related to Carino-
tetraodon, contain species which show a few of the same
features (arched vertebral column, well-developed
haemal spines on the abdominal vertebrae) of similarity
between Carinotetraodon and Canthigaster, but, to a
much lesser degree, than in Carinotetraodon. Neverthe-

parasphenoid
prefrontal

Figure 245.— Spfcoeroide* mat

(atturlateral view of head,

97.5 mm SL, Virginia.

x^^^-ff

^A \\ \ Xs^mplecc

''^"""y orticulor \

\ \\ -"<.p.erygo,d
interoperculum \\ quadrate

angula

ectoplerygoid

Figure 246.— Colomesue
psUtacus : lateral view of
head, 178 mm SL, Surinam.

frontal

paraspheno,d\
prefrontal \ J\

hyomandibul
sphenotic i

frontal ^A^

^--

von,er ^:<C L ^

-—-"'^^ \

X"

e.Kn,o,d\: u^ .---v>

5x \

/

palot,ne\^;. ^A . ^ \/

r\ -^

yLx:

basioccipitoi

Figure 247.— Xmftlyr/iync/iofes
riehei: lateral view of head,
59.4 mm SL, New Zealand.

^roperculu
preoperculum
ectopterygoid \ rnetcpterygoid
symptectic

less, this adds a confirmatory note to the hypothesized
ancestry of Canthigaster as being from among a group of
Carinotetraodon-Mk^ fishes.

In contrast to the origin and anatomical speciali-
zations of the Canthigasterinae discussed above, the
diversity of the Tetraodontinae is far greater and more
complex. Before a consideration of the anatomical diver-
sity between genera can be undertaken, it is first
necessary to determine the diversity to be expected

within genera, especially those with relatively large
numbers of species. With this in mind, as many species
of the systematically relatively well known and speciose
genera Sphoeroides and Lagocephalus have been ex-
amined as possible (11 of Sphoeroides and 6 of Lago-
cephalus, the recognition of the Atlantic species of the
former being greatly aided by the recent works of Shipp,
see literature cited). Sphoeroides occurs in the eastern
Pacific and Atlantic (mostly in the western Atlantic),

Figuit24&.—Fuguchrytop8: lateral
view of head, 98.6 mm SL, India.

Figure 249.— fu^ oblongus: lateral
view of head, 46.2 mm SL, India.

and while they are primarily shallow-water coastal
forms, one deepwater species (pachygaster) occurs on
both sides of the Atlantic as well as in the Indo-Pacific.
Lagocephalus is world wide in distribution and tends to
be found further offshore than Sphoeroides, with at least
one of its species {lagocephalus) relatively oceanic and
circumtropical in distribution.

The basic structure of the vertebral column does not
vary greatly among the species of Sphoeroides, there
always being eight abdominal vertebrae modally, with
most species {angusticeps, dorsalis, greeleyi, tricho-
cephalus, lobatus, sechurae, spengleri) having a total of
17 vertebrae, with others {annulatus, nephelus, pachy-

gaster, testudineus) a total of 18, one (maculatus) a total
of 19, and another (marmoratus) a total of 20 (all figures
modal). Lowered numbers of vertebrae can be considered
a specialization.

In contrast to the vertebral column, the range in skull
structure in Sphoeroides is so diverse that if only the two
species at the extremes of the osteological range were
studied one would receive the definite impression that
they were not at all closely related and could not be con-
tained in the same genus.

To contrast the two extremes, S. dorsalis has: 1) a
narrow skull £ind long slender snout region; 2) the
palatines are not widely divergent and the jaws are rela-

interoperculum

mesopterygoic

Figure 250.— Arothron
atellatus : lateral view of
head, 292 mm SL. India.

hyomandibulor
proolk I pterotic
sphenottc /|

Figure 251. —Tctroodonm6u: lateral
view of head, 47.7 mm SL, Congo.

nteroperculum

ectoplerygoi

lively narrow; 3) the ethmoid and vomer are extremely
elongate; 4) the prefrontals are placed in about the mid-
dle of the skull; 5) the sphenotics form only the rear
margin of the orbit; and 6) the frontals are wide poste-
riorly over the skull but very narrow interorbitally and of
constantly decreasing width anteriorly. By comparison,
S. annuiatus has: 1) a broad skull and short wide snout
region; 2) the palatines are widely divergent and the jaws

are relatively wide; 3) the ethmoid is of moderate length
and the dorsal surface of the vomer is exceptionally
short; 4) the prefrontals are placed anteriorly well in
front of the middle of the skull; 5) the sphenotics are
prolonged anterodorsally to form not only the rear but
also part of the dorsal margin of the orbit; 6) the frontals
are laterally expanded anteriorly to the extent that they
are wider anteriorly than posteriorly.

Figure 252.— Chelonodon

fluviatilia: lateral view of head,

84.7 mm SL. Thailand.

metapterygoid
mesopterygoid

Between these two extremes of dorsalis and annulatus,
however, are other species of Sphoeroides which form an
almost continuous series of intermediate conditions.
Thus, S. lobatus, angusticeps, and nephelus have only
slightly wider skulls and slightly less elongate snout
regions than S. dorsalis and otherwise are similar to it,
while a number of other species (S. spengleri, maculatus,
and greeleyi) have progressively wider skulls and less
elongate snout regions, leading to species such as S.
trichocephalus and pachygaster with wider frontals,
progressively less slender or tapering anteriorly, and
eventually to S. testudineus in which the frontals are
only slightly tapered anteriorly above the orbit, not being
greatly narrower there than posteriorly, and only taper-
ing to a point far anteriorly and abruptly at the rear edge

of the prefrontals. The above series of progressively wider
skulls does not necessarily indicate precisely similar
specific relationships.

Among the species of Sphoeroides the condition of the
skull in testudineus is that which most closely ap-
proaches that of annulatus. In fact, these two species are
usually considered to be closely related geminates to
each side of the Central American isthmus {annulatus in
the eastern Pacific and testudineus in the Gulf of Mexico
and the Caribbean) which have retained relatively simi-
lar color patterns of pale circles and reticulations unique
in the genus.

One would suspect that the extremes of dorsalis and
annulatus both represent specializations, and that a
generalized skull structure in tetraodontids would

prerosphenoid hyomandibular

parasphenoid
frontal

Figure 254.— Chonerhinos

modestus: lateral view of head,

31.0 mm SL, Borneo.

Figure 255.— Xenopterug
naritua: lateral view of head,
143 mm SL, Bay of Bengal.

feature a moderate skull width and snout region length,
with the frontals wide posteriorly but gradually and
evenly narrowed anteriorly, and the sphenotics not
prolonged anterodorsally to form more than the rear
margin of the orbit, these being conditions as found in
the ancestral triodontids. In this view, species such as S.
maculatus, trichocephalus, and pachygaster have
relatively generalized skulls in which the frontals are of
moderate width and taper gradually anteriorly, while,
going in one direction, greeleyi is perhaps very slightly

specialized in at least this one respect by the slightly
abrupt narrowing of the frontals at the level of the
prefrontals, while testudineus is more specialized by the
lesser anterior tapering of the frontals and the more
abrupt narrowing at the level of the prefrontals. The
testudineus-Uke condition could be ancestral to that of
annulatus through an increased expansion of the ante-
rior region of the frontals and an anterodorsal expansion
of the sphenotics.
The monotypic Guentheridia in the eastern Pacific

Figure 256.— Carinotetraodon lorteti: lateral
view of head, 35.9 nun SL, locality i

differs externally from Sphoeroides only in having the in-
ternal epithelium of the nasal sac deeply pitted rather
than smooth or with ridges, and the gill rakers anteriorly
on the fourth arch less well developed. The single
species, formosa, is often placed in Sphoeroides (as by
Fraser-Brunner 1943). The skull structure of formosa is
remarkably similar to that of annulatus, being an even
more specialized version of it, with the frontals more
laterally expanded anteriorly and the sphenotics even
more anterodorsally expanded to form at least half of the
dorsal margin of the orbit. Only in annulatus among the
species of Sphoeroides does the parasphenoid have a dor-
sal flange in the interorbital septum which meets the
frontal, and formosa also has a similar flange. The color
pattern of formosa can be considered a variation on that
of annulatus, with more numerous light reticulations and
circles and larger dark spots. It is obvious that annulatus
is a specialized derivative of a testudineus-Vike stock, and
that formosa is a specialized derivative of an annulatus-
like stock. Thus, formosa should be included in Sphoe-
roides as the most extreme example of frontal widening
and sphenotic anterior prolongation, for the pitted
epithelium of its nasal sac and smaller gill rakers on the
fourth arch certainly are not sufficiently distinctive
features to warrant even subgeneric recognition of for-
mosa within Sphoeroides.

Two of the examined species of Sphoeroides in which
the skull is only slightly less wide and the snout region
only slightly less elongate than in the extreme condition
of dorsalis share with dorsalis one external peculiarity
not otherwise found among the tetraodontids. In the
western Atlantic S. dorsalis and in the eastern Pacific S.
lobatus and angusticeps there are a pair of dark dermal
flaps which are relatively constant in position on the dor-

sum (sometimes absent as an intraspecific variation),
while Shipp (1974:59) reported that this pair of flaps is
also present in the eastern Atlantic S. marmoratus, not
studied here. These three species (four including mar-
moratus) undoubtedly share a close common ancestry,
with the ancestral species becoming divided into two
isolated populations with the emergence of the Central
American isthmus. The Atlantic population evolved into
the highly specialized (in narrowness of skull) dorsalis
(and probably marmoratus), while the eastern Pacific
population diverged into lobatus and angusticeps with a
slightly lesser narrowing of the skull, differing from one
another mainly in angusticeps having a plainer colora-
tion and having lost all the spines.

Sphoeroides nephelus, which also has a relatively
narrow skull, is probably also related to the line which
gave rise to dorsalis, lobatus, and angusticeps, while its
closest relative is maculatus (see Shipp and Yerger
1969a, b, who show that parvus, not studied here,
nephelus, and maculatus constitute a closely related
species complex). On the basis of its skull structure,
spengleri, with a moderate skull width, is not far remov-
ed from that complex, while pachygaster, with a
somewhat wider skull, may be derived from a spengleri-
like form.

The deepwater S. pachygaster, with its thick, flabby
spineless skin, seems externally to be one of the most dis-
tinctive and specialized species of Sphoeroides. However,
internally pachygaster is only moderately specialized by
a slightly more than moderate width of the skull, a slight
anterodorsal prolongation of the sphenotics, a less firm
ossification of the skull, and an unusually great amount
of sculpturing on the dorsal surface of the skull, especial-
ly on the frontals, and by the loss of teeth on the first

pharyngobranchial, pachygaster being the only species of
Sphoeroides examined lacking at least minute teeth on
this pharyngobranchial. The characteristics of the skull
of pachygaster were first well described and illustrated
by Kuronuma (1943). The possible relationships of S.
pachygaster to Lagocephalus is discussed subsequently.

In addition to the anatomical diversity of Sphoeroides
discussed above (number of vertebrae, skull shape, rare
presence of a dorsal flange of the parasphenoid, rare loss
of teeth on the first pharyngobranchial), a few other
items can be mentioned. A dorsal and ventral hypohyal
are present in most species, but only a single hypohyal is
present in S. greeleyi and trichocephalus (and
Guentheridia formosa). An interhyal is absent in most
species, but it is present in greeleyi, maculatus, and
pachygaster. All species have two or more (most
numerous in annulatus, in a disorganized plate) tritura-
tion teeth to either side of the midline in the upper jaw,
but most species do not have trituration teeth in the
lower jaw, these being present only in S. angusticeps, an-
nulatus, dorsalis, and greeleyi (and Guentheridia for-
mosa) as three or more units, often tending to form a
platelike structure (especially with increasing specimen
size?) to either side of the midline. Remnants of the dor-
sal roof of the myodome are present in a few species. In S.
dorsalis medial prongs of the prootics are well de-
veloped, almost as well as described and illustrated
here for Lagocephalus laevigatus, while smaller prongs
are present in S. angusticeps and greeleyi. In S.
maculatus most specimens have only a very weak indica-
tion of a medial projection from the prootics, but lome
have moderately developed prongs about as large ..s in
angusticeps and greeleyi.

The amount of variation in the structure of the dorsal
surface of the skull that is correlated with specimen size
can be judged from the illustration of four specimens of
S. maculatus ranging from 12.4 to 201 mm.

While the above gives an idea of the anatomical diver-
sity to be expected within a speciose genus and at least a
broad view of the species relationships with one another,
a far more detailed analysis based on numerous external
features of all of the Atlantic species is given by Shipp
(1974), including several newly described species not
studied here.

All of the six species of Lagocephalus have been ex-
amined to provide an additional survey of the anatomical
diversity to be expected within a genus. The variation in
skull structure in Lagocephalus is only slightly less than
that found in the far more numerous species of Sphoe-
roides examined.

Three of the six species of Lagocephalus have relative-
ly deep chunky bodies and drab coloration and seem to
be closely related on the basis of external features alone,
differing mainly in scale pattern on the dorsum; lunaris
having scales from the snout to the dorsal fin, spadiceus
from the rear of the snout to no more than half way back
the distance between the head and dorsal fin, and iner-
mis with spmes totally absent. These three species have
the olfactory epithelium folded over the entire inner sur-
face of the sac, usually with two of the folds on the rear

wall much larger than the others and themselves bearing
additional smaller folds. Two other species are somewhat
more elongate and conspicuously patterned and seem
closely related, laevigatus with a pattern of crossbars on
the dorsum and lagocephalus, slightly more elongate
than laevigatus, with a dark dorsum, often with black
spotting, and gradually paler ventrum. Both of these
species lack spines on the back but have them well
developed on the abdomen, and both tend to have only
one of the folds on the rear wall of the nasal sac much
larger than the others and itself bearing smaller folds.
The most distinctive species of Lagocephalus is scler-
atus, with an elongate and depressed body and distinctly
depressed caudal peduncle, and a color pattern of a dark
dorsum with pale and dark spots and reticulations

Figure 257 .Sphoeroidet maculatus: dorsal

views of skulls to show changes in shape, amounts

of cartilage viaible, and degree of auturing with

increasing specimen sizes, as indicated,

12.4-201 mm SL, New Jersey and Virginia.

abruptly changing on the middle of the side of the body
to plain pale below. It also has the olfactory epithelium
less folded than in the other species, the folds confined to
the rear wall of the sac and the enlarged folds not them-
selves bearing elaborate folds, and fewer dorsal and anal
fin rays. In scleratus the dorsal rays are usually 10 to 12
and in the other species 13 to 15, while the anal fin usual-
ly has one less ray than the dorsal in all of the species.
While the depressed body and caudal peduncle are
specializations in some way associated with its probably
rapid swimming, and its color pattern unique in the
genus, the less elaborate folding of the olfactory
epithelium is probably a more generalized condition than
in the other species. Since the majority of tetraodontids
have relatively short-based fins with relatively few rays,
as does Triodon, I would think that the moderate number
of rays in scleratus is more generalized than the larger
number in the other species.

There are more numerous internal differences between
the species of Lagocephalus. A specialized dorsal flange
of the parasphenoid in the interorbital septum that
meets and sutures with the ventral surface of the frontals
is present in laevigatas, lagocephalus, lunaris, and
spadiceus but is absent in inermis and scleratus. In
laevigatus and lagocephalus the pteroaphenoid is in con-

Figure 258.— Dorsal views of skulls of:

A, Sphoeroidet dortalit, 155 mm SL, Florida;

B,S. nepheliu, 128 mm SL, Bahamas:

C, 5. lobattu, 67.5 mm SL, Panama (Pacific).

tact with the dorsal flange of the parasphenoid, but not
in the other two species with a dorsal flange. A single
hypohyal is present in all species except laevigatus,
which retains the more generalized condition of both a
dorsal and ventral hypohyal. The first pharyngo-
branchial is toothless in laevigatus, lagocephalus, and
spadiceus, but in lunaris it bears a few minute teeth
(much smaller than those on the second and third
pharyngobranchials) and in inermis and scleratus
numerous minute teeth, the retention of numerous teeth
being the most generalized condition.

A feature of the skull unique to Lagocephalus among
the tetraodontids is the development of a posterolateral
wing of the frontal which meets or closely approaches the
posterodorsal surface of the pterotic to completely or par-
tially roof over a temporal fossa. These specialized wings
are best developed in lunaris, laevigatus, spadiceus, and
lagocephalus (in respective increasing order of greatest
size and length of the wing), in which their wings usually
make direct contact with the pterotic to completely roof

over the fossa, but the wings are relatively shorter in iner-
mis and scleratus and do not meet the posterodorsal sur-
face of the pterotic and thus only incompletely roof
over the fossa. The lesser development of the wings in m-
ermis and scleratus would seem to be the most generaliz-
ed condition. The posterolateral region of the pterotic is
directed more or less laterally in laevigatas, lunaris,
scleratus, and spadiceus, but it is prolonged postero-
laterally in inermis and, especially greatly so, in lago-
cephalus. The ventral flange along the middle of each
side of the rear of the skull formed by the pterotic and ex-
occipital is much longer in lagocephalus than in the other
species. A larger portion of this flange in Lagocephalus is
formed by the exoccipital than in most, but not all, other
tetraodontids (see, for example, illustration of Tetraodon
lineatus).

The shape of the anterior portion of the frontals in
Lagocephalus is probably of as much phylogenetic in-
terest as is that of the posterolateral wing development.
In scleratus the frontal tapers gradually and evenly
toward the front and is of moderate width, this being
similar to the condition in the more generalized species of
Sphoeroides. In inermis the frontal is somewhat broader
over the orbit than in scleratus and abruptly begins to
taper to a point at the rear of the prefrontals. In lunaris
the configuration is much the same as in inermis, except

Figure 259.— Dorsal views of skulls of:

A, Sphoeroides anguaticepa, 172 mm SL. Galapagos;

B, S. epengleri, 92.8 mm SL, Nicaragua;

C, S. maculattu, 179 mm SL, Virginia.

that there is greater abruptness in the tapering, the
length of the broad region of the frontal above the orbit is
shorter, and the lateral edge of the prefrontal forms a
proportionally greater amount of the edge of the orbit. In
laevigatus, lagocephalus, and spadiceus the frontal is
even wider over the orbit than in lunaris, and is similarly
abruptly tapered, with the prefrontal forming a large
part of the relatively straight edge of the orbit. Since a
gradually tapered frontal of moderate width, as found in
the ancestral triodontids, can be considered the
generalized condition, the above series is one of in-
creasing specialization. The most generalized species in
this respect, scleratus, is also one of the two species with
the most generalized condition of the posterolateral wing
of the frontal, while the other species {inermis) with a
generalized posterolateral wing has the next most
generalized condition of anterior tapering.

The extreme posterolateral region of the dorsal surface
of the pterotic in most species of Lagocephalus is uprais-
ed into a flange, with which the posterolateral wing of the
frontal articulates in lunaris, spadiceus, and laevigatus.

or is directed toward in scleratus and inermis. No such
flange is present in lagocephalus, but the area of contact
between the especially long posterolateral prolongations
of both the frontal and pterotic is much greater than in
the other species.

The ethmoid in Lagocephalus is relatively long in all
six species, but its width varies from moderate in
scleratus and lunaris to wide (the other four species).
Anteriorly the ethmoid always broadly sutures with the
dorsal surface of the vomer, and, as explained for
laevigatas in the detailed osteological description, this
articulation sometimes becomes fully fused in large
specimens. In laevigatas the ethmoid and the haemal
spines of the sixth and seventh, and, to a lesser extent, of
the eighth, caudal vertebrae become swollen or hypero-
stotic in large adults. The only other species studied here
as a large adult specimen, a 214 mm lagocephalus, does
not show any hyperostosis, but the specimens oi inermis,
lunaris, scleratus, and spadiceas studied are all too small
to have hyperostotic parts even if that should be the
norm for large adults.

The supraneural element is relatively long in most
species of Lagocephalus, while it is extremely long in
lagocephalus and only moderately long in inermis.
Medial prongs from the prootic in the rear of the orbit,
representing the remains of the dorsal roof of the
myodome, are well developed in scleratus and
laevigatus, reaching almost to the midline, and are
moderately developed in lunaris and spadiceus, while

Figure 260.— Dorsal views of skulls of:

A, Sphoeroides greeteyi, 60.8 mm SL, locality

unknown: B, S. trichocephalus, 57.1 mm SL, Panama (Pacific);

C, S. pachygaster. 117 mm SL, Mozambique.

they are absent in inermis and lagocephalus. The
amount of surface sculpturing (not fully shown in the il-
lustrations) of the top of the skull varies from slight in
scleratus to moderate or well developed in the other
species. This, however, may be subject to intraspecific
variation or change with increasing specimen size, for
Kuronuma (1943) reported that lunaris and spadiceus, in
contrast to inermis, had little surface sculpturing, mostly
confined to the ethmoid.

The modal number of vertebrae in Lagocephalus is 17
{lunaris, scleratus, spadiceus), 18 (inermis, lago-
cephalus), and 19 (laevigatus), with the lower numbers
more specialized. Neural prezygapophyses, especially on
the caudal and more posterior abdominal vertebrae, are
extremely well developed as anterolaterally directed
prongs in lagocephalus and scleratus, the two most
streamlined and probably rapid swimming species, but
are only moderately developed in the other four species.
Likewise associated with muscle attachment are keels or
ridges along the dorsolateral region of the neural arches
(and, to a lesser extent, along the ventrolateral region of
the haemal arches of the caudal vertebrae). These keels
are well developed in lagocephalus and scleratus,
moderately developed in laevigatus and spadiceus, and

scarcely developed at all in inermis and lunaris. The
neural and haemal spines of the last few caudal vertebrae
are shorter in scleratus than in the other species,
associated with the depressed caudal peduncle.

A feature unique to Lagocephalus among the tetra-
odontids is the posterior prolongation of the distal ends of
the last basal pterygiophores of the dorsal and anal fins
as prongs above and below the neural and haemal spines
of the vertebrae just behind the bases of these fins. The
prongs are moderately developed in inermis and lago-
cephalus, slightly longer in laevigatas and lunaris, and of
greatest length in spadiceus (dorsal basal pterygiophore
only, the anal basal pterygiophore having a prong of only
moderate length) and in scleratus.

All of the species of Lagocephalus have two or three
rounded trituration teeth in a single series to each side of
the midline of the upper jaw but none in the lower jaw,
and all possess an interhyal, while none of the abdominal
vertebrae possess complete haemal arches. The first ray
in both the dorsal and anal fins of Lagocephalus is excep-
tionally short (slightly longer in young than adults) and
sometimes lacks cross-striations, and it is shorter than
the first ray of any of the other tetraodontids examined.
In all of the species of Lagocephalus the form of the
epural is characteristic. In other tetraodontids the epural
varies from a more or less square block of bone to an
elongate rod, but the orientation of the bone is more or
less vertical or obliquely anteroventral to posterodorsal,

Figure 261 .—Dorsal views of skulls of:

left, Sphoeroides testudineua, 68.5 mm SL, Venezuela;

right, S. annulatus, 174 mm SL. locality unknown.

while in Lagocephalus the epural is dorsoventrally flat-
tened and prolonged anteroposteriorly, in which direc-
tion it is oriented, an apparently specialized condition
unlike that of any other gymnodont. The presence of
two lateral lines on the body in Lagocephalus can also be
considered a specialization, since only a single lateral
line is present in triodontids and the basal triacan-
thodids.

Fraser-Brunner (1943) believed that Lagocephalus is
the most primitive tetraodontid and that Canthigaster
was derived from it, but he gave no reasons for so think-
ing. As discussed earlier, Canthigaster probably evolved
from a Carinotetraodon-like line. I find no particular
similarities between Canthigaster and Lagocephalus.
Fraser-Brunner thought that the skull of molids was
"almost exactly similar to that of Lagocephalus, even
possessing the postero-lateral limbs of the frontals
characteristic of that genus." In fact, the frontals of
molids do not have prominent posterolateral wings and
to my eyes the skulls of molids show no particular
similarities to those oi Lagocephalus, but, rather, as dis-
cussed under the Molidae, they do show similarities to
that of Triodon. One of the few similarities that at least a
few species of Lagocephalus have to Triodon not found in

other tetraodontids are the exceptionally well-developed
neural prezygapophyses as anterolaterally projecting
prongs found in lagocephalus and scleratus, which, while
not as large as in Triodon, at least approach those of
Triodon in size. However, these prongs and the keels on
the neural arches seem to be specialized structures
associated with the musculature of a strongly powered
caudal peduncle for rapid swimming and are probably
independently evolved in Triodon and in a few of the
species of Lagocephalus.

Lagocephalus cannot be reasonably considered as an
especially generalized tetraodontid. It possesses many
unique features not found elsewhere among the tetra-
odontids or the ancestral triodontids and triacanthodids
(most importantly the posterolateral wings of the fron-
tals partially or completely enclosing a temporal fossa,
the horizontal epural, especially short first dorsal and
anal fin rays, posterior prolongation of the distal ends of
the last dorsal and anal fin basal pterygiophores) and
other features that, while not unique to it, can still be
considered specialized (highly folded olfactory
epithelium, two lateral lines).

It seems more reasonable to me to consider Lago-

Fifcure 262.— Dorsal views of skulls of:

left, Guentheridia formoaa, 175mmSL, Panama

(Pacific); right, Colomeeus peittacus, 179 mm SL,

Surinam.

cephalus a specialized derivative of a Sp/ioerotdes-like
group, for the following reasons. Lagocephalus scleratus
must be considered, overall, the most generalized species
of the genus, because, in spite of a few specializations
probably mostly associated with its mode of swimming
(depressed body and caudal peduncle, 17 vertebrae,
single hypohyal, exceptionally well-developed neural
prezygapophyses and last basal pterygiophore prongs) it
has a lesser degree of development of the specialized
features that characterize the genus (especially the even-
ly tapered frontals of moderate width and the relatively
poorly developed posterolateral wings which do not meet
the pterotic and thus only partially enclose the temporal
fossa) as well as many other generalized features (first
pharyngobranchials with numerous teeth, no dorsal
flange on the parasphenoid contacting the frontals in the
orbit, well-developed prootic prongs representing the
remains of the dorsal roof of the myodome, moderate

325

number of dorsal and anal fin rays, least highly folded
olfactory epithelium, minimal surface sculpturing on
top of the skull, spines on the back as well as the bel-
ly), a greater combination of generalized features than
any of the other species.

A scleratus-like stock, minus its specializations, prob-
ably gave rise to inermis, which alone among Lago-
cephalus shares a few of the generalized features of
scleratus (first pharyngobranchial with numerous teeth,
no dorsal flange on parasphenoid) and has the second
most generalized configuration of the cranium, with the
posterolateral wing of the frontal similar to scleratus and
the anterior region of the frontal only slightly more
specialized than in scleratus by its greater width and
more abrupt tapering. It is probable that this line leading
to inermis also gave rise, by further specializations and
before the scales were totally lost, to lunaris, and that a
lunaris-like stock is ancestral to spadiceus, laevigatus,
and lagocephalus. I would guess that in general
spadiceus has remained closer to this ancestral line than
have laevigatus and lagocephalus, with lagocephalus a
specialized more open water derivative of a laevigatus-
like stock.

If scleratus can be accepted as the most generalized
species of Lagocephalus, then it is probable that Lago-
cephalus is an originally deepwater derivative of a
Sp/ioeroides-like ancestral group, for the cranium of

Figure 263.— Dorsal views of skulls of:

A, Lagocephalus inermis, 52.2 mm SL, Bay of Bengal;

B, L. scleratus, 80.9 mm SL, Philippines;

C, L. lunaris, 62.3 mm SL, Bay of Bengal.

scleratus is not markedly different from that of several
species of Sphoeroides, including that of pachygaster,
which alone among Sphoeroides has a deepwater habitat,
a highly folded olfactory epithelium, and a general
countenance similar to that of the deep, chunky bodied
species of Lagocephalus, such as inermis, the closest
relative of scleratus and a species which may have
retained the general appearance of the ancestral line
leading to it and scleratus, which scleratus lost as it
became specialized for more rapid swimming. In S.
pachygaster the frontals are about as wide as in L.
scleratus, and they are similarly evenly tapered anteri-
orly, and the ethmoid-vomerine regions are remarkably
similar. Moreover, while not prolonged posterolaterally,
the broad posterior end of the frontal projects out over
the underlying epiotic and pterotic as a short roof over
this region. In many other species of Sphoeroides,
both specialized (e.g., dorsalis and testudineus) and
generalized (e.g., spengleri and maculatus) the posterior
end of the frontal is slightly prolonged posterolaterally
over the underlying epiotic and pterotic as a short roof.
An only slightly increased posterolateral prolongation of

Figure 264.— Dorsal

views of skulls of:

left, Lagocephalu» apadiceus,

98.2 mm SL, Mozambique;

right, L. lagocephalus,

214 mm SL, Malpelo Islands.

the frontal in pachygaster or any of the other species of

Sphoeroides with a moderately wide, evenly tapered
frontal would lead to exactly the condition of the frontal
found in L. scleratus.

These species of Sphoeroides also share with L.
scleratus the generalized features of the absence of a dor-
sal flange of the parasphenoid in the orbit, while most of
them have numerous small to minute teeth on the first
pharyngobranchial, except for S. pachygaster in which it
is a toothless plate similar to that of most Lagocephalus.
Most of these species of Sphoeroides have several small
rounded trituration teeth in a single row to either side of
the midline and none in the lower jaw, just as in Lago-
cephalus, and most also have both a dorsal and ventral
hypohyal, a probable condition of the ancestral stock of
Lagocephalus since L. laevigatus retains two hypohyals.
A few of these species of Sphoeroides have moderately to
well -developed prootic prongs in the rear of the orbit
representing the remains of the dorsal roof of the
myodome, and a few have interhyals, all conditions to be
expected in the ancestral stock of Lagocephalus.

In short, Lagocephalus seems relatively clearly to have
been derived from a Sp/ioeroides-like ancestral group,
the connection between the two being most clearly seen
in L. scleratus and in any number of relatively gener-
alized species of Sphoeroides, as well as particularly in S.
pachygaster. I would suspect that it was the same line of

Sphoeroides radiation leading to the deepwater pachy-
gaster that gave rise to Lagocephalus, at a time before
the few specializations of pachygaster (loss of teeth on
first pharyngobranchial, loss of spines, complete loss of
remnants of the dorsal roof of the myodome) had become
established, this line leading to a form with two lateral
lines, moderately folded olfactory epithelium, and the
skeletal structure of scleratus and external character-
istics ofinermis (except with spines on the belly and dor-
sum), with the scleratus -like line diverging in external
and a few internal features associated with its more rapid
swimming and the inermis-like line developing a more
specialized skull structure, this latter line being
ancestral to most of the other species of Lagocephalus.

Perhaps Sphoeroides was more widespread in the Indo-
Pacific than at present, and has subsequently become
confined to the Atlantic and eastern Pacific as it met
superior competition in shallow Indo-Pacific waters from
the plethora of genera of tetraodontids evolving there,
and was able to remain in the Indo-Pacific only in the
form of a single deepwater species.

Fraser-Brunner (1943) associated Amblyrhynchotes
and Torquigener (including Fugu) with Sphoeroides (as
the Sphoeroidinae) and Lagocephalus (the Lago-
cephalinae, together with the Sphoeroidinae comprising
the Lagocephalidae), on the basis of their having the
sphenotics small and separated from the prefrontals by

327

the frontals, in contrast to Colomesus (family
Colomesidae), another genus with two nostrils, in which
the sphenotics are large and prolonged anteriorly to meet
the prefrontals, excluding the frontals from the orbital
margin.

I agree that the generalized condition of the nasal sac
with two nostrils probably does link the "Lagoceph-
alidae" and "Colomesidae" in a natural group as ap-
posed to the "Tetraodontidae," and that Sphoeroides is
distinctive among them in retaining the generalized con-
dition of the lateral line. However, when numerous
species are examined the osteological differences
between the groups are not as clear cut as previously
thought, and they scarcely merit familial recognition.

For example, Colomesus does have the sphenotics
prolonged anteriorly further than in any other tetra-
odontid, and the posterior recurving of its anterolateral
region is also distinctive. However, the sphenotics
become fully prolonged anteriorly to meet or very nearly
meet the prefrontals and exclude the frontals from the
margin of the orbit only in specimens of about 100 mm
and larger (see Tyler 1964 for illustrations of the skull at
various sizes). At sizes of about 50 mm the skulls of the
two species of Colomesus bear a strong resemblance to
those of Sphoeroides annulatus and formosa, both of
which, like Colomesus, have the frontal narrower poste-

Figure 265.— Dorsal views of skulls of: left,
Amblyrhynchotes honckenii, 97.4 mm SL, Mozambique;
right, A. richei, 59.4 mm SL, New Zealand.

riorly than anteriorly and the sphenotics prolonged ante-
riorly, although not as far forward in adults of S. an-
nulatus and formosa as in adults of Colomesus. In both
S. annulatus and formosa the anterior region of the
sphenotics is laterally expanded, although the expansion
is not recurved as in Colomesus. Among Sphoeroides
(including Guentheridia) only annulatus and formosa
have a dorsal flange of the parasphenoid meeting the
frontals in the interorbital septum, and one (psittacus,
along the coast of northern South America) of the two
species of Colomesus also has a dorsal flange of the para-
sphenoid while the other (asellus, fresh water of northern
South America) does not.

Both species of Colomesus have two hypohyals, as does
S. annulatus, while formosa has only one. In Colomesus
there are numerous trituration teeth in both jaws, some-
times consolidated into paired plates, and both S. an-
nulatus and formosa have the same arrangement,
somewhat unusual among Sphoeroides. Colomesus has
minute teeth on the first pharyngobranchial and lacks an
interhyal, as do S. annulatus and formosa, along with

most other Sphoeroides . Colomesus has 19 vertebrae, as
does S. annulatus, while formosa has a more specialized
reduced number of 17. In Colomesus medial prongs of the
prootic are present representing the remains of the dorsal
roof of the myodome, although these are less well-
developed in psittacus than asellus, while neither S. an-
nulatus nor formosa retain these remnants, a slightly
more specialized condition. In both S. annulatus and for-
mosa there are eight dorsal rays and seven anal rays,
while Colomesus has a perhaps slightly more gener-
alized number of 10 or 11 rays in these fins.

In short, there are striking similarities between the os-
teological makeup of Colomesus, S. annulatus and for-
mosa, and the conversion of an annulatus-formosa-\ike
fish into one like a Colomesus would require only the
retention of remnants of the myodome and a few more
dorsal and anal fin rays along with the further anterior
prolongation and recurving of the sphenotics and the
development of a second lateral line, short segments of
which are present in some specimens of many species of
Sphoeroides, including S. annulatus and formosa. It
seems likely to me that Colomesus arose from the same
line of Sphoeroides radiation as that which gave rise to
annulatus and formosa from a testudineus-\ike ancestry.
Perhaps this happened somewhere between the an-
nulatus and formosa levels of organization, with
Colomesus becoming far more specialized than either an-
nulatus or formosa. Fraser-Brunner (1943) also con-

Figure 266.— Dorsal views of skulls of:
left, Torquigener pleurogramma, 113 mm S
Australia; right, Amblyrhynchotes piosae,
33.8 mm SL, Australia.

sidered Colomesus to be a derivative of a Sphoeroides-
like stock, but for unstated reasons, which I would guess
were zoogeographic.

While Lagocephalus and Colomesus are relatively easi-
ly distinguished from the ancestral Sphoeroides-
( including Guentheridia here and following) -like stock,
the relationships of the Indo-Pacific Amblyrhynchotes,
Torquigener, and Fugu are less clear. Fraser-Brunner
(1943) said that Sphoeroides has a long ethmoid, with
the frontals well removed from the premaxillary pedicels,
17 vertebrae, and no lower lateral line, while Ambly-
rhynchotes, Torquigener, and Fugu have the ethmoid
shorter, the frontals reaching far forward, close to the
premaxillary pedicels, 20 to 21 vertebrae, and a lower
lateral line.

Sphoeroides does tend to have a longer ethmoid, a
longer ethmoid-vomerine region, and the frontal further
removed from the anterior end of the ethmoid-vomerine
region than in Amblyrhynchotes, Torquigener, and
Fugu, in most species of which the portion of the
ethmoid-vomerine region that is exposed dorsally is rela-
tively short so that the frontals do closely approach the
anterior end of the ethmoid-vomerine region. However,

in Torquigener pleurogramma, for example, the ethmoid
is no longer than in, for example, Sphoeroides greeleyi,
nor is the entire ethmoid-vomerine region of T. pleuro-
gramma shorter than in S. greeleyi, and the frontals are
about equally far removed from the anterior end of the
ethmoid-vomerine region. Fugu rubripes and Ambly-
rhynchotes honckenii likewise have relatively long
ethmoid-vomerine regions in contrast to the other species
of those genera examined, while these regions are still
shorter than in any species of Sphoeroides. The usually
greater length of the ethmoid-vomerine region in Sphoe-
roides helps to distinguish that genus from the other
three, and this moderate length in Sphoeroides can be
considered more generalized, for it more closely cor-
responds to that of triodontids and triacanthodids than
does the usually foreshortened ethmoid-vomerine region
oi Amblyrhynchotes, Torquigener, and Fugu.

While the modal number of vertebrae in Sphoeroides
ranges from 17 to 20, only two species have 19 or 20. In
Amblyrhynchotes the modal number in the three species
studied is 19, and using Fraser-Brunner's (1943) figure of
20, the range is 19 to 20. In Torquigener the modal
number in the three species studied ranges from 19 to 21.
In Fugu the modal number ranges from 20 to 23. Thus,
the number of vertebrae tend to be one to several units
higher in most species of Amblyrhynchotes, Torqui-
gener, and Fugu than in most species of Sphoeroides.

Figure 267.— Dorsal views of skulls of: left,
Torquigener pleuroatictus, 87.1 mm SL, Australia:
right, Fugu ehrysope, 98.6 mm SL, Japan.

Fraser-Brunner (1943) said that Amblyrhynchotes had
the frontal evenly narrowed anteriorly, 20 vertebrae,
prickles on the side of the body at least behind the pec-
toral fin and no ridge along the lower lateral line, while
Torquigener (including the species assigned to Fugu) had
the frontals much wider behind the prefrontals than
between them, 21 vertebrae, no prickles on the side of the
body and a ridge along the lower lateral line. As previous-
ly discussed, the ridge along the lower lateral line is pre-
sent in some species of Amblyrhynchotes as well as in
Torquigener and Fugu, although perhaps more common-
ly in Torquigener and Fugu. The modal vertebral
numbers do not significantly differ, except that Fugu has
proportionally more species with more than 20 vertebrae
than does Torquigener. At least the species of Ambly-
rhynchotes studied here have prickles on the side of the
body which are absent in all of the species of Torqui-
gener and Fugu studied.

In the three species of Amblyrhynchotes studied the
frontals are of moderate width and gradually taper ante-
riorly in honckenii and piosae, very much as in the spe-
cies oi Sphoeroides with moderately wide frontals, except

that the ethmoid-vomerine region is at least somewhat
shorter, while in richei the frontals over the rear of the or-
bit are very wide, at least as wide as more posteriorly,
and evenly taper anteriorly to bluntly rounded ends, in
front of which is an extremely short ethmoid-vomerine
region. The two species of Torquigener studied are about
as different in cranial configuration as A. richei is in com-
parison to A. honckenii and piosae. In T. pleurogramma
the frontal over the orbit evenly tapers to the posterior
edge of the prefrontal, and then more abruptly tapers
medial to the prefrontal, with a general configuration not
much different than that of, for example, Sphoeroides
testudineus, except for having a wider ethmoid. In T.
pleurostictus the frontal is laterally expanded over the
orbit, much wider there than more posteriorly, and only
begins to taper abruptly to a point at the posterior edge
of the prefrontal. The sphenotic in pleurostictus is more
anterodorsally expanded than in pleurogramma. The
major features of the skull in 10 species of Fugu has been
described and illustrated by Kuronuma (1943), who com-
pared their configuration to that of three species oiLago-
cephalus and one of Sphoeroides (pachygaster) . The
structure of three of the species of Fugu studied here
(rubripes, oblongus, and chrysops) more or less encom-
passes the diversity found by Kuronuma. In rubripes and
oblongus the sphenotic is somewhat anterolaterally ex-
panded; the frontal is laterally expanded over the orbit,

Figure 268.— Dorsal views of skulls of:
left, Fugu oblongus, 46.2 mm SL, India;
right. F. rubripes, 151 mm SL, Japan.

usually gently tapers to the posterior edge of the prefron-
tal, and then abruptly tapers medial to the prefrontal.
The skull of chrysops differs from the others by being ex-
ceptionally wide and having relatively straight lateral
margins, the frontal being especially wide.

In all of the species oi Amblyrhynchotes, Torquigener,
and Fugu studied the parasphenoid has a dorsal flange in
the orbit meeting the frontals, there are small to minute
teeth on the first pharyngobranchial, a supraneural is
present but the interhyal is absent and there are essen-
tially no prootic medial prongs representing the remains
of the dorsal roof of the myodome. In all of the species of
these three genera studied there are one to three small
rounded trituration teeth in a single row to either side of
the midline of the upper jaw but none in the lower jaw,
except for A. richei, which had no trituration teeth in
either jaw. Amblyrhynchotes piosae has both dorsal and
ventral hypohyals, while the other two species (honckenii
and richei) of that genus have only the ventral hypohyal.
Torquigener pleurogramma has both hypohyals but
pleurostictus has only one. In Fugu all three species
studied {chrysops, oblongus, rubripes) had a single
hypohyal. Amblyrhynchotes richei is unique among the

331

^L\'<^i.j^.:'^i

tetraodontids studied in lacking even a trace of a mesop-
terygoid.

On the basis of the material examined, I am unable to
distinguish Fugu from Torquigener, and I suspect that
when more species of Torquigener are examined the dis-
tinction would be even more difficult to make. Ambly-
rhynchotes is obviously closely related to Torquigener
and Fugu, differing from them by the tendency to have
the frontal evenly narrowed anteriorly and to have
prickles on the side of the body. Much more remains to
be done on the external and internal morphology of these
Indo-Pacific genera with two nostrils and, most usually,
two lateral lines.

It is pure surmise, but I would guess that an early
Sp/ioeroides-like stock with a generalized moderately
wide skull and evenly tapered frontals in the Indo-Pacific
gave rise to Amblyrhynchotes, Torquigener, and Fugu,
with the major changes from the ancestral type being the
usual development of a lower lateral line, the usual shor-
tening of the ethmoid-vomerine region, the constant
development of the dorsal flange of the parasphenoid,
and the frequent anterodorsal prolongation of the sphen-
otics. If this hypothesis is correct, the ancestral Sphoe-
roides-like group has subsequently become extinct in the
shallow waters of the Indo-Pacific, and is represented
there only by a single deepwater species.

All of the genera discussed above with two nostrils (the
"Lagocephalidae" and "Colomesidae") were said by

Figure 269.-Dorsal views of skulls of: left,
Arothron atellatus, ca. 420 mm SL, Seychelles;
right, A. armilla, 61.3 mm SL, Australia.

Fraser-Brunner (1943) to have the prefrontals separated
on the dorsal surface of the skull mainly by the frontals,
while in Ephippion, Arothron, Monotreta, Chelonodon,
Carinotetraodon, and Tetraodon (the "Tetraodonti-
dae"), all with a single nostril, the prefrontals were said
to be separated mainly by the ethmoid (and the sphen-
otics separated from the prefrontals by the frontals, as in
the "Lagocephalidae").

Whether one considers the ethmoid or the frontal to be
the main element separating the prefrontals is often a
highly subjective decision. In only a minority of species
of both groups can it be said clearly that the prefrontals
on the dorsal surface of the skull are nearly exclusively
separated by either the frontals or the ethmoid. In the
great majority of species of both groups the ethmoid ex-
tends posteriorly approximately to the level of the pos-
terior end of at least the main body of the prefrontals,
although much of this posterior region of the ethmoid is
often overlaid by the frontals and thus cannot be seen in
dorsal view. In Sphoeroides, Lagocephalus, and Colome-
sus, for example, the prefrontals are separated about half
by the ethmoid and half by the frontals, and there is
much variation between species. In Amblyrhynchotes

332

the prefrontals tend to be separated mainly by the fron-
tals, while in Fugu it is variously mostly by the frontals
or about half by the ethmoid and half by the frontals. In
Torquigener the two species examined have the prefron-
tals separated about half by the ethmoid and half by the
frontals. Again, however, this is more a simple matter of
how broadly the frontals cover the dorsal surface of the
ethmoid rather than a characteristic of fundamental im-
portance.

In Ephippion, Arothron, Monotreta, Chelonodon, Ca-
rinotetraodon, and Tetraodon the ethmoid tends to form
a greater proportion of the prefrontal separation than
does the frontal, and a greater proportion than in most
(but not all) species of the genera with two nostrils. In
Arothron, for example, most species have the prefron-
tals separated almost entirely by the ethmoid, but in one
species {armilla) the prefrontals are separated almost ex-
actly half and half by the ethmoid and frontals. In Tetra-
odon, Carinotetraodon, and Chelonodon the separation
is mostly if not entirely by the ethmoid, while in Mono-
treta the separation in most species is about half and half
by the ethmoid and frontals, but almost exclusively by
the frontals in one species (gularis).

Fraser-Brunner said that Arothron (as the Arothroni-
nae) had the sphenotics not laterally expanded beyond
the rest of the orbital roof formed by the frontals, the
orbital roof strongly arched, the prefrontal curved down

Figure 270.— Dorsal views of skulls of: left,
Tetraodon lineatiu, 222 mm SL, French Equi-
torial Africa; right, T. mbu, 47.7 mm SL, Congo.

before the eye and enclosing the olfactory foramen, the
ethmoid relatively narrow and compressed, and a single
lateral line, while in Ephippion, Monotreta, Chelono-
don, Carinotetraodon, and Tetraodon (the Tetraodon-
tinae) the sphenotics are laterally expanded beyond the
frontal as a prominent lobe, the orbital roof is not much
arched, the prefrontal not curved down before the eye
and not enclosing the olfactory foramen, the ethmoid
broad and a lower lateral line present which joins the up-
per.

It is true that the approximately six species of Aroth-
ron form a distinctive subgroup of Indo-Pacific puffers
that is recognizable by external features alone: by a cer-
tain difficult to describe sameness in configuration
(short, heavy bodies), as well as by the normally bifid
tentacle with pitted inner surfaces and the single lateral
line. In the relatively normal members of Arothron
(represented in species studied here by hispidus, nigro-
punctatus, stellatus) the sphenotics are only slightly if at
all expanded beyond the edge of the frontal, primarily
because the frontal is exceptionally wide at the posterior
region of the orbit (about 3 times as wide as more poste-
riorly in front of the supraoccipital), and the orbital roof

Figure 271 .—Dorsal view of skull o( Ephippion
guttifer, 101 mm SL, Guinea.

is at least moderately arched, with the anterolateral end
of the prefrontals curved down in front of the eye, more so
in adults than in the young, and enclosing the olfactory
foramen, and the ethmoid of moderate to narrow width
(less wide than in Tetraodon and Ephippion but no less
wide than in Chelonodon, Carinotetraodon, and most
species of Monotreta).

Not all of these typical features are confined to Aroth-
ron, for the prefrontal entirely encloses the olfactory fo-
ramen in the related Ephippion and in Chelonodon
patoca (but not C. fluviatilis), while in the related Tetra-
odon lineatus (but not T. mbu) the prefrontal encloses all
but the medial edge of a large olfactory foramen. The fo-
ramen completely encloses the prefrontal in several spe-
cies of the more distantly related genera Amblyrhyn-
chotes, Torquigener, and Fugu, although in none of these
do the prefrontals have such large downcurved antero-
lateral regions as found in most adult Arothron. The
sphenotics are not laterally expanded beyond the fron-
tals in Monotreta gularis, and are only slightly to mod-
erately expanded beyond them in the other species of
Monotreta.

Moreover, one species, originally described as Tetrao-
don armilla McCuUoch, that has the general external
configuration and look of an Arothron, including a single
lateral line, differs greatly from the typical Arothron
skull plan. Externally, armilla differs from Arothron in
having the nasal apparatus represented by a single flap
of skin on each side of the head, the outer surface smooth
and the inner only very slightly irregular, while in Aroth-
ron there is always a bifid tentacle whose inner surfaces
are pitted by circular olfactory organs.

In the three typical species of Arothron studied {hispi-
dus, nigropunctatus, stellatus), the parasphenoid either
does not have a dorsal flange in the orbit, or does not
have it well enough developed to reach dorsally far
enough to contact the frontals, while in armilla the dor-
sal flange is well developed and contacts the prefrontals
as well as the frontals. While the typical species of Aroth-
ron have minute teeth on the first pharyngobranchial, in
armilla they are relatively well developed, being only
slightly smaller than those of the second and third
pharyngobranchials. In typical Arothron there is little or
no evidence of prootic prongs in the rear of the orbit
representing the remains of the dorsal roof of the myo-
dome, while the prootic prongs are well developed in ar-
milla, reaching almost to the midline. In typical Aroth-
ron several trituration teeth are present in a single series
to either side of the midline of the upper jaw, but none
are present in the lower jaw, while in armilla, even at the
small size of the examined individual, the numerous tri-
turation teeth in both the upper and lower jaws are con-
solidated into large composite plates to either side of the
midline. Like the typical species o( Arothron, armilla has
only a ventral hypohyal and no interhyal.

The most dramatic and perhaps phylogenetically in-
teresting way, however, in which armilla differs from
typical Arothron is in the shape of the frontals and pre-
frontals. Instead of having the frontals greatly laterally
expanded above the orbit, as in typical Arothron, the
frontals of armilla are only slightly wider over the rear of
the orbit than more posteriorly, and gradually and evenly
taper to bluntly rounded ends anteriorly, much as in
most of the more generalized species of genera with two
nostrils. The prefrontals of armilla are not enlarged and
laterally expanded to the same degree as the frontals at
the rear of the orbit, and, although they do completely
enclose the olfactory foramen, they are only slightly
downcurved anterolaterally. The prefrontals in armilla
are separated from one another about half and half by
the ethmoid and frontals, a probably more generalized
condition than having them separated mostly by the eth-
moid as in typical Arothron.

Although in two features (a single nasal flap and a dor-
sal flange of the parasphenoid meeting the frontals) ar-
milla is more specialized than the typical species of
Arothron, it has a far more generalized condition of skull
configuration, especially of the frontals and prefrontals.
In fact, if the several species of Arothron not examined
for this work have skull structures similar to that of the
three typical species examined here and not intermedi-
ate between them and armilla, then armilla may be con-

sidered sufficiently distinct from Arothron to merit ge-
neric recognition, the name Omegophora Whitley being
available for it (by monotypy and original designation,
even though not sufficiently diagnosed).

More importantly, it seems likely to me that a form
like armilla which still retained a bifid nasal tentacle and
the dorsal flange of the parasphenoid not yet fully devel-
oped, could represent the ancestral group from which the
typical Arothron arose, for armilla has the most general-
ized condition of the frontals (moderate width, evenly
and gradually tapering anteriorly), sphenotics (not es-
pecially laterally expanded and not expanded antero-
dorsally) and prefrontals (moderate size, not greatly
downcurved before eye) found among the ArothronAike
species.

It is possible that the pitted cup inner olfactory epi-
thelium, as found in typical Arothron, indicates a rela-
tionship with Chelonodon, one of whose species {fluvia-
tilis) is the only other one among these related genera
(Fraser-Brunner's "Tetraodontidae") to have a pitted
cup epithelium. However, the skull structure in typical
Arothron and Chelonodon is not especially similar and
the partial similarly in olfactory epithelium may be for-
tuitous, as is, I would guess, the slight similarity in skull
structure between C. patoca and A. armilla.

Figure 272.— Dorsal views of skulls of: left,
Chelonodon patoca, 70.6 mm SL, New Guinea;
right, C. fluviatilia, 84.7 mm SL, Thailand.

The monotypic Ephippion of the west coast of Africa is
tolerant to fresh and brackish water when young, but
adults seem to be found primarily in marine coastal hab-
itats. Since Ephippion has a decidedly bifid nasal tenta-
cle, as do many species of Tetraodon, several of which
live in African fresh waters, and both genera have an up-
per lateral line which, usually, is joined by a lower lat-
eral line, it has usually been thought that Ephippion is a
close relative of Tetraodon with specialized scales. As
previously described, the scales of Ephippion even in
juvenile stages are larger than those of any other tetrao-
dontids except Xenopterus and Chonerhinos, while with
increasing specimen size the scales oi Ephippion become
enlarged into a partial carapace over the body (except
ventrally and to some extent dorsally) between the head
and the dorsal and anal fins, the scales of adults being far
larger there than in any other plectognaths except os-
tracioids.

The skull of Ephippion differs from that in the two
species of Tetraodon studied mostly by having the ante-
rolateral wings of the sphenotics heavier and thicker, and

less dorsoventrally compressed into a plate. The frontal
in Ephippion is about as laterally expanded over the
middle of the orbit as in Tetraodon, but it tapers some-
what more evenly and gradually to a point anteriorly
than in Tetraodon. In Ephippion the prefrontals are
separated about half and half by the ethmoid and fron-
tals, while in Tetraodon they are mostly separated by the
ethmoid. The ethmoid of Ephippion is slightly less wide
than in Tetraodon. The vertebrae in Ephippion are 8 +
12 = 20, while in the three species of Tetraodon for which
vertebral counts are presented the range is 8 + 9 = 17
to 8 + 11 = 19. In £p/iippion there are several tritura-
tion teeth in a single row to either side of the midline in
the upper jaw, but none in the lower; while in the two
species of Tetraodon cleared and stained, there are no
trituration teeth in either jaw. In both Ephippion and
Tetraodon there are minute teeth on the first pharyngo-
branchial and the dorsal flange of the parasphenoid in
the orbit meets the frontals, but there are no remnants of
the dorsal roof of the myodome, no interhyal, and only a
single hypohyal.

Until recently, the categories Monotreta, Chelonodon,
and Tetraodon were badly in need of revision. They are
defined, as subgenera of Tetraodon, by Fraser-Brunner
(1943) entirely on nasal and scale pattern characteristics

Figure 273.— Dorsal views of skulls of: 1
Monotreta gularis, 46.8 mm SL, Burma;
right. M. leiurus. 61.5 mm SL, Thailand.

that are less than adequate. These categories have nearly
always been used at the generic level, for failure to do so
leaves Tetraodon an unmanageable assemblage of
numerous species, some obviously far more closely
related than others. The use of these three categories also
presents problems, which will probably only be solved
when each of the species not discussed here has had its
external and internal features compared with those de-
scribed here. Dekkers (1975) has provided an excellent
revision of these Asiatic and mainly freshwater puffers
based on external characters, placing them all in one
genus, with five unnamed subcategories for 15 species,
two of which (erythrotaenia and waandersii) contain sin
gle species not studied here. For purposes of compara
five discussion in this monograph, Tetraodon, Chelono-
don, and Monotreta are recognized at the generic level
In Monotreta (nasal apparatus a tube with a single
nostril, the aperture of the tube with or without lips or
flaps) three of the four species studied are rather similar
to one another, and possess one feature unique to the

tetraodontids, while the other species does not share
these similarities. All four species studied are rather sim-
ilar externally, with the nasal apparatus about the same
in all and varying only in the length of the tube, the up-
per lateral line joining the lower, at least a few prickles
on the side of the body, and a color pattern featuring at
least one large black spot on the side of the body.

In M. gularis (Dekkers 1975:95, considered gularis a
synonym of cutcutia, but the single specimen I have
identified and studied as gularis is not conspecific with
cutcutia), the sphenotic does not extend laterally beyond
the edge of the frontal, the frontal is much expanded lat-
erally over the middle of the orbit, the posterolateral
region of the jjrefrontal is not much less wide than the
average width of the interorbital region, the lateral edge
of the prefrontal is gently curved, the ethmoid is very
short and relatively wide, the palatine-vomerine strut
supporting the upper jaw is relatively short, and the pre-
maxillary pedicel is of moderate length (as in most other
tetraodontids).

However, in the other three (leiurus, cutcutia, palem-
bangensis) species of Monotreta studied the sphenotics
extend laterally beyond the edge of the frontal (at least
as much as illustrated for leiurus; in palembangensis the
sphenotic is even wider and further extended laterally),
the frontal is only moderately laterally expanded over

Figure 274.— Dorsal views of skulls of: left,
Chonerhinos modestus, .54.5 mm SL, Borneo;
right. Xenoplerus naritus, 143 mm SL, Bay of
Bengal, with the barred band representing the
cut surface of the thickened frontal, removed
to expose the portions of the sphenotic and
pterotic which it overlies (dashed line marks
lateral edge of removed portion of frontal).

the middle of the orbit and is strongly tapered anteriorly
to the posterolateral region of the prefrontal, which is
much less wide than the average width of the interor-
bital region, the lateral edge of the prefrontal is rela-
tively straight and produced anterolaterally to a point,
the ethmoid is very long and of moderate width, the pal-
atine-vomerine strut supporting the upper jaw is long
relative to gularis (but not relative to a large number of
other tetraodontids), and the premaxillary pedicel is of
relatively great length.

The open space enclosed between the dorsomedial
edges of the two premaxillaries in M. leiurus, cutcutia,
and palembangensis is much greater than in gularis or
any other tetraodontid, partially because of the rela-
tively great length of the premaxillary pedicels, but
equally important because the open space extends far
further forward beyond the level of the anterior ends of

337

y^m^.:

Figure 275.— Dorsal view of skull of Carinotetra-
odon lorteti, 33.1 mm SL, locality unknown.

the palatines than in any other tetraodontids, i.e., the
medial edges of the premaxillaries in leiurus, cutcutia,
and palembangensis have a much longer concave region
than in any other tetraodontids.

The four species of Monotreta studied are alike in that
there are small to minute teeth on the first pharyngo-
branchial, no trituration teeth in either jaw, no evidence
of prootic prongs representing remnants of the dorsal roof
of the myodome, no interhyal, a single hypohyal (one
specimen of leiurus with two on one side and one on the
other), and no supraneural element. The only other
tetraodontids in which the supraneural is absent are
members of the closely related genera Tetraodon (in
which it is absent in mbu but present in Uneatus),
Chelonodon (absent in fluuiatilis but present in patoca),
and Carinotetraodon (absent in the single species
studied). In one of the species oi Monotreta (leiurus) the
first five abdominal vertebrae have bifid neural spines,
and the neural spine of the sixth abdominal vertebra is
bifid anteriorly but single posteriorly, while in the other
three species the neural spines of only the first four verte-
brae are bifid. In most other tetraodontids the first three
abdominal vertebrae have bifid neural spines, and the
neural spine of the fourth vertebra is bifid anteriorly but
single posteriorly. In Arothron, however, the neural

spines of the first four abdominal vertebrae are bifid, and
at least a short portion of the anterior end of the neural
spine of the fifth vertebra is bifid. Thus, Arothron and
Monotreta tend to have a slightly greater development of
bifid neural spines than do the other genera of tetraodon-
tids, a specialization.

In Monotreta the vertebrae are 18 in cutcutia and
gularis, but 20 or 21 in leiurus and palembangensis.

Even though the ensemble of subsidiary characters, in-
ternal and external, does not serve to separate gularis
from the other three species of Monotreta studied, the
differential shapes and sizes of the frontal, prefrontal,
ethmoid, vomer, palatine, and, especially, premaxillary
clearly indicate that there are two lines of diversification
within the species of Monotreta studied. If additional
species of Monotreta not studied here are eventually
shown not to bridge the gap between the two groups, they
should be recognized as at least subgenerically distinct
(the identification of the specimen here studied as
gularis needs verification). One suspects that the ex-
amination of additional species will bridge the gap. For
example, Carinotetraodon lorteti, otherwise assignable
to Monotreta because of the nasal tube with a single nos-
tril (but with the upper and lower lateral lines not joining
and both reaching the tail), has the length of the eth-
moid about intermediate between gularis and the other
three species of Monotreta, the palatine-vomerine shaft
about as long as in the three specialized species, the
length of the premaxillary pedicel intermediate between
gularis and the other three species, and the size of the
space between the concave posteromedial regions of the
two premaxillaries very similar to that of gularis. The
anterolateral wing of the sphenotic in Carinotetraodon is
about like that of one of the three specialized species
(palembangensis) of Monotreta, but there is no struc-
ture in any of the four species of Monotreta similar to the
posterolateral wing of Carinotetraodon.

It seems obvious that Carinotetraodon and thus
Canthigaster, as discussed earlier in this section, have
their closest relationships among the Afom)tre(a-like
tetraodontids, but that a clearer understanding of the
relationships of these fishes awaits the time when the
critical internal and external features can be compared
for a larger number of species than it has been possible to
do here. Such a further study must also obviously in-
clude Chelonodon and Tetraodon, the generic limits of
neither of which are presently clear.

For example, in Tetraodon the lateral line system is
highly variable, but it is probably basically represented
by a lower line which joins the upper and continues to the
tail, this being modified in some species by the loss of the
connection between the upper and lower lines, or even
the loss of the lower line, and perhaps of much of the up-
per line as well. The bifid nasal tentacle tends to be deep-
ly split, almost to the base, and the inner epithelium
smooth. The skulls of the two species studied are
relatively similar, differing mainly that in Uneatus the
sphenotics are more anteriorly expanded than in mbu
(probably partially attributable to the much greater size
of the study individual of Uneatus) and that the postero-

Figure 276.— Ventral views of

skulls of: left, Sphoeroides

/atus, 179 mm SL, Virginia

right, Arothron stellatus.

ca. 420 mm SL, Seychelles.

branchiostegal rays

Figure 277. — Dorsal views of terus naritus, lOS mm SL, Bay

branchial arches (extended on of Bengal: (right)

lower side) and lateral views Arothron nigropunctatua,

of hyoid arch of: (left) Xenop- .i6.8 mm SL, Solomon Islands.

ventral flange of the pterotic in lineatus has a goodly con-
tribution from the exoccipital but in mbu next to none.
Tetraodon mbu is unique among the tetraodontids
studied in completely lacking a pterosphenoid. The eth-
moid-vomer-palatine region is essentially similar in both
lineatus and mbu, and the prefrontals differ mainly in
that of lineatus having an anteromedial arm helping to
enclose the olfactory foramen. The skull shape of Ephip-
pion is not far removed from that of Tetraodon and both
probably have a close common ancestral group, perhaps
as close as that which unites Tetraodon with Monotreta
and Chelonodon.

The only two species usually assigned to Chelonodon
differ slightly more than do the two of Tetraodon ex-
amined. In C. fluviatilis, which has a deeply pitted ol-
factory epithelium, the frontals abruptly end anteriorly
in a more or less straight transverse line at the rear edge
of the prefrontals, the latter being relatively thin and en-
tirely separated by the ethmoid. In C. patoca the frontals
become narrower anteriorly over the orbit, but they do
not stop abruptly at the level of the posterior edge of the
prefrontals, continuing anteriorly as rapidly tapering
points between the prefrontals, the latter being relatively
thick and separated by both the ethmoid and frontals.
The sphenotics in patoca are less anterodorsally ex-
tended than in fluviatilis. The skull condition in patoca
is slightly more generalized than that of fluviatilis, and
patoca retains a supraneural which is lost by fluviatilis.
Both species lack any remnants of the dorsal roof of the
myodome, and there is no interhyal and only one hypo-
hyal. The parasphenoid has a dorsal flange in the orbit
meeting the frontals and there are minute teeth on the
first pharyngobranchial, while fluviatilis has numerous
trituration teeth to either side of the midline in both the
upper and lower jaws, and patoca has only two or three
teeth in each series in the upper jaw and none in the
lower jaw. The vertebrae are modally 18 in fluviatilis and
19 in patoca.

On the basis of the differences in the olfactory epithe-
lium, and of fluviatilis having the lobes of the olfactory
apparatus oriented parallel to the body versus at a right
angle to it in patoca, Le Danois (1959) generically
separated fluviatilis (as Dichotomy cterus) from patoca
(Chelonodon). The validity of such minor nasal dif-
ferences alone in distinguishing genera is highly ques-
tionable, but should other species assignable to
Chelonodon other than the commonly collected
fluviatilis and patoca come to light (a few valid species
probably lurk among the numerous synonyms usually
listed for each of these names) and fall into one or the
other of the fluviatilis or patoca skull plans and not into
intermediate types, there may be justification for es-
tablishing two genera for what is now Chelonodon.

The precise relationship of Chelonodon to Tetraodon
and Monotreta is not clear on the basis of the present
data.

As previously discussed, Chonerhinos and Xenopterus
are highly specialized tetraodontids which have second-
arily increased the number of vertebrae and of dorsal and
anal fin rays, elaborated the lateral line system until

there are approximately three lines on the body, in-
creased the size and amount of folding of the olfactory
epithelium in the open cup nasal apparatus, and in-
creased the size of at least some of the spines, mostly
those of the belly. The greater increase in numbers of
vertebrae and dorsal and anal fin rays in Xenopterus in-
dicates that it is the more specialized of the two, and this
is also borne out in the structure of the skull. While both
species are unique among the tetraodontids in the loss of
the prefrontals (Fraser-Brunner 1943 said that the pre-
frontals are very small, but I find no trace of them at all),
the skull of Chonerhinos otherwise is not markedly dif-
ferent from that of many of the species of the Mono-
treta-Chelonodon-Tetraodon group. The frontals of
Chonerhinos are slightly wider over the middle of the or-
bit than more posteriorly and are tapered gradually and
evenly to gently rounded points anteriorly that broadly
overlie the relatively broad ethmoid, which is of
moderate length, while the sphenotics are well extended
anteroventrally, projecting out beyond the edges of the
frontals and forming about the rear half of the upper edge
of the orbit.

In Xenopterus the frontals are much more laterally ex-
panded and thickened than in Chonerhinos, forming a
large plate over most of the dorsal surface of the skull. In
the two smaller specimens studied the frontals are nor-
mally articulated by interdigitation in the midline, but
in the largest specimen the two frontals are indis-
tinguishably fused to one another in about the middle
third of their lengths. While the sphenotic oi Xenopterus
is about as anterolaterally extended as in Chonerhinos, it
is nearly entirely overlain by the frontal and only its ex-
treme distal end appears in dorsal view, projecting
slightly beyond the edge of the frontal in about the mid-
dle of the upper edge of the orbit. The main body of the
supraoccipital in Xenopterus is less wide than in Chone-
rhinos, but the supraoccipital crest is wider and heavier
in the former than in the latter. In Xenopterus the neural
and haemal spines of the penultimate vertebra become
hyperostotic, but this does not occur in Chonerhinos.

The ethmoid in Xenopterus is shorter but broader than
in Chonerhinos and in both genera the anterodorsal end
of the vomer tends to fully fuse with the anteroventral
end of the ethmoid. Both genera are also similar in hav-
ing the pterosphenoids in contact with the parasphenoid
in the rear of the orbit, the pterosphenoid, para-
sphenoid, and prootic forming a more elaborate and mas-
sive structure there than in any other tetraodontids. This
is perhaps functionally similar to the bracing strut of a
dorsal flange from the parasphenoid in the middle of the
orbit to the under surface of the frontals as found in
many other tetraodontids, but not in Chonerhinos and
Xenopterus. The only other tetraodontids in which the
pterosphenoid meets the parasphenoid are several
species of Lagocephalus, but this is associated with the
dorsal flange of the parasphenoid in the middle of the or-
bit and thus is not analogous to that in Chonerhinos and
Xenopterus. In Xenopterus the supraneural is larger and
deeper bodied than in Chonerhinos. Both genera lack any
remnants of the dorsal roof of the myodome, there is no

interhyal and only one hypohyal, the first pharyngo-
branchial has small to minute teeth and there are no
trituration teeth in either jaw. A sesamoid articular was
not found in the lower jaw of any of the specimens of
Chonerhinos and Xenopterus examined, and these
genera may be at least unusual, if not unique, among the
tetraodontids in the loss of this element.

In short, the skull in the more generalized of these two
specialized genera bears its greatest similarity to that of
some of the species of Monotreta, Chelonodon, and
Tetraodon, perhaps especially to the latter, but the pre-
cise relationship of Chonerhinos to any of them is un-
clear. One would expect Chonerhinos to have arisen from
a line having a nasal sac with a single nostril, a well-
developed lateral line system with at least one and a half
or two lines on the body, a tendency to increase the num-
ber of dorsal and anal fin rays above the perhaps
generalized number of about 10 to 12 and a tendency to
increase the number of vertebrae above the generalized
number of 20, all of which features are found in varying
degrees among one or the other of Monotreta,
Chelonodon, and Tetraodon, and the ancestry of
Chonerhinos is probably shared at one point with that of
one or more of those three closely related genera.

Chonerhinos and, progressively more so, Xenopterus are
probably derived from the ancestry of the Monotreta-
Chelonodon-Tetraodon group, but their more precise
relationships remain unknown.

Canthigaster is deemed to be sufficiently anatomi-
cally distinct from the other tetraodontids to be recog-
nized as subfamilially distinct (Canthigasterinae) from
them (Tetraodontinae), although the author is biased
toward the conservative approach of only subfamilial
recognition for Canthigaster, since recently it usually is
given full familial rank. Almost as much anatomical dis-
tinctiveness is present between Chonerhinos-Xenop-
terus and the other tetraodontins as there is between the
canthigasterins and tetraodontins. Moreover, there is no
genus known as closely intermediate between
Chonerhinos-Xenopterus and the other tetraodontins as
Carinotetraodon is between the canthigasterins and the
other tetraodontins. Thus, there is no compelling reason
to recognize Canthigaster as even subfamilially distinct,
and it is done here more as a matter of personal
preference than on the basis of persuasive anatomical
evidence.

The argument against subfamilial recognition of
Canthigaster could also take into account the situation

Summary of generic relationships and intra-
familial classiflcation. — Because of the complexity of
the external and osteological diversity of the tetraodon-
tids, most of the analyses of the generic relationships
within the family are given in the preceding section on
anatomical diversity, which need be only summarized
here.

The genera having a nasal sac with two nostrils seem to
form a natural group whose more generalized represen-
tatives are more generalized than most of those genera in
which a single nostril is present at the end of a tube or in
which there is a tentacle or an open cup nasal ap-
paratus. Among the genera with two nostrils,
Sphoeroides is the only one with a single lateral line, and
its osteology seems to be overall the most generalized,
with Lagocephalus and Colomesus derived from a
Sp/ioeroides-like ancestral group, but from rather dif-
ferent lines of radiation within that group. Ambly-
rhynchotes, Fugu, and Torquigener may also have been
derived from an early Sphoeroides-like group, and are
probably more closely related to one another than to any
of the other genera with two nostrils, but their more
precise relationships are not clear on the basis of the pre-
sent work.

The genera in which the nasal sac has opened up by
the loss of the separation between the two nostrils are
probably derived from an ancestral group with two nos-
trils and a single lateral line. A Carinotetraodon-Vike
form is probably ancestral to Canthigaster, and
Carinotetraodon itself is probably most closely related to
Monotreta. Ephippion is probably a close derivative of a
Tetraodon-like form, and the relationship between
Monotreta, Chelonodon, and Tetraodon undoubtedly is
close but the details of those relationships are not clear
on the basis of the present work. The highly specialized

Canthigaster

Carinotetraodon

Ephippion

Figure 278.— Hypothesized
phylogenetic relationships of
the genera of Tetraodontidae.

341

in the Monacanthidae and in the triacanthodid sub-
family Triacanthodinae. In the monacanthids the highly
specialized genus Psilocephalus could be considered as
subfamilially distinct (as it sometimes has been in the
past) from the monacanthids, and is not so recognized
here mainly because some of the intermediates between
its structure and that of more generalized monacanthids
are found in some of the Alutera-Vike fishes. Similarly,
the highly specialized monacanthid Pseudaluteres could
be considered subfamilially distinct, and is not so recog-
nized here mainly because it is clear that Oxymona-
canthus is intermediate between it and the more
generalized monacanthids, and an Oxymonacanthus-\ike
form gave rise to it. In the triacanthodids, the two long-
snouted highly specialized genera of the Recent sub-
family Triacanthodinae could each be recognized as a
distinct subgroup (tribe) and are not so recognized here
mainly because a different genus is intermediate
between each of the two long-snouted genera and the
more generalized triacanthodins, i.e., a Bathyphylax-like
line is ancestral to Halimochirurgus and one like Tyde-
mania to Macrorhamphosodes.

Relationships to the Diodontidae and to the Other
Tetraodontoidei. — The ways by which a triodontidlike
ancestral group probably gave rise to two major lines of
radiation, one leading to the tetraodontids and dio-
dontids and the other to the molids, are discussed under
the Triodontidae. The molids undoubtedly branched off
the common ancestral line at a more generalized level of
triodontidlike organization than did the tetraodontoids,
for molids retain a significant number of important
generalized features found in triodontids but not in tetra-
odontoids, even though molids superficially seem far
removed from triodontids and almost totally specialized.
In fact, the tetraodontoids are at least as highly specializ-
ed and as anatomically far removed from the ancestral
triodontidlike fishes as are the molids, as further discuss-
ed under the Triodontidae.

In the vast majority of ways that tetraodontids differ
from diodontids, tetraodontids retain more generalized,
triodontidlike, conditions than do diodontids. But there
are important exceptions in which diodontids have the
more generalized conditions, indicating that the division
of the ancestral line into tetraodontids and diodontids
took place early in the evolution of the superfamily
(Eocene, on the basis of the fossil record), with diodon-
tids retaining a few generalized features while overall
becoming far more specialized than the tetraodontids.

As previously discussed, it seems reasonable to assume
that the ancestral eoplectins that gave rise to the trio-
dontids and other gymnodont lines did so at a time when
the premaxillaries and dentaries were not yet fused to
their opposite members, but only interdigitated to them,
and with the teeth in the biting edge of the jaws small
rounded units incorporated into the matrix of the bone
but retaining much of their individual identity. It is here
assumed that the line of triodontidlike fishes which gave
rise to the tetraodontoids had the premaxillaries and
dentaries unfused and that the fusion of these bones to

their opposite members in diodontids and molids has
been independent. It is possible that the triodontidlike
fishes which gave rise to the molids already had the den-
taries fused, as in Triodon and at least Zignoichthys of
the eoplectins, and that molids became further specializ-
ed by having the premaxillaries fused as well (along with
numerous other specializations, including the loss of dis-
tinct teeth in the biting edge of the jaws).

While diodontids have the more specialized condition
of fused premaxillaries and dentaries in relation to tetra-
odontids having both of these bones articulated to their
opposite members by interlocking emarginations, the fu-
sion of these two jaw bones is a relatively uncomplex
event in comparison to the changes in the dental units in-
corporated into the matrix of the bone. Diodontids retain
the generalized condition of small rounded dental units
in the biting edge, just as in triodontids and eoplectins,
while tetraodontids have extremely specialized teeth,
perhaps the most specialized among teleosts, these being
long, slender rods lying obliquely transverse to the body
axis, parallel to the biting edge and incorporated into the
matrix. The large trituration plates present in the upper
and lower jaws of diodontids are of about the same size as
those of Triodon, even though there are usually fewer
series of teeth involved in the formation of the composite
plates (see discussion under the diodontid Chilomyc-
terus orbicularis). However, the form of the trituration
teeth and plates has probably been highly variable in
most groups of gymnodonts, correlated with changes of
diet in newly occupied habitats, and not too much im-
portance can be attached to the at least superficial
similarity of the trituration plates in Triodon and
diodontids, i.e., the trituration plates of diodontids are
not necessarily to be considered generalized.

Other than the form of the teeth of the biting edge of
the jaws, diodontids are more generalized and Triodon-
like than tetraodontids only by: 1) the more frequent
presence of both a dorsal and ventral hypohyal; 2) the
presence of a prominent anteriorly directed prong on the
suboperculum attached by ligament to the interoper-
culum; 3) having none of the basal pterygiophores of the
dorsal and anal fin interdigitated to one another distally;
and 4) the presence of a prominent lateral flange on the
fused hypural plate, if this is homologous with the
hypurapophysis of Triodon, which is debatable.

In all of the other far more numerous characters listed
here in the comparative diagnoses of the tetraodontids
and diodontids, the conditions found in the tetra-
odontids are the more generalized and the least far
removed from those in the ancestral triodontids (pre-
Recent Triodon level of organization).

The diodontids must thus be considered an early
Eocene offshoot of the basal tetraodontids, with most of
the diodontid specializations centering around a more
highly defensively armoured exoskeleton variously of
long erectile quills or short nonerectile spines with
massive bases forming an open quilted carapace around
a relatively slow swimming body with a reduced caudal
region and usually relatively more massive jaws with a
greater crushing and grinding function.

Family Diodontidae

Comparative diagnosis (contrast with that of the
Tetraodontidae). — Teeth incorporated into the matrix
of the biting edge of the jaws as small and more or less
rounded units; premaxillaries and dentaries fully fused
to their opposite members in the midline; lateral surface
of the maxillary with a deeply indented or laterally flang-
ed surface, the thickened ridges increasing its strength;
the jaws massive; a large trituration plate always present
in the upper and lower jaws; first and second pharyngo-
branchials with minute teeth, third pharyngobranchial
with minute teeth or toothless, and sometimes absent;
dorsal and ventral hypohyal usually both present; inter-
hyal never present; anterior edge of ectopterygoid rela-
tively straight and only slightly concave; ethmoid and
vomer greatly reduced in size and fused together into a
relatively thin and functionless plate; palatine not
notched posteriorly, but broadly sutured thereto, and
taking its main support from, the frontal; prefrontal
small or absent; frontals exceptionally wide and massive;
anterior end of parasphenoid exceptionally wide and
deeply concave, the concavity not being a place of ar-
ticulation for other bones; rear margin of the orbit form-
ed by the frontal alone; sphenotic with a long, slender,
laterally directed sturdy prong from its anterolateral
edge; frontal broadly in contact posteroventrally in the
rear of the orbit with the prootic, and, to a lesser extent,
with the pterosphenoid, but not in contact with the
sphenotic; suboperculum with a prominent anteriorly
directed prong attached by a short ligament to the pos-
terior end of the interoperculum, the latter not articu-
lating with the operculum; supracleithrum positioned
horizontally almost in the same line as the axis of the
body; postcleithrum a single short piece, no longer than
about the distance along the scapula to the lowest ac-
tinost; supraoccipital crest dorsoventrally compressed
and entirely in a horizontal plane, wider than deep
throughout its length; exoccipital condyles poorly
developed; all of the vertebrae anterior to the first basal
pterygiophore of the dorsal fin and a few of those poste-
rior to the last basal pterygiophore of the dorsal fin with
bifid divergent neural spines; many of the more poste-
rior abdominal vertebrae and several of the more ante-
rior caudal vertebrae with prominent lateral flanges from
the ventrolateral surfaces of the centra; neural spines of
the vertebrae supporting the basal pterygiophores of the
dorsal fin short and broad, not slender shafts and not
penetrating deeply the interspaces between the pteryg-
iophores; a supraneural element never present; none of
the basal pterygiophores of the dorsal and anal fins inter-
digitated with one another; none of the abdominal
vertebrae with haemal arches, complete or incomplete;

several of the vertebrae posterior to the bases of the last
basal pterygiophores of the dorsal and anal fins antero-
posteriorly compressed, much shorter in centrum length
than those more anteriorly; abdominal vertebrae always
greater in number than the caudal vertebrae; dorsal and
anal fins more posterior in position; only two vertebrae
fully posterior to that whose haemal spine is the last sup-
port of the last anal fin basal pterygiophore; caudal fin
supporting skeleton with no free epural, no free hypurals,
and no free parhypural, and with the haemal spine of the
penultimate vertebra fused to its centrum; a prominent
lateral flange present on the fused hypural-centrum
plate; haemal canal not penetrating the last vertebral
complex; caudal fin rays modally either 9 or 10, in the
ventral region of the fin only the single lowermost ray un-
branched, perhaps 11 caudal rays in the Eocene
Prodiodon; scales always relative massive, whether long
erectile quills or shorter spines borne on a large triradiate
plate, or some combination of erectile and fixed spines.

Detailed description of Diodon holocanthus.

Material examined: — Five cleared and stained
specimens, 12.3-113 mm; one dry partially disarticu-
lated skeleton, 124 mm.

SKULL.

Occipital Region.

Basioccipital. — A short column, expanded antero-
laterally; cartilage filled at its anterior and anterolateral
edges; articulates by interdigitation posterolaterally with
the exoccipitals, anterolaterally with the prootics, and
anteriorly with the overlying parasphenoid. In all of the
specimens examined a certain amount of externally visi-
ble cartilage persists at the region of junction of the
basioccipital, exoccipitals, and prootics, as it also does at
the junction of the exoccipital, prootic, and pterotic and
at the junction of the posterior portions of the para-
sphenoid and prootics. Extremely large specimens can be
expected to have more extensive interdigitation and less
cartilage visible in these areas than shown in the illus-
trations. The rim of the round concave posterior end of
the basioccipital articulates by fibrous tissue with the
rim of the concave anterior face of the first vertebra. The
posterodorsal surface of the basioccipital forms the lower
wall of the foramen magnum.

Exoccipital. — Cartilage filled at all of its edges of
articulation with the other cranial bones; articulates by
interdigitation anterodorsally with the epiotic, laterally
with the pterotic, anteroventrally with the prootic, and

Figure 279.— Range of diversity in body form

in the Diodontidae: Diodon holocanthus (left)

and Chilomycterua achoepfi (right).

343

Figure 280.— Dtodon holocanthus (center),

with Chilomycterus schoepfi (above)

for comparison : lower left, nasal

region as seen externally and scales

from upper middle region of body of

C. schoepfi, and, lower right, nasal

region as seen externally and scales

from upper middle region of body of

D. holocanthus; dotted line on Figure

of C. schoepfi shows the course of the

lateral line canals and their major

pores, as deciphered by placing drops of

ink on each pore found by microscopic

search of a partially drying specimen.

Otic Region.

Pterotic. —Expanded posterolaterally into a
flattened, almost vertical, supporting strut for the pec-
toral girdle; broadly cartilage filled along its medial
edges; articulates by interdigitation anterodorsally with
the sphenotic, anteroventrally with the prootic, postero-
ventrally with the exoccipital, and dorsomedially with
the epiotic. At a concavity on the posterior surface of its
posterolaterally expanded portion, the pterotic articu-
lates by fibrous tissue with the anterior end of the supra-
cleithrum. Along the anterolateral edge of its ventral sur-
face the pterotic supports the hyomandibular. This
articulation between the pterotic and hyomandibular is
by fibrous tissue anteriorly, but posteriorly the two bones
become extensively interdigitated with one another

Sphenotic. —Confined to the dorsal and lateral
portions of the skull and not entering into the formation
of the wall of the orbit; cartilage filled along all. of its
edges of articulation with the other cranial bones; ar-
ticulates by interdigitation anteriorly with the frontal,
medially with the supraoccipital, posteromedially with
the epiotic, posterolaterally with the pterotic, and ven-
trolaterally with the prootic. The region of articulation
between the sphenotic and prootic forms a concave sur-
face to which the anterior half of the dorsal edge of the
hyomandibular is held by fibrous tissue. The dorso-
lateral edge of the sphenotic possesses a stout, pronglike,
lateral process which serves as the place of origin for
muscles which insert on the operculum.

Epiotic— A thin, more or less squarish plate;
cartilage filled along all of its edges of articulation with
the other cranial bones; articulates by interdigitation
anteriorly with the sphenotic, medially with the supra-
occipital, posteriorly with the exoccipital, and laterally
with the pterotic.

ventromedially with the basioccipital. Posteromedially
the dorsal edges of the exoccipitals interdigitate with one
another so that the dorsal as well as the lateral walls of
the foramen magnum are formed by the exoccipitals.
From the posterior end of its ventrolateral edge the exoc-
cipital possesses a pair of short but sturdy processes
which overlie and articulate by fibrous tissue with the
anterolateral region of the first vertebra, these processes
being the modified exoccipital condyle.

Supraoccipital. —A flat plate anteriorly, but drawn
out posteriorly into a much depressed spine which
reaches to about the level of the end of the first vertebral
centrum; articulates by interdigitation anteriorly with
the frontals, anterolaterally with the sphenotics, and
posterolaterally with the epiotics and exoccipitals. The
ventral surface of the anterior half of the supraoccipital
spine interdigitates with the dorsomedial surfaces of the
exoccipitals.

Prootic. — Cartilage filled along all of its edges of
articulation with the other cranial bones, except an-
teriorly; articulates by interdigitation anterodorsally
with the frontal, anteromedially with the ptero-
sphenoid, anteroventrally with the parasphenoid,
posteromedially with the basioccipital, posteriorly with
the exoccipital, posterolaterally with the pterotic, and
laterally with the sphenotic. Medially on its ventral sur-
face the prootic articulates with the lateral edge of the
posterior end of the parasphenoid. This articulation is
usually through cartilage, and only in large specimens do
the two bones become interdigitated to any appreciable
extent. Cartilage persists in this region long after it has
at least mostly disappeared from external view between
the region of junction of the basioccipital, exoccipital,
and prootic and the region of junction of the prootic,
exoccipital, and pterotic. There is no evidence of a
myodome. Along the concave area at the region of ar-
ticulation between the prootic and sphenotic, the
hyomandibular is supported through fibrous tissue. This
support is strengthened, however, by the extensive inter-

digitation of the anterolateral edge of the prootic with the
anterodorsal edge of the hyomandibular.

Orbital Region.

Frontal. — Extremely large and laterally ex-
panded; articulates by interdigitation posteromedially
with the supraoccipital, which it slightly overlies, and
posterolaterally with the sphenotic. Medially along its
ventral surface in the region of the orbital cavity the fron-
tal interdigitates with the pterosphenoid. At its anterior
end the frontal is extensively interdigitated medially
with the palatine and laterally with the prefrontal, the
latter bone being broadly overlain by the frontal. Along
the posterior half of its medial edge the frontal interdigi-
tates with its opposite member, but more anteriorly the
medial edges of the two frontals articulate with one
another only by fibrous tissue. Between the regions where
the medial edge of the frontal is in close contact with its
opposite member and where it interdigitates with the
palatine, the frontal overlies the ethmoid cartilage.

Prefrontal. — A thin plate of extremely variable
size, but always broadly overlain by the frontal and
hence apparently very small as seen dorsally. The pre-
frontal articulates by extensive interdigitation medially
with the palatine and posteriorly and laterally with the
frontal.

Parasphenoid. — More or less in the form of a cross,
with the longer shaft becoming increasingly deeper and
wider toward its anterior end. The anterior end of the
parasphenoid is so deeply concave that the cavity ex-
tends back posteriorly almost to the level of the shorter
transverse arms of the parasphenoid. The parasphenoid
articulates by interdigitation posterolaterally along the
dorsal surface of its transverse arms with the prootics,
while at its extreme posterior end the parasphenoid over-
lies and interdigitates with the basioccipital. Between its
regions of interdigitation with the basioccipital and with
the prootic, the parasphenoid has its lateral edges in con-
tact with the cartilage that is present along the medial
edges of the prootics. Anterolaterally the parasphenoid
interdigitates with the medial edge of the palatine. The
highly concave anterior end of the parasphenoid has its
dorsal surface held to the ventral surface of the ethmoid
cartilage. The posterior end of the platelike ossification
that occurs between the ethmoid cartilage and the dor-
sal surface of the parasphenoid, to be described below as
the ethmoid-vomer, is slightly overlain by and interdigi-
tated with the anterior concave end of the parasphenoid.
A very shallow ventral flange or keel is present along the
medial portion of the ventral surface of the para-
sphenoid in the region under the orbital cavity.

Ethmoid Region.

Ethmoid-vomer. — All that remains of the ossifica-
tions of the ethmoid region is a plate of bone which is
very thin throughout its length, except at its anterior
edge where it becomes substantially thicker. The plate of
bone lies between the ventral surface of the ethmoid car-
tilage and the anterodorsal portion of the parasphenoid.
The posterior end of the plate fits under and interdigi-
tates with the inner surface of the dorsal half of the deep-
ly concave anterior end of the parasphenoid, while the
anterior edge of the plate lies anterior to the ethmoid car-
tilage. There is no evidence in any of the specimens ex-
amined that any of the edges of the plate are cartilage
filled, but even if the plate were entirely an endochon-
dral ossification the bone is so thin that one would not
expect to find cartilage filled edges. Only by histological
examination of the developing bone would it be possible
to state whether the bone is of dermal or endochondral
origin or of a combination of the two. From the position
of the anterior portion of the plate (between the frontals
and in at least close apposition with the ethmoid car-
tilage) one would suspect that the bone is the ethmoid.
However, the posterior end of the bone fits into the con-
cavity of the parasphenoid, a characteristic of the vomer
in most plectognaths. Until histological examination of
the development of the bone sheds light on its nature, I
presume that it probably represents the fused rudiments
of both the ethmoid and vomer.

Mandibular Region.

Hyomandibular. — Expanded dorsally, tapering to a
stout shaft anteroventrally; cartilage filled at its antero-
ventral edge and along the middle of its dorsal edge; ar-
ticulates dorsally by fibrous tissue with the concavity
along the ventrolateral edges of the prootic and sphenotic
and along the anterior half of the anteroventral edge of
the pterotic. The support of the hyomandibular is fur-
ther strengthened by the extensive interdigitation of its
anterodorsal end with the anterolateral edge of the
prootic and of its posterodorsal end with the ventro-
lateral edge of the pterotic. Anteriorly the hyoman-
dibular articulates by fibrous tissue with the posterior
ends of the metapterygoid and symplectic. In large
specimens the articulation between the hyomandibular
and the metapterygoid becomes slightly interdigitated.
Along the ventral half of its posterior edge the hyoman-
dibular articulates by fibrous tissue with the preoper-
culum. Immediately behind its uppermost point of con-
tact with the preoperculum, the hyomandibular
possesses an upraised region whose posterior surface is
hollowed out to receive and hold by fibrous tissue the an-
terior rounded articular facet of the operculum.

Pterosphenoid. — Relatively elongate dorsoven-
trally; much wider dorsally than ventrally; cartilage
filled along all of its edges, except medially; articulates
by interdigitation dorsally with the frontal and ventrally
with the prootic.

Quadrate. — Wide posteriorly, tapering to a knob
anteriorly for articulation with the articular in the lower
jaw; only a very short posterior process present from its
ventral edge between the symplectic and preoperculum;
cartilage filled at its posterior edge; articulates by inter-

digitation anterodorsally with the ectopterygoid and
posteroventrally with the symplectic, which it broadly
overlies; articulates by fibrous tissue ventrally with the
preoperculum and through cartilage posteriorly with the
metapterygoid. The anterior knoblike end of the quad-
rate articulates against the groove on the posterior edge
of the articular in the lower jaw, with the fibrous tissue
articulation strengthened by a short anteromedially
directed process from the anterior edge of the quadrate
just above its knoblike anteroventral end.

Metapterygoid. —A large plate of bone; cartilage
filled along its anteroventral edge; articulates by inter-
digitation anterodorsally with the ectopterygoid and
mesopterygoid and anteroventrally with the symplectic,
all three of which bones the metapterygoid broadly over-
lies; articulates posteroventrally by fibrous tissue and
slight interdigitation with the hyomandibular, while an-
teriorly it articulates through cartilage with the quad-
rate.

Opercular Region.

Operculum. — A large flat plate with a high, well-
developed, lateral flange which ends dorsally by project-
ing over the hyomandibular to serve as a place of attach-
ment for muscles originating on the lateral wing of the
sphenotic; articulates by fibrous tissue ventrally with the
suboperculum. The upper anterior edge of the oper-
culum is laterally expanded into a stout facet for fibrous
tissue articulation with the concave surface of the up-
raised area on the posterior edge of the hyomandibular.

Suboperculum. — A relatively flat plate, except for a
lateral thickening along its edge of fibrous tissue ar-
ticulation with the anterior edge of the operculum. From
its anterodorsal region the suboperculum is prolonged
anteriorly as a narrow shaft to a level slightly forward of
the posterior edge of the preoperculum. The anterior end
of this shaftlike process connects by a short ligament
with the posterior end of the interoperculum.

Symplectic. — Large; broadly overlain by both the
metapterygoid and quadrate; cartilage filled at its pos-
terior end; articulates through cartilage with the an-
terior end of the hyomandibular, and by interdigitation
dorsally with the metapterygoid and anteriorly with the
quadrate. Ventrally the symplectic articulates by fi-
brous tissue with the preoperculum.

Palato-Pterygoid Region.

Palatine. — Expanded dorsally into a horizontal
plate; articulates by interdigitation dorsally with the
frontal, ventrally with the ectopterygoid, and postero-
dorsally with the parasphenoid. Posteroventrally the
palatine broadly overlies and articulates by slight inter-
digitation with the mesopterygoid. The anterior edge of
the portion of the palatine which overlies the mesoptery-
goid is cartilage filled in all but the larger study
specimens. Anterodorsally the rounded edge of the
palatine articulates by fibrous tissue with the concave ar-
ticular facet of the maxillary, while a medially directed
prong from the anterior edge of the palatine articulates
by fibrous tissue with the posteromedial edge of the pre-
maxillary.

Ectopterygoid. — An almost straight shaft of bone,
except for a thin posterodorsally expanded portion that
interdigitates with the palatine, mesopterygoid, and
metapterygoid; articulates by extensive interdigitation
dorsally with the palatine and ventrally with the quad-
rate.

Mesopterygoid. — Somewhat variable in shape, but
always broadly overlain by the palatine and metaptery-
goid; articulates anterodorsally by slight interdigitation
with the palatine, while anteroventrally and posteroven-
trally it articulates by more extensive interdigitation
with, respectively, the ectopterygoid and metaptery-

Interoperculum. — A stout rod, with a large ventral
flange in about the middle of its length; articulates by
ligaments anteriorly with the angular in the lower jaw
and posteriorly with the anterior end of the suboper-
culum. The medial surface of the ventral flange of the in-
teroperculum articulates by fibrous tissue with the
lateral surface of the epihyal and the posterolateral sur-
face of the ceratohyal.

Preoperculum. — Large; expanded posteroven-
trally; its dorsal edge somewhat laterally expanded to
form a broad surface of fibrous tissue articulation with
the ventral edges of the quadrate, symplectic, and
hyomandibular.

Upper Jaw.

Premaxillary. — The two premaxillaries are indis-
tinguishable fused in the midline and together with the
fused teeth form a massive crushing plate; anterior edge
of premaxillary forming the border of the upper jaw for
about the dorsal two-thirds of that border, ventral to
which the maxillary forms the anterior edge of the upper
jaw. In the middorsal line the fused premaxillaries form a
short posteromedial process which articulates by fibrous
tissue with the palatines and the ethmoid-vomer.
Posterolaterally the premaxillaries are broadly overlain
by and extensively interdigitated with the maxillaries.
The fused teeth of the premaxillary are closely similar to
those described for Triodon. They are small, discrete, in-
dividual pieces formed in the relatively small pulp cavity
of the premaxillaries. They move forward toward the
edge of the jaw and become closely packed against one
another in the bony matrix that surrounds them. By the
time they reach the leading edge of the jaw they are so
densely packed together that they have lost much of their
individuality, at least in large specimens. Whether the
teeth at the edge of the jaw ever truly fuse indistinguish-
ably with themselves and the premaxillary is again a

matter for histological study. The remains of approxi-
mately 20 of the formerly separate dental units can be
seen along the edge of the premaxillary to either side of
the midline and to a depth of 5 to 10 units posteriorly
from the edge. At the posterior end of the tooth bearing
region, the sockets surrounding the individual pri-
mordia open to the exterior on the outer surface of the
premaxillaries by small pores, similar to those described
for Triodon. A large trituration plate is present in the
dorsomedial region of the inner surface of the fused pre-
maxillaries. The trituration plate is divided into right
and left halves whose medial edges are in close contact.
Each plate has its relatively flat surface marked by ap-
proximately five parallel grooves which show the limits of
the originally separate dental plates that are now fused
together in a solid mass. These dental plates are formed
in a large pulp cavity, above the trituration plate, which
is separated from the pulp cavity in which the teeth of
the jaws are formed by a thin bony partition and which is
broadly open to the exterior posteriorly to either side of
the midline. The newly developed platelike trituration
teeth are obliquely placed in an anterodorsal to postero-
ventral plane at about a 45° angle to the surface of the
trituration plate. At first, only the posteroventral edge of
the new tooth plate is exposed at the surface of the
trituration plate, but as the tooth is moved forward and
downward, its whole substance is gradually worn away.
As the tooth plates are moved forward they become more
and more closely packed together, until they lose most of
their individual identity, at least in larger specimens.

Maxillary. — Expanded ventrally where it forms the
lower one-third of the anterior margin of the upper jaw;
broadly overlies and extensively interdigitates with the
premaxillary. The medial surface of its free ventral end
articulates by fibrous tissue with the dorsolateral surface
of the dentary. Posterodorsally the maxillary possesses a
medially directed process which runs under the anterior
end of the palatine to make fibrous tissue contact with
the lateral edge of the posteromedial projection of the
fused premaxillaries. The concavity on the posterodor-
sal edge of the maxillary articulates by fibrous tissue
with the anterior end of the palatine. In about the middle
of its length the maxillary is laterally expanded into a
heavy flange which serves as a place for muscle at-
tachment.

Lower Jaw.

Dentary. —The two dentaries are indistin-
guishably fused in the midline, and, like the premaxil-
laries, form a huge crushing beak in conjunction with the
fused teeth. The concave posteromedial surface of the
dentary interdigitates with the lateral surface of the ar-
ticular. Posteroventrally the dentary interdigitates with
the angular. A large trituration plate, divided into right
and left halves in close contact with one another, is pre-
sent in the medial region of the inner surface of the fused
dentaries. The structure and development of the dental
units of the trituration plate and biting edge of the lower

jaw are exactly like that described for the premaxillary,
except that the internal cavity for the dental pulp is
somewhat smaller. The dorsolateral surface of the den-
tary articulates by fibrous tissue with the ventromedial
surface of the maxillary.

Articular. — More or less triangular in shape; its
posterior edge laterally expanded basally and bearing a
groove for fibrous tissue articulation with the quadrate;
cartilage filled for only a short distance anteriorly where
it is continuous with the remains of Meckel's cartilage;
articulates by extensive interdigitation anteriorly with
the dentary and posteroventrally with the angular. The
sesamoid articular is a nubbin of bone, of variable size,
which is usually interdigitated with the articular just
above and posterior to the remains of Meckel's cartilage.

Angular. —A small bone interdigitated anteriorly
with the dentary and dorsally with the eu'ticular.
Posteriorly the angular attaches by ligament to the inter-
operculum.

BRANCHIAL APPARATUS.

Hyoid Arch and Branchiostegal Rays.

Hypohyals. — Dorsal and ventral hypohyals
present; the ventral hypohyal about twice as large as the
dorsal hypohyal; dorsal hypohyal cartilage filled at its
ventral edge, ventral hypohyal cartilage filled at its pos-
terior edge; articulate through cartilage with one an-
other and with the anterior end of the ceratohyal; ar-
ticulate by fibrous tissue medially with their opposite
members.

Ceratohyal. —Somewhat expanded posteriorly;
cartilage filled at its anterior and posterior edges; ar-
ticulates through cartilage anteriorly with the hypo-
hyals, and through cartilage and slight interdigitation
posteriorly with the epihyal, which it overlies. All six
branchiostegal rays articulate by fibrous tissue with the
ceratohyal, the platelike first ray with a vertical groove
on the medial surface in about the middle of its length, as
explained below.

Epihyal. — Round and broadly overlain by the pos-
terior end of the ceratohyal, with which it articulates
through cartilage and slight interdigitation. The lateral
surface of the epihyal articulates by fibrous tissue with
the medial surface of the ventral flange of the inter-
operculum.

Branchiostegal rays. — Six in number; the first
branchiostegal ray a large plate, of increasing width pos-
teriorly, whose lateral edge is sharply downturned and
whose medial edge is more gradually upturned. Anterior-
ly the downturned lateral edge of the first branchio-
stegal becomes thickened into a vertical articular face
which fits against the vertical groove on the medial sur-
face of the ceratohyal. The second branchiostegal ray is a

347

normally tapering shaft, articulating with a slight inden-
tation on the ventromedial surface of the ceratohyal just
posterior to the articulation of the first branchiostegal.
The third to sixth branchiostegal rays articulate one just
behind the other with the lower half of the posterolateral
surface of the ceratohyal. The third branchiostegal is
more laterally expanded in the anterior third of its length
than are the other rays.

Branchial Arches. — All the elements are cartilage
filled at their edges of articulation with the other
elements of the series, and the articulations are usually
through cartilage and fibrous tissue. The branchial
arches are composed of three basibranchials, three pairs
of hypobranchials, five pairs of ceratobranchials, four
pairs of epibranchials, and three pairs of pharyngo-
branchials. Three gills are present; the fourth arch has no
gill and there is no slit between the fourth arch and the
lower pharyngeal.

First arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. First basi-
branchial the largest of the basibranchial elements; dis-
placed forward so that it articulates posteriorly with the
second basibranchial and posterolaterally with the first
hypobranchials. First hypobranchial the longest of the
hypobranchial elements; articulates ventrally with the
region of articulation between the first and second basi-
branchials and dorsally with the first ceratobranchial.
First ceratobranchial the shortest of the ceratobranchial
elements, which, except for the first ceratobranchial, in-
crease in length posteriorly in the series; possesses weak-
ly developed flanges from its anteroventral and postero-
ventral edges, similar flanges being increasingly well
developed on the second and third ceratobranchials but
not present on the fourth or fifth ceratobranchials; ar-
ticulates ventrally with the first hypobranchial and dor-
sally with the first epibranchial. First epibranchial a
slender rod; articulates dorsally with an upraised area on
the anterodorsal surface of the first pharyngobranchial.
First pharyngobranchial a relatively flat plate; wider
laterally than medially; bearing over all of its ventral
surface innumerable, small, villiform teeth, slightly
larger in the posterior region than anteriorly; articulates
at an upraised area on its anterodorsal surface with the
first epibranchial and at an upraised area on its postero-
dorsal surface with the second epibranchial.

Second arch. — Basi-, hypo-, cerate-, epi-, and
pharyngobranchial elements present. Second basi-
branchial the smallest of the basibranchial elements; ar-
ticulates anteriorly with the first basibranchial, antero-
laterally with the first hypobranchials, posteriorly with
the third basibranchial, and posterolaterally with the
second hypobranchials. Second hypobranchial the short-
est of the hypobranchial elements; articulates ventrally
at the region of articulation between the second and third
basibranchials and dorsally with the second cerato-
branchial. Second ceratobranchial articulated dorsally
with the second epibranchial. Second epibranchial a long

slender rod; like the first epibranchial, but with a slight
expansion on its posterior edge for fibrous tissue articu-
lation with the third epibranchial; articulates dorsally
principally to the upraised area on the second pharyngo-
branchial, only secondarily with the upraised area on the
posterodorsal surface of the first pharyngobranchial. Sec-
ond pharyngobranchial the longest of the pharyngo-
branchial elements; its toothed surface narrower but
longer than that of the first pharyngobranchial; in-
numerable villiform teeth present, like those on the first
pharyngobranchial, but with those along the posterior
edge distinctly larger than more anteriorly; articulates at
an upraised area on its dorsal surface principally with the
second epibranchial but also with the anterodorsal end of
the third epibranchial.

Third arch. — Basi-, hypo-, cerato-, epi-, and
pharyngobranchial elements present. Third basi-
branchial somewhat wider posteriorly than anteriorly;
articulates anteriorly with the second basibranchial,
anterolaterally with the second hypobranchials, pos-
terolaterally with the third hypobranchials, and poste-
riorly with the fourth ceratobranchials. Third hypo-
branchial almost as long as the first hypobranchied;
directed anteroventrally under the second hypo-
branchial, with its narrow anterior end articulated by fi-
brous tissue with the region of the first basibranchial and
dorsal hypohyals. Third ceratobranchial articulated ven-
trally with the third hypobranchial and dorsally with the
third epibranchial. Third epibranchial thicker than
those anterior to it; expanded dorsally into two short
flanges, the more anterior of which articulates with the
upraised area on the second pharyngobranchial, while
the more posterior flange articulates with the anterior
edge of the fourth epibranchial. Third pharyngo-
branchial an elongate, thin, flat plate; toothless; held
in place primarily by fibrous tissue to the posterior
edge of the second pharyngobranchial, without any
special connection with any of the other branchial
elements.

Fourth arch. —Cerato- and epibranchial elements
only. Fourth ceratobranchial a stout shaft; without ven-
tral flanges; the longest of the ceratobranchial elements;
articulates ventrally with the third basibranchial and
dorsally with the fourth epibranchial. Fourth epi-
branchial by far the longest and stoutest of the epi-
branchial elements; somewhat produced into a thin
flange posteriorly along the lower two-thirds of its length;
articulates ventrally with the fourth ceratobranchial,
while dorsally it ends in the general fibrous tissue that
binds the dorsal surfaces of the pharyngobranchial
elements to the ventral surfaces of the parasphenoid and
prootics.

Fifth arch. —Ceratobranchial (lower pharyngeal)
element only. Fifth ceratobranchial large; slightly ex-
panded laterally along most of its length; toothless; ar-
ticulates ventrally with the base of the fourth cerato-
branchial.

PAIRED FIN GIRDLES.

Pectoral Fin.

Supracleithrum. — In position almost parallel to the
axis of the body; a round shaft anteriorly, but becoming
wider and dorsoventrally depressed posteriorly; articu-
lates at its rounded anterior end by fibrous tissue with
the concavity on the posterior surface of the laterally
directed wings of the pterotic. The posterior portion of
the supracleithrum overlies and articulates by fibrous
tissue with the dorsolateral surface of the cleithrum and
postcleithrum.

Cleithrum. — Wide dorsally, but tapering to a
narrow shaft in the lower third of its length; much ex-
panded posterolaterally throughout the middle of its
length. The cleithrum articulates by fibrous tissue dor-
sally with the overlying supracleithrum and with the
anterior half of the ventral edge of the postcleithrum,
which it overlies. Posterodorsally the cleithrum articu-
lates by fibrous tissue and slight interdigitation with the
anterior edge of the scapula. Along its posterior edge, in
about the middle of its length, the cleithrum articulates
by fibrous tissue with the anterodorsal and antero-
ventral edges of the coracoid. The anterior ends of the
two cleithra articulate with each other by fibrous tissue
in the midventral line of the body between the medial
edges of the platelike portions of the first branchiostegal
rays at the level of the end of the ceratohyal.

Postcleithrum. — A rounded shaft anteriorly, but
becoming compressed posteriorly into a thin sheet whose
outline is highly irregular from specimen to specimen. In
none of the specimens is there any evidence of two
separate elements making up the postcleithrum, for the
dorsal and ventral postcleithra apparently have fused
into one piece. The postcleithrum articulates by fibrous
tissue anteriorly with the cleithrum and supra-
cleithrum.

Coracoid. — Constricted in the middle, but
expanded dorsally and even more expanded ventrally;
cartilage filled along its dorsal and anteroventral edges;
articulates by fibrous tissue anterodorsal ly and antero-
ventrally with the cleithrum, while dorsally it articu-
lates through cartilage with the base of the scapula and
bases of the second to fourth actinosts.

Scapula. — Scapular foramen not entirely enclosed
by the scapula, but, rather, with the lower half of its
anterior edge closed by the cleithrum; cartilage filled at
its dorsal and ventral edges; articulates by interdigi-
tation anteriorly with the cleithrum, posterodorsally
with the anterior edge of the first actinost and postero-
ventrally with the anteroventral edge of the second ac-
tinost. Ventrally the scapula is supported through car-
tilage by the coracoid.

Actinosts. — Four elements; all cartilage filled at
both ends, except for the reduced first actinost, which is
cartilage filled only dorsally; first actinost a triangular
wedge interdigitating with the dorsal half of the poste-
rior edge of the scapula and the anterodorsal edge of the
second actinost; second, third, and fourth actinosts more
or less constricted in the middle, but with decreasing
amounts of constriction from the second to the fourth.
The second, third, and fourth actinosts interdigitate with
one another at their ventral edges of contact, and the
anteroventral edge of the second actinDst interdigitates
with the posteroventral edge of the scapula. Dorsally the
first, second, and third actinosts interdigitate with one
another. The fourth actinost articulates anterodorsally
with the third actinost by fibrous tissue in some
specimens but by interdigitation in others. The second to
fourth actinosts are supported basally through cartilage
by the coracoid. Distally the actinosts and scapula sup-
port by fibrous tissue most of the fin rays.

Fin rays. — Twenty-two to twenty-four fin rays
present in most specimens; the first ray very small and
divided into two halves throughout its length, the medial
half much larger than the lateral half. The first ray ar-
ticulates by fibrous tissue with the scapula, whereas the
other rays articulate with the actinosts. The first two
rays and the last ray are unbranched; all the others are
branched. Fiist ray without cross-striations, the others
cross-striated.

VERTEBRAL COLUMN. —All vertebrae with
biconcave centra, except for the last, which ends poste-
riorly in fusion with the parhypural and hypurals.

Abdominal Vertebrae.

First vertebra. — Neural spine bifid; neural arch
with a bony roof over the neural canal, the shelflike
medial processes from each side of the neural arch
meeting and interdigitating in the midline; articulates
by fibrous tissue over the upper half of the anterolateral
surface of its centrum with the short exoccipital con-
dyles, while the rim of the concave anterior end of the
centrum rests against the rim of the concave posterior
end of the basioccipital. The deeply concave postero-
lateral edge of the neural arch articulates by fibrous tis-
sue with the anteriorly projecting neural prezygapophy-
sis of the second vertebra.

Other abdominal vertebrae. — Twelve abdominal
vertebrae in 10 specimens. The 1st to the Uth ab-
dominal vertebrae have bifid and divergent neural
spines. The bifid neural spines are of about the same
height on all the vertebrae, but the anteroposterior
length of the spines increases from the 1st to about the
4th abdominal vertebra and then decreases from the 5th
to the 11th abdominal vertebra. Shelflike medial projec-
tions from each side of the neural arch meet and inter-
digitate (or even fuse) in the midline to form the bony

roof over the neural canal on the first to ninth ab
dominal vertebrae. The 10th and Uth abdominal verte
brae have no such bony roof over the neural canal, the
canal being roofed over between the divergent neura
spines of these two vertebrae only by fibrous tissue. The
dorsomedial region of the bony roof over the neural cana
becomes upraised into what is, in effect, a third neura
spine on some of the abdominal vertebrae. There is only
a trace of such a dorsomedial flange on the third ab
dominal vertebra but on the fourth abdominal vertebra it
is distinctly present, and by the fifth and sixth ab
dominal vertebrae it reaches its maximum develop
ment, then becomes progressively less well developed on
the seventh and eighth abdominal vertebrae until by the
ninth abdominal vertebra it is no longer present. At its
greatest development, on the fifth and sixth abdominal
vertebrae, this dorsomedial flange is prolonged an-
teriorly to such an extent that it protrudes over the pos-
terior region of the neural arch of the preceding verte-
bra, with fibrous tissue attaching it to the bony roof over
the neural canal. Neural prezygapophyses are well
developed on all of the abdominal vertebrae and ar-
ticulate by fibrous tissue with the posterodorsal regions
of the lateral surfaces of the centrum just anterior to
them. Haemal arches are not present on any of the ab-
dominal vertebrae, but from the 9th to 12th abdominal
vertebrae progressively larger ventrolateral flanges are
present from the ventrolateral edges of the centra. These
flanges are the place of attachment of large muscle
masses. The 12th abdominal vertebra differs from the
other abdominal vertebrae in that its neural arch does
not possess a bifid and divergent neural spine. From the
base of its neural arch area two broad plates project dor-
sally and slightly medially to meet the bases of the first
two dorsal fin basal pterygiophores. There is no distinct
bony roof over the neural canal of the 12th abdominal
vertebra, the functional roof being formed by the fibrous
tissue articulation of the ventral ends of the basal pteryg-
iophores with the dorsal ends of the lateral walls of the
neural arches.

Caudal Vertebrae. — Nine caudal vertebrae in 10
specimens; the first six caudal vertebrae variously sup-
porting the dorsal and anal fin basal pterygiophores. The
first three caudal vertebrae are structurally similar to the
last abdominal vertebra, but their ventrolateral flanges
become progressively shorter, and delicate posteroven-
trally directed processes from the ventrolateral surfaces
of their centra are present which articulate by fibrous tis-
sue with the dorsal ends of the anal fin basal pterygio-
phores. As is the case with the last abdominal vertebra,
the lateral walls of the neural arches of the first three
caudal vertebrae articulate by fibrous tissue dorsally
with the ventral ends of the dorsal fin basal pterygio-
phores. The fourth caudal vertebra differs from the
preceding caudal vertebrae by the only slight develop-
ment of its ventrolateral flange, and by the presence
along the posterior half of the dorsal surface of its neural
arch of a bifid and divergent neural spine. The anterior
half of the dorsal surface of this neural arch possesses the

dorsomedially directed bony plates which articulate by
fibrous tissue with the bases of the last two dorsal fin
basal pterygiophores, in the same way as described for
the preceding caudal vertebrae. From its posteroventral
edge the fourth caudal vertebra possesses a short pos-
terior process which articulates by fibrous tissue with an
anteriorly directed prong from the haemal process of the
fifth caudal vertebra. The fifth caudal vertebra possesses
a bifid and divergent neural spine between which is sup-
ported by fibrous tissue the last dorsal fin basal ptery-
giophore, with no bony roof present over the neural canal
between the medial edges of its divergent neural spine.
Delicate ventral processes are present from the ventro-
lateral edges of the fifth caudal vertebra which support
the dorsal end of the last anal fin basal pterygiophore,
while a bifid and divergent haemal process is present
ventrally to either side of the upper end of the last anal
fin basal pterygiophore. There is no special bony roof
over the haemal canal, except the functional one formed
by the ventral processes of the fifth caudal vertebra
meeting with the dorsal end of the last anal fin basal
pterygiophore. The anterior edge of the bifid haemal
process is produced into a short prong which articulates
by fibrous tissue with the posterior process from the
fourth caudal vertebra. The sixth and seventh caudal
vertebrae are similar to one another, their centra being
relatively short anteroposteriorly and both possessing
bifid neural and haemal spines. No bony roofs are present
between these bifid neural and haemal spines, so that the
neural and haemal canals are roofed over only by fibrous
tissue. The anterior edges of the neural and haemal
spines of the sixth caudal vertebra support by fibrous tis-
sue, respectively, the last dorsal fin basal pterygiophore
and the last anal fin basal pterygiophore. The eighth
caudal vertebra has its neural and haemal spines bifid
and divergent anteriorly, but as single medial plates pos-
teriorly. The undivided posterior portions of the neural
and haemal spines of this vertebra are posteriorly ex-
panded, and they interdigitate with the anterior edge of
the plate formed by the last vertebra and associated
caudal fin supporting elements. From the lateral surface
of the haemal spine of the eighth abdominal vertebra, a
posteriorly directed process is present which articulates
by fibrous tissue with the anteriorly directed process
from the last vertebral segment.

Caudal Skeleton. — The caudal fin supporting struc-
tures are consolidated into a single plate which in adults
shows only meager traces of areas of suturing or fusion of
the originally separate elements. In specimens smaller
than about 100 mm, however, at least one of the elements
retains its individuality, this being the epural. The
elongate epural is interdigitated with the posterior edge
of the neural spine of the eighth caudal vertebra and with
the anterodorsal edge of the main portion of the plate
supporting the caudal fin. In larger specimens the sutur-
ing between these two elements becomes obscure and
true fusion may take place. Anteroventrally on the
caudal plate in the smaller study specimens there is some
indication of an originally separate parhypural having

fused with the main body of the plate, but even in the
12.3 mm specimen this thinner parhypural region is con-
tinuous with the main body of the plate. The rest of the
plate is composed of the last vertebral centrum fused
with all of the hypural elements. In about the anterior
two-thirds of its length the caudal plate possesses a lateral
flange whose depth increases anteriorly. The anterior end
of the lateral flange articulates by fibrous tissue with the
posterolateral process from the haemal spine of the
penultimate vertebra. The posterior edge of the caudal
plate supports the caudal fin rays.

Caudal fin rays. — Nine in number; the uppermost
ray and the lowermost ray unbranched, the other rays
becoming increasingly branched toward the middle rays,
which are branched in triple dichotomies. The bifid
bases of the rays articulate by fibrous tissue with the pos-
terior edge of the caudal plate.

DORSAL A^^D ANAL FINS.

Fin rays and pterygiophores. — Fifteen fin rays
usually present, the first two rays and the last ray un-
branched, the others branched in single or double dicho-
tomies. Distal pterygiophores either absent or unos-
sified. The fin rays are supported through fibrous tissue
at their bifid bases by 12 basal pterygiophores. The
pterygiophores are cartilage filled at both ends, and all of
them, with the exception of the last, are stout rods which
decrease in width posteriorly in the series. The last
pterygiophore is rodlike anteriorly, but posteriorly it is
expanded into a flat plate. The ventral edge of the last
pterygiophore articulates by fibrous tissue with the bony
roofs over the neural canal of the fourth and fifth caudal
vertebrae. The other pterygiophores articulate ventrally
by fibrous tissue with the irregular dorsal edges of the
neural arch plates of the 12th abdominal to 4th caudal
vertebrae. The pterygiophores articulate by fibrous tis-
sue dorsally with one another and with the dorsal fin
rays.

Anal Fin.

Fin rays and pterygiophores. — Thirteen or fourteen
fin rays usually present; the first two rays and the last
ray unbranched, the others branched in single or double
dichotomies. Distal pterygiophores either absent or unos-
sified. The fin rays are supported through fibrous tissue
at their bifid bases by nine basal pterygiophores. The
basal pterygiophores decrease slightly in length pos-
teriorly in the series and are all, with the exception of the
last pterygiophore, stout rods. The last pterygiophore is
rodlike anteriorly, but slightly expanded posteriorly into
a flat plate. The amount of expansion of the last anal fin
pterygiophore is not equal to that of the last dorsal fin
pterygiophore, but in large specimens the difference is
not great. The pterygiophores are cartilage filled at both
ends. The dorsal ends of the pterygiophores articulate by
fibrous tissue with the irregular ventral edges of the

haemal processes of the first to fifth caudal vertebrae. At
their distal ends the pterygiophores articulate with one
another by fibrous tissue.

Anatomical diversity.— The few genera of Recent
diodontids form an anatomically compact family of
gymnodonts that has probably changed little in overall
configuration since the Eocene, with the exception of an
increase in the size of the scales and in certain rearrange-
ments of the dentition and jaw supporting structures.

The few relatively complete fossil diodontids are from
the Eocene of Monte Bolca, Italy, and are referable to
Diodon tenuispinus Agassiz and D. erinaceus Agassiz.
Agassiz (1844b) simply said that D. erinaceus was larger
and rounder, and had shorter and heavier spines, than in
the more fully described D. tenuispinus. The types of D.
tenuispinus are in Paris, and Le Danois (1955, 1959)
proposed the new generic name Prodiodon for
tenuispinus on the basis of the supposedly anteriorly
elongate sphenotics broadly entering the edge of the or-
bit above and of the supposedly sutured (rather than ful-
ly fused) premaxillaries which would distinguish
tenuispinus from all Recent species.

The three type-specimens of P. tenuispinus (see
Material Examined) have been examined for this work.
None of the three show much detail of the jaws, skull
structure, vertebral column, and armament. A dorsal
view illustration of the only one of these three specimens
that is in counterpart, and considered at the Paris
museum to be the holotype, is presented by Le Danois
(1959:215, fig. 184). The illustration, in my opinion, is
wildly inaccurate and impressionistic, being rendered
with heights of artistic license. I find it difficult to
decipher any pertinent details in the dorsoventrally com-
pressed and probably fractured cranium, but Le Danois
showed the shapes and sizes of most of the bones there,
with the sphenotic being quite unlike that of any Recent
species, lacking a lateral prong and being prolonged an-
teriorly to form much of the lateral edge of the upper or-
bit, much as in the tetraodontid Colomesus. I seriously
doubt that the degree of detail shown in the illustration
by Le Danois is actually present.

Le Danois described and showed the upper jaws with
the premaxillaries sutured as in tetraodontids, with in-
terlocking emarginations, while saying that the dentaries
are either fused to one another or at least closely at-
tached. To my eyes the jaws that are exposed in the three
specimens seem to be fully fused medially without any
evidence of sutured right and left halves. Several trans-
versely elongate trituration teeth are present to either
side of the midline on what is apparently the inner sur-
face of the lower jaw in the specimen in counterpart. Le
Danois showed the pectoral and dorsal fins far more
clearly than they exist on the specimen. What vague in-
dication remains of the dorsal fin in the fossil is far
shorter based than illustrated by Le Danois, with the
true fin corresponding only to the posterior half of the il-
lustrated fin. The illustration clearly shows 23 verte-
brae, but I would estimate that the much vaguer out-
lines represent about 18 to 20 vertebrae.

351

The extremely long spine shown in the illustration by
Le Danois of P. tenuispinus behind the pectoral fin base
is hinted at in the specimen in counterpart, but it is not
as definite as shown, and could represent either a large
spine or the postcleithrum. Most of the spines that are
apparent seem to have been relatively short and without
massive three-rooted bases, perhaps having been mostly
erectile and like those of the Recent Diodon except
shorter. The spines just behind the pectoral fin base
seem to have been somewhat longer than the others. The
placement of the branchiostegals is clear, but the first,
presumedly platelike, branchiostegal is not evident.

In the specimen in counterpart there seems to have
been 11 caudal fin rays, one more than in any Recent
species, although some of the rays are displaced and the
count is not certain. In two of the three specimens there

</.-{/
•<///

prefrontal cartilage pterosphenoid sphenotic epiotic

ethmoid-vomer
palatine
premaxillary

supraoccipital

^- exoccipital
pterotic

basioccipital

hyomandibular

interoperculun^ quadrate metapterygoid parasphenoid

is some indication that the hypurals were not as fully fus-
ed together as in Recent species, with it being possible
that there were two more or less separate plates of about
equal size, one above and one below, but, again, it is only
clear in this regard that there was some sort of groove or
less ossified narrow horizontal region in the middle of the
caudal plate posterior to the region of the last vertebral
centrum.

Four additional specimens (see Material Examined)
that are identified as P. tenuispinus in European collec-
tions add little to what is known on the basis of the type-
specimens. In MCSNV B.18, 31.5 mm SL, the only items
of interest to the present discussion are that, as in the
type-specimens, there seem to be 11 caudal fin rays and
the trituration teeth are paired, at least one transversely
elongate tooth being placed to each side of the midline,
and that the spines just behind the pectoral fin base (up
to 1.1 mm long) are a little more than twice the length of
the spines elsewhere (0.3-0.4 mm long), while the basal
plate of the spine has up to six radiations, these being
relatively small, about one-half the length of the pro-
jecting spine. The scales are similar to this in MCSNV
B.19, 40.5 mm SL.

In the type-specimen of D. erinaceus some of the
scales, especially posteriorly on the body, do seem to be
three rooted and thus unerectile, although the roots are
relatively short and do not approach the massiveness of
those found in Chilomycterus. It is possible, however,
that the larger size of the roots in the type of P. erina-
ceus (77.5 mm SL) in comparison to those of the types of
P. tenuispinus (12.6-46.5 mm SL) is at least partially ac-
counted for by the differential specimen sizes. The spines
just behind the region of the pectoral fin (which is not
clear) base are longer (3.3 mm) than those elsewhere (up

Figure 282.— Diodon holocanthus:
lateral view of head, composite based c
several specimens, 12. .3-124 mm SL,
Gulf of Mexico and Caribbean.

to L6 mm) on the body of the type of P. erinaceus and
may have been two rooted and erectile, although their
impressions are so poor that this is difficult to tell for
sure.

Other specimens labeled P. erinaceus in European col-
lections (see Material Examined) variously show the
transversely elongate trituration teeth in a single series of
one or two units to each side of the midline, just as in P.
tenuispinus, and spines slightly longer behind the pec-
toral fin base than elsewhere and of a small size similar
to that of P. tenuispinus, some seeming to have short
multiradiate bases and others only biradiate bases, and
thus probably some combination of small erectile and
nonerectile spines. The number of caudal fin rays and
the structure of the last few caudal vertebrae are not
clear in any of the specimens assigned to P. erinaceus,
but in one of them (IGUP 8670-71) there is a strong indi-
cation that the first branchiostegal ray is an enlarged
horizontal plate just as in Recent species.

Since there is the possibility that the Eocene diodon-
tids assigned to P. tenuispinus, and by implication to the
at least closely related P. erinaceus, differ from the Re-
cent species by having 11 caudal fin rays and the
hypurals not fully fused to one another, the generic term
Prodiodon is retained for them, especially in light of the
fact that the numerous diodontid jaws known from the
Eocene onward have been given generic nomenclatural
status separate from that of the Recent species, as ex-
plained below.

353

* i S_•-
Q i » O

I 2 S|

nil

1 1 1 1 1 1 1 1

5lf r &1|J

354

Being especially strong and massive, isolated diodon-
tid jaws and fragments have been widely found in every
period from the Eocene onward, and the confused
nomenclature given to these parts has been extensively
and astutely reviewed by Tavani (1955), who recognized
several new genera on the basis of dental structure, as
only briefly summarized here. Tavani found that Eodio-
don, from the middle and upper Eocene of Europe, dif-
fered from all other diodontid jaws by having a uniform
bony mass not differentiated into small teeth at the bit-
ing edge and larger elements in an internal trituration

plate, as well as a difference in the place of articulation
with the palatine.

Tavani (1955) recognized Eodiodon as the family
Eodiodontidae, in contrast to four genera with distinct
and differentially formed teeth in the biting and tritura-
tion regions. Of the latter four genera, Tavani differenti-
ated Progymnodon (see Dames 1883), from the middle
and upper Eocene of Europe, and Kyrtogymnodon, from
the Pliocene of Europe, from the remaining two genera,
Oligodiodon from the Oligocene and Miocene of Europe
and North America, and Diodon, from the Miocene of
Europe to Recent, on the basis of the depth of the osse-
ous jaw tissue separating the small teeth in the biting
edge from the large teeth in the trituration plate. Pro-
gymnodon and Kyrtogymnodon, as the subfamily Pro-
gymnodontinae, have only a thin band of bony tissue
separating the two, and Oligodiodon and Diodon, as the
Diodontinae, have a broad band.

The difference between the Progymnodon and Oligo-
diodon type spatial arrangement as so carefully de-
scribed by Tavani (1955) undoubtedly represents the
evolutionary change that took place leading to the
arrangement as found in the Recent species, in which the
biting edge and trituration teeth are well separated by a
deep layer of bone, although less so in young specimens
than adults. However, I am skeptical about the reported
dental condition in Eodiodon, of it not having any teeth
differentiated from the osseous tissue of the jaws. My
skepticism is primarily because in the two groups most
basal in the ancestry of the gymnodonts, the eoplectin
triacanthodids and triodontids, the teeth are clearly dif-
ferentiated from the bone, as they are in all diodontids
from the Eocene to Recent, with the possible exception of
Eodiodon.

I suspect that the apparent lack of discrete teeth in
Eodiodon alone is some artifact of preservation of the few
known jaw fragments assigned to Eodiodon, or that the
biting edge teeth and those of the trituration plate were
lost sometime between the process of preservation and
the eventual recovery of the abraded fragments. With the
true nature of the dental units of Eodiodon perhaps still
unknown, and with no knowledge of what the structure of
the fish behind the jaws was like in Eodiodon, Progym-
nodon, Kyrtogymnodon, and Oligodiodon, and only the
slightest knowledge of the body of Prodiodon, I feel no
compulsion for recognizing Eodiodon as familially dis-
tinct from the other diodontids, and none for recogniz-
ing subfamilies on the basis alone of the width of the os-
seous tissue separating the biting and trituration teeth,
while still being keenly aware of the great value of
Tavani's work.

With the exception of Tavani's (1955) careful analysis
of the dental structure in fossil diodontid jaws, and of the
few hints about the general configuration of the body in
Prodiodon, our knowledge of the anatomical diversity of
the family rests essentially on that of the Recent species.
The diodontids differ externally mainly in the form of
the nasal apparatus (two nostrils or one) and spines
(erectile or nonerectile). In times past, only two genera of
diodontids were usually recognized, Chilomycterus for

ceratobranchials

epibranchials

pharyngobranchials

CDQ C=~> basibranchials

hypobranchials

Figure 285.— Diodon holocanthua:
dorsal view of branchial arches
(extended on lower side); lateral
(above) and medial (below, upside down)
views of hyoid arch; composite based
on several specimens 12.3-124 mm SL,
Gulf of Mexico and Caribbean.

those species in which all of the spines have massive tri-
or quadriradiate bases and are nonerectile and Diodon
for those species in which at least some of the spines have
less massive and only biradiate bases and are erectile.
However, in the two most recent revisions of diodontid
systematics, three genera have been recognized by
Fraser-Brunner (1943) and Le Danois (1959), entirely on
the basis of the nasal apparatus and spines, and none of
the three genera are entirely comparable between the two
works.

Fraser-Bruner ( 1943) recognized Diodon for those spe-
cies with two nostrils and with at least some of the spines
erectile (principally hystrix, holocanthus, jaculiferus,

calori), Dicotylichthys for those with a bifid nasal tenta-
cle and with at least some of the spines erectile (princi-
pally nicthemerus, pilatus, punctulatus, and diuersi-
spinis), and Chilomycterus for those with two nostrils
(the division between which might be fragile and easily
broken) and with all of the spines three rooted and non-
erectile (numerous species).

Two or three subgenera were recognized by Fraser-
Brunner (1943) in each of the three genera but these have
not been used by subsequent workers. The three subgen-
era of Diodon and Dicotylichthys have exactly the same
characters, one subgenus in each for those species all of
whose spines are two rooted and erectile and two sub-

Figure 286.— Diodon holocanthus
lateral view of pectoral girdle,
composite based

specimens, 12.3-124 mm SL, Gulf
of Mexico and Caribbean.

Figure 2H7.— Diodon holocanthus:
lateral view of caudal fin supporting
structures, composite based on several
specimens, 12.3-124 mm SL, Gulf of
Mexico and Caribbean.

genera in each for those species which have most of the
spines three rooted and nonerectile, differing in whether
the few erectile spines are behind the pectoral fin or on
the head. This seems too fine a splitting, especially on
the basis of one character. The division of Chilomyc-
terus by Fraser-Brunner into two subgenera was on the
basis of having two spines above the eye, one on the mid-
dle of the forehead, and a nasal tube with two nostrils
(Cyclichthys for antennatus, antillarum, echinatus, orbi-
cularis, schoepfi, spinosus) versus three spines above the
eye, none on the forehead and "nasal papilla adpressed,

division between nostrils feeble, easily broken" (Chilo-
mycterus for affinis, atinga, tigrinus = reticulatus, and
calif orniensis, a synonym oi affinis).

As discussed subsequently, the nasal differences
between the subgenera Cyclichthys and Chilomycterus
are valid, as modified, and those species with an open
pitted cup do lack a spine in the middle of the forehead
while those with two nostrils always have a spine there.
The number of spines above the eye is not consistently
different between the two groups, and much depends on
how one interprets the position of the third spine in the

series over the upper edge of the orbit. In most species of
the subgenus CycUchthys the third spine has its exter-
nal process placed distinctly behind the eye, but in orbi-
cularis it is placed just above the rear margin of the eye,
while in the subgenus Chilumycterus, only affinis has the
external process just above the rear margin of the eye,
while in atinga and tigrinus it is placed just behind the
rear margin of the eye and in reticulatus distinctly
behind the rear margin of the eye.

Some of the difficulties in this arrangement of genera
are that: 1) Diodon contains species such as jaculiferus
in which most of the spines have massive triradiate bases
and only a few just behind the pectoral fin base are es-
pecially elongate and erectile, the bases of these spines
being only biradiate, or with a triradiate base in which
one of the radiations (that normally directed anteriorly
when there is a triradiate base and nonerectile spine) is
shortened and directed strongly dorsally or ventrally
toward one of the two well-developed radiations; 2) Dio-
don contains species such as calori in which only the
spines on the forehead are two rooted and erectile;
3) those species of Diodon such as hystrix and holocan-
thus that have all or most of the spines essentially two
rooted and erectile in young stages have most of the
spines nonerectile in large adults, the base of the spine
tending to become massive and triradiate with the devel-
opment of an anteriorly directed root; 4) in Diodon at
least two species, jaculiferus and calori, sometimes do
not have two nostrils, the septum between the nostrils in
a minority of specimens (about lO'^^c in jaculiferus) ap-
parently being resorbed (and not damaged by tearing);

5) in Dicotylichthys, just as in Diodon, there are species
such as pilatus and punctulatus in which most of the
spines are triradiate and nonerectile, those that are es-
sentially biradiate and erectile being, respectively, just
behind the pectoral fin base and on the head and belly;

6) in Dicotylichthys at least one species, nicthemerus,
usually has a nasal tube with two nostrils in young spec-
imens, the partition between the two nostrils apparently
usually being resorbed in adults to produce a bilobed
tube or tentacle, and Le Danois (1959:227) reported a
specimen with both conditions, one on either side; 7) in
Chilomycterus there are two types of nasal apparatuses,
most species having two nostrils and a relatively smooth
or only slightly folded epithelium on the inner surface of
the tube, but three species, reticulatus, tigrinus, and af-
finis having the nasal apparatus as an open low cup (i.e.,
one large nostril) with a pitted epithelium, perhaps at all
sizes, and another species, atinga, in which the nasal
apparatus is tubular with two nostrils and a relatively
smooth inner epithelium in young specimens, but with
the great majority of large specimens having converted
this into an open cup with a pitted surface by the resorp-
tion of the partition originally separating the two nos-
trils, the remains of the partition apparent even in large
adults as an unpitted ridge across the middle of the
pitted cup.

Le Danois (1959) recognized Diodon for those species
with either two nostrils or a bifid tentacle and with at
least some of the spines two rooted and erectile, this be-

ing the combination of Fraser-Brunner's (1943) Diodon
and Dicotylichthys, while Fraser-Brunner's Chilomyc-
terus, containing all the species in which all the spines
are three rooted and nonerectile, in Le Danois' system is
split into Chilomycterus for reticulatus (and tigrinus),
with the nasal apparatus an open cup with a pitted sur-
face, and Atinga for all the other species, supposedly all
with two nostrils and an unpitted inner epithelium. Le
Danois' system has all of the drawbacks of Fraser-
Brunner's in that Diodon contains species with a cover-
ing of almost as fully nonerectile, triradiate spines as in
Fraser-Brunner's Chilomycterus, and of two different
types of nasal tentacles, while Le Danois' Chilomycterus
contains a single species (she considered tigrinus and
reticulatus synonymous) with a pitted cup nasal appa-
ratus (as well as one poorly preserved Miocene species,
acanthodes, which seems to have had some two-rooted
erectile spines and an unknown nasal apparatus, and
would thus be assignable by her own system to Diodon)
and her Atinga contains species having a nasal tube with
two nostrils except for two species, affinis, which has a
pitted cup at all stages as far as known, and atinga, in
which adults develop a pitted cup nasal apparatus exact-
ly as found in Chilomycterus reticulatus, tigrinus, and
affinus.

Neither of the above two systems are tenable, but
internal characters do little to clarify the situation, even
though a large number of species were examined for this
work. For purposes of discussion, the generic categories
of Fraser-Brunner are used below. The nomenclature and
diagnostic characters of many of the species of Chilo-
mycterus are confused, and one or two of the names used
here may be synonymous. For example, C. reticulatus,
the type-species of Chilomycterus, is known only from
very large adults (300 to over 600 mm), and C. tigrinus
may be the younger adult of it, since both are extremely
similar osteologically. They differ mainly in tigrinus hav-
ing fewer but larger dark spots and few if any larger dark
blotches.

Before passing on to the osteological diversity of the
diodontids, one additional external feature is of syste-
matic significance. As outlined by Tyler (1970b:25), the
great majority of diodontids have 9 caudal fin rays, while
a few species of Chilomycterus (sensu Fraser-Brunner)
have 10. The four species presently known with 10 rays
are affinis, atinga, reticulatus, and tigrinus, all four of
which have the nasal apparatus as a pitted cup, at least
in adults. Since the ancestral tetraodontids and at least
some of the Eocene diodontids have 11 caudal rays, the
four species of Chilomycterus with 10 caudal rays are
more generalized in this respect than are the other spe-
cies of diodontids, all of which have the caudal rays fur-
ther reduced to 9, at least modally. Conversely, these
four species are more specialized than other diodontids in
their nasal apparatus, for at least the more generalized of
the ancestral tetraodontids, as well as the triodontids
and triacanthodids, have two nostrils to each side.

One unusual internal feature is shared by the four spe-
cies of Chilomycterus with the combination of 10 caudal
rays, no spines on the forehead, and a pitted cup nasal

apparatus. All four have the specialized condition of the
absence of the prefrontal. However, the absence of a pre-
frontal is not entirely confined to these four species, for
C. orbicularis lacks the prefrontal but has nine caudal
rays, a spine on the forehead, and a nasal tube with two
nostrils. In C. orbicularis the palatines are more elon-
and less widely spaced than in the other species of
Chilomycterus, and the frontals are deeply indented in
the region where prefrontals ordinarily occur.

The minor variation in the caudal fin supporting struc-
tures in diodontids is discussed by Tyler (1970b), this in-
volving the degree of fusion between the epural and the
neural spine of the penultimate vertebra and between
the parhypural variously to the fused hypural-centrum
plate or to the haemal spine of the penultimate verte-
bra. These regions are so difficult to interpret in the spec-
imens examined that no information of diagnostic or
phylogenetic usefulness was found.

In many diodontids, as in all other gymnodonts except
Triodon, the fifth ceratobranchial is toothless. However,
in C. affinis, orbicularis, and tigrinus the fifth cerato-
branchial bears a small patch of minute teeth. In the sin-
gle large specimen of C. reticulatus examined a small
patch of minute teeth is present on the fifth ceratobran-
chial on one side but not on the other. The two species of
Dicotylichthys examined (nicthemerus and punctula-
tus) both have a well-developed elongate patch of minute
teeth on the fifth ceratobranchials. These small teeth
could either be remnants of those that probably occurred
on the fifth ceratobranchial of the ancestral tetraodon-
tid line, retained from its triodontid ancestry before the
fourth gill was lost, or a de novo acquisition in a few dio-
dontids from an immediate ancestry lacking teeth on the
fifth ceratobranchial. The latter seems more reasonable,
for the loss of teeth on the fifth ceratobranchial is un-
doubtedly associated with the loss of the fourth gill and
of the gill slit between the fourth and fifth arches, which
event occurred in the ancestral line common to both the
tetraodontids and diodontids, and both families prob-
ably have a long history of lacking teeth on the fifth cer-
atobranchial while still retaining the ability to occasion-
ally develop teeth there as dietary changes took place in
various evolutionary lines making use of slightly differ-
ing foods.

Figure 289.— Dorsal views of skulls of: left, Chilomycterus schoepfi,
60.1 mm SL, Louisiana; right, C. affinis, 310 mm SL, Galapagos.

Figure 290.— Dorsal views of skulls of: left, DiodonjacuUferus,
5.3.4 mm SL, Australia; right, Chilomycterua orbicularia, 72.9 mm SL, Somalia.

Figure 291.— Dorsal view of skull of

Dicotylichthya punctulatus, 226 mm SL,

Australia.

pharyngobranchials

epihyal

branchiostegal rays

basibranchials
hypobranchials

ceratobranchials

dorsal hypohyal

ventral hypohyal
ceratohyal

Figure 292.— Chi/omyctcnts achoepfi:

dorsal view of branchial arches (extended oi

lower side) and lateral view of hyoid arch,

168 mm SL, Florida.

Figure 293.— Nasal apparatuses of representative

diodontids: A, Chilomycterus mauretanicus;

B, C. orbicularis; C, C. schoepfi; D, C. antennatus,

1 being the normal condition of two nostrils

in each tube and a ridge along the inner

surface, while 2 and 3 are variations of

abnormal conditions, 2 having a single nostril

on one side but the normal two nostrils on the

other side (USNM 249592, 75.2 mm), and 3 having

a single nostril on both sides (USNM 249592,

85.7 mm); E, C. reticulatus; F. C. atinga.

1 being the normal condition of an open nasal

sac (one nostril) on each side and 2 being an

abnormal condition of a closed sac with two

nostrils on one side but an open sac on the

other (ANSP 103900, 278 mm); G, Diodon hystrix;

H. D. holocanthus, 1 being the normal condition

of two nostrils in each nasal sac and 2 being

an abnormal condition of a single nostril on

one side but two nostrils on the other (ANSP

105586, 186 mm); l.D.jaculiferus, 1 being

the normal condition of two nostrils in

each nasal sac and 2 being an abnormal

condition of a single nostril on one side

but two nostrils on the other (USNM 176952,

42.2 mm): J, Dicotylichthys punctulatus;
K, D. nicthemerus, a bifid tentacle in adults
but two nostrils in a tube in young specimens.

0 f IP '^7 ^"a

1/:^:^

Figure 294. —Ventral viewsof upper jaws to

show trituration tooth plates: A, Chilomycterus

orbicularis, 77.0 mm SL, Somalia, and 158 mm SL,

Philippines: B, C. schoepfi, 62..'j mm SL, Texas, and 168 mm SL,

Florida, the latter species with trituration

plates more or less typical of all

diodontids except C. orbicularis.

Figure 29.5. — Chilomycterus schoepfi: upper
left, lateral view of upper jaw, with inset
showing detail of lateral edge of crushing
beak with its individual dental units
incorporated into but distinct from the
bony matrix of the premaxillary; lower
left, cross section through middle of one

of the individual flattened disklike

trituration teeth; right, outline of about

one-half of a trituration tooth, with the

detail showing the irregular surface;

ie8 mm SL, Florida.

The dentition of the pharyngobranchials is also slight-
ly variable in diodontids. The pharyngobranchials of the
first and second arches always bear small to minute
teeth, while the pharyngobranchial of the third arch,
when present, is always smaller than those preceding it
and is nearly always toothless. The pharyngobranchial of
the third arch is absent in Chilomycterus affinis, atinga,
reticulatus, and tigrinus, and it is especially small and
toothless in C. antillarum, mauretanicus, orbicularis,
spinosus, and Dicotylichthys nicthemerus . In C. schoepfi
this pharyngobranchial is as small as in the preceding
species, but it sometimes bears minute teeth even though
usually it is toothless. In D. punctulatus this pharyngo-
branchial is toothless but of moderate size. In Diodon
this pharyngobranchial is toothless and of moderate
(hystrix, jaculiferus) to relatively large (holocanthus)
size.

Gill rakers along the anterior edge of the fourth cer-
atobranchial are present in Chilomycterus affinis,
atinga, reticulatus, tigrinus, orbicularis, Diodon holo-
canthus, hystrix, jaculiferus, and in Dicotylichthys nic-
themerus and punctulatus, while they are absent in C.
antennatus, antillarum, mauretanicus, schoepfi, and
spinosus. The loss of gill rakers along the anterior edge of
the fourth arch must be considered a specialization, since
in the only other family without a fourth gill, the tetrao-
dontids, rakers are always present here.

Both the dorsal and ventral hypohyals are present in
all of the diodontids examined except for Chilomycterus
atinga, reticulatus, and tigrinus, which lack (in the
single specimen examined of each species) the dorsal ele-
ment.

The trituration teeth internal to the biting edge of the
upper and lower jaws always form a large plate divided
into right and left halves, each half formed by a single
series of large wide flattened tooth plates whose medial

edges are straight and the other edges rounded, in all spe-
cies examined except for C. orbicularis. In C. orbicularis
the trituration plate is smaller than in the other species,
the division into right and left halves is less distinct, and
there are two or more series of far smaller and less wide
teeth in each half of the plate. In fact, the individual
teeth in the trituration plate of C. orbicularis are not
much larger than those of the biting edge, at least in
adults, in which the number of series of teeth increases
greatly over the two or three to each side found in young
specimens, while at the same time the width of the rows,
especially of the innermost row, is greatly decreased.

Having numerous series of trituration teeth to each
side of the midline and not greatly different in size from
those in the biting edge would seem to be a more
generalized condition than that of a single series of large
wide plates to each side of the midline and greatly dif-
ferent from the teeth of the biting edge. Such numerous
series of small trituration teeth are also found in the
triodontids, the most generalized gymnodonts. However,
C. orbicularis is highly specialized in certain other
features of the skull, and tetraodontids, the closest
relatives of the diodontids, are more generalized than
diodontids and usually have only a single series of small
trituration teeth to each side of the midline, when they
occur at all. Moreover, the trituration teeth in the
earliest known diodontids, in the Eocene, are large
transversely elongate units that occur in a single series to
each side of the midline, just as in all Recent species ex-
cept C. orbicularis. I suspect that the numerous tritura-
tion teeth of C. orbicularis are a de novo acquisition of
this single specialized diodontid and thus not the reten-
tion of an ancestral condition such as found in triodontids.
The numerous small trituration teeth of C. orbicularis
can be considered just as specialized in their own way as
the far larger trituration teeth in a single series to either
side of the midline found in all other diodontids.

The jaws of C. orbicularis tend to be less massive and
to have a sharper biting edge than in other diodontids,
and this, in conjunction with the differences in the tri-
turation plates, probably indicates that C. orbicularis
has a diet of less hard bodied food than do the other dio-
dontids.

In the large trituration plates as found in the great
majority of diodontids, the wide individual tooth plates
are formed at the base of the pulp cavity and have a
slightly concave smooth upper surface and a slightly con-
vex papillate lower surface. The individual plates mi-
grate toward the exposed surface of the trituration plate
to replace those worn away through use as new plates are
continuously being formed at the edge of the pulp cavity.
As they move toward the exposed trituration surface the
individual plates become increasingly flattened and
more closely cemented together. The number of individ-
ual plates in the single series to each side of the midline
usually increases greatly with increasing specimen size,

Figure 296.— Diodon hystrix: cross section of
upper (above in each set) and lower jaws just
to one side of the midline, showing the numbers

of dental units in the biting edges of the jaws
and in the trituration plates, increasing greatly
in number with increasing specimen size:
A, 88.3 mm SL, B, 228 mm SL, and C, 505 mm S

Figure 297 .—Diodon holocanthus: cross sectioi

of upper (above in each set) and lower jaws

just to one side of the midline, showing

the numbers of dental units in the biting

edges of the jaws and in the trituration

plates, increasing greatly in number with

increasing specimen size: A, 63.4 mm SL;

B, 122 mm SL; C, 183 mm SL; D. 375 mm SL.

but there is little difference between species in the num-
ber of plates at any given size, based on the limited sur-
vey made (Fig. 302).

However, the largest specimen of a diodontid exam-
ined for internal trituration teeth, a single 545 mm spec-
imen of C. reticulatus, had only about 18 plates, ap-
proximately the number to be expected from the other
species examined at sizes between about 200 and 220 mm.
The total number of plates in the series carmot be counted
externally because only about a third to a half or fewer
have any of their surfaces exposed, and the total number
can only be seen by carefully sawing through the jaw just

to one side of the midline, a laborious procedure. The
trituration surface is somewhat oblique to the broad plane
of the individual plates, and the exposed surface is
covered by several overlapping plates.

In most diodontids there are a large pair of apertures
posteromedially on the dorsal surface of the premaxil-
laries just to either side of the midline which open into
the large posterior edge of the pulp cavity, but in Chilo-
mycterus affinis and orbicularis the apertures on the dor-
sal surface of the premaxillaries are smaller, more
numerous, and less precisely arranged than in the other
diodontids. This character may be particularly prone to
change with increasing specimen size.

The numbers of vertebrae in the diodontids examined
show no differences useful in recognizing natural groups.
In Chilomycterus the total number varies modally from
18 to 23, with 10 to 13 abdominal and 8 to 10 caudal,
while in Diodon there are 19 to 21 total, with 10 to 12 ab-
dominal and usually 9 caudal. In the single specimens of

364

Figure 29S.—Chilomycterus schoepfi: cross

section of upper (above in each set) and

lower jaws just to one side of the midline,

showing the numbers of dental units in

the biting edges of the jaws and in the

trituration plates, increasing greatly

in number with increasing specimen size:

A, 97.6 mm SL and B, 166 mm SL.

the two species of Dicotylichthys examined, both had 9
caudal vertebrae, one with 11 abdominal and the other
with 12.

The uppermost pectoral fin ray in diodontids is usually
relatively short and formed of two approximately equal
unsegmented halves, but in the single specimen of Chi-
lomycterus affinis examined, a large adult of 310 mm,
this ray is a single piece of bone, the two halves that can
be expected to be found in smaller specimens apparently
having completely fused, or one of the halves having been
resorbed.

Generic relationships. — It is difficult to discuss ge-
neric relationships without first establishing a natural
division of the family into genera, and as pointed out in
the preceding section, the two most recent revisions
recognized three noncomparable genera on the basis of
nostril and spine characteristics, with each of the revi-
sions having serious limitations. The osteology of the spe-
cies examined here helps to some extent, but leaves
much left to be done.

Figure 299.—ChUomycterug antillarum: cross

section of upper (above) and lower jaws just

to one side of the midline, showing the numbers of

dental units in the biting edges of the jaws

and in the trituration plates, 116 mm SL.

Figure 300.— Chilomycterus atinga: cross
section of upper (above) and lower jaws just
0 one side of the midline, showing the numbers

of dental units in the biting edges of the
jaws and in the trituration plates, 173 mm SL.

Within Chilomycterus, by far the most speciose genus
of diodontids, three groups are recognizable. Chilomyc-
terus affinis, atinga, reticulatus, and tigrinus are unique
among the diodontids by having the third pharyngo-
branchial absent, a pitted open cup nasal apparatus (at
least as adults), no spines on the forehead and 10 caudal
fin rays. Three of these four species are also unique by
the presence of only a single hypohyal. All four species
have the prefrontal absent and similar skull shapes, the
prefrontal otherwise being absent in diodontids only in

5201 O a
• D.
DC.
■ c.

VC

▼ c,
Oc

holocanthus

Figure 301. — Chitomycterus mauretanicus: cross

section of upper (above) and lower jaws just to

one side of the midline, showing the numbers of

dental units in the biting edges of the jaws

and in the trituration plates, 92.2 mm SL.

C. orbicularis, which has a much different skull shape
and arrangement of the trituration teeth. All four spe-
cies have gill rakers along the anterior edge of the fourth
arch, as does C. orbicularis (as well as Diodon tind
Dicotylichthys), all of these species being from the Indo-
Pacific with the exception of C. affinis, from the Atlan-
tic, while the other species of Chilomycterus {anten-
natus, antillarum, mauretanicus, schoepfi, and
spinosus), all from the Atlantic, do not have gill rakers
along the anterior edge of the fourth arch. The only
species of Chilomycterus with teeth on the fifth
ceratobranchial are orbicularis and three {affinis,
reticulatus, tigrinus) of the four species with a pitted cup
nasal apparatus, no spines on the forehead and 10 caudal
rays.

The evidence indicates that C. affinis, atinga, reticu-
latus, and tigrinus are all closely related, and that C.
antennatus, antillarum, mauretanicus, schoepfi, and
spinosus are a distinct subgroup of equally closely related
species.

The Indo-Pacific C. orbicularis would seem to have its
closest relatives among the primarily Indo-Pacific former
group than to the latter Atlantic group, for it shares a few
differential characters (prefrontal absent, gill rakers
present on anterior edge of fourth arch, teeth on fifth cer-
atobranchial) with them and none with the Atlantic
group. Most of the differential features of C. affinis,
atinga, reticulatus, and tigrinus are specializations
(pitted open cup nostril, loss of prefrontal and pharyn-
gobranchial of third arch, usual presence of teeth on the
fifth ceratobranchial, and usual loss of the dorsal hypo-
hyal), and their retention of 10 caudal fin rays is the only
way in which they are more generalized than C. orbicu-
laris and the other speciose subgroup of Chilomycterus
(and other diodontids). The presence in C. orbicularis of
a toothless small pharyngobranchial of the third arch,
two hypohyals, and a nasal apparatus of a tube with two
nostrils is less specialized than in C. affinis, atinga, retic-
ulatus, and tigrinus, but its other features that distin-
guish it from them are specialized (narrowness of the
region around the palatine and anterior end of the fron-
tal, numerous rows of trituration teeth, nine caudal fin
rays), while it shares with them the specialized features
of the loss of the prefrontal and the presence of teeth on
the fifth ceratobranchial.

The exclusively Atlantic subgroup of Chilomycterus
is specialized only in the loss of gill rakers on the anterior

10 15 20 2!

Number of tritufation tooth plates i

Figure 302.— Chart showing the increased

number of trituration plates in the upper

jaw with increasing specimen size, with

the same relationship holding for the lower

jaw also, in Diodon holocanthus, D. hyslrix,

Chilomycterus schoepfi, C. mauretanicus,
C. atinga, C. antillarum, and C. reticulatus.

edge of the fourth arch and in having nine caudal fin
rays, while being more generalized than either of the
other two subgroups by always retaining a prefrontal,
and more generalized than the C. affinis, atinga, reticu-
latus. and tigrinus subgroup by the constant presence of
two hypohyals and a pharyngobranchial on the third
arch, the constant absence of teeth on the fifth
ceratobranchial, and the normal presence of a nasal tube
with two nostrils. The only specialized feature that the
exclusively Atlantic subgroup shares with C. orbicularis
is the presence of nine caudal fin rays, but otherwise the
Atlantic subgroup is far more generalized than C. orbicu-
laris.

Whether the three subgroups within Chilomycterus
should be recognized subgenerically is a matter of opin-
ion, but on a practical level I doubt that such categories
would be generally used and none are here proposed.

In summary, C. orbicularis is most closely related to
the C. affinis, atinga, reticulatus, and tigrinus subgroup
and in its own way is about as specialized as them,
probably having diverged from a common ancestral line
at a time when the nasal apparatus was still a tube with
two nostrils. The C. antennatus, antillarum, maure-
tanicus, schoepfi, and spinosus subgroup is more distant-
ly related to the other two and in most ways is more
generalized than them, probably having diverged from
the line leading to the other two subgroups at a time
when there were still 10 caudal fin rays and gill rakers on
the anterior edge of the fourth arch, with the subse-
quent loss of one of the caudal rays and of those gill
rakers while in general becoming less differentiated from
the basal stock than did the other two subgroups.

The three species of Diodon studied, two of which
(holocanthus and hystrix) have all of the spines erectile
and the other (jaculiferus) with only a few behind the
pectoral fin erectile, scarcely differ osteologically, the
prefrontal in jaculiferus being slightly smaller than in the
other two. The two species oi Dicotylichthys studied, one
(nicthemerus) with all the spines erectile and the other
(punctulatus) with only those on the head erectile, are
equally similar osteologically and can be distinguished
from Diodon internally only by the presence of a spe-
cialized well-developed band of teeth on the fifth cera-
tobranchial, this being entirely toothless in Diodon,
while externally they differ only by Diodon having a

nasal tube with two nostrils versus the specialized bi-
lobed tentacle of Dicotylichthys. I do not think that this
combination of two characters is of sufficient magnitude
to warrant generic recognition, it being no greater, and in
actuality less, than that between the three unnamed sub-
groups of Chilomycterus. It is suggested that Dico-
tylichthys be placed in the synonymy of Diodon.

Moreover, the skeletal structure of Diodon and Dicoty-
lichthys is nearly indistinguishable from that of the en-
tirely Atlantic subgroup of Chilomycterus. Both Diodon
and Dicotylichthys differ from it only by the presence of
at least a few erectile spines and of gill rakers on the
anterior edge of the fourth arch, and Dicotylichthys ad-
ditionally by the presence of teeth on the fifth cerato-
branchial, a rather unimpressive array of differences.

Diodon, Dicotylichthys, and the entirely Atlantic sub-
group of Chilomycterus have apparently diverged little
from the generalized ancestral stock. The presence of a
full covering of erectile spines without massive radiating
bases would seem to be the most generalized condition in
diodontids, as it is the closest to the normal condition in
tetraodontids. In this view the development of a huge
triradiate base to the scale, which makes it nonerectile, is
a specialization, and the species of Diodon and Dicoty-
lichthys with all of the spines erectile are more generaliz-
ed than those with only a minority of them erectile, while
by the same token Chilomycterus, all of whose species
have all of the spines with huge triradiate bases, is the
most specialized of all.

In short, Diodon and Dicotylichthys do not seem
anatomically different enough to warrant separate
generic recognition, and while they are more generalized
and easily distinguishable osteologically from the two

Chilomycterus : orbicularis

Chilomycterus: affinis, atinga,
reticulatus, tigrinus

Figure 303.— Hypothesized
phylogenetic relationships of
the genera of Diodontidae.

Dicotylichthys

Chilomycterus : antennatus, antillarum,
mauretanicus, schoepfi, spinosus

Prodiodon

more specialized subgroups of Chilomycterus (the af-
finis, atinga, reticulatus, (igrinuA- subgroup, and the or-
bicularis subgroup) they differ from the more generalized
subgroup of Chilomycterus {antennatus, antillarum,
mauretanicus, schoepfi, spinosus) only slightly, pri-
marily by retaining at least some erectile spines without
huge triradiate bases.

Relationship to the other Tetraodontoidei.— The

relationship of the Diodontidae as a specialized offshoot
of the line of gymnodonts leading to the Tetraodontidae
is discussed under the latter and under the Trio-
dontidae, with additional remarks under the Molidae.

SUPERFAMILY MOLOIDEA

Comparative diagnosis (contrast with that of the
Tetraodontoidea), which is also that of its only con-
tained family, the Molidae.— No inflatable diver-
ticulum of the gut; first branchiostegal ray without an
inturned and enlarged dorsal edge, articulated to the
ventral edge of the ceratohyal and not forming a pump-
ing plate; air bladder absent, at least in adults; four gills,
greatly expanded dorsally above and beyond the sup-
porting arches; a gill slit present between the gill bearing
fourth arch and the gill-less fifth arch, with gill rakers
present along both the posterior edge of the fourth arch
and the anterior edge of the fifth arch; gill rakers present
along the anterior edge of the first gill slit; pharyngo-
branchials with extremely large teeth, the three ele-
ments being those of the second to fourth arches; cera-
tohyal and epihyal not sutured to one another; interhyal
and dorsal hypohyal always present; teeth in biting edge
of jaws apparently indistinguishably incorporated into
the bony matrix of the preraaxillary and dentary;
basisphenoid present; caudal fin aborted, either absent
altogether or represented by only a few rays in the cen-
tral region of the pseudocaudal fin formed by posteriorly
migrated soft dorsal and anal fin rays and supported
mostly by equally posteriorly migrated basal pterygio-
phores from the soft dorsal and anal fins; dorsal, anal,
caudal, and pectoral fin rays often extensively branched
but with extremely few cross-striations, those present
only at the extreme distal ends of the rays; sesamoid ar-
ticular absent; postcleithrum with an anteriorly di-
rected process toward or over the actinosts; three acti-
nosts; supracleithrum extremely elongate, broadly ar-
ticulated over the anterior one-third to one-half of its
length with the pterotic; coracoid long and slender,
without a posterodorsal prong below the lower actinost;
operculum and suboperculum of greatly simplified struc-
ture and reduced lateral surface area; interoperculum
absent as an ossification or present as a simple slender
rod of bone not extending posteriorly beyond the level of
the epihyal; basioccipital greatly prolonged dorsally
behind the exoccipitals to border the foramen magnum
to the exclusion of the exoccipitals; exoccipitals without

Figure 304.— Range of diversity in body

form in the Molidae: Mola mola (left)

and Romania laevis (right).

condyles and not in contact with the first vertebra, which
articulates anteriorly only with the basioccipital; epiotic
with a high dome or ribbonlike posterodorsal prolonga-
tion; pterotic greatly prolonged posteriorly, well past the
level of the end of the basioccipital; bony canal for the
nerves and blood vessels running from the orbit to the
nasal region complete, entirely surrounded by the pre-
frontal; palatine receiving its main support dorsally by a
broad articulation with the ethmoid, vomer (if present),
prefrontal, and parasphenoid; no vertebrae with bifid
neural spines projecting dorsally or dorsolaterally on
each side of the neural arch; centra of the abdominal
vertebrae without any ventral or ventrolateral processes,
the ventral and ventrolateral surfaces of the centra
perfectly smooth; dorsal and anal fin rays widely
separated from their basal pterygial supports by a large
block of cartilage; scales small but the basal plates in
more or less close contact and forming a continuous
covering over the entire body, with none of the scales in
the form of prominent prickles or spines; two minute
nostrils flush with the surface of the skin; lateral line
either absent or extremely inconspicuous (not discerni-
ble to me on either cleared and stained or intact alcohol
preserved specimens at 30 magnifications).

Detailed description of Mola mola.

Material examined. — Two cleared and stained speci-
mens, 306 mm and 310 mm; two sets of jaws from large
specimens of unknown length, one dry and one alcohol
preserved.

Occipital Region.

Basioccipital. — A short column, expanded at the
anterior end and prolonged anterodorsally into a pair of
sturdy prongs much like those of the first vertebra;
broadly cartilage filled along the expanded anterior edge

Figure 305.— Mota mola: upper left, nasal

region as seen externally: lower left, scales

from upper middle region of body.

of the columnar portion; articulates through cartilage
anteroventrally with the prootics and anterodorsally with
the exoccipitals. The posterior end of the parasphenoid
overlies the anterior half of the midventral surface of the
basioccipital and is held to it by fibrous tissue, the
basioccipital remaining largely cartilaginous in the re-
gion overlain by the parasphenoid. The anterodorsally
directed prongs of the basioccipital articulate by fibrous
tissue anteriorly with the exoccipitals and posteriorly
with the bifid neural spine of the first vertebra. The
medial edges of the anterodorsally directed prongs form
the lateral walls of the foramen magnum, while the dor-
sal wall is formed by cartilage and the ventral wall by the
dorsal surface of the columnar portion of the basiocci-
pital. The rim of the round concave posterior end of the
basioccipital articulates by fibrous tissue with the rim of
the concave anterior end of the first vertebra.

Exoccipital. — A large, nearly flat plate; cartilage
filled along all of its edges of articulation with the other
cranial bones; articulates through cartilage dorsomedial
with the supraoccipital, dorsolaterally with the epiotic,
laterally with the sphenotic and pterotic, ventrally with
the prootic, and medially with the basioccipital. Along
the middle region of its medial edge the exoccipital ar-
ticulates by fibrous tissue with the anterodorsal prong of
the basioccipital. The exoccipital does not articulate

with the first vertebra, nor does it enter into the forma-
tion of the walls of the foramen magnum.

Supraoccipital. — A rounded plate ventrally, but
produced dorsally into a high and stout supraoccipital
crest; cartilage filled along all of its ventral edges; ar-
ticulates through cartilage laterally with the epiotics and
posteriorly with the exoccipitals. Anteriorly the base of
the supraoccipital is overlain by the frontals, with which
it articulates by fibrous tissue.

Otic Region.

Pterotic. — Broadly cartilage filled along its
anterior edge; articulates through cartilage antero-
laterally with the sphenotic, anteromedially on its dorsal
surface with the epiotic and exoccipital and antero-
medially on its ventral surface with the prootic. Along
most of the length of its dorsal surface the pterotic pos-
sesses an upraised flange just lateral to which the supra-
cleithrum articulates by fibrous tissue. Along most of the
length of its lateral surface a similar flange is present just
lateral to which the hyomandibular articulates by
fibrous tissue.

Sphenotic. — Cartilage filled along its medial and
anterior edges; articulates through cartilage posteriorly
with the pterotic, dorsomedially with the epiotic and ex-
occipital, and ventromedially with the prootic and
pterosphenoid. Anteriorly the sphenotic articulates by
fibrous tissue with the overlying frontal. At its postero-
dorsal edge the sphenotic possesses a posterior process
which articulates closely by fibrous tissue with the dorsal
surface of the pterotic and which supports the antero-
medial edge of the supracleithrum. Posteriorly along its
ventral surface the sphenotic articulates by fibrous tissue
with the anterior edge of the hyomandibular.

Epiotic. — A more or less rounded plate ventrally,
but prolonged posterodorsally into a spine which con-
nects by a tendon to the muscle mass of the dorsal fin;
cartilage filled along all of its ventral edges; articulates
through cartilage medially with the supraoccipital, pos-
teriorly with the exoccipital and laterally with the
sphenotic. Anteriorly the epiotic articulates by fibrous
tissue with the overlying frontal.

Prootic. — Cartilage filled along all of its edges of
articulation with the other cranial bones; articulates
through cartilage anterodorsally with the pterosphenoid,
anterolaterally with the sphenotic, posterolaterally with
the pterotic, and posteriorly with the exoccipital and
basioccipital. The anteroventral edge of the prootic ar-
ticulates by fibrous tissue with the posterodorsal wing of
the parasphenoid, while more posteriorly the medial edge
of the prootic similarly articulates with the lateral edge
of the parasphenoid. Along its anteroventral edge the
prootic is produced into a laterally expanded wing which
articulates by fibrous tissue with the anteromedial edge
of the dorsal end of the hyomandibular.

Orbital Region.

Frontal. — A relatively flat plate; expanded pos-
terolaterally to overlie and articulate by fibrous tissue
with the supraoccipital, epiotic, and sphenotic. The
medial edges of the two frontals are broadly concave and
enclose between themselves the anterior portion of the
cartilaginous tissue of the otic and occipital regions. The
dorsomedial edges of the two frontals articulate with one
another by fibrous tissue. Along its posteroventral edge
the frontal overlies and articulates by fibrous tissue
with the pterosphenoid and sphenotic. Anteriorly the
rounded end of the frontal broadly overlies and artic-
ulates by fibrous tissue with the ethmoid and prefron-
tal.

Prefrontal. — A more or less vertical column pos-
terolaterally, but expanded into a thin vertical sheet
anteriorly; cartilage filled along the middle region of its
medial surface; articulates by fibrous tissue posterodor-
sally with the overlying frontal, anteromedially with the
ethmoid, and anterolaterally with the palatine. A canal
for the olfactory nei-ve and vessels is present through the
substance of the prefrontal, rather than being present at
the interface of the prefrontal, frontal, and ethmoid car-

Basisphenoid. — A large rounded plate anteriorly,
but laterally expanded posteriorly into a process whose
concave posterior surface articulates by fibrous tissue
with the medial edge of the anteroventral region of the
prootic. The ventral edge of the basisphenoid articulates
by fibrous tissue with the longitudinal concavity on the
dorsal surface of the parasphenoid. The anterior and dor-
sal edges of the upper half of the basisphenoid are carti-
lage filled. Dorsally the basisphenoid articulates by
fibrous tissue with the ventral edge of the ptero-
sphenoid. A small gap is present between the otherwise
closely apposed edges of the pterosphenoid and basi-
sphenoid which serves as a foramen for the exit of vessels
from the cranial cavity.

Ethmoid Region.

Ethmoid. — Thick and squarish; continuous pos-
teroventrally with the remains of the ethmoid cartilage;
articulates by fibrous tissue along the anterior half of its
ventral surface with the vomer and along the posterior
half of its ventral surface with the dorsal surface of the
concave anterior end of the parasphenoid. The lateral
surfaces of the ethmoid articulate by fibrous tissue ante-
riorly with the palatines and posteriorly with the pre-
frontals.

Parasphenoid. — Elongate; expanded ventrally
along most of its length into a stout, but not particularly
deep, flange. About two-thirds the way back its length
the parasphenoid gives rise to a pair of short posterodor-
sally directed wings which articulate by fibrous tissue
with the anteroventral edges of the prootics. Posterior to
these dorsal wings, the parasphenoid articulates by fi-
brous tissue laterally with the posteromedial edges of the
prootics and posteriorly with the anteromedial surface of
the basioccipital. Just anterior to its dorsal wings, the
dorsal surface of the parasphenoid becomes concave for a
short distance to receive and hold by fibrous tissue the
ventral edge of the basisphenoid. Along the lateral sur-
face of about the anterior one-fourth of its length the
parasphenoid articulates by fibrous tissue with the thin
posterior portion of the palatine. The anterior end of the
parasphenoid is deeply concave and filled with gela-
tinous tissue. The posterior end of the vomer does not fit
into this concavity; rather, the ventral edge of the con-
cave anterior end of the parasphenoid supports by
fibrous tissue the posterior edge of the vomer. The dorsal
portion of the concave anterior edge of the parasphenoid
articulates by fibrous tissue with the ventral surface of
the ethmoid.

Pterosphenoid. —A relatively flat plate; cartilage
filled along all of its edges, except medially; articulates
through cartilage posteroventrally with the prootic and
posterodorsally with the sphenotic. Dorsally the
pterosphenoid is overlain by the frontal, with which it ar-
ticulates by fibrous tissue. The ventral edge of the
pterosphenoid articulates by fibrous tissue with the dor-
sal edge of the basisphenoid.

Vomer.— A small squarish bone articulated by
fibrous tissue dorsally with the anterior half of the ven-
tral surface of the ethmoid, laterally with the medial
edges of the palatines and posteriorly with the ventral
edge of the concave anterior end of the parasphenoid.
The anterior edges of both the vomer and ethmoid ar-
ticulate by fibrous tissue with the posteromedial edge of
the upper jaw.

Mandibular Region.

Hyomandibular. — Expanded posterodorsally, but
tapering anteroventrally to a thick rounded shaft; carti-
lage filled at its anteroventral edge and along the ante-
rior one-third of its dorsal edge. The hyomandibular ar-
ticulates by fibrous tissue along most of the length of its
dorsal edge with the ventrolateral surface of the pterotic,
but anterodorsally it is also supported by the medial por-
tion of the posteroventral edge of the sphenotic and by
the ventrolateral flange of the prootic. Along the ventral
half of its posterior edge the hyomandibular articulates
by fibrous tissue with the posterior half of the dorsal edge
of the preoperculum. Dorsally along its anteroventral end
the hyomandibular articulates by fibrous tissue with the
posteroventral edge of the metapterygoid. Just above the
posterodorsal end of the preoperculum, the hyomandi-
bular possesses a thickened area whose concave poste-
rior face articulates by fibrous tissue with the concave
anterior end of the operculum.

Quadrate. — Wide posteriorly, but tapering to a
knob anteriorly for articulation with the articular in the
lower jaw; a short posteriorly directed process present

370

from its posteroventral region; slightly cartilage filled
along its posterior edge; articulates by fibrous tissue
anterodorsally with the ectopterygoid and ventrally with
the preoperculum. Posteriorly the quadrate articulates
through cartilage with the metapterygoid. The inner sur-
face of the posteroventral portion of the quadrate that
overlies the symplectic forms a concave groove in which
the symplectic is held by fibrous tissue. The articulation
of the knoblike anterior end of the quadrate with the ar-
ticular in the lower jaw is strengthened by the fibrous tis-
sue attachment of an anteromedial projection from the
anterior edge of the quadrate with the posteromedial
edge of the articular.

Metapterygoid. — Broad anteriorly, but tapering to
a point posteriorly; cartilage filled along its anterior and
anteroventral edges; articulates through cartilage ante-
riorly with the quadrate and by fibrous tissue anteroven-
trally with the symplectic and posteroventrally with the
hyomandibular. The cartilaginous region between the
hyomandibular, metapterygoid, and symplectic is the
place of fibrous tissue articulation of the dorsal end of the
interhyal. The anterodorsal edge of the metapterygoid
articulates by fibrous tissue with the posterior end of the
mesopterygoid.

Symplectic. — A long rod, slightly wider posteriorly
than anteriorly; cartilage filled at its posterior and ante-
rior edges; articulates by fibrous tissue dorsally with the
metapterygoid and anteroventrally with the concavity on
the inner surface of the quadrate.

Palato-Pterygoid Region.

Palatine. — Wide anteriorly, but becoming a flat
vertical plate posteriorly; articulates by fibrous tissue
anteromedially with the ethmoid and vomer, postero-
medially with the parasphenoid, ventrally with the ec-
topterygoid and mesopterygoid, and posterodorsally with
the prefrontal. The anterodorsal edge of the palatine,
which is produced medially as a slender prong, articu-
lates by fibrous tissue with the posterodorsal edge of the
maxillary.

Ectopterygoid. — Only slightly V-shaped, with the
apex directed posteriorly; articulates by fibrous tissue
dorsally with the slightly overlying palatine, ventrally
with the quadrate, and posteriorly with the mesoptery-
goid, which it slightly overlies.

Mesopterygoid. — More or less squarish; articu-
lates by fibrous tissue anteriorly with the slightly overly-
ing palatine and ventrally with the cartilaginous region
between the metapterygoid and quadrate.

Opercular Region.

Operculum. — A straight shaft of bone, becoming
laterally compressed posteriorly; articulates at its con-
cave anterior edge by fibrous tissue with the concave

facet on the posterior edge of the hyomandibular, while
ventrally it overlies and articulates by fibrous tissue with
the suboperculum.

Suboperculum. — Lx)ng and flat; widest in about the
middle of its length and becoming much thinner both
anteriorly and posteriorly; articulates by fibrous tissue
posteriorly with the overlying operculum and anteriorly
with the long ligament in which the interoperculum is
embedded.

Interoperculum. — A long delicate needle of bone
entirely embedded within the ligament which runs
between the angular in the lower jaw and the anterior
end of the suboperculum. At the level of the posterior end
of the interoperculum, the ligament is particularly firm-
ly held to the general connective tissue around the lat-
eral surface of the epihyal.

Preoperculum. — More or less flat throughout its
length, except along its dorsal edge where it is slightly ex-
panded to form a broadened surface for fibrous tissue ar-
ticulation posteriorly with the hyomandibular and ante-
riorly with the quadrate. In about the middle of its length
the dorsal edge of the preoperculum articulates by
fibrous tissue with the cartilaginous region between the
hyomandibular, metapterygoid, and symplectic.

Upper Jaw.

Premaxillary. — The two premaxillaries ars indis-
tinguishably fused in the midline and together with the
fused teeth form a large nibbling plate; anterior edge of
premaxillary forming about the dorsal two-thirds of the
border of the upper jaw, the lower third of the border be-
ing formed by the anterior edge of the maxillary. The
premaxillaries articulate laterally by fibrous tissue with
the broadly overlying maxillaries. Dorsally, however, the
expanded upper ends of the maxillaries fit into deep con-
cavities in the surface of the premaxillaries and are held
there by fibrous tissue. The premaxillaries thus broadly
overlie the dorsal ends of the maxillaries. The posterior
edge of the fused premaxillaries is overlain by the maxil-
laries along all of its length, except for a very short dis-
tance medially. The pulp cavity of the fused premaxil-
laries is small, shallow, and confined to the dorsal region
of the bone. The dorsal roof over the pulp cavity is thin
and smooth internally, but the ventral floor of the cavity
is thick and highly irregular. The pulp material extends
into and between the innumerable bony lamellae and
spikes rising from the floor of the eavity. No discrete
teeth or dental units of any kind are visible there in
either of the two cleared and stained study specimens, or
in the two additional sets of much larger jaws, nor can
any such structures be distinguished in the hard sub-
stance of the biting edge of the fused premaxillaries. A
large trituration plate is present on the medial third of
the under surface of the fused premaxillaries. In the two
relatively small cleared and stained specimens, the tri-
turation plate is distinctly divided into closely apposed

right and left halves, each of which contains four or five
elongate teeth. The teeth are set in deep sockets and re-
main distinct and separate from one another but become
increasingly intimately held to the bony matrix ante-
riorly. As an old tooth at the anterior end of the tritura-
tion plate is worn away or resorbed, a new tooth devel-
ops in a new socket at the posterior end of the plate. In
the two sets of jaws from specimens apparently at least
several times as large as the two cleared and stained
specimens, the teeth of the trituration plate are far more
numerous, less regularly arranged, relatively smaller and
placed more anteriorly on the undersurface of the pre-
maxillaries. The trituration teeth apparently undergo ex-
tensive change with increasing size.

Maxillary. —Straight and rather flattened, except
dorsally where it becomes expanded into a large rounded
head which fits into the deep concavity on the dorsal sur-
face of the premaxillary. Below the latter region the max-
illary broadly overlies and articulates by fibrous tissue
with the dorsal surface of the premaxillary. Posterodor-
sally the medial ends of the two maxillaries and the
medial edge of the fused premaxillaries articulate by a
fibrous tissue sheet with the anterior edges of the eth-
moid and vomer. The ventromedial surface of the maxil-
lary articulates by fibrous tissue with the dorsolateral
surface of the dentary.

medial surface of the dentary. Posteroventrally the
medial surface of the articular is concave and forms a
groove in which the dorsal end of the angular is held by
fibrous tissue. In about the middle of its posterior edge
the articular possesses a transverse groove which articu-
lates by fibrous tissue with the quadrate. A sesamoid ar-
ticular is absent.

Angular. — A short rod; cartilage filled at its dorsal
edge; articulates by fibi-ous tissue anteriorly with the
dentary and dorsally with the groove on the medial sur-
face of the articular. Posteriorly the angular connects
with the ligament that encloses the interoperculum.

BRANCHIAL APPARATUS.

Hyoid Arch and Branchiostegal Rays.

Hypohyals. — Dorsal and ventral hypohyals
present, the ventral slightly larger than the dorsal; the
ventral and posterior edges of the dorsal hypohyal and
the dorsal and posterior edges of the ventral hypohyal
cartilage filled; articulate through cartilage with one
another and with the anterior edge of the ceratohyal. The
anterodorsal end of the dorsal hypohyal articulates by
fibrous tissue with the middle of the lateral surface of the
basibranchial. The anteromedial edges of the hypohyals
articulate by fibrous tissue with their opposite members.

Dentary. — The two dentaries are indistinguish-
ably fused in the midline and, with the fused teeth, form
a large nibbling plate like that of the upper jaw. The pos-
teromedial surfaces of the fused dentaries are deeply con-
cave to receive and articulate by fibrous tissue with the
anterior ends of the articulars. Posteroventrally the den-
tary articulates by fibrous tissue with the anterior end of
the angular. Posterodorsally along its lateral surface the
dentary articulates by fibrous tissue with the medial sur-
face of the maxillary. The biting edge of the dentary is a
hard and apparently completely homogenous substance,
like that of the upper jaw. A trituration plate composed
of separate, but closely apposed, right and left halves is
present on the medial third of the inner surface of the
fused dentaries in the two relatively small cleared and
stained specimens. Each side contains two series of
separate and distinct teeth set in deep sockets. The inner
series contains four elongate teeth, while the other series
contains three somewhat rounded teeth. The pulp cav-
ity and replacement of the teeth of the trituration plate
are the same as described for the fused premaxillaries,
and the two sets of jaws from much larger specimens sim-
ilarly have far more numerous and smaller teeth placed
more anteriorly on the bone.

Articular. — More or less triangular in shape; car-
tilage filled for a short distance anteriorly on its medial
surface where it is continuous with the remains of
Meckel's cartilage. The tapering anterior portion of the
articular articulates by fibrous tissue with the concave

Ceratohyal. — Short, but wide; more expanded
anteriorly than posteriorly; cartilage filled at its anterior
knd posterior edges; articulates through cartilage ante-
riorly with the hypohyals and posterodorsally with the
epihyal. The posterior half of the ventral edge of the cera-
tohyal possesses two successive indentations separated
by a ventrally directed prong. The ceratohyal supports
the first three branchiostegal rays.

Epihyal.— Small; cartilage filled along its anterior
and ventral edges; articulates anteriorly and ventrally
through cartilage with the ceratohyal, while posterodor-
sally it articulates by fibrous tissue with the base of the
interhyal, and laterally supports the last few branchios-
tegal rays. The lateral surface of the epihyal articulates
by fibrous tissue with the ligament enclosing the inter-
operculum at the level of the posterior end of the latter.

Interhyal. — A short, thick column; deeply carti-
lage filled at both ends; articulates by fibrous tissue ven-
trally with the epihyal and dorsally with the cartilagi-
nous area between the hyomandibular, ectopterygoid,
sympletic, and preoperculum.

Branchiostegal rays. —Six in number; the first five
rays long, flat, and relatively straight; the sixth branchi-
ostegal wide for a short distance anteriorly, but rapidly
tapering posteriorly to a long, thin, slender shaft. The
branchiostegal rays articulate by fibrous tissue with the
hyoid arch elements as follows: first and second branchi-
ostegals with the successive concavities on the ventral

edge of the ceratohyal; third branchiostegal with a slight
concavity on the posterior edge of the ceratohyal; fourth
branchiostegal with the cartilage between the epihyal
and ceratohyal; fifth and sixth branchiostegals with the
lateral surface of the epihyal.

Branchial Arches. —The elements are cartilage filled
at their edges of articulation with the other elements of
the series, and the articulations are usually through car-
tilage and fibrous tissue. The branchial arches are com-
posed of three basibranchials, three pairs of hypobran-
chials, five pairs of ceratobranchials, four pairs of epi-
branchials, and three pairs of pharyngobranchials. Four
gills are present, with a large slit between the fourth arch
and the lower pharyngeal. The gills are enormously
enlarged and only about the lower third of their lengths
are supported by the branchial arches; these were de-
scribed long ago by Alessandrini (1839) and most recent-
ly, in great detail, by Adeney and Hughes (1977). The
anterior edges of the upper two-thirds of their lengths are
held by fibrous tissue to the ventral surfaces of the pos-
terior end of the cranium and the first several vertebrae.

First arch. — Basi-, hypo-, cerato-, and epibran-
chial elements present. First basibranchial laterally
compressed anteriorly; about the same length as the
third basibranchial and somewhat shorter than the sec-
ond basibranchial; with a ventral keel medially on its
ventral surface; displaced forward so that it articulates
posteriorly with the second basibranchial and postero-
laterally with the first hypobranchials. First hypobran-
chial the longest of the hypobranchial elements, which
decrease in length posteriorly in the series; articulates
dorsally with the first ceratobranchial, ventrally with the
second basibranchials and anteroventrally with the first
basibranchial. First ceratobranchial slightly longer and
deeper than the other ceratobranchials, which are about
equal in length but which decrease in depth posteriorly
in the series; articulates ventrally with the first hypo-
branchial and dorsally with the first epibranchial. First
epibranchial a slender rod articulating dorsally with the
fibrous tissue sheet between the dorsal edges of the
metapterygoid and hyomandibular and the ventral edge
of the parasphenoid.

Second arch. — Basi-, hypo-, cerato-, epi-, and pha-
ryngobranchial elements present. Second basibranchial
the longest and widest of the basibranchial elements;
with a ventral keel medially on its ventral surface; artic-
ulates anteriorly with the first basibranchial, anterolat-
erally with the first hypobranchials, posterolaterally with
the second hypobranchials, and posteriorly with the
third basibranchial. Second hypobranchial only slightly
shorter and narrower than the first hypobranchial; artic-
ulates ventrally with the anterolateral edge of the third
basibranchial and posterolateral edge of the second basi-
branchial, while dorsally it articulates with the second
ceratobranchial. Second ceratobranchial articulated dor-
sally with the second epibranchial. Second epibranchial
a very narrow rod articulating dorsally with the middle of

the toothless edge of the second pharyngobranchial. Sec-
ond pharyngobranchial a rounded plate bearing about
four long teeth set in sockets along its ventral edge. The
bases of the teeth are expanded and articulate with their
sockets by fibrous tissue.

Third arch. — Basi-, hypo-, cerato-, epi-, and pha-
ryngobranchial elements present. Third basibranchial
rectangular in shape; its ventral surface without a keel;
articulates anteriorly with the second basibranchial,
anterolaterally with the second hypobranchials, and pos-
terolaterally with the dorsal ends of the third hypobran-
chials and the ventral ends of the third ceratobran-
chials, while posteriorly it supports the cartilaginous
area to which are attached the fourth and fifth cerato-
branchials. Third hypobranchial a short column of bone,
almost vertical in position and sunken below the general
connective tissue that lies between the rest of the basi-
branchials and hypobranchials; articulates dorsally with
the ventral surface of the cartilage between the postero-
lateral end of the third basibranchial and the ventral end
of the third ceratobranchial, while its ventral end con-
nects by a band of fibrous tissue with the ventral keel of
the second basibranchial and, continuing anteriorly,
with the ventral keel of the first basibranchial. Third
ceratobranchial articulated dorsally with the third epi-
branchial. Third epibranchial a stout rod; articulates
dorsally with the third pharyngobranchial. Third pha-
ryngobranchial the largest of the pharyngobranchial ele-
ments; bears about six long, sharp-pointed teeth whose
expanded bases articulate by fibrous tissue with their
sockets on the ventral edge of the pharyngobranchial; ar-
ticulates ventrally with the third epibranchial and ante-
riorly and posteriorly with, respectively, the second and
fourth pharyngobranchials.

Fourth arch. — Cerato-, epi-, and pharyngobran-
chial elements present. Fourth ceratobranchial articu-
lated ventrally with the third basibranchial and dorsally
with the fourth epibranchial. Fourth epibranchial a large
flattened plate, wider ventrally than dorsally; articu-
lates ventrally with the dorsal ends of the fourth and fifth
ceratobranchials and dorsally with the fourth pharyngo-
branchial. Fourth pharyngobranchial similar to the sec-
ond pharyngobranchial; bearing four teeth in a single

Fifth arch. — Ceratobranchial (lower pharyngeal)
element only. Fifth ceratobranchial like the others in the
series, but not as deep; toothless; articulates ventrally
with the base of the fourth ceratobranchial.

PAIRED FIN GIRDLES.

Pectoral Fin.

Supracleithrum. —Long and flat, somewhat
expanded ventrally where it overlies the cleithrum; in
position more or less parallel to the posterior end of the
skull; articulates by fibrous tissue posteromedially with

373

the cleithrum and posterodoreally with the postcleith-
rum. Anteriorly the medial edge of the supracleithrum
articulates by fibrous tissue with the entire length of the
dorsolateral surface of the pterotic and with the postero-
lateral edge of the sphenotic.

Cleithrum. —Widest dorsally, constricted at the
level of the scapula and becoming a thin flat plate ven-
trally. Above the level of the scapula, the posterior region
of the medial surface of the cleithrum articulates by
fibrous tissue dorsolaterally with the broadly overlying
supracleithrum, while the dorsal edge of the portion of
the cleithrum overlain by the supracleithrum articu-
lates with the anteroventral edge of the postcleithrum.
Along its posterior edge the cleithrum possesses a broad
but shallow concavity in which the anterodorsal edge of
the scapula is held by fibrous tissue. The posterior edge
of the cleithrum below the scapula articulates by fibrous
tissue with the anterior edge of the coracoid. The ventro-
medial surfaces of the two cleithra articulate with one
another by fibrous tissue.

Postcleithrum. — Approximately Y-shaped; with
the bifurcate portion at the level of the actinosts and the
long arm articulated by fibrous tissue anterolaterally
with the supracleithrum and anteroventrally with the
cleithrum. In neither of the two cleared and stained spec-
imens is there any evidence of the postcleithrum being
composed of two elements, the dorsal and ventral post-
cleithra apparently having fused into a single piece. The
lateral surface of the portion of the postcleithrum that
lies posterior to the cleithrum is embedded in the thick,
firmly collagenous subdermal tissue of the skin. The
medial surface of the anterior fork of the inverted Y over-
lies and articulates by fibrous tissue with the bases of the
three actinosts.

Coracoid. — A flattened plate throughout its length,
except ventrally where it becomes more rounded and
shaftlike; slightly expanded anteriorly in the middle
region of its length and more expanded anteriorly at its
dorsal end; cartilage filled at its anterodorsal edge; artic-
ulates anterodorsally through cartilage with the base of
the scapula, while its dorsal edge articulates through car-
tilage and fibrous tissue with the bases of the three ac-
tinosts. The anterior edge of the coracoid articulates by
fibrous tissue with the posterior edge of the cleithrum.

Scapula. — Widest dorsally, constricted in the mid-
dle and only becoming slightly expanded ventrally;
scapular foramen not entirely enclosed by bone, for the
anteroventral edge of the scapula is well separated from
the posterior edge of the cleithrum. A sheet of fibrous tis-
sue, bearing a small pore representing the scapular fora-
men, connects these edges of the scapula and cleithrum.
The scapula is deeply cartilage filled at its dorsal and
ventral edges. The posterodorsal edge of the scapula is
somewhat expanded posteriorly and overlies the antero-
dorsal edge of the anteriormost actinost, with which it ar-
ticulates by fibrous tissue. The posteroventral edge of the

scapula articulates by fibrous tissue with the anteroven-
tral edge of the anteriormost actinost, while ventrally the
scapula articulates through cartilage with the coracoid.

Actinosts. —Three actinosts present; decreasing in
size and degree of central constriction posteriorly in the
series. The reduced first actinost of other plectognaths is
not present as a separate piece, although it is possible
that the posterodorsal portion of the scapula represents
the remains of this actinost indistinguishably fused to
the scapula. The three actinosts articulate with one
another dorsally and ventrally by fibrous tissue and with
the coracoid through cartilage and fibrous tissue. The ac-
tinosts are cartilage filled at their dorsal and ventral

Fin rays. — Twelve fin rays present, the first three
rays and the last ray unbranched, the others so exten-
sively branched distally that up to 20 terminal segments
(incomplete quintuple dichotomies) are present on some
fin rays. The medial and lateral halves of each fin ray are
not closely apposed to one another except distally, a large
amount of connective tissue otherwise separating the two
halves for most of their lengths. For the vast majority of
their lengths the rays lack cross-striations, which are
only present extremely distally on some of the branches
as poorly defined articulations. The first ray differs from
the others by possessing a slender anteroventral process
from the anterior edge of the medial half of the fin ray.
The deeply bifurcate bases of the rays surround plugs of
cartilage whose ventral regions are calcified either as a
single piece or as a pair (lateral and medial) of calcifica-
tions (not shown in the figures).

VERTEBRAL COLUMN. —All vertebrae with bicon-
cave centra, except for the last, which ends posteriorly
with a relatively flat surface.

Abdominal Vertebrae.

First vertebra. — Neural spine bifid; no bony roof
over the neural canal. The rim of the concave anterior
face of the centrum articulates by fibrous tissue with the
rim of the concave posterior end of the basioccipital. The
anterolateral edges of the bifid neural spine articulate by
fibrous tissue with the medial surfaces of the anterodor-
sal wings of the basioccipital. The dorsal edges of the
bifid neural spine are overlain by the anterior edge of the
neural spine of the second vertebra, while its posterolat-
eral edges are indented to accommodate the short prongs
from the anterolateral edges of the neural spine of the
second vertebra.

Other abdominal vertebrae. — Eight abdominal ver-
tebrae in two specimens; neural spines increasing in
length from the second to the fifth abdominal vertebrae;
neural spines of the fifth to eighth abdominal vertebrae
about equal in length; second to eighth abdominal ver-
tebrae with bony roofs over the neural canal. The neural
arches from either side of the second abdominal verte-

bra do not fuse with one another above the neural canal,
but, rather, have their medial surfaces closely apposed
and articulated with one another by fibrous tissue. This
dorsomedial region of apposition is thickened and pro-
longed anteriorly and posteriorly into a neural spine
whose right and left halves are likewise closely apposed
medially and held together by fibrous tissue. The ven-
tral surface of the anterior prolongation of the neural
spine of the second vertebra overlies the dorsal edge of
the bifid neural spine of the first vertebra and effectively
roofs over the neural canal in this area. The posterior pro-
longation of the neural spine of the second vertebra over-
lies the anterior edge of the neural arch and base of the
neural spine of the third vertebra. The neural arches and
spines of the third and fourth abdominal vertebrae are
likewise only closely apposed, with fibrous tissue hold-
ing the right and left halves closely together. The neural
spines of the third and fourth abdominal vertebrae are
prolonged only posteriorly, and overlie the neural arch
and spine of the vertebra just posterior to them. The pos-
terodorsal end of the neural spine of the fourth abdom-
inal vertebra articulates by fibrous tissue with the ven-
tral edge of the first dorsal fin basal pterygiophore. The
right and left sides of the neural arches of the fifth to
eighth abdominal vertebrae fuse to one another in the
midline above the neural canal and are prolonged pos-
terodorsally into long, stout, undivided neural spines.
The lower posterior edges of these neural spines overlie
the anterior edges of the neural arches of the vertebra
just posterior to them. More distally the neural spines of
the fifth to eighth abdominal vertebrae articulate by
fibrous tissue between the more anterior of the dorsal fin
basal pterygiophores.

Caudal Vertebrae. — Nine caudal vertebrae in two
specimens. With the exception of the last two vertebrae,
all of the caudal vertebrae have well-developed and un-
divided neural and haemal spines. The neural and hae-
mal spines decrease slightly in stoutness, but not partic-
ularly in length, posteriorly in the series. The degree of
anteroposterior expansion of the neural arch and base of
the neural spine decreases from the first to the sixth cau-
dal vertebrae so that the neural spines of the sixth and
seventh caudal vertebrae are slender shafts throughout
their entire lengths. The neural spines of the first to sev-
enth caudal vertebrae support by fibrous tissue the more
posterior of the dorsal fin basal pterygiophores. The dis-
tal three-fourths of the haemal spines of the first and sec-
ond caudal vertebrae articulate with one another by fi-
brous tissue and the thick shaft thus formed is overlain
laterally and anteriorly by the deeply concave posterior
surface of the enlarged first anal fin basal pterygiophore.
The posterior edge of the haemal spine of the second cau-
dal vertebra articulates with the anterior edge of the sec-
ond anal fin basal pterygiophore. The haemal spines of
the third to seventh caudal vertebrae support the other
anal fin basal pterygiophores. The size of the neural and
haemal canals through the neural and haemal arches of
the caudal vertebrae decreases posteriorly in the series
until these canals are very narrow in the seventh caudal

vertebra. The short neural spine of the eighth caudal ver-
tebra arises directly from the centrum and contains no
canal whatsoever. The haemal process, however, of the
eighth caudal vertebra contains an extremely narrow
canal through its substance in one of the specimens but
not in the other. The neural and haemal spines of the
eighth caudal vertebra curve anteriorly and their distal
ends articulate with, respectively, the posterior edges of
the neural spine and of the haemal spine of the seventh
caudal vertebra. The ninth caudal vertebra is a simple
shaft which ends posteriorly with a flat surface abutting
against a calcified cartilage, explained below.

DORSAL. ANAL, AND PSEUDOCAUDAL FINS. —

While a larval caudal fin fold without rays is present in at
least the molids Masturus and Ranzania, this is lost in
development and the adult pseudocaudal fin in molids is
apparently a secondary formation from dorsal and anal
fin rays (Gudger 1937a, b; Raven 1939a; Tyler 1970b).
The fin rays of the pseudocaudal fin are continuous with
those of the dorsal and anal fin series and are supported
basally by modified basal pterygiophores of the dorsal
and anal fins. Only an arbitrary distinction can be made
between the dorsal, pseudocaudal, and anal fins, but
such a distinction is perhaps advisable for the sake of dis-
cussion. The pseudocaudal fin, composed of postero-
medially migrated dorsal and anal fin rays, is here con-
sidered to include all those fin rays supported by the
modified basal pterygiophores posterior to the neural and
haemal spines of the seventh caudal vertebra.

The 15 dorsal fin basal pterygiophores and the 9 anal
fin basal pterygiophores anterior to the seventh caudal
vertebra have a normal relationship with the neural and
haemal spines of the vertebrae to which they articulate.
The ventral ends of these dorsal fin basal pterygiophores
are cartilage filled while the distal ends of all but the first
dorsal fin basal pterygiophore are irregularly cartilage
filled and closely apposed to the ventral surface of the
large cartilaginous plate that intervenes between them
and the dorsal fin rays. The first dorsal fin basal pteryg-
iophore tapers to a point distally and articulates by fi-
brous tissue posteriorly with the neural spine of the fifth
abdominal vertebra and with the anterior edge of the sec-
ond basal pterygiophore. Along the dorsal third of its
anterior edge the first basal pterygiophore articulates by
fibrous tissue, and possibly slight interdigitation, with
the posteroventral edge of the supraneural. The supra-
neural tapers to a blunt point anteriorly and is embedded
between the right and left muscle masses of the dorsal
fin. The large cartilaginous plate that intervenes between
the basal pterygiophores and the dorsal fin rays is calci-
fied irregularly in the two cleared and stained specimens.
The posterior one-third of the plate is the best calcified
portion, but calcareous areas are also present at the ante-
rior end and along much of the ventral portion of the
plate. The plate is homogenous posteriorly, but shows
some evidence anteriorly of being composed of succes-
sive cartilaginous blocks closely applied to one another.
The lateral surface of the plate possesses upraised areas
between which run the ligaments from the fin rays to the

dorsal fin muscle mass. There are 18 dorsal fin rays pres-
ent anterior to the first dorsal fin ray supported by a
modified dorsal fin basal pterygiophore posterior to the
neural spine of the seventh caudal vertebra. The first five
dorsal fin rays are unbranched, and the sixth ray is only
slightly branched at its extreme distal end. The 7th to
16th fin rays are extensively branched distally, so much
so that some of the rays have over 50 terminal segments
(incomplete sextuple dichotomies). The right and left
halves of each of the rays are well-separated by fibrous
tissue from one another medially throughout their
lengths, except distally where they begin to branch. As
with the pectoral fin rays, cross-striations are only pres-
ent extremely distally on some of the branches as poorly
defined articulations. The bifurcate bases of the dorsal
fin rays enclose plugs of calcified cartilage which articu-
late by fibrous tissue with the cartilaginous plate below
them. There are thus three series of pterygial elements
supptorting the dorsal fin rays: 1) the ossified basal
pterygiophores; 2) the partially calcified cartilaginous
plate representing the closely apposed or confluent
medial pterygiophores; and 3) the calcified cartilagi-
nous plugs or distal pterygiophores at the bifurcate bases
of the fin rays.

Except for the anteriormost pterygiophore, the nine
anal fin basal pterygiophores are like those of the dorsal
fin, except longer. They have concave proximal ends
which are cartilage filled and they articulate by fibrous
tissue with the haemal spines of the caudal vertebrae and
with one another. Their distal ends are irregularly carti-
lage filled and closely apposed or continuous with the
partially calcified cartilaginous plate that intervenes
between them and the anal fin rays. The first anal fin
basal pterygiophore is deeply concave along the poste-
rior surface of the upper two-thirds of its length where it
overlies and articulates by fibrous tissue with the hae-
mal spines of the first two caudal vertebrae. Distally the
first basal pterygiophore is bifurcate into two parts which
fit against the medial pterygial plate. The medial ptery-
gial plate is, like the dorsal fin medial pterygial plate,
irregularly calcified. Sixteen anal fin rays are present
anterior to the first anal fin ray supported by a modified
basal pterygiophore posterior to the haemal spine of the
seventh caudal vertebra. The first five anal fin rays are
unbranched, and the sixth is only slightly branched dis-
tally. The 7th to 15th anal fin rays are branched in the
same extensive manner as the corresponding rays of the
dorsal fin. The 16th anal fin ray is much less branched
than those anterior to it. The anal fin rays have the same
structure as those of the dorsal fin and likewise have
plugs of calcified cartilage, or distal pterygiophores,
embedded between their bifurcate bases.

Posterior to the neural spine of the seventh caudal ver-
tebra there are seven modified dorsal fin basal pterygio-
phores which distally support the seven dorsal fin rays of
the upper half of the pseudocaudal fin. The uppermost, '
or first, modified basal pterygiophore is placed at about a
right angle to the vertebral column, but the second to
seventh modified basal pterygiophores are progressively
more obliquely placed, until the seventh is almost

parallel to the vertebral column. Except for the first,
these modified basal pterygiophores are cartilage filled at
both ends and articulate by fibrous tissue proximally
with the neural spines of the seventh or eighth caudal
vertebrae and distally with the blocks of calcified carti-
lage that form the medial pterygiophores of the pseudo-
caudal fin. The proximal ends of the second to fifth mod-
ified basal pterygiophores articulate with the posterior
edge of the neural spine of the seventh caudal vertebra,
while the proximal ends of the sixth and seventh modi-
fied basal pterygiophores articulate with the posterior
edge of the neural spine of the eighth caudal vertebra.
The proximal end of the first modified basal pterygio-
phore tapers to a narrow shaft articulated by fibrous tis-
sue with the posterior edge of the last unmodified dorsal
fin basal pterygiophore. The distal end of the first modi-
fied basal pterygiophore articulates with the postero-
ventral edge of the large, partially calcified, cartilagi-
nous mass that makes up the medial pterygiophore of the
dorsal fin. The medial pterygiophore above the second
modified basal pterygiophore is also continuous with the
medial pterygiophore of the dorsal fin, being delimited
from it only by a constricted region of cartilage. The cal-
cified cartilages of the medial pterygiophores at the dis-
tal ends of the third to sixth modified basal pterygio-
phores remain separate from one another in the two
cleared and stained specimens. The medial pterygio-
phore of the modified seventh basal pterygiophore is con-
nected with the cartilage at the end of the ninth caudal
vertebra and with the medial pterygiophore of the eighth
modified anal fin basal pterygiophore. It is to be expect-
ed that in larger specimens more of the medial pterygio-
phores of the pseudocaudal fin become continuous with
one another, because even in the two relatively small
study specimens, the short, thin processes of the fifth
and sixth medial pterygiophores almost make contact
with the medial pterygiophores below them. The distal
pterygiophores of the pseudocaudal fin, whose postero-
lateral surfaces are overlain by the bifurcate bases of the
fin rays, are blocks of calcified cartilage which increase in
size in the series from that distal to the first to that dis-
tal to the seventh modified basal pterygiophore. The dis-
tal pterygiophores are thickest at their ends which artic-
ulate by fibrous tissue with the medial pterygiophores.
The seven fin rays supported by the distal pterygio-
phores increase slightly in length in the series from that
attached to the first to that attached to the seventh
pterygiophore. The fin ray above the first modified basal
pterygiophore is branched in a double dichotomy, but
the other rays of the pseudocaudal fin are simply divided
into two terminal segments. None of the fin rays are
cross-striated. In the two study specimens, bony ossicles
are just beginning to be evident around the bifurcate dis-
tal ends of a few of the more medial rays of the pseudo-
caudal fin, these ossicles being well developed in adults.
The anal fin portion of the pseudocaudal fin has basi-
cally the same structure as the dorsal fin portion. How-
ever, since eight modified anal fin basal pterygiophores
are present, rather than seven modified basal pterygio-
phores as in the dorsal portion, there are eight anal fin

rays that can be assigned to the anal portion of the pseu-
docaudal fin. The lowermost, or first, modified anal fin
basal pterygiophore is placed at about a right angle to
the vertebral column, but the second to eighth modified
basal pterygiophores are progressively more obliquely
placed so that the eighth is placed about parallel to the
vertebral column. The first modified basal pterygio-
phore articulates distally with the inner edge of the car-
tilaginous plate that forms the medial pterygiophore of
the anal fin, while proximally it articulates by fibrous tis-
sue with the posterior edge of the last unmodified anal fin
basal pterygiophore. The second modified basal pteryg-
iophore ends distally at its medial pterygiophore, which
is continuous with the medial pterygiophore of the anal
fin and is delimited from it only by a slightly constricted
region of cartilage. The anterior edge of the second mod-
ified basal pterygiophore articulates proximally by
fibrous tissue with the posterior edge of the first modi-
fied basal pterygiophore. The third to fifth modified
basal pterygiophores articulate proximally by fibrous tis-
sue with the posterior edge of the haemal spine of the
seventh caudal vertebra, while the sixth to ninth modi-
fied basal pterygiophores articulate with the posterior
edge of the haemal spine of the eighth caudal vertebra.
Distally the modified basal pterygiophores connect with
the calcified cartilage of their medial pterygiophores.
The medial pterygiophores distal to the third to the sev-
enth modified basal pterygiophores are separate from
one another, but that distal to the eighth modified basal
pterygiophore is continuous with the calcified cartilage
at the end of the ninth caudal vertebra and with the
medial pterygiophore distal to the seventh modified
basal pterygiophore of the dorsal portion of the pseudo-
caudal fin. The distal pterygiophores of calcified carti-
lage in the anal portion of the pseudocaudal fin have the
same structure as those of the dorsal portion. The lower-
most fin ray of the anal portion of the pseudocaudal fin is
branched in a double dichotomy, but the other seven fin
rays of the anal portion are simply divided distally into
two terminal segments. None of the rays are cross-
striated, and bony ossicles have just begun to form, just
as in the dorsal portion, in the two study specimens.

The pseudocaudal fin, as here defined, thus consists of
15 fin rays, the uppermost ray and lowermost ray of
which are branched in double dichotomies, while the
other 13 rays are simply bifurcate distally. The upper
seven rays belong to the dorsal fin and are supported by
seven bony modified dorsal fin basal pterygiophores, as
well as by calcified cartilages which form a medial and a
distal series. The lower eight rays belong to the anal fin
and are supported by eight bony modified anal fin basal
pterygiophores, as well as by calcified cartilages which
form a medial and a distal series. The dorsal and anal fin
rays form a continuous series around the posterior end of
the body, with those dorsal to the ninth caudal vertebra
being considered as dorsal fin rays and those ventral to
that vertebra being considered as anal fin rays. The divi-
sion of the dorsal and anal fin rays into those of the dor-
sal and anal fins proper and those of the pseudocaudal
fin is arbitrary, but convenient.

Anatomical diversity. —While three genera of molids
are presently recognized, there is some doubt regarding
the number of species in two of them, as most recently
revised (Fraser-Brunner 1951). Ranzania is monotypic
{laevis = truncatus). Masturus is probably monotypic,
Fraser-Brunner stating that the two forms he tenta-
tively recognized as specifically distinct were probably
only the sexes of a single species (lanceolatus =oxyu-
ropterus). Fraser-Brunner believed that Mola was repre-
sented by two species, the worldwide M. mola being
largely replaced in the south Pacific by M. ramsayi, the
latter differing from M. mola mainly in having several
more fin rays in the pseudocaudal fin, larger bony ossi-
cles at the distal ends of these rays, and no band of
smaller scales along the base of the pseudocaudal fin.

Subsequent to Fraser-Brunner's (1951) revision, a new
genus and species of molid has been described,
Pseudomola lassarati Cadenat (1959), based on a single
specimen from west Africa. The specimen was thought to
be unusual by the presence of a large white abdominal
region and additional white spots and reticulations. It
otherwise was said to differ from Mola only in lacking, at
its large size (1,090 mm total length), bony ossicles at the
distal ends of at least some of the rays of the pseudo-
caudal fin and to differ from Masturus only by lacking
the posteriorly prolonged lobe of the pseudocaudal fin.
This lobe is often lost (bitten off?) in Masturus and even-
ly healed over, while little is known about variation in
the time of development of the ossicles in Mola.
Pseudomola lassarati could easily be a Mola mola that
has not yet developed the bony ossicles along the edge of
the pseudocaudal fin or a Masturus lanceolatus that has
lost the posterior lobe of the pseudocaudal fin. However,
some specimens of M. lanceolatus are known to have a
white abdominal region and additional white spots or
blotches (see photographs in Gudger 1937a, b) very simi-
lar to those of P. lassarati, and it is probably with M.
lanceolatus that P. lassarati is synonymous.

A fossil species of molid, Mola pileatus (Beneden), is
known from the Miocene on the basis of the bony ossicles
that are present in the skin above and below the mouth,
just as in the Recent species of Mola (these being the
remains of the postlarval skin spines, according to
Fraser-Brunner 1951:110) and of a few fused premaxil-
laries and dentaries that are more molidlike than diodon-
tidlike (see Leriche 1907 and 1926, and contained refer-
ences, and Deinse 1953).

Fraser-Brunner (1951) distinguished Ranzania (as the
Ranzaniinae) from Mola and Masturus (Molinae) as
follows: Ranzania more elongate; vertebrae 8 -I- 10 or 11
(vs. 9 -(• 8); carapace of hexagonal plates (vs. a colla-
genous carapace); lips produced and funnellike (vs. lips
normal); gill rakers free (vs. concealed in thick skin); 5
branchiostegal rays (vs. 6); no secondary postlarval meta-
morphosis (vs. the development of prominent spiny proc-
esses from the body during a Molacanthus stage). Mas-
turus was distinguished from Mola by having the middle
fin rays of the pseudocaudal fin not supported basally
by basal pterygiophores and none of the pseudocaudal
fin rays developing bony ossicles at their distal ends.

377

^■^v

supraoccipital epiotic

pterosphenoid frontal
basisphenoid

exoccipital

cartilage
prefrontal
ethmoid
palatine
ectopterygoid

interopercuiuin

operculum
suboperculum

, . , V preoperculum ^

symplectic \ \ "^ "^

quadrate metapterygold parasphenoid /

mesopterygoid

I would slightly modify some of the above statements.
The body depth in Ramania is about 3 times in the stan-
dard length in the young, but large adults are deeper
bodied, the depth being only twice, or slightly less than
twice, in the standard length. In Mola and Masturus the
depth is usually between one and one and a half in the
standard length, with the young slightly less deep bodied
than large adults. The individual scale plates in Ran-
zania are thicker and more angular (often being hexa-
gonal) than in Mola and Masturus, whose scale plates are
more or less circular. However, the carapace of Ranzania
is relatively thin, for in Mola and Masturus the thinner
scale plates rest upon a thick layer of collagenous connec-
tive tissue. The funnellike mouth in Ramania, which
when closed forms a vertical slit, is unique among the
plectognaths. The gill rakers in the specimens of Mola
and Masturus examined here, none of which are large
adults, do not have the gill rakers concealed in thick
skin, although a flap of skin from the leading edge of the
raker to the gill arch strengthens their articulation to the
arch, more so than in Ramania. It is apparently only in
large adults of Mola and Masturus that the gill rakers
become more concealed. In the four specimens of Ran-
zania either cleared and stained or radiographed, the
vertebrae were 8 -(- 10 = 18, while in two specimens of
Mola they were 8 + 9 = 17 and in two oi Masturus 8-1-8
= 16, and it would appear that eight is the normal

Figure 307.— MoJa mola: lateral view of
head, composite based on two specimens,
306 and 310 mm SL, California.

number of abdominal vertebrae in molids and that there
is a modal difference of one caudal vertebra between the
three genera. It is possible that Fraser-Brunner's counts
of nine abdominal vertebrae in Mola and Masturus were
made from radiographs and that the anterodorsal prongs
of the basioccipital were mistaken for the first vertebra.
A few additional external characters distinguish Ran-
zania from Mola and Masturus. In Ramania the pectoral
fin is relatively longer and more falcate, and fits into a
shallow concavity on the carapace when folded back,
while in Mola and Masturus the pectoral fin is shorter
and more rounded, and there is no concavity for it on the
carapace. In Ramania the rays of the pseudocaudal fin
are highly branched distally, while in Mola and
Masturus they are usually not branched in more than
single to triple dichotomies. In young Ranzania the distal
edge of the pseudocaudal fin is relatively rounded and
not set off distinctly from the dorsal and anal fins, much
as in both young and adult Mola and Masturus. In adult
Ranzania, however, the distal edge of the pseudocaudal
fin becomes relatively straight and slightly oblique, and
is distinctly set off from the dorsal and anal fins.

379

Ill

i! .-S U

III

■s ^ ;;

380

Mi

5 I 2 •= : > o

S S S S

■= f £

i|.|rH|s
I- iSc«t"

liii

III"-

T S. B "^

'•"S S

i.J.

: dorsal v
extended o
iew of hyoi

two speci.

California

m§^-

'olam
arche
latera
based
mmS

^ «^ «i2

10.-
nchi
side
posii
nd3

" £ ^ E a

1^ !i

i .2 » » f -^

Most of the internal differences between the three
genera of molids clearly indicate that Mola and
Masturus are much more closely related to one another
than either is to Ranzania, as widely supposed in the
past, mostly on the basis of external appearances.

In Mola and Masturus the mesopterygoid is a small
squarish bone which does not make contact posteriorly
with the metapterygoid, while in Ranzania it is larger
and more elongate, and contacts the metapterygoid. In
Mola and Masturus the epiotics are dome-shaped and
only moderately prolonged posterodorsally, while in Ran-
zania they are greatly elongate posterodorsally as thin
but wide bands of bone. In Mola and Masturus the
supraoccipital crest is short and broad, while in Ran-
zania it is long and narrow, extending back as far beyond
the head to contact the anterior prolongation of the first
basal pterygiophore of the dorsal fin. In Mola and Mas-
turus the sphenotic only slightly overlies the anterolat-
eral surface of the pterotic, while in Ranzania it broadly
overlies it. In Mola and Masturus the vomer is ossified,
while it is unossified in Ranzania. In Mola and Mastu-
rus the basisphenoid is placed in the posterior half
of the orbit and articulates with the prootic and ptero-
sphenoid as well as with the parasphenoid, while in Ran-
zania the basisphenoid is placed in the middle of the or-
bit and articulates with the frontals above and the pala-
tines and parasphenoid below. In Mola and Masturus the
anterodorsal prolongations from either side of the basioc-
cipital do not meet one another in the midline above the
neural canal, being separated there by the long anterior
extension of the neural spine of the first vertebra, while
in Ranzania the anterodorsal prolongations of the basioc-
cipital meet and fuse or suture in the midline above the
neural canal, being only slightly separated from one
another posteriorly by the short anterior extension of the
neural spine of the first vertebra. In Mola and Masturus
there are six branchiostegal rays, the fifth about the
same size as the fourth and the sixth much more slender
than any of the others, while in Ranzania there are only
five branchiostegals, the sixth either being lost or fused
with the fifth, which is much larger than any of the other
branchiostegals. In Mola and Masturus the cerato-
branchials are exceptionally deep bodied, but in Ran-
zania they are not. In Mola and Masturus the operculum
is shorter and wider than in Ranzania. In Mola and
Masturus the suboperculum retains a posterior prolonga-
tion behind or below the operculum as well as an ante-
rior portion directed internal to the rear of the preoper-
culum, while the suboperculum of Ranzania is repre-
sented only by the anterior portion, which is longer than
in Mola and Masturus. In Masturus the interoperculum
is unossified and in Mola it is a short rod well separated
from the suboperculum, while in Ranzania it is a long rod
which contacts the suboperculum. In Mola and Masturus
none of the actinosts are sutured to one another, while in
Ranzania all of them are sutured or interdigitated to one
another to some extent. In Mola and Masturus the post-
cleithrum is not greatly expanded posteriorly as a
flattened plate, while it is in Ranzania. In Mola and
Masturus the parasphenoid is about as wide anteriorly as

it is toward the rear of the orbit, while in Ranzania it is
narrower anteriorly than more posteriorly. In Mola and
Masturus the posterior portion of the palatine articu-
lates broadly to the lateral surface of the parasphenoid,
while in Ranzania it articulates more to the dorsal sur-
face of the parasphenoid. In Mola and Masturus the
symplectic is of moderate length, extending back poste-
riorly only about half the length of the ventral edge of the
metapterygoid, while in Ranzania it is elongate and ex-
tends back nearly the whole length of the ventral edge of
the metapterygoid. In Mola and Masturus the neural
arch and spine of the second vertebra is prolonged ante-
riorly and posteriorly well beyond the levels of the ante-
rior and posterior edges of the centrum, while in Ran-
zania they are not prolonged. In Mola and Masturus the
neural arches of the abdominal vertebrae are not inter-
digitated with one another, while in Ranzania they are
broadly and deeply interdigitated. In association with
their deeper bodies, the basal pterygiophores of the dor-
sal and anal fins in Mola and Masturus are less elongate,
and the neural and haemal spines and the pterygio-
phores they support are less obliquely placed than in
Ranzania. In Mola and Masturus only the first two
haemal spines of the caudal vertebrae are decidedly
oriented toward one another, while in Ranzania most of
the successive sets of two haemal spines from the caudal
vertebrae are decidedly oriented toward one another.
The first basal pterygiophore of the dorsal fin is a rela-
tively simple shaft in Mola and Masturus, and it is
broadly in contact with the second basal pterygiophore,
while in Ranzania the first basal pterygiophore is a com-
plex element obviously representing the fusion of two
basal pterygiophores, and it is not in contact with the se-
cond basal pterygiophore. A short but deep supraneural
is present in Mola, but the similarly shaped region
anterodorsally on the first basal pterygiophore of the dor-
sal fin in the single specimen of Masturus studied is fully
fused to the pterygiophore. In Ranzania the long ante-
rior extension from the complex first basal pterygio-
phore may represent a supraneural fully fused to the
pterygiophore. In Mola and Masturus the posterior sur-
face of the proximal end of the first basal pterygiophore
of the anal fin is deeply concave and encloses the distal
end of the haemal spines of the first two caudal
vertebrae, while in Ranzania this pterygiophore is not
concave proximally and articulates along the anterior
edge of the haemal spine of the first caudal vertebra. In
Mola and Masturus most of the basal pterygiophores of
the dorsal and anal fins are anteroposteriorly expanded
and platelike, while in Ranzania they are mostly long
slender rods. In Mola and Masturus the bases of the first
dorsal and anal fin rays are only slightly, if at all, larger
than those just behind them, while in Ranzania these
bases are greatly enlarged anteriorly. In Mola and
Masturus the ossification of the skeleton is less complete
than in Ranzania.

In only a few ways are either Mola or Masturus more
like Ranzania than one another. The trituration teeth in
the upper jaw of all three genera occur as a single series of
elongate teeth to either side of the midline. In the lower

$»-4th abdominal vertebrae

Figure 317. — Maaturus lanceolatus: ventral

(left) and dorsal (right) views of skull,

127 mm SL. Flonda.

Figure 318.— Ra
laevU: dorsal (left) and
ventral (right) views of skull.
65.1 mm SL, Hawaii.

Figure 3\9.—Masturus lanceolatua: lateral
view of head, 127 mm SL, Florida.

Figure 320.— Ranzania laevi*: lateral
view of head, 65.1 mm SL, Hawaii.

basibranchials
hypobranchials

ceratobranchials
pharyngobranchials '^^^ epibranchials

i,i9^

Figure 322. — Masturus lanceolatus: dorsal
view of lower jaw, 127 mm SL, Florida.

Figure 321.— Dorsal views of branchial arches

(extended on lower side) and lateral views of hyoid

arch of: left, Ramania laevis, 66.9 mm SL, Hawaii;

right, Masturus lanceolatus, 127 mm SL, Florida.

Figure 323.— Ramania laevis: A, dorsal view (left) of

lower jaw and ventral view (right) of upper jaw,

493 mm SL, Hawaii, to show the more numerous

trituration teeth in a larger specimen,

in comparison to; B, dorsal view of lower jaw,

65.1 mm SL, Hawaii.

Figure 32i.—Ranzania laevis: diagrammatic
representation of the relationships between
the basal pterygiophores of the dorsal, anal.

and pseudocaudal fins and the neural and
haemal spines of the supporting vertebrae,
65.1 mm SL, Hawaii.

Figure 325.— iJonzonia laevu: scales

from upper middle region of body, to

show the approximately hexagonal scale

plates, articulated to one another by

extensive minute interdigitations

along all of their edges of contact

(greatest length of largest scale

plate 6.0 mm), 493 mm SL, Hawaii.

jaw of Mola and Ranzania the trituration teeth are less
elongate than in the upper jaw and occur in a major
series to either side of the midline as well as in a series of
smaller teeth to either side lateral to the major series. In
Masturus the trituration teeth in the lower jaw are not
divided into right and left sets, but rather has a single
major series placed medially, lateral to which are two
series of smaller teeth on each side.

In general configuration the pseudocaudal fin and its
supporting elements are much more similar in Mola and
Masturufi than in Ranzania. While a normally shaped
and placed larval caudal fin fold without rays is known to
be present in at least Masturus and Ranzania, this is lost
in development and the adult pseudocaudal fin in molids
is a secondary formation from posteromedially migrated
dorsal and anal fin rays and their basal pterygial sup-
ports, although it is possible that the rays in the central
nipple of the pseudocaudal fin of Masturus are remnants

of true caudal fin rays (as believed by Fraser-Brunner
1951; see Tyler 1970b for discussion).

In Ranzania the pseudocaudal fin is supported by the
last two caudal vertebrae. The last vertebra is a simple
rod seemingly representing a single centrum, which Leis
(1977) has shown to be a fusion product of two anlagen in
larval stages. The penultimate vertebra has relatively
well-developed neural and haemal spines which ante-
riorly support the last basal pterygiophores of the dor-
sal and anal fins and which posteriorly support the radial
elements of the pseudocaudal fin. Most of the radials
probably are posteromedially migrated basal pterygio-
phores (Tyler 1970b:27). There are seven radials above
and eight below in two of the three specimens of Ranza-
nia examined, but only six above and seven below in the
other (as also illustrated by Leis 1977:fig. 11), and a total
of 19 or 20 fin rays in the pseudocaudal lobe, while
Fraser-Brunner (1951) recorded 22 rays as the norm.

Contrary to Tyler (1970b:29), at least the more exten-
sively branched dorsal, anal, and pectoral fin rays of all
molids, and in Ranzania the pseudocaudal rays as well,
often bear cross-striations distally, the number of cross-
striations probably increasing with increasing specimen
size. The molid cross-striations, however, are fewer in
number and more distally placed than in the fins of other
plectognaths.

In Mola the pseudocaudal fin is supported by the last
three caudal vertebrae, of which the last is a simple rod
representing only a centrum. The penultimate vertebra
has short neural and haemal spines which curve ante-
riorly to be supported at their distal ends by the fully
developed neural and haemal spines of the antipenul-
timate vertebra, while their posterior edges support the
more medially placed radial elements. But it is the
neural and haemal spines of the antipenultimate
vertebra which support most of the radial elements.
There are seven radial elements above and eight below,
and 15 pseudocaudal fin rays, in both specimens examin-
ed. Fraser-Brunner (1951) gave 16 rays as the norm for
the south Pacific M. ramsayi and 12 for the worldwide M.
mola, with which distinction the two California
specimens of M. mola examined here are not in agree-
ment.

Masturus has one less caudal vertebra than Mola, and
it is obvious that Masturus has lost what is the last
vertebra oi Mola, the simple rod representing a centrum.
With the exception of Masturus lacking this rodlike cen-
trum the pseudocaudal supports of Mola and Masturus
are rather similar, the ultimate vertebra oi Masturus be-
ing homologous and functionally corresponding to the
penultimate of Mola, and the penultimate of Masturus
to the antipenultimate of Mola.

The most diagnostically important difference between
Masturus and Mola in this region is that Masturus has
more (20 or more) rays in the pseudocaudal fin than does
Mola, the additional rays mostly being accounted for by
those in the medial lobe that is present in Masturus but
absent in Mola, and that bony ossicles do not develop
distally on any of the pseudocaudal rays in Masturus but
do in Mola. Whereas at least most of the rays in the up-
per and lower halves of the pseudocaudal fin of both
Masturus (exclusive of the medial lobe) and Mola are in-
dividually supported by radial elements, at least most of
the rays of the medial lobe of Masturus are not supported
by radial elements but rather take support from the last
vertebral centrum. In the single specimen of Masturus
examined there are about (it being difficult to decide
where the pseudocaudal rays begin) 7 rays in the upper
half of the pseudocaudal, 10 below, and 6 in the medial
lobe, with the dorsal fin having 20 rays and the anal fin
19, a total dorsal, pseudocaudal, and anal count of 62.
Fraser-Brunner (1951) gave a total count of 60 to 62, with
eight (rarely seven or nine) rays in the medial lobe for M.
lanceolatus, and a total of 55 to 57, with four (rarely three
or five) in the medial lobe for M. oxyuropterus. Obvious-
ly, more extensive data is needed for these various fin
counts in both Mola and Masturus, and such can be easi-
ly ascertained only from cleared and stained specimens,
or from very finely radiographed specimens.

In short, Mola and Masturus differ primarily mainly in
the details of the pseudocaudal fin and its sup-
ports: 1) Mola has a ninth caudal vertebral centrum
that is lost by Masturus; 2) Masturus has a group of rays
in a medial lobe of the pseudocaudal fin that are sup-
ported by the last vertebral centrum rather than by
radials, this medial lobe not being present in Mola and
perhaps being the remnants of true caudal fin rays;

3) Mola develops bony ossicles at the distal end of many
of the pseudocaudal rays while Masturus never does; and

4) Mola has a better developed neural spine on the eighth
caudal vertebra than does Masturus. Other differences
are: 1) Mola has a supraneural that appears in
Masturus to be fully fused to the first basal pterygio-
phore of the dorsal fin; 2) Mola has the interoperculum
ossified, while it is unossified in the single small
specimen of Masturus examined; 3) the trituration teeth
in the lower jaw of Mola are divided into right and left
series, while Masturus has a major medial series.

The above differences between Mola and Masturus
seem to justify the retention of them as distinct genera,
even if they should be found to be represented by a single
species each. The far more numerous differences listed
between Ranzania and Mola-Masturus reflect in part the
ease of distinguishing between subgroups composed of
only a few species. In the present case each of the three
genera either has only a single species or only a single
species was examined for this work, so there is, in es-
sence, no intrageneric variation and the comparisons
made are only between three units. Thus, it seems to me
that the extensive differences listed between Ranzania
and Mola-Masturus do not necessarily justify the recog-
nition of two subfamilies within the family Molidae, and,
as a personal opinion, I do not think the differences
between the two groups are of sufficient magnitude for
subfamilial recognition.

Generic relationships. — With it being clear that
Mola and Masturus are far more closely related to one
another than to Ranzania, the most pertinent remaining
question is whether one or the other of these two lines of
molid evolution is more generalized than the other and
anatomically closer to the triodontidlike ancestral group.
The only other question of consequence is whether Mola
or Masturus is the more generalized within its line of
radiation.

To address the last question first. Mola is more
generalized than Masturus by: 1) the retention of the
ninth caudal vertebra, separate supremeural, and ossified
interoperculum; 2) by the better development of the
neural spine of the eighth caudal vertebra; 3) by the
retention of the trituration teeth in right and left series to
either side of the midline in the lower jaw. Conversely,
Masturus is more generalized than Mola by: 1) the
possible retention of true caudal fin rays in the medial
lobe of the pseudocaudal fin; and 2) the lack of ossicles at
the ends of any of the pseudocaudal fin rays. Overall, it
would seem that Mola has remained slightly more gener-
alized than Masturus.

Many of the differences between Ranzania and Mola-
Masturus indicate that Ranzania is more specialized

Ranzania

Figure 326.— Hypothesized phylogenelic
relationships of the genera of Molidae.

than the other two genera, these specializations often
associated with the streamlining of the body for what is
probably more rapid and sustained swimming.

The specialized features of Ranzania (i.e., the features
further removed in comparison to Mola-Masturus from
those of the hypothetical pre-Triodon ancestral group,
discussed under Triodon) are: 1) the funnellike lips,
closing to a vertical groove; 2) the falcate pectoral fins,
resting in an indented region on the carapace; 3) the
thickened, angular scale plates of the carapace; 4) the
pseudocaudal fin distinctly set off from the dorsal and
anal fins in adults; 5) the great posterior prolongation of
the epiotics and of the supraoccipital crest; 6) the un-
ossified vomer; 7) the anterior displacement of the basi-
sphenoid; 8) the greater anterodorsal prolongations of
the basioccipital, meeting in the midline over the neural
canal; 9) the strong interdigitation of the neural arches
and spines of the abdominal and more anterior caudal
vertebrae leading to greater rigidity of the vertebral
column within a slightly less flexible carapace; 10) the
loss of the sixth branchiostegal and great enlargement of
the fifth; 11) the loss of the posterior portion of the sub-
operculum; 12) the elongate, rodlike operculum; 13) the
interdigitation of the actinosts to one another; 14) the
great platelike expansion of the postcleithrum beneath
the carapace in the region of the pectoral fin; 15) the
decreased width of the anterior end of the parasphenoid;
16) the articulation of the posterior region of the palatine
with the dorsal surface of the parasphenoid; 17) the
elongation of the basal pterygiophores, and the greater
obliqueness of them and their neural and haemal spine
supports; 18) the grouping in pairs of most of the
successive sets of two haemal spines of the caudal
vertebrae; 19) the complexity of the first basal pterygio-
phore of the dorsal fin; and 20) the anterior expansion of
the bases of the first dorsal and anal fin rays.

Conversely, a lesser number of features of Romania
can be considered more generalized than in Mola-
Masturus, these being: 1) the lesser reduction in the
number of caudal vertebrae, the centrum of the 10th be-
ing retained; 2) the branching of the rays of the pseudo-
caudal fin not much less than that of the dorsal and anal
fins; 3) the better development of the interoperculum
and mesopterygoid; 4) the lack of a secondary larval

stage; 5) the lack of a deep collagenous layer under the
scales, probably associated with the better ossification of
the skeleton; 6) the gill rakers perhaps less concealed in
adults; 7) ceratobranchials not as deep bodied; and
8) the second abdominal vertebra without prominent
anterior and posterior extensions of its neural arch and
spine.

On the whole, Ranzania has far more specializations
than do Mola and Masturus, and Ranzania must be con-
sidered to be the more specialized of the two lines of
molid diversification. The fewer specializations of MoUx
and Masturus (the opposite of the generalized condi-
tions of Ranzania discussed above) can be assumed to
have arisen after the divergence of the early molids into
the two lines leading on the one hand to Mola and the
slightly more specialized Masturus and, on the other
hand, through even greater specialization, to Ranzania.

Raven (1939b), on the basis of the musculature and
skeleton, also concluded that Ranzania was more
specialized than Mola and Masturus, while Fraser-
Brunner (1951), on more superficial and fewer char-
acters, believed Ranzania, in spite of certain specializa-
tions, to be the most generalized of the molids, and
Masturus to be slightly more generalized than Mola,
primarily because of the reputed retention of true caudal
fin rays in the medial lobe of the pseudocaudal fin.

Relationships to the other Tetraodontoidei.— As dis-
cussed under the Triodontidae, molids have their closest
anatomical affinity among the gymnodonts with the trio-
dontids, retaining a number of generalized features (e.g.,
fourth gill and gill slit; basisphenoid; unmodified first
branchiostegal ray; uninflatable gut; general configura-
tion of the bones of the snout) from their triodontid
ancestry while at the same time becoming remarkably
specialized with the abortion of the rear end of the body
and the encasement of the relatively slow swimming
body in a firm or thickened skin. While molids are most
closely related to triodontids, they undoubtedly evolved
from them at a level of organization somewhat less
specialized than that of the Recent Triodon, as discussed
more fully under the latter.

Molids have sometimes been thought to be closely
related to diodontids, since both the premaxillaries and
dentaries are fully fused to their opposite members only
in these two families of gymnodonts. However, as dis-
cussed under Triodon, the fusion of the premaxillaries
and dentaries is not that complex an event and the
phylogenetic usefulness of it is meager within the gymno-
donts. While it is true that the generalized biting edge
dentition of diodontids (with numerous small units)
could have given rise to that of molids (discrete units es-
sentially absent), the structure of the rest of the head and
body of diodontids is extremely unlike that of molids,
there being no similarities such as are found between
molids and triodontids. In one other superficial respect
there is a minor similarity between molids and diodon-
tids. Diodontids have a short caudal peduncle and a
reduced number of caudal vertebrae, especially of those
posterior to the dorsal and anal fin bases, while in molids

392

this latter region is even more reduced. In light of the fact
that molids have their closest anatomical affinity with
triodontids and that diodontids and molids are
anatomically extremely different, there is every reason to
believe that the reduction in the bony structure in the
caudal region of diodontids has taken place indepen-
dently of that in molids, and that it in no way indicates

that the diodontid caudal peduncle was a precursor to
the abortion of the rear end of the body in molids. The
diodontid caudal peduncle has probably become reduced
from that found in most of the ancestral tetraodontoids
concomitant with the development of the huge spines
that guard the body in diodontids as a slow swimming
highly defensive mode of life was evolved.

Material Examined

Length in millimeters is always standard length, and,
unless otherwise stated, the Recent specimens listed
below are cleared and alizarin stained preparations. In-
stitutional abbreviations are as follows: ANSP,
Academy of Natural Sciences of Philadelphia; BMNH,
British Museum (Natural History), London; GVF,
George Vanderbilt Foundation, fishes now at the Cali-
fornia Academy of Sciences, San Francisco; JLBSII,
J.L.B. Smith Institute of Ichthyology, Grahamstown,
South Africa; SU, Stanford University, Calif, (now at the
California Academy of Sciences, San Francisco); TABL,
Tropical Atlantic Biological Laboratory, U.S. Bureau of
Commerical Fisheries (now National Marine Fisheries
Ser\'ice), Miami, Fla.; USNM, United States National
Museum (National Museum of Natural History), Wash-
ington, D.C.; WAM, Western Australian Museum,
Perth.

Recent Species
PLECTOGNATHI
Triacanthodidae

Atrophacanthus japonicus (Kamohara): ANSP 102138,
2, 9.1-10.4 mm, Celebes; ANSP 102139, 5, 35.6-43.1
mm, Celebes; BMNH 1960. 5.4.1, 1, 34.7 mm, Celebes.

Halimochirurgus alcocki Weber: USNM 93474, 1, 124
mm, Philippines.

Halimochirurgus centriscoides Alcock: ANSP 100828,
2, 99.2-113 mm, Bay of Bengal.

Hollardia hoUardi Poey: ANSP 103303, 1, 84.7 mm,
Puerto Rico; ANSP 100867, 2, 62.7-95.5 mm, Hon-
duras, ANSP 103299, 3, 72.0-129 mm, Nicaragua;
ANSP 97654, 1, 174 mm, Venezuela.

Hollardia meadi Tyler: ANSP 97396, 1, 64.5 mm.

Johnsonina eriomma Myers: ANSP 97750, 2, 75.5-85.5
mm, Panama; ANSP 103149, 2, 60.1-66.3 mm,
Panama; ANSP 103300, 2, 67.0-68.0 mm, Puerto Rico.

Macrorhamphosodes platycheilus Fowler: ANSP
100862, 1, ca. 85 mm (posterior part of body absent),
Bay of Bengal; ANSP 100865, 2, 80.3-85.3 mm. Bay of
Bengal; USNM 93482, 1, too disintegrated to measure,
Philippines.

Macrorhamphosodes uradoi (Kamohara): ANSP
101254, 2, 69.9-114 mm, Japan.

Parahollardia lineata (Longley): ANSP 100473, 1, 45.7
mm, Florida; ANSP 93375, 1, 62.2 mm, Florida; ANSP
102144, 1, 82.0 mm, Florida; ANSP 102145, 1, 86.1
mm. South Carolina; ANSP 97637, 3, 79.4-84.4 mm,
Louisiana.

Parahollardia schmidti Woods: ANSP 100866, 5, 48.7-

62.5 mm, Honduras; ANSP 103302, 2, 62.0-74.0 mm,
Nicaragua; ANSP 100128, 4, 56.8-77.8 mm, Nicara-
gua.

Paratriacanthodes retrospinis Fowler: ANSP 103285, 1,
85.3 mm, Mozambique.

Triacanthodes anomalus (Schlegel): ANSP 101257, 4,
50.6-59.7 mm, Japan; ANSP 101256, 1, 71.5 mm,
Japan; SU 49435, 1, 91.0 mm, Formosa.

Triacanthodes ethiops Alcock: USNM 93486, 2, 55.3-
63.7 mm, Philippines; ANSP 103286, 1, 76.1 mm,
Kenya.

Tydemania navigatoris Weber: ANSP 100861, 6, 58.8-
67.0 mm. Bay of Bengal; ANSP 100863, 2, 60.1-62.2
mm, Bay of Bengal; ANSP 102137, 1, 63.3 mm. Bay of
Bengal; USNM 93471, 1, ca. 95 mm, dry skeleton,
Philippines.

Triacanthidae

Pseudotriacanthus strigilifer (Cantor): SU 41732, 1,

79.0 mm, India; ANSP 89387, 4, 119-145 mm,
Thailand.

Triacanthus biaculeatus (Bloch): SU 27746, 2, 113-118
mm, Borneo; ANSP 76585, 1, 124 mm, China; SU
41730, 1, 25.6 mm, India; GVF reg. no. 2655, 3, 10.9-

19.1 mm, Thailand.

Triacanthus nieuhofi Bleeker: ANSP 102982, 1, 108
mm, possibly Australia.

Tripodichthys angustifrons (Hollard): ANSP 98719, 1,
137 mm, Australia.

Tripodichthys blochi (Bleeker): ANSP 103298, 3, 55.1-
88.9 mm, locality unknown; ANSP 103301, 3, 69.4-97.3
mm, locality unknown; ANSP 63429-54, 11, 18.4-50.9
mm, Philippines; ANSP 63476, 1, 81.0 mm, Philip-
pines; ANSP 63421-25, 5, 24.3-105 mm, Philippines;
SU 26930, 2, 92.0-107 mm, Philippines.

Tripodichthys oxycephalus (Bleeker): ANSP 102325, 1,
125 mm. Bay of Bengal.

Trixiphichthys weberi (Chaudhuri): ANSP 102136, 1,

29.6 mm. Bay of Bengal; ANSP 101389, 3, 101-119
mm, Bay of Bengal.

Balistidae

Abalistes stetlatus (Lacepede): ANSP 111538, 1, 87.6
mm, India.

Balistapus undulatus Mungo Park: ANSP 102146, 1,
124 mm, unknown western Pacific locality; ANSP
102147, 1, 122 mm, unknown western Pacific locality;
ANSP 102152, 3, ca. 120 mm, unknown western Pacific
locality, wet partially disarticulated skeletons.

Batistes capriscus Gmelin: ANSP 109437, 2, 38.4-46.1
mm, Texas; ANSP 109438, 1, 44.7 mm, Texas; ANSP
109534, 1, 127 mm, Mexico, dry partially disarticu-
lated skeleton; ANSP 109535, 1, 93.7 mm, Louisiana,
dry partially disarticulated skeleton; USNM 12983, 1,
ca. 360 mm, Gulf of Mexico.

Balistes forcipatus Gmelin: ANSP 103217, 1, 164 mm,
Guinea.

Batistes potytepis Steindachner: ANSP 88972, 1, 56.1
mm, Galapagos; ANSP 109530, 1, ca. 390 mm, Mexico.

Balistes vetuta Linnaeus: ANSP 109439, 1, 59.8 mm,
Colombia.

Balistoides conspiciltum (Bloch and Schneider): ANSP
105836, 1, 214 mm, locality unknown.

Balistoides viridescens (Bloch and Schneider): ANSP
106793, 1, 87.8 ram, Borneo.

Canthidermis maculatus (Bloch): ANSP 100085, 1,

80.1 mm, Alabama; ANSP 101364, 1, 217 mm, Mexico.
Hemibatistes bursa (Bloch and Schneider): ANSP

109441, 2, 141-160 mm, locality unknown.

Hemibalistes chrysopterus (Bloch and Schneider):
ANSP 68600, 1,90.1 mm. New Hebrides.

Melichthys niger (Bloch): ANSP 109440, 2, 147-168
mm, locality unknown.

Melichthys vidua (Solander): ANSP 109442, 1, 202
mm, locality unknown.

Odonus niger (Ruppell): ANSP 100991, 1, 173 mm.
New Guinea.

Rhinecanthus aculeatus (Linnaeus): ANSP 97373, 1,
72.8 mm, Samoa; ANSP 109443, 2, 172-207 mm, locali-
ty unknown.

Rhinecanthus rectangulus (Bloch and Schneider):
ANSP 101777, 1, 37.4 mm. Phoenix Islands; ANSP
113259, 1, 70.4 mm, Chagos Archipelago.

Rhinecanthus verrucosus (Linnaeus): ANSP 72220, 1,

98.2 mm, Borneo.

Sufflamen frenatus (Latreille): ANSP 106816, 1, 74.3
mm, Somalia; ANSP 109436, 1, 180 mm, locality un-
known.

Sufflamen verres (Gilbert and Starks): ANSP 100279,
2, 30.9-39.5 mm, Panama.

Xanthichthys lineopunctatus (Hollard): ANSP 91806,
1, 181 mm, Hawaii.

Xanthichthys ringens (Linnaeus): ANSP 74895, 1, 40.7
mm. Lesser Antilles; ANSP 113814, 2, 57.6-62.2 mm,
Bahamas.

Monacanthidae

Acanthaluteres spilomelanurus (Quoy and
Gaimard): ANSP 109810, 1, 63.7 mm, Australia.

Alutera heudetotii Hollard: ANSP 109455, 2, 107-114

mm, Florida; ANSP 109536, 2, ca. 190 mm, Florida,

wet completely disarticulated skeletons.
Alutera monoceros (Linnaeus): ANSP 111505, 2, 111-

136 mm, India.
Alutera schoepfi (Walbaum): ANSP 105131, 1, 82.8

mm, Colombia; ANSP 109454, 1, ca. 200 mm, Florida,

wet partially disarticulated skeleton.
Alutera scripta (Osbeck): ANSP 100109, 2, 46.2-73.3

mm, Florida.
Amanses scopas (Cuvier): ANSP 109739, 1, 167 mm,

Saipan.
Brachaluteres trossulus (Richardson): SU 20566, 2,

43.2-55.5 mm, Australia.
Cantherhines pardalis (Ruppell): ANSP 104784, 1, 79.4

mm, Madagascar.
Cantherhines putlus (Ranzani): ANSP 109444, 1, 56.2

mm, Bahamas; ANSP 101613, 3, 35.1-42.6 mm,

Bahamas; ANSP 97383, 1, 45.9 mm. Lesser Antilles;

ANSP 109445, 2, 68.7-83.2 mra; Louisiana.
Cantherhines sandwichiensis (Quoy and

Gaimard): ANSP 10032, 1, 92.5 mm, Hawaii.
Chaetoderma spinosissimus (Quoy and

Gaimard): ANSP 109812, 1, 33.0 ram, Malaya.
Laputa cingalensis Fraser-Brunner: ANSP 100831, 1,

51.9 mm. Bay of Bengal.
Monacanthus chinensis (Bloch): ANSP 90145, 1, 79.4

mm, Moluccas; ANSP 89557, 2, 84.4-101 mm,

Thailand.
Monacanthus ciliatus (Mitchill): ANSP 109446, 2,

51.3-58.2 mm, Colombia; ANSP 109447, 2, 75.7-81.3

mm, Florida; ANSP 109448, 2, 52.5-78.4 mm, Florida;

ANSP 109449, 3, 57.2-88.1 mm, Florida; TABL un-

catalogued {Silver Bay station 1268), 1, 74.5 mm.

North Carolina; TABL uncatalogued (Oregon station

5402), 1, 38.3 mm, Jamaica; TABL uncatalogued

(Oregon station 3589), 1, 119 mm, Panama.
Monacanthus mylii (B. de Saint-Vincent): ANSP

117147, 1, 60.8 mm. New Guinea.
Monacanthus tuckeri Bean: ANSP 84488, 3, 28.4-31.9

mm, Bahamas; ANSP 110521, 1, 70.4 mra, Baharaas.
Navodon setosus (Waite): ANSP 97904, 1, 31.0 mm.

New Zealand; ANSP 96426, 1, 69.2 mm. New Zealand.
Oxy monacanthus tongirostris (Bloch and

Schneider): ANSP 109434, 1, 26.5 mm, Seychelles;

ANSP 104808, 5, 18.0-56.7 ram, Seychelles; SU 8884, 2,

63.5-73.2 mm, Samoa.
Paraluteres prionurus (Bleeker): ANSP 102875, 1, 46.4

mm, Seychelles.
Paramonacanthus barnardi Fraser-Brunner: ANSP

109808, 1, 42.5 mm. South Africa.
Paramonacanthus cryptodon (Bleeker): ANSP 96729,

1, 40.3 mm, Philippines; ANSP 63130, 1, 68.5 mm,

Siam; USNM 169045, 2, 64.1-68.9 mm, Philippines.
Paramonacanthus curtorhynchus (Bleeker): ANSP

100821, 3, 44.7-47.4 mm. Bay of Bengal.
Pervagor metanocephalus (Bleeker): ANSP 104788, 2,

57.4-74.4 mm, Mauritius.
Pervagor spilosomus (Lay and Bennett) : ANSP 109451,

1, 103 mm, locality unknown; ANSP 109450, 3, 67.7-

83.2 mm, Hawaii; ANSP 112743, 1, 54.4 mm, Hawaii;

ANSP 84774, 4, all ca. 65 mm, Hawaii, dry partially

disarticulated skeletons.
Pseudaluteres nasicornis (Schlegel): SU 26926, 2, 50.8-

108 mm, Philippines.
Psilocephalus barbatus (Gray): ANSP 109648, 1, 142

mm, Australia; SU 35714, 2, 137-139 mm, Singapore.
Rudarius ercodes Jordan and Fowler: ANSP 29477-85,

3, 37.7-44.9 mm, Japan.
Rudarius minutus Tyler: ANSP 109788, 1, 17.9 mm,

Borneo.
Stephanolepis auratus (Castelnau): ANSP 106265, 1,

112 mm, Ghana.
Stephanolepis cirrhifer (Schlegel): ANSP 76598, 1, 50.2

mm, Hong Kong.
Stephanolepis hispidus (Linnaeus): ANSP 109452, 2,

42.8-50.4 mm, Florida; ANSP 109543, 1, 80.2 mm,

Florida.
Stephanolepis setifer (Bennett): ANSP 94610, 1, 113

mm, Bahamas.

Aracanidae

Aracana aurita (Shaw): ANSP 109572, 1, 87.3 mm,

Tasmania; ANSP 98627, 1, ca. 110 mm, Australia, dry

partially disarticulated skeleton.
Aracana flavigaster Gray: ANSP 33169, 2, 45.0-70.0

mm, Australia, dry partially disarticulated skeletons;

perhaps the female of A. ornata.
Aracana ornata Gray: ANSP 109570, 1, 70.5 mm,

Australia; ANSP 33173, 1, 67.5 mm, Australia.
Caprichthys gymnura McCulloch and Waite: WAM

P7711, 1, 74.1 mm, Australia.
Capropygia unistriata Kaup: ANSP 109575, 1, 87.9

mm, Australia.
Kentrocapros aculeatus (Houttuyn): SU 53434, 2, 90.0-

90.7 mm, Japan.
Strophiurichthys robustus Fraser-Brunner: ANSP

109569, 1, 150 mm, Australia.

89.9-107 mm, Panama; ANSP 98626, 4, 105-193 mm,

locality unknown, dry partially disarticulated

skeletons.
Lactoria cornuta (Linnaeus): ANSP 91662, 1, 43.0 mm,

Guam; ANSP 98621, 1, 114 mm, Philippines; ANSP

98622, 1, 88.2 mm, Philippines; ANSP 98623, 1, 104

mm, Philippines; ANSP 98620, 1, 119 mm, China Sea.
Lactoria fornasinii (Bianconi): ANSP 89016, 2, 12.5-

18.0 mm, Hawaii; ANSP 83863, 1, 65.2 mm, Hawaii;

ANSP 104837, 1, 15.4 mm, Madagascar; ANSP

104866, 2, 50.3-66.2 mm, Seychelles.
Ostracion lentiginosum Bloch and Schneider: ANSP

112738, 2, 24.4-33.2 mm, Hawaii; ANSP 98624, 1, 88.8

mm, Hawaii; ANSP 106667, 1, 68.2 mm, southwest

Pacific; ANSP 104875, 2, 95.9-115 mm, Mauritius.
Ostracion tuberculatus Linnaeus: ANSP 104830, 1,

29.8 mm, Seychelles; ANSP 112942, 1, 54.5 mm,

Aldabra Atoll; ANSP 112881, 1, 38.7 mm, Australia;

ANSP 112904, 2, 96.7-122 mm, Australia.
Rhinesomus bicaudalis (Linnaeus): ANSP 85199, 1,

83.7 mm, Haiti; ANSP 70147, 1, 77.4 mm, Honduras.
Rhinesomus triqueter (Linnaeus): ANSP 74890, 3,

15.5-36.9 mm, Florida; ANSP 98625, 2, 88.7-136 mm,

Puerto Rico; ANSP 91385, 1, 134 mm, Colombia.
Rhynchostracion rhinorhynchus (Bleeker): ANSP

90158, 1, 88.2 mm, Java; ANSP 95825, 1, 133 mm,

Java.
Tetrosomus concatenatus (Bloch): ANSP 104755, 1,

43.4 mm, Thailand.
Tetrosomus gibbosus (Linnaeus): ANSP 10081, 1, 96.3

mm, Burma.

Triodontidae

Triodon macropterus Lesson: SU 13747, 1, 391 mm,
Philippines; ANSP 98917, 1, 463 mm, Japan.

Tetraodontidae

Ostraciidae

Acanthostracion guineensis Bleeker: ANSP 102874, 1,
178 mm, Ivory Coast.

Acanthostracion notacanthus (Bleeker): ANSP 102909,
1, 148 mm, St. Helena.

Acanthostracion polygonius Poey: ANSP 83840, 1, 117
mm, Bahamas; ANSP 80008, 1, 92.5 mm. Lesser Antil-
les.

Acanthostracion quadricornis (Linnaeus): ANSP
98614, 1, 67.1 mm, Florida; ANSP 98616, 1, 58.2 mm,
Mexico; ANSP 98615, 1, 77.7 mm, Florida; ANSP
98617, 2, 8.2-15.3 mm, Louisiana; ANSP 98618, 1, 73.0
mm, Texas; SU 51172, 3, 17.5-25.0 mm, Guiana;
ANSP 98843, 2, 121-152 mm, Surinam; ANSP 102749,
1, 254 mm, Colombia; ANSP 103506, 1, 350 mm,
Venezuela; ANSP 98619, 3, 130-163 mm, locality un-
known, dry partially disarticulated skeletons.

Lactophrys trigonus (Linnaeus): ANSP 49179-80, 2,

Amblyrhynchotes honckenii (Bloch): ANSP 11507, 1,

70.7 mm, India; ANSP 104778, 1, 97.4 mm, Mozambi-
que.

Amblyrhynchotes piosae (Whitley): ANSP 113990, 1,

33.8 mm, Australia.

Amblyrhynchotes richei (Freminville): USNM ace. no.

244693, 1, 59.4 mm. New Zealand; ANSP 109916, 1,

50.8 mm, Australia.
Arothron armilla (Waite and McCulloch): ANSP

109790, 1, 61.3 mm, Australia.
Arothron hispidus (Linnaeus): ANSP 104798, 1, 40.8

mm, Seychelles.
Arothron nigropunctatus (Bloch and

Schneider): ANSP 106817, 1, 56.8 mm, Solomon

Islands; ANSP 104737, 1, 182 mm, Thailand.
Arothron stellatus (Bloch and Schneider): ANSP

109529, 1, ca. 420 mm, Seychelles, dry partially

disarticulated skull; ANSP 78242, 1, 292 mm, India.
Canthigaster amboinensis (Bleeker): ANSP 82355, 1,

80.3 mm, Hawaii.

Canthigaster bennetti (Bleeker): ANSP 104825, 1, 58.3

mm, Seychelles.
Canthigaster jactator Jenkins: ANSP 89255, 1, 54.2

mm, Hawaii.
Canthigaster janthinoptera (Bleeker): ANSP 104806, 2,

42.6-45.5 mm, Seychelles.
Canthigaster margaritata (Ruppell): ANSP 104829, 1,

46.0 mm, Seychelles; ANSP 109471, 1, 76.6 mm,
locality unknown.

Canthigaster punctatissima (Gunther): ANSP 109472,

2, 13.1-20.2 mm, Costa Rica; ANSP 109473, 2, 19.6-

46.1 mm, Costa Rica; ANSP 81467, 3, 42.5-49.5 mm,
Mexico.

Canthigaster rivulata (Schlegel): ANSP 84675, 1, 101

mm, Hawaii.
Canthigaster rostrata (Bloch): ANSP 113827, 3, 10.0-

18.2 mm, Virgin Islands; ANSP 102141, 1, 59.6 mm,
Texas; ANSP 102142, 1, 55.2 mm, Texas; ANSP
102143, 1, 96.5 mm, Texas, dry partially disartic-
ulated skeleton.

Canthigaster ualentini (Bleeker): ANSP 99975, 4, 33.9-
72.9 mm, Seychelles; ANSP 99948, 9, 30.5-72.8 mm,
Seychelles; ANSP 99962, 10, 18.1-70.7 mm, Seychel-
les; ANSP 104790, 1, 80.3 mm, Mauritius.

Carinotetraodon lorteti (Tirant): ANSP 109469, 1, 25.1
mm, locality unknown; ANSP 109470, 2, 33.1-35.9
mm, locality unknown.

Chelonodon fluviatilis (Hamilton-Buchanan): ANSP

109456, 4, 21.1-30.2 mm, locality unknown; ANSP

109457, 3, 70.1-84.7 mm, Thailand.

Chelonodon patoca (Hamilton-Buchanan): ANSP

117140, 1, 70.6 mm. New Guinea.
Chonerhinos modestus (Bleeker): SU 33563, 3, 31.0-

54.5 mm, Borneo.
Colomesus asellus (MuUer and Troschel): ANSP 98840,

3, 27.3-33.6 mm, Peru; ANSP 98908, 2, 50.6-50.8 mm,
Brazil; ANSP 76254, 1, 113 mm, Venezuela; ANSP
37899, 1, 128 mm, Venezuela.

Colomesus psittacus (Bloch and Schneider): ANSP
75845, 2, 49.7-52.3 mm, Trinidad; ANSP 98869, 2,
76.9-112 mm, Surinam; ANSP 98907, 1, 179 mm,
Surinam; ANSP 97658, 2, 178-180 mm, Surinam, dry
skeletons.

Ephippion guttifer (Bennett): ANSP 103236, 1, 101
mm, Guinea; ANSP 103245, 1, 108 mm, Guinea.

Fugu chrysops (Hilgendorf): ANSP 97548, 1, 98.6 mm,
Japan.

Fugu oblongus (Bloch): ANSP 77245, 1, 46.2 mm, In-
dia.

Fugu rubripes (Temminck and Schlegel): ANSP 95480,
1, 83.2 mm, Japan; USNM 26455, 1, 151 mm, Japan.

Guentheridia formosa (Gunther): ANSP 95552, 1, 175
mm, Panama.

Lagocephalus inermis (Temminck and
Schlegel): ANSP 100832, 1, 52.2 mm, Bay of Bengal.

Lagocephalus laevigatus (Linnaeus): ANSP 103120, 1,
61.4 mm, Alabama; ANSP 102155, 2, 143-166 mm,
Florida; ANSP 102153, ca. 290 mm, Uruguay, wet
disarticulated skeleton; ANSP 102154, 1, ca. 290 mm,
Uruguay, dry skeleton.

Lagocephalus lagocephalus (Linnaeus): ANSP 70299,

1, 214 mm, Malpelo Islands, eastern Pacific.
Lagocephalus lunaris (Bloch and Schneider): ANSP

111509, 1, 62.3 mm. Bay of Bengal.
Lagocephalus scleratus (Gmelin): ANSP 96685, 1, 80.9

mm, Philippines.
Lagocephalus spadiceus (Richardson): ANSP 104787,

1, 98.2 mm, Mozambique.
Monotreta cutcutia (Hamilton-Buchanan): ANSP

63132, 1, 45.4 mm, Thailand; ANSP 59928-37, 1, 73.9

mm, Thailand.
Monotreta gularis (Hamilton-Buchanan): USNM

44811, 2, 46.8-47.1 mm, Burma.
Monotreta leiurus (Bleeker): ANSP 59928-37, 2, 61.5-

66.3 mm, Thailand.
Monotreta palembangensis (Bleeker): ANSP 27770, 1,

103 mm, Sumatra.
Sphoeroides angusticeps (Jenyns): ANSP 95841, 1, 172

mm, Galapagos.
Sphoeroides annulatus (Jenyns): ANSP 109459, 3,

15.0-21.4 mm, California; ANSP 109461, 3, 10.9-15.8

mm, California; ANSP 109460, 1, 13.5 mm, California;

ANSP 109458, 1, 174 mm, locality unknown.
Sphoeroides dorsalis Longley: ANSP 109462, 1, 131

mm, Florida; ANSP 109519, 1, 139 mm, Georgia, dry

skeleton; ANSP 109520, 1, 155 mm, Florida, dry

skeleton.
Sphoeroides greeleyi Gilbert: ANSP 109463, 3, 24.2-

60.8 mm, locality unknown.
Sphoeroides lobatus (Steindachner): ANSP 97332, 1,

67.5 mm, Panama.
Sphoeroides maculatus (Bloch and Schneider): ANSP

109466, 1, 126 mm, Virginia; ANSP 109465, 3, 31.0-

57.8 mm, Virginia; ANSP 109464, 2, 97.5-106 mm,

Virginia; ANSP 84307, 3, 9.2-10.2 mm. New Jersey;

ANSP 102043, 1, 12.4 mm. New Jersey; ANSP 109467,

1, 201 mm, Florida; all the following 11 specimens are

dry and partially to completely disarticulated

skeletons from Virginia of between ca. 50 and 2(X) mm,

ANSP 109516-18, 109521-28.
Sphoeroides nephelus (Goode and Bean): ANSP

105222, 1, 48.5 mm, Mississippi; ANSP 110533, 1, 128

Sphoeroides pachygaster (Ranzani): ANSP 109468, 2,

134-137 mm, Florida; ANSP 101322, 1, 80.6 mm,

Nigeria; ANSP 104781, 1, 117 mm, Mozambique.
Sphoeroides spengleri (Bloch): ANSP 10531, 1, ca. 90

mm, Florida, dry skeleton; ANSP 10532, 1, 92.8 mm,

Nicaragua, dry skeleton.
Sphoeroides testudineus (Linnaeus): ANSP 107328, 1,

68.5 mm, Venezuela.
Sphoeroides trichocephalus (Cope): ANSP 109742, 1,

57.1 mm, Panama.
Tetraodon lineatus Linnaeus: ANSP 77913, 1, 222 mm,

French Equitorial Africa.
Tetraodon mbu Boulenger: ANSP 112236, 1, 47.7 mm,

Belgian Congo.
Torquigener pleurogramma (Regan): ANSP 98690, 1,

113 mm, Australia.

Torquigener pleurostictus (Giinther): ANSP 89848

87.1 mm, Australia.
Xenopterus naritus (Richardson): ANSP 100814,

108-143 mm. Bay of Bengal.

jaws only, from a large specimen of unknown length
from Massachusetts.
Ramania laeuis (Pennant): ANSP 109435, 2, 65.1-66.9
mm, Hawaii; ANSP 109561, 1, 493 mm, Hawaii.

Diodontidae

Chilomycterus affinis Gunther: ANSP 103806, 1, 310

mm, Galapagos.
Chilomycterus antennatus (Cuvier): ANSP 76412, 1,

70.5 mm, Trinidad.
Chilomycterus antillarum Jordan and Rutter: ANSP

97512, 1, 66.4 mm, Surinam.
Chilomycterus atinga (Linnaeus): ANSP 103900, 1, 278

mm, Florida.
Chilomycterus mauritanicus Le Danois: ANSP 103209,

1, 82.1 mm, Guinea.
Chilomycterus orbicularis (Bloch): ANSP 108827, 2,

72.9-77.0 mm, Somalia; ANSP 98008, 1, 158 mm,

Philippines.
Chilomycterus reticulatus (Liimaeus): ANSP 102574,

1, 545 mm, Sierra Leone.
Chilomycterus schoepfi (Walbaum): ANSP 109476, 2,

20.3-22.6 mm, Louisiana; ANSP 109477, 1, 60.1 mm,

Louisiana; ANSP 109478, 1, 62.5 mm, Texas; ANSP

109474, 1, 159 mm, Louisiana; ANSP 109479, 1, 35.2

mm, Louisiana; ANSP 109475, 1, 168 mm, Florida;

ANSP 109513, 1, ca. 80 mm, Texas, dry partially disar-
ticulated skeleton; ANSP 109514, 1, ca. 125 mm,

Florida, dry partially disarticulated skeleton.
Chilomycterus spinosus (Linnaeus): ANSP 109669, 1,

51.0 mm, Brazil.
Chilomycterus tigrinus (Cuvier): ANSP 101341, 1, 165

mm, Somalia.
Dicotylichthys nicthemerus (Cuvier): USNM 201630,

1, 51.8 mm, Australia.
Dicotylichthys punctulatus Kaup: USNM 47923, 1, 226

mm, Australia.
Diodon holocanthus Linnaeus: ANSP 102151, 1, 12.3

mm, Louisiana; ANSP 109480, 1, 53.5 mm, Texas;

ANSP 102149, 1, 108 mm, Florida; ANSP 102148, 1,

81.8 mm, Florida; ANSP 102150, 1, 113 mm, Florida;

ANSP 102162, 1, 124 mm. Virgin Islands, dry partially

disarticulated skeleton.
Diodon hystrix Linnaeus: ANSP 104740, 2, 80.6-84.4

mm, Somalia; ANSP 106756, 2, 45.1-42.2 mm, Florida;

ANSP 109515, 1, ca. 200 mm, Florida, dry partially

disarticulated skull.
Diodon jaculiferus Cuvier: ANSP 98272, 1, 53.4 mm,

Australia.

Molidae

Masturus lanceolatus (Li6nard): USNM 117330, 1, 127
mm, Florida.

Mola mola (Linnaeus): SU 16438, 1, 306 mm, Califor-
nia; SU 16441, 1, 310 mm, California; USNM 27238,
dry jaws only from a large specimen of unknown length
from Massachusetts; USNM 102086, alcohol preserved

NONPLECTOGNATHS

Acanthuridae

Acanthurus sp. (acronurus stage): ANSP 86416, 1, 53.6

mm. Society Islands,
Acanthurus gahhm (Forskal): ANSP 109483, 2, 155-158

mm, unknown western Pacific locality.
Acanthurus leucosternon Bennett: ANSP 108204, 1,

114 mm, Seychelles.
Acanthurus lineatus (Linnaeus): ANSP 109484, 2, 103-

118 mm, unknown western Pacific locality; ANSP

109482, 1, ca. 150 mm, unknown western Pacific

locality, dry partially disarticulated skeleton; ANSP

109481, 1, ca. 160 mm, unknown western Pacific

locality, dry partially disarticulated skeleton.
Acanthurus nigoris Cuvier: ANSP 109487, 1, 127 mm,

locality unknown.
Acanthurus nigrofuscus (Forskal): ANSP 109485, 4,

62.3-73.3 mm, unknown, western Pacific locality;

ANSP 109486, 3, 77.3-131 mm, locality unknown.
Acanthurus olivaceus Bloch and Schneider: ANSP

109488, 1, 109 mm, locality unknown; ANSP 109480, 1,

152 mm, locality unknown.
Acanthurus sandvicensis Streets: ANSP 109490, 1, 69.0

mm, Hawaii.
Acanthurus triostegus (Linnaeus): ANSP 83246, 2,

29.1-29.8 mm, Marquesas Islands; ANSP 87546, 3,

27.9-35.1 mm, Marquesas Islands; ANSP 109491, 6,

40.6-68.9 mm, locality unknown.
Ctenochaetus striatus (Quoy and Gaimard): ANSP

109492, 4, 74.8-123 mm, locality unknown.
Ctenochaetus strigosus (Bennett): ANSP 108208, 1,

93.0 mm, Seychelles.
Naso sp.: ANSP 108416, 1, 107 mm, Seychelles; ANSP

108419, 1, 72.0 mm, Seychelles.
Naso sp., probably N. annulatus (Quoy and

Gaimard): MCZ uncatalogued, 1, ca. 370 mm,

locality unknown, dry skull.
Naso breuirostris (Cuvier and Valenciennes): ANSP

109493, 1, 209 mm, locality unknown; ANSP 88537, 1,
228 mm, Tuamotus.

Naso fageni Morrow: ANSP 103532, 1, 515 mm,

Seychelles.
Naso hexacanthus (Bleeker): ANSP 109495, 1, 158 mm,

locality unknown; ANSP 109494, 1, 188 mm, locality

unknown.
Naso literatus (Bloch and Schneider) : ANSP 109497, 2,

111-209 mm, unknown western Pacific locality; ANSP

109496, 1, 191 mm, locality unknown.
Naso thynnoides (Cuvier and Valenciennes): JLBSII

616d, 1, 191 mm, Tanganyika.
Naso unicornis (Forskal): ANSP 89114, 1, 267 mm,

Hawaii.

Paracanthwus hepatus (Linnaeus): ANSP 108441, 1,

21.2 mm, Seychelles; ANSP 108444, 1, 31.6 mm,

Seychelles; ANSP 90446, 1, 179 mm, Java.
Prionurus punctatus Gill: ANSP 81238, 3, 35.0-47.9

mm, Galapagos.
Prionurus scalprum Cuvier and Valenciennes: ANSP

109779, 1, 44.3 mm, Japan; ANSP 109553, 1, 111 mm.

Japan.
Zebrasoma flavescens (Bennett): ANSP 109498, 1, 60.0

mm, unknown western Pacific locality; ANSP 109499,

2, both 101 mm, locality unknown.
Zebrasoma rostratus (Gunther): ANSP 108278, 1, 94.7

mm, Seychelles.
Zebrasoma veliferum (Bloch): ANSP 109500, 1, 156

mm, locality unknown.

Zanclidae

Zanclus cornutus (Linnaeus): ANSP 109504, 1, ca. 70
mm, locality unknown; ANSP 109501, 1, 58.0 mm,
locality unknown; ANSP 109502, 1, 66.4 mm, locality
unknown; ANSP 109503, 1, 70.4 mm, locality un-
known.

Chaetodontidae

Chaetodon citrinellus Broussonet: ANSP 109506, 1,

81.6 mm, unknown western Pacific locality.
Chaetodon octofasciatus Bloch: ANSP 109508, 1, 67.1

mm, unknown western Pacific locality.
Chaetodon striatus Linnaeus: ANSP 91118, 1, 44.6

mm, Bahamas; ANSP 91611, 1, 57.4 mm. Virgin

Islands.
Chaetodon triangulum Cuvier and Valencien-
nes: ANSP 73227, 1, 87.2 mm, Philippines.
Chelmon rostratus (Linnaeus): ANSP 109511, 3, 48.7-

93.4 mm, unknown western Pacific locality.
Chaetodontoplus mesoleucus (Bloch): ANSP 109510, 3,

59.2-76.3 mm, unknown western Pacific locality.
Centropyge multispinis (Playfair): ANSP 108469, 3,

19.0-69.0 mm, Seychelles.
Centropyge vrolicki (Bleeker): ANSP 109505, 1, 90.4

mm, unknown western Pacific locality.
Forcipiger longirostris (Broussonet): ANSP 109512, 1,

114 mm, locality unknown; ANSP 108393, 2, 139-140.0

mm, Seychelles.
Holacanthus ciliaris (Linnaeus): ANSP 91094, 2, 35.7-

54.4 mm, Bahamas; ANSP 104989, 1, 26.0 mm.

Cayman Islands.
Holacanthus tricolor (Bloch): ANSP 104970, 2, 40.2-

68.4 mm, Cayman Islands.
Pomacanthodes imperator (Bloch): ANSP 108461, 1,

27.2 mm, Seychelles.
Pomacanthus aureus (Bloch): ANSP 91609, 1, 56.6

mm, Florida.

Fossil Species

Institutional abbreviations for the repositories of the

fossils examined are as follows: AMNH, American
Museum of Natural History, New York; BMNH, British
Museum (Natural History), London; IGPUB, Istituto di
Geologia e Paleontologia della University di Bologna;
IGPUP, Istituto di Geologia e Paleontologia della
Universita di Pisa; IGPUR, Istituto di Geologia e Paleon-
tologia della University di Roma; IGUN, Institute de
Geologie, Universite de Neuchatel; IGUP, Istituto di
Geologia della Universita di Padova; MCSNV, Museo
Civico di Storia Naturale, Verona; MNHN (IP),
Museum National d'Histoire Naturelle, Institute de
Paleontologie, Paris; NSKG, Naturwissenschaftliche
Sammlungen des Kantons Glarus, Switzerland.

PLECTOGNATHI
Triacanthodidae

Eoplectus bloti Tyler: MCSNV NS 52-53, in counter-
part, 65.2 mm, Eocene of Monte Bolca, Italy, holotype.

Protobalistum imperiale (Massalongo): MCSNV T9-
10, in counterpart, 522 mm. Eocene of Monte Bolca,
Italy, holotype of Ostracion imperialis.

Spinacanthus cuneiformis (de Blainville): MNHN (IP)
10918, single plate, 104 mm. Eocene of Monte Bolca,
Italy, holotype of Blennius cuneiformis.

Zignoichthys oblongus (Zigno): IGUP 6789, single
plate, ca. 161 mm. Eocene of Monte Bolca, Italy,
holotype of Ostracion oblongus.

Triacanthidae

Acanthopleurus collettei Tyler, new species: all
specimens from the Oligocene of Canton Claris,
Switzerland; NSKG 2689, single plate, 110 mm,
holotype; NSKG 146, single plate, 96.2 mm, para-
type; NSKG 56, single plate, ca. 120 mm, paratype;
NSKG 6b, single plate, 82.7 mm, paratype; NSKG
uncatalogued, single plate, last few vertebrae missing
and length unmeasurable, paratype; BMNH P 524,
single plate, ca. 82 mm, paratype; IGUN uncata-
logued, single plate, 108 mm, paratype.

Acanthopleurus serratus (Agassiz): all specimens from
the Oligocene of Canton Claris, Switzerland; BMNH P
454, single plate, ca. 91 mm, cotype of Pleuracanthus
serratus; BMNH P 3974, single plate, front of head
missing and length unmeasurable, cotype of Pleura-
canthus serratus; BMNH P 1892, single plate, ca. 78
mm; NSKG 222 a-b, in counterpart, 114 mm; NSKG
7, single plate, 93.3 mm; NSKG 158 b, single plate, 153
mm; NSKG 8, single plate, 109 mm; NSKG uncata-
logued, single plate, 120 mm; NSKG uncatalogued,
single plate, 117 mm; NSKG uncatalogued, single
plate, 127 mm; IGUN uncatalogued, single plate, last
few vertebrae missing and length unmeasurable.

Acanthopleurus sp.: three plates representing poor
impressions or distorted and incomplete specimens
unidentifiable to species, all from the Oligocene of

Canton Glarus, Switzerland; BMNH uncatalogued,
ca. 135 mm; BMNH P 4522, ca. 120 mm; BMNH P
1893, unmeasurable.
Protacanthodes ombonii (Zigno): IGUP 10.901-902, in
counterpart, 112 mm. Eocene of Monte Bolca, Italy,
holotype of Protobalistum ombonii.

Balistidae

Balistomorphus orbiculatus (Heer): NSKG 2688, single
plate, 65.6 mm, Oligocene of Canton Glarus,
Switzerland, holotype of Acanthoderma orbicula-
tum.

Balistomorphus ovalis (Agassiz): both specimens from
the Oligocene of Canton Glarus, Switzerland; IGUN
228, single plate, 121 mm, holotype of Acanthoderma
ovale; NSKG 178b, single plate, 119 mm.

Balistomorphus spinosus (Agassiz); both specimens
from the Oligocene of Canton Glarus, Switzerland;
BMNH P. 3973, single plate, ca. 90 mm, holotype of
Acanthoderma spinosum; NSKG 189 a-b, in counter-
part, too indistinct to measure.

Aracanidae

Proaracana dubia (de Blainville): all three specimens
from the Eocene of Monte Bolca, Italy; MNHN (IP)
10974-75, in counterpart, 54.5 mm, holotype oiBalistes
dubius; MCSNV T8 and T63, in counterpart, 31.4
mm; IGPUP uncatalogued, single plate, ca. 52 mm.

Ostraciidae

Eolactoria sorbinii Tyler: MCSNV T6-7, in counter-
part, 15.5 mm, Eocene of Monte Bolca, Italy, holo-
type.

Triodontidae

Triodon antiquus Leriche: BMNH P 12629, large frag-
ment of right premaxillary, Eocene of Barton, Hants.,
England; BMNH P 15193, large fragment of right
premaxillary. Eocene of Barton, Hants., England;
BMNH P 15726, large fragment of fused dentaries,
Eocene of Schaerbeek, Belgium; BMNH P 46696, frag-
ment of middle of fused dentaries. Eocene of the
London Clay, England; BMNH P 28899, small frag-
ment of teeth from edge of jaw. Eocene of the London
Clay, England; BMNH P 25776-81, five fragments of
fused dentaries and one right premaxillary. Eocene of
Barton Hants., England.

Tetraodontidae

"Tetraodon" lecointrae Leriche: IGPUP Miocene of
Piromafo, Italy, no. 13, three fragments of the elongate
dental lamellae at edge of jaw.

Eotetraodon pygmaeus Zigno: IGUP 6890-91, in counter-
part, 18.2 mm. Eocene of Monte Bolca, Italy.

Diodon scillae Agassiz: IGPUP Miocene of Terra Ros-
sa, Italy, no. 1, one complete trituration plate with
right and left halves; IGPUP Miocene of Livello ad
Aturia, Italy, no. 3, one half of a trituration plate;
IGPUP Miocene of Terreno Agrario, Italy, no. 27, one
half of a trituration plate.

Diodon tenuispinus Agassiz: seven specimens from the
Eocene of Monte Bolca, Italy; MNHN (IP) 10976-77,
in counterpart (dorsoventral), 19.7 mm, holotype;
MNHN (IP) 10978, single plate (dorsoventral), 12.6
mm; MNHN (IP) 10979, single plate (dorsoventral),
46.5 mm; MCSNV B. 18, single plate (dorsoventral),
31.5 mm; MCSNV B. 19, single plate (dorsoventral),
40.5 mm; BMNH P 11168, in counterpart (dorso-
ventral), 22.9 mm; BMNH P 12360, single plate
(dorsoventral), 13.5 mm.

"Diodon" sp.: eight specimens from the Eocene of
Monte Bolca, Italy; MCSNV T 10, single plate (dorso-
ventral), 11.8 mm, Italy; MCSNV T 11, single plate
(dorsoventral), 35.1 mm; IGUP 12810-11, in counter-
part (dorsoventral), 27.3 mm; IGUP 6894, single plate
(dorsoventral), 17.4 mm; IGUP 6861, single plate
(dorsoventral), 11.8 mm; MCSNV MS 50-51, in
counterpart (dorsoventral), 68.9 mm; MCSNV T2,
single plate (dorsoventral), 18.4 mm; BMNH P 10735,
in counterpart (dorsoventral), 25.8 mm.

Eodiodon erinaceus (Agassiz): five specimens from the
Eocene of Monte Bolca, Italy; BMNH P 3873, single
plate (dorsoventral), 77.5 mm, holotype of Diodon
erinaceus; BMNH P 10426, single plate (dorso-
ventral), 96.5 mm; IGUP 8855-56, in counterpart
(dorsoventral), 50.3 mm; IGUP 8853-54, in counterpart
(dorsoventral), 68.8 mm; IGUP 8670-71, in counterpart
(dorsoventral), ca. 64 mm.

Kyrtogymnodon capellinii (de Stefano): IGPUP
Miocene of Livello ad Aturia, Italy, no. 4, right and left
halves of a trituration plate; IGPUP Miocene of Ter-
rene Agrario, Italy, no. 28, fragment of a large tritura-
tion plate.

Oligodiodon acanthodes (Sauvage): IGPUP uncata-
logued, most of left premaxillary and top of maxillary,
Oligocene of Monti Livomesi, Italy.

Oligodiodon platyodus (Portis): IGPUR uncatalogued,
one half of a trituration plate, 15.4 mm greatest length,
Oligocene of Montecchio Maggiore, Italy, type
specimen of Diodon platyodus.

Progymnodon hilgendorfi Damea: BMNH P 11089,
large fragment of jaw with teeth and trituration plate.
Eocene of Fayum, Egypt.

Progymnodon gigantodus (Portis): IGPUR uncata-
logued, right and left halves of a trituration plate, 41.8
mm greatest length of entire plate. Eocene of Castel
Madama, Italy, type specimen of Diodon gigantodus;
IGPUP Miocene of Terra Rossa, Italy, no. 1, two
separate halves of the same trituration plate; IGPUP
Miocene of Livello ad Aturia, Italy, no. 3, half of a
large trituration plate; IGPUP Miocene of Terreno
Agrario, Italy, no. 27, two halves of a trituration plate.

NONPLECTOGNATHS

Acanthuridae

(Identifications under revision)

Acanthurus sp.: three specimens from the Eocene of
Monte Bolca, Italy; BMNH P 16130 and P 17020, in
counterpart, the two halves separately catalogued, 120
mm; MCSNV T 3, single plate, 25.6 mm; MCSNV T 1,
single plate, 18.8 mm, possibly a young Naseus
nuchalis.

Acanthurus ovalis Agassiz: both specimens from the
Eocene of Monte Bolca, Italy; MCSNV T 4-5, in
counterpart, 127 mm; IGUP 6874, single plate, 76.1
mm.

Acanthurus tenuis Agassiz: BMNH P 11176, single
plate, 86.6 mm. Eocene of Monte Bolca, Italy.

Naseus intermedius Zigno: IGUP 6917-18, in counter-
part, 105 mm. Eocene of Monte Bolca, Italy, holotype.

Naseus nuchalis Agassiz: eight specimens from the
Eocene of Monte Bolca, Italy; MNHN (IP) 10910-11,
in counterpart, 155 mm, holotype; BMNH P 19059,
single plate, 212 mm; BMNH P 11174, in counterpart,
151 mm; BMNH P 11098, in counterpart, 177 mm;
IGUP 12062-63, in counterpart, 190 mm; IGUP 11609-
10, in counterpart, 152 mm; MCSNV T 12, single
plate, 109 mm; MCSNV B 1965.13, single plate, 106
mm.

Naseus rectifrons Agassiz: 18 specimens from the
Eocene of Monte Bolca, Italy; MNHN (IP) 10908-09,
in counterpart, 189 mm, holotype; BMNH P 10427,
single plate, 205 mm; BMNH P 11173, in counterpart,
112 mm; BMNH P 41890, single plate, 171 mm;
BMNH P 9831, single plate, 67.5 mm; BMNH P 21393,
single plate, ca. 200 mm, incomplete; BMNH P 43490,
in counterpart, 38.4 mm; IGUP 11895, single plate,
83.0 mm; IGUP 25103, single plate, 160 mm; IGUP
25101, single plate, 189 mm; IGUP 8701-02, in counter-

part, 185 mm; IGUP 8698-99, in counterpart, 97.0 mm;
MCSNV Vin C 58-59, two specimens with the same
catalogue number, both in counterpart, 92.4 and 195
mm; MCSNV Vm C 56-57, in counterpart, 88.8 mm;
MCSNV MS. C 50, single plate, 144 mm; MCSNV VHI
C 70, single plate, 108 mm; MCSNV Vm C 68, single
plate, 187 mm; AMNH 9531 A-B, in counterpart, 212
mm.
Naso deani (Hussakof): AMNH 7483, single plate, 231
mm, Antigua, West Indies, uncertain age, holotype of
Zebrasoma deani.

Acanthonemidae

Acanthonemus subaureus (de Blainville): 10 speci-
mens from the Eocene of Monte Bolca, Italy;
MNHN (IP) 10904-05, in counterpart, 192 mm,
holotype of Chaetodon subaureus; MNHN (IP) 10906-
07, in counterpart, 99.5 mm; BMNH P 16201, single
plate, 126 mm; BMNH P 9940, single plate, 165 mm;
BMNH P 16200-01, in counterpart, 126 mm; IGUP
25100, single plate, ca. 123 mm; IGUP 25102, single
plate, 167 mm; IGUP 11606-07, in counterpart, 126
mm; IGUP 6884, single plate, 62.8 mm; MCSNV VD
108-109, ca. 72 mm.

Zanclidae

Eozanclus brevirostris (Agassiz): MNHN (IP) 10740-
41, in counterpart, 101 mm. Eocene of Monte Bolca,
Italy, holotype of Zanclus brevirostris.

Teleostei Incertae Sedis

Protriacanthus gortanii d'Erasmo: IGPUB uncata-
logued, single plate, 20.5 mm. Lower Cenomanian of
Comen, near Trieste, Italy, holotype.

EPILOGUE

Remaining Problems

The more outstanding remaining problems on the
anatomy, classification, and phylogeny of the plectog-
naths of which I am aware are the need for: 1) system-
atic surveys of other anatomical systems to complement
that on the myology by Winterbottom (1974) and on the
osteology and external features presented here; 2)
systematic worldwide familial revisions of the balis-
toids, ostracioids, tetraodontoids, and molids, to expand
the coverage of groups begun by Tyler (1968) for triacan-
thoids; 3) future diggings to increase the number of perti-
nent fossils available and tell us more about the

phylogeny and anatomical diversity of the order and its
relatives; 4) rigorous comparisons between the fossil and
Recent acanthurids and their chaetodontoidlike rela-
tives with the fossil and Recent plectognaths to defini-
tively determine if these groups share a late Cretaceous
or early Eocene ancestral stock; 5) a reconciliation of
any conflicting classificatory schemes for plectognaths
based on similar hypothesized phylogenies; and 6) clari-
fication of specific identifications and nomenclature of
many Indo-Pacific tetraodontids, diodontids, and mona-
canthids.

ACKNOWLEDGMENTS

This monograph is a much revised and expanded out-
growth of a work begun at Stanford University in 1958 as
a doctoral dissertation under the. guidance of George S.
Myers; his instructive and encouraging influences
prevailed over so many aspects of the germinal
stages of this monograph that it would be impossible to
list them, and I will simply express my lasting appreci-
ation for his help and innumerable kindnesses devoted to
my education.

Stanley H. Weitzman, then also of Stanford Univer-
sity, was equally helpful to the early formative aspects of
this work, and he has continued to the present to freely
make available on many occasions his osteological,
systematic, and phylogenetic expertise.

The late Margaret H. Storey of Stanford's Natural
History Museum was of constant curatorial and biblio-
graphic assistance, while the interchange of ideas, tech-
niques, and data with my fellow students was most in-
valuable and is here gratefully acknowledged, these then
student friends being Warren C. Freihofer, Hugh H.
DeWitt, and Tyson R. Roberts.

The majority of this work was done while a member of
the curatorial staff of the Academy of Natural Sciences
of Philadelphia, and I am especially grateful to H.
Radclyffe Roberts, until 1973 its Director, for encour-
aging this research during my time there. His under-
standing of the vagaries of research is much appreci-
ated. My associates there in the Department of Ichthy-
ology and Herpetology constantly assisted in the
preparation of this monograph, these coworkers being
Jane H. Baker, Ann L. Berkes, Warren N. Berkes,
Eugenia B. Bohlke, James E. Bohlke, Charles C. G.
Chaplin, Kathy Depolito, Lynn Keane, Mark D. Lange,
and Judith Silver; I would like to especially single out
Judith Silver and Lynn Keane, successive secretary-
major domos of the department, for unstinting coopera-
tion in every aspect of this work, from typing to proof-
reading, and from bibliographic searching to the
considerate care of the researcher. I have been blessed
by a succession of equally cooperative Eind artistic
scientific illustrators whose skillful techniques form
a major portion of any merit to this monograph and
who were exceptionally pleasant to work closely with:
Gail Costanzo, Mary H. Fuges, Olivia B. Hall, Helga
0. Kumpera, and Joanne R. Schatz. Additionally,
Stephen P. Gigliotti generously gave his advice on il-
lustrative matters and James E. Bohlke did so on
systematic questions.

Other members of the Academy of Natural Sciences of
Philadelphia were immeasurably helpful in a variety of
ways. C. Willard Hart, then editor of Academy publica-
tions, provided many helpful suggestions in coordin-
ating the manuscript and illustrations, as did the
Academy's engraver, Hal Talman, president of
Phototype of Philadelphia. Larry Herbert, the late Oliver
Bradley, and Joseph Green of the Academy engineering
department were always on call to build or repair, often
on short notice, the mechanical and electrical contriv-

ances essential to this project. Beverly Mowbray and
Nancy Steele of the publications staff at the Academy
could always see a lighter side to a heavy scientific tale.
Jesse J. Freese and Mary E. Sink kept the convolutions
of budgetary and administrative rules and reports to the
efficient minimum. Charles Sibre, and his sons John and
Michael Sibre, of the photographic department, grace-
fully as well as proficiently provided the 35 mm nega-
tives of the cleared and stained specimens that formed
the basis for the initial traced outlines for most of the os-
teological illustrations presented here. Their good humor
in the face of messy containers of glycerine scattered
about their studio was remarkable.

The introductory and concluding portions of this
monograph, and of the updating and revisions of the en-
tire manuscript, were written while serving as the Direc-
tor of the Lerner Marine Laboratory, Bimini, Bahamas,
of the American Museum of Natural History, and while
on the staff of the Office of Resource Research of the
National Marine Fisheries Service, Washington, D.C.,
where Lamarr B. Trott, Chief of the then Division of Re-
search Management, provided many kindnesses, and
then as a member of the Southeast Fisheries Center of
that service, Miami, Fla., where the support of Harvey R.
Bullis, Jr., the then Center Director, and William J.
Richards, Miami Laboratory Director, and Laurie
Smith, secretary to the Atlantic Bluefin Tuna Program,
were especially helpful.

Most of the 10 years of research at the Academy of
Natural Sciences of Philadelphia leading to the major
portion of this monograph were supported in large part
by the National Science Foundation through grants
GB-1248, GB-5102, GB-6859X, and GB-16190. J.
Frances Allen, Richard F. Johnston, William Sievers,
and J. T. Spencer, the officers of the National Science
Foundation then administering these grants, are sincere-
ly thanked for their support, and for doing so with a
minimal amount of bureaucratic paper work. The
thoughtfulness and trust of the National Science Foun-
dation and of their advisory boards and ad hoc com-
mittees of colleagues mostly unknown to me is a true
pleasure to acknowledge.

The success of any osteological survey of a speciose
worldwide group of fishes such as the plectognaths is
dependent on obtaining for dissection specimens repre-
senting as many species as possible. The great majority
of specimens studied here were either loaned or donated
to Stanford University and the Academy of Natural
Sciences of Philadelphia by the following cooperative
colleagues: Tokiharu Abe, Frederick H. Berry, Harvey
R. Bullis, Jr., Daniel M. Cohen, H. Adair Fehlmann,
Leslie W. Knapp, Giles W. Mead, George S. Myers, and
Robert R. Rofen. Without the specimens from them, my
coverage of the plectognaths would be so meager as not to
warrant publication.

Photographic prints of many of the illustrations, used
in preference to constant handling for over a decade of
the final ink drawings, generously were provided by Vic-

tor Kumpera and Vlastimil Wagner, of Sao Paulo,
Brazil.

The cross sections of the jaws of diodontids, showing
the trituration plates and teeth of the biting edge, all
were skillfully cut and polished by Edgar R. Lawrence,
dental surgeon of West Chester, Pa.

More generally, Helga 0. Kumpera was a mainspring
in the laborious mechanism leading to the completion of
this monograph, as the culmination of a 20-year study.
Her help ran the gamut from the typing of hundreds of
pages of manuscript to translating into English numer-
ous articles in several foreign languages, and from thou-
sands of hours spent in stippling illustrations to equally
numerous hours spent in proofreading.

The following colleagues have generously provided
specimens for examination or made available data about
or assistance with specimens in their care, and their vital
help is greatly appreciated (affiliations listed as of the
time the help was rendered, and therefore sometimes
outdated at this time): Tokiharu Abe, Tokaiku Fish-
eries Research Laboratory, Tokyo; B. Accordi, Istituto di
Geologia e Paleontologia dell'Universita di Roma; Gerald
R. Allen and John E. Randall, Bernice P. Bishop
Museum, Honolulu; Roberto D. Andreucci, Escola
Paulista de Medicina, Sao Paulo; James E. and Ethyl H.
Atz, Gareth J. Nelson, Donn E. Rosen, Bobb Schaeffer,
and C. Lavett Smith, American Museum of Natural
History, New York; Reeve M. Bailey and Teruyo Uyeno,
University of Michigan, Ann Arbor; Keith E. Banister,
P. Humphrey Greenwood, Alison Longbottom, Rosemary
McConnell, G. Palmer, Colin Patterson, Ethelwynn
Trewavas, Charlotte Welch, and Peter J. Whitehead,
British Museum (Natural History), London; M. L.
Bauchot, M. Blanc, Jacques Blot, and J. Guibe,
Museum National d'Histoire Naturelle, Paris; Frederick
H. Berry, National Marine Fisheries Service, Miami; E.
Bertelsen, Marinebiologisk Laboratorium, Charlotten-
lund Slot; M. Boeseman and G.F. Mees, Rijksmuseum
van Natuurlijke Historic, Leiden; James E. Bohlke,
Charles C. G. Chaplin, and Neal R. Foster, Academy of
Natural Sciences of Philadelphia; Niels Bonde, Mineral-
ogisk-Geologisk Institut, K(/benhavn; Margaret G. Brad-
bury, San Francisco State College; Heraldo A. Britski,
Museu de Zoologia da Universidade de Sao Paulo;
Harvey R. Bullis, Jr., and Richard A. Waller, National
Marine Fisheries Service, Miami and Washington, D.C.;
Vernon E. Brock, Warren E. Burgess, and William A.
Gosline, University of Hawaii, Honolulu; Jean Cadenat,
Institute Frangais d'Afrique Noire, Gofee, Senegal;
David K. Caldwell, University of Florida, St. Augustine;
Edgar Casier, Institut Royal des Sciences Naturelles de
Belgiques, Brusselles; William Chan, Fisheries Research
Station, Aberdeen, Hong Kong; Edward Chin, Richard
L. Haedrich, and Andrew Konnerth, Woods Hole
Oceanographic Institution, Woods Hole, Mass.; Eugenie
Clark, University of Maryland, College Park; Glenn H.
Clemmer and Royal D. Suttkus, Tulane University, New
Orleans; Daniel M. Cohen and Bruce B. Collette,
National Systematics Laboratory, National Marine
Fisheries Service, Washington, D.C.; Walter R.

Courtenay, Florida Atlantic University, Boca Raton;
Giambattista Dal Piaz, Istituto di Geologia, Paleon-
tologia e Geologia Applicata dell'Universita di Padova;
William P. Davis, Aquatic Sciences, Inc., Boca Raton;
L. F. de Beaufort, H. Nijssen, and P. J. H. van Bree,
Zoologisches Museum, Amsterdam; KurtDeckert, Hum-
boldt Universitat, Berlin; Thomas Devany, C. Richard
Robins, and David G. Smith, University of Miami; Hugh
H. DeWitt, University of Maine, Walpole; Myvanwy M.
Dick, Alfred S. Romer, and Tyson R. Roberts, Museum
of Comparative Zoology, Cambridge; Pedro Duarte
Bello, Universidad de La Habana; Alfred W. Ebeling,
University of California, Santa Barbara; Paul R. Ehrlich
and Donald Kennedy, Stanford University; Donald S.
Erdman, Department of Agriculture, San Juan, Puerto
Rico; William N. Eschmeyer, W. I. FoUett, Warren C.
Freihofer, George S. Myers, and Pearl M. Sonoda,
California Academy of Sciences, San Francisco; H. Adair
Fehlmann and Leslie W. Knapp, Smithsonian Oceano-
graphic Sorting Center, Washington, D.C.; P. Four-
manoir. Centre Office de la Recherche Scientifique et
Technique Outre-Mer, Paris; A. J. Fraser and R. J.
McKay, Western Australian Museum, Perth; A. Fraser-
Brunner and John C. Marr, Food and Agriculture
Organization of the United Nations, Rome; K. C.
George, Central Marine Fisheries Research Substation,
Kerala; Robert R. Gibbs, Edgar N. Gramblin, Robert H.
Kanazawa, Ernest A. Lachner, Leonard P. Schultz, Vic-
tor G. Springer, William R. Taylor, and Stanley H.
Weitzman, National Museum of Natural History,
Washington, D.C.; Carter R. Gilbert, University of
Florida, Gainesville; J. R. Grindley, South African
Museum, Cape Town; Bruce W. Halstead, World Life
Research Institute, Colton, Calif.; Carl L. Hubbs, John
E. McCosker, and Richard H. Rosenblatt, Scripps
Institution of Oceanography; J. Jenny, Naturwissen-
schaftliche Sammlungen des Kantons Glarus; Robert S.
Jones, College of Guam, Agana; S. Jones, Central
Marine Fisheries Research Institute, Mandapam Camp,
India; Paul Kahsbauer, Naturhistorisches Museum,
Wien; Toshiji Kamohara, Kochi University; Wolfgang
Klausewitz, Natur-Museum und Forschungs-Institute,
Frankfurt; Adolf Kotthaus, Biologische Anstalt
Helgoland, Hamburg; Richard J. Kresja, California
State Polytechnic College, San Louis Obispo; Louis A.
Krumholz, University of Louisville; V. M. Makushok
and T. S. Rass, Institute of Oceanology, Academy of
Sciences of the U.S.S.R., Moscow; E. Marchal, Centre
de Recherches Oceanographiques, Abidjan; G. E. Maul,
Museu Municipal do Funchal; Giles W. Mead, Los
Angeles County Museum of Natural History; A. G. K.
Menon, Zoological Survey of India, Calcutta; Robert V.
Miller, National Marine Fisheries Service, Washington,
D.C.; Martin A. Moe, Oceanographic Mariculture In-
dustries, Riviera Beach, Fla.; Donald Moore, National
Marine Fisheries Service, Galveston; Basil Nafpaktitis,
University of California, Los Angeles; J(/rgen Nielsen,
Universitetets Zoologiske Museum, K<rt)enhavn; Ross F.
Nigrelli, New York Zoological Society, Brooklyn; John R.
Paxton, Frank H. Talbot, and Gilbert P. Whitley,

Australian Museum, Sydney; Michael J. Penrith, State
Museum, Windhoek; Craig Phillips, U.S. National
Aquarium, Washington, D.C.; E. Postel, Faculte des
Sciences, Rennes; Jiirgen Remane and J. P. Schaer,
Universite de Neuchatel; Robert R. Rofen, Aquatic
Research Institute, Hayward, Calif.; Robert L. Shipp,
University of South Alabama, Mobile; J. L. B. and
Margaret M. Smith, Rhodes University, Grahamstown;
Lorenzo Sorbini, Museo Civico di Storia Naturale,
Verona; Walter A. Starck, Chino, Calif.; F. Strauch,
Universitat Koln; Guido Tavani, Licio Giannelli, and
Andrea Cerrina Feroni, Museo di Paleontologia dell
'Universita di Pisa; Donald A. Thompson, University of
Arizona, Tucson; Luiz R. Tommasi, Institute
Oceanografico da Universidade de Sao Paulo; E. Tor-
tonese, Museo Civico di Storia Naturale, Genoa; Vittorio
Vialli, Istituto di Geologia e Paleontologia dell'Univer
sita di Bologna; Vlastimil Wagner, Sao Paulo, Brazil
Frank Williams, Guinean Trawling Survey, Lagos
Richard Winterbottom, Queen's University, Kingston,
Ontario; Donald Wohlschlag, University of Texas, Port
Aransas; Loren P. Woods, Field Museum of Natural
History, Chicago; Ralph W. Yerger, Florida State
University, Tallahassee.

Preliminary arrangements toward the publication of
this monograph were facilitated by a special grant from
the Lerner Fund for Marine Research of the American
Museum of Natural History.

The manuscript for this work has benefited greatly
from the helpful criticism of several colleagues who
volunteered their valuable time to peruse it: first,
Richard Winterbottom and Bruce B. Collette, and then

C. Lavett Smith and Stanley H. Weitzman. Their
generosity is much appreciated and their suggestions
mostly accepted and used.

Richard Winterbottom, while at Queens University,
Kingston, Ontario, and at both the Academy of Natural
Sciences of Philadelphia and the U.S. National Museum
of Natural History, Washington, D.C., provided many
highly valuable hours of discussion on the anatomy,
classification, and phylogeny of plectognath fishes and
freely shared his research data with me. His interest in,
and information given toward furthering, the work
leading to this monograph has been invaluable, even
though he is not necessarily in agreement with all of the
phylogenetic conclusions set forth here and probably
does not approve of the style of classification adopted
here. He is an esteemed colleague of impeccable creden-
tials and more than proven worth in the study of myology
and of plectognaths, as well as being a dear friend with
whom I happen to disagree on some points of how to
divide and group the supra- and subfamilial (but not
familial, on which we are mostly in agreement)
categories of Tetraodontiformes.

The errors of omission and commission that remain in
this monograph, after eluding my strenuous efforts to
eliminate them, are entirely my own and are not at all at-
tributable even in part to any of my so freely cooperating
colleagues acknowledged above.

This monograph is dedicated to the four friends most
intimately associated at the personal level with its
preparation: Helga Olga Schrader, Milan Kumpera
Tyler, Marisa Alenna Tyler, and Antoinette Carmela De
Fazio.

LITERATURE CITED

ABE, T.

1942. Taxonomic studies on the puffers (Tetraodontidae, Teleostei)
from Japan and adjacent regions— I. Vertebral variation. Palao
Trop. Biol. Stn. Stud. 2:477-496.

1944. Taxonomic studies on the puffers (Tetraodontidae, Teleostei)
from Japan and adjacent regions— II. Variation of dorsal fin.
Annot. Zool. Jpn. (Zool. Soc, Tokyo) 22:200-234.

1949a. Taxonomic studies on the puffers (Tetraodontidae, Teleostei)
from Japan and adjacent regions— III. Variation of anal fin.
Bull. Jpn. Soc. Sci. Fish. 14:123-144.

1949b. Taxonomic studies on the puffers (Tetraodontidae, Teleo-
stei) from Japan and adjacent regions — IV. Variation of caudal
fin. Bull. Jpn. Soc. Sci. Fish. 15:19-27.

1949c. Taxonomic studies on the puffers (Tetraodontidae, Teleo-
stei) from Japan and adjacent regions — V. Synopsis of the puf-
fers from Japan and adjacent regions. Bull. Biog. Soc. Jpn. 14:
1-15; 14:89-140.

1950-51. Taxonomic studies on the puffers (Tetraodontidae, Teleo-
stei) from Japan and adjacent regions — VI. Vtu-iation of pectoral
fin. Jpn. J. Ichthyol. 1:198-206 (1950); 1:272-283 (1951).

1952. Taxonomic studies on the puffers (Tetraodontidae, Teleostei)
from Japan and adjacent regions— VH. Concluding remarks, with
the introduction of two new genera, Fugu and Boesemanichthys.
Jpn. J. Ichthyol. 2:35-44; 2:93-97; 2:117-127.

1954. Taxonomic studies on the puffers from Japan and adjacent

regions— Corrigenda and addenda. Part I. Jpn. J. Ichthyol. 3:
121-128.
1960. Taxonomic studies on the puffers from Japan and adjacent
regions— Corrigenda and addenda. Part II. Jpn. J. Ichthyol.

ADENEY, R. J., and G. M. HUGHES.

1977. Some observations on the gills of the oceanic sunfish, Mola

mola. J. Mar. Biol. Assoc. U.K. 57:825-837.
AGASSIZ, L.

1833. Plecherches sur les poissons fossiles. Vol. 2, Part 1, p. 1-48.

Neuchatel.
1839. Recherches sur les poissons fossiles. Plate 38 of Vol. 5.

Neuchatel.
1842. Recherches sur les poissons fossiles. Plate 75 of Vol. 2.

Neuchatel.
1844a. Recherches sur les poissons fossiles. Vol. 1, p. 1-188.

Neuchatel.
1844b. Recherches sur les poissons fossiles. Vol. 2, Part 2, p. 73-

336. Neuchatel.
1844c. Recherches sur les poissons fossiles. Vol. 5, Part 1, p. 33-

122. Neuchatel.
1857. [Orthagoriscus mola, on grounds of internal anatomy, does

not belong in the same family with Diodon and Tetraodon, but is

the type of a distinct family). Untitled paper, Proc. Am. Acad.

Arts Sci. (Boston) 3:319.

403

1860. On the various existing systems of classification of fishes.

Proc. Am. Acad. Arts Sci. (Boston) 4:8-9.
ALESSANDRINI, A.

1839. De piscium apparatu respirationis turn speciatim Orthragor-

isci = Orthragoriscus alexandrini Ranzani. Nuovi Coram. Acad.

Sci. Inst. Bonon. (Accad. Sci. Istit. Bologna) 3:359-382.
AL-HUSSAINI, A. H.

1947. The feeding habits and the morphology of the alimentary

tract of some teleosts living in the neighbourhood of the Marine

Biological Station, Ghardaqa, Red Sea. Mar. Biol. Stn., Ghar-

daqa, Publ. 5:1-61.
ALLEN. G. R., and J. E. RANDALL.

1977. Review of the sharpnose pufferfishes (Subfamily Can-

thigasterinae) of the Indo-Pacific. Rec. Aust. Mus. 30:475-

517.
ANDREUCCI, R. D.

1966a. Aspectos morfologicos, macro e raicroscopicos, dos dentes de

Spheroides testudineus (L., 1735) (baiacu). Thesis ("Tese de

Doutoramento'"), Univ. Sao Jose dos Campos, 47 p.
1966b. Aspectos estruturais dos dentes de Spheroides testudineus.

Rev. Assoc. Paulista Cirurg. Dent. (Sao Paulo) 20:1-7.
1967a. Aspectos anatomicos dos dentes de Spheroides testudineus.

Prog. V Cong. Brasil. Anat. (Sao Paulo), p. 9.
1967b. Aspectos da odontogenese de Spheroides testudineus.

Cienc. Cult. (Sao Paulo) 19:360.
1968a. Macroscopic and microscopic morphology of Spheroides

testudineus teeth (checkered puffer). Acta Anat. (Basel) 71:448-

471.
1968b. Aspectos anatomicos dos dentes de Spheroides testudineus.

Rev. Assoc. Med. Brasil. (Sao Paulo) 14:245-246.

1969. Dos dentes de Spheroides testudineus (L., 1735) (baiacu-
mirim). Estudo autoradiografico e a luz polarizada. Thesis
("Tese de Livre Docencia"), Univ. Sao Jose dos Campos, 51 p.

1970. Sobre a estrutura das lojas denticulares de Spheroides testu-
dineus (baiacu-mirim). Rev. Brasil. Pesq. Med. Biol. (Sao Paulo)
3:63-67.

ANDREUCCI, R. D., and G. BLUMEN.

1971. Radioautographic study of Spheroides testudineus denticles
(checkered puffer). Acta Anat. (Basel) 79:76-83.

ANDREUCCI, R. D., and H. A. BRITSKI.

1968a. Aspectos anatomicos dos dentes de baiacu-de-espinho,
Chilomycterus spinosus. Prog. VI Cong. Brasil. Anat. (Campi-

Studies in Indian puffers or globe fishes, m. The i
of Tetrodon oblongus, Bloch. J. Bombay Univ. 5(pt. V):44-

57.

13.

J Mola t

Cienc. Cult. (Sao

1968b. Aspectos da odontogenese c

Paulo) 20:346-347.
1969a. Macroscopic and microscopic morphology of the Mola mola

teeth (sun fish). Rev. Brasil. Pesq. Med. Biol. (Sao Paulo) 2:

159-162.
1969b. Aspectos macro e microscopic dos dentes de Lagocephatus

laeuigatus (L., 1758) e de Colomesus asellus (Mueller & Troschell,

1848). Cienc. Cult. (Sao Paulo) 21:483-484.

1970. Sobre os dentes de Liosaccus intermedius, Ribeiro, 1904 e
Masturus lanceolatus (Lienard, 1840). Prog. XXH Reun. Anu.
Soc. Brasil. Pesq. Cienc. (Salvador), p. 305.

1971. Dos dentes de "Tetraodontidae". Prog. Vm Cong. BrasU.
Anat. (Pelotas), p. 36.

ARBOCCO, G.

1957. Intomo all'apptu-ato digerente dei Monacantidi (Pesci Pletto-
gnati). Boll. Zool. (Unione Zool. Ital., Torino) 23:779-783.
ARSUFFI, E.

1939. BeitrSge zur vergleichenden Histologie und Histogenese der
Zahne (Untersuchungen am Gebiss von Labridae, Sparidae und
Gymnodontes). Z. Zellforsch. Mikrosk. Anat. (Berl.), Ser. A,
29:670-693.
ARTEDl, P.

1738. Petri Artedi sueci genera piscium. Ichthyologiae, pars HI.
Lugduni, 88 p.
AWATI, P. R., and D. W. BAL.

1933. Studies in Indian puffers or globe fishes: I. The endo-skeleton
of Tetrodon oblongus— B\och. J. Bombay Univ. l(pt. V):74-95.

1934. Studies in Indian puffers or globe fishes. II. The blood vascu-
lar system of Tetrodon oblongus (Bloch). J. Bombay Univ. 2(pt.
V):58-74.

BARNARD, K. H.

1927. A monograph of the marine fishes of South Africa. Ann. S.

Afr. Mus. 21:419-1065.
1935. Notes on South African marine fishes. Ann. S. Afr. Mus.
30:645-658.
BARTOLINI, M.

1941. Ricerche suUa dentatura di alcuni plettognati. Thalassia
(Rovigno, Italy) 5:1-11.
BEAUREGARD, M.

1893. Contribution a I'etude de Orthagoriscus truncatus (Flem.).
Bull. Soc. Sci. Nat. Quest. Fr. (Nantes) 3:229-246.
BENL, G. VON.

1957. Carinotetraodon chlupatyi nov. gen., nov. spec, ein Kugel-

fisch mit Kamm und Kiel. Opusc. Zool. (Munich) 5:1-4.
1959. Zur Nomenklatur siamesischer Kugelfischformen. Aquar.-u.
Terrar.-z. (Stuttgart) 12:42-44.
BENL, G. VON, and P. CHLUPAT\'.

1957. Carinotetraodon, eine neue Kugelfischgattung. Aquar.-u.
Terrar.-z. (Stuttgart) 10:227-229.
BERG, L. S.

1940. Classification of fishes both Recent and fossil. Trav. Inst.
Zool. Acad. Sci. URSS (Leningrad) 5:87-517.
BERRY, F. H., and W. J. BALDWIN.

1966. Triggerfishes(Balistidae)ofthe eastern Pacific. Proc. Calif.
Acad. Sci., Ser. 4, 34:429-474.
BERRY, F. H., and L. E. VOGELE.

1961. Filefishes (Monacanthidae) of the western North Atlantic.
U.S. Fish Wildl. Serv., Fish. Bull. 61:61-109.

BLAINVILLE, H. M.

1818. Nouveau dictionnaire d'historie naturelle. Pla-por (Vol. 27),
586 p. Paris.
BLEEKER, P.

1852a. Bijdrage tot de kennis der Blootkakige [Gymnognathenl
visschen van den Soenda-Moljkschen Archipel, met beschrijving
van eenige nieuwe soorton. Verb. Batavia Genoot. Kunsten Wet.
(Batavia) 24(10):l-26.
1852b. Bijdrage tot de kennis der Balistini en Ostraciones van den
Indischen Archipel. Verb. Batavia Genoot. Kunsten Wet. (Bata-
via) 24(ll):l-38.

1865. Atlas ichthyologique des Indes Orientales Neerlandaises.
Vol. 5 - Baudroies, Ostracions, Gymnodontes, Balistes, p. 1-152.
Amsterdam.

1866. Systema Balistidorum, Ostracionidorum, Gymnodontidor-
umque revisum. Ned. Tijdschr. Dierkd. (Amsterdam) 3:8-19.

BLISS, R., JR.

1872. On the osteology of the anterior vertebrae in Doras niger, with
a comparison of the structure of the dorsal fm in Doras and Ba-
tistes. Proc. Boston Soc. Nat. Hist. 14:3-12.
BLOCH, M. E.

1786. Naturgeschichte der auslandischen Fische. Vol. 2, p. 1-160.
Berlin.
BLOCH, M. E., and J. G. SCHNEIDER.

1801. Systema ichthyologiae iconibus ex illustratum. Berolini,
584 p.
BLOT, J.

1969. Studie e richerche sui giacimenti Terziari di Bolca. I. Les
poissons fossiles du Monte Bolca. Mem. Mus. Civ. Stor. Nat.
Verona, Mem. fuori, Ser. 2, 525 p.
BOESEMAN, M.

1962. Triodon macropterus versus Triodon bursanus; an attempt
to establish the correct name and authorship. Zool. Meded. R.
Nat. Hist. Leiden 38:77-85.

BOHLKE, J. E., and C. C. G. CHAPLIN.

19(38. Fishes of the Bahamas and adjacent tropical waters. Acad.
Nat. Sci. Phila.. 771 p.
BONAPARTE, C. L.

1832. Saggio d'una distribuzione metodica degli animali vertebrati

a sangue freddo. 86 p. Rome.
1841a. Iconografia della fauna Italica per le quattro classi degli

404

animali vertebrati. Vol. 3. Pesci. Unpaginaged text with 78 plates.

1841b. A new systematic arrangement of vertebrated
Trans. Linn. Soc. Lend. 18:247-304.
BONNATERRE, J.

1788. Tableau encyclopedique et methodique des trois rfegnes de
la nature. Ichthyologie. Panckoucke, Paris, 215 p.
BORCEA, J.

1907. Observations sur la musculature branchiostegale des t^le-
osteens. Ann. Sci. Univ. Jassy 4:202-225.
BORN, G.

1827. Beraerkungen uber den Zahnbau der Fische. Z. Organ.
Phys. (Eisenach) 1:182-206.
BOULENGER, G. A.

1904a. A synopsis of the suborders and families of teleostean fishes.

Ann. Mag. Nat. Hist., Ser. 7., 13:161-190.
1904b. Teleostei. In Harmer, S. F., and A. E. Shipley (editors),
Fishes. Vol. VII, p. 539-727. MacMillan and Co., Ltd., London.
1907. Zoology of Egypt: the fishes of the Nile. 578 p. London.
BREDER, C. M., JR., and E. CLARK.

1947. A contribution to the visceral anatomy, development, and
relationships of the Plectognathi. Bull. Am. Mus. Nat. Hist. 88;
287-319.
BRIDGE, T. W.

1896. The mesial fins of ganoids and teleosts. J. Linn. Soc. Lond.,
Zool. 25:530-602.
BRITSKI, H. A., and R. D. ANDREUCCI.

1971. Sobre os dentes dos peixes Molidae. Prog. XXm Reun.
Anu. Soc. Brasil. Pesq. Cienc. (Curitiba), p. 287.
BROUSSONET, P. M.

1780. Memoire sur les differentes especes de chiens de mer. Mem.
Acad. R. Sci. (Paris), 1780:641-680.
BROHL, C. B.

1856. Osteologisches aus dem Pariser Pflanzengarten. Wien:A-G

-H - 76 -I- i-xliii.
1880. Zootomie aller Thierklassen, nach Autopsien skizzirt. Lief.

24. Wien (160 plates in toto published between 1877-1888).
1891. Balistes-Skelet. Wien (separate plate).
BURKENROAD, M. D.

1931. Notes on the sound-producing fishes of Louisiana. Copeia
1931:20-28.
CADENAT, J.

1959. Notes d'ichthyologie ouest-africaine. XXTV.— Molidae ouest-
africains avec description d'une espece nouvelle: Pseudomola las-
sarati de Cote d'lvoire. Bull. Inst. Fr. Afr. Noire (Dakar), Ser. A.,
21:1112-1118.
CLARK, E., and H. A. F. GOHAR.

1953. The fishes of the Red Sea: Order Plectognathi. Mar. Biol.
Stn., Ghardaqa, Red Sea, Publ. 8:1-80.
CLELAND, J.

1862. On the anatomy of the short sun-fish (Orthagonscus mola).
Nat. Hist. Rev. (Lond.), New Ser., 2:170-185.
CLOTHIER, C. R.

1939. The trigger mechanism of a trigger fish {Capnscus polylepis).
Calif. Fish Game 25:233-236.
COPE, E. D.

1871a. Observations on the systematic relations of the fishes. Am.

Nat., 5:579-593.
1871b. (On Plectognathi and Lophobranchii). Untitled paper.
Proc. Acad. Nat. Sci. Phila., 1871:157-158.
CUNNINGHAM, J. T.

1910. On the marine fishes and invertebrates of St. Helena. Proc.
Zool. Soc. Lond., 1910:86-130.
CUVIER, G.

1805. Lemons d'anatomiecomparee. 5 vol. Paris. [Vol 1, 518 p.;
Vol. n, 697 p.; Vol. HI, 558 p.; Vol. IV, 539 p.; Vol. V, 368 p.]

1817. Le regne animal. Vol. U, 532 p. Paris.

1818. Sur les Diodons, vulgairement orbes-epineux. Mem. Mus.
Natl. Hist. Nat. 4:121-138.

1829. Le r^gne animal. Ed. 2., Vol. U, 406 p. Paris.
DAMES, W.

1883. Uber eine tertiare Wirbelthierfauna von der westlichen Insel
des Birket-el-Qurun im Fajum (Aegypten). Sitzungsber. K.

Preus. Akad. Wiss. Beri., Phys.-Math. CI., 1883, Erst. Halb., p.

129-153.
DANILCHENKO, P. G.

1960. Bony fishes of the Maikop deposits of the Caucasus. Izdatel'-

stro Akad. Nauk. SSSR, Moskva. (Jerusalem, Israel Prog. Sci.

Transl., 1967, 247 p.]
DARESTE, C.

1849. Observations sur I'osteologie du poisson appele Triodon mac-
roptere. Ann. Sci. Nat. (Paris), Zool., Ser. 3, 12:68-83.

1850. Recherches sur la classification des poissons de I'ordre des
Plectognathes. Ann. Sci. Nat. (Paris), Zool., Ser. 3, 14:106-
133.

1872a. Sur les affinit^s naturelles des poissons de la famille des

Batistes. C. R. Acad. Sci. (Paris) 74:1527-1530.
1872b. On the natural affinities of the Balistidae. Ann. Mag. Nat.

Hist., Ser. 4, 10:68-70.
1872c. Etudes sur les types ostiologiques des poissons osseux.

C. R. Acad. Sci. (Paris) 75:942-946, 1018-1021, 1086-1089, 1172-

1175, 1253-1256.
DE BEAUFORT, L. F.

1926. On a collection of marine fishes from the Miocene of South

Celebes. Jaarb. Mijwezen Ned. -Indie (Batavia), Verb., 1925:

115-166.
1962. The fishes of the Indo-Australian Archipelago. Vol. XI,

481 p. Leiden.
DEINSE, A. B. VAN.

1953. Fishes in upper Miocene and lower Pleistocene deposits in the

Netheriands. Med. Geol. Sticht. (Geol. Dienst Haarlem), New

Ser., 7:5-11.
DEKKERS, W. J.

1975. Review of the Asiatic freshwater puffers of the genus Tetrao-

don Linnaeus, 1758 (Pisces, Tetraodontiformes, Tetraodontidae).

Bijdr. Dierkd. 45:87-142.
D'ERASMO, G.

1946. L'ittiofauna Cretacea dei dintomi di Comeno nel Carso tries-

tino. Atti R. Accad. Sci. Fis. Math. (Soc. R. Napoli), Ser. 3, 2:1-

136.
DUMERIL, a. H.

1855. Note sur un travail infedit de Bibron relatif aux poissons
Plectognathes Gymnodontes (Diodons et Tetrodons). Rev. Mag.
Zool. Pure AppL, Ser. 2, 7:274-282.

DUMERIL, A. M.

1806. Zoologie analytique, ou methode naturelle de classification
des animaux. 344 p. Paris.

1856. Ichthyologie analytique au essai d'une classification naturelle
des poissons, a I'aide de tableaux synoptiques. 511 p. Paris.

DUNKER, G., and E. MOHR.

1931. Die Fische der Siidsee - Expedition der hamburgischen wis-
senschafthchen Stiftung 1908-1909. 3 Teil. Mitt. Zool. Staat.
Zool. Mus. (Hamburg) 44:57-84.

EBINA, K.

1932. Sexual dimorphism in a file-fish, Canthennus modestus
(Giinther). J. Imp. Fish. Inst. 27:15-18.

EIGENMANN, C. H.

1885. A review of the genera and species of Diodontidae found in
American waters. Ann. New York Acad. Sci. 3:297-311.
FISH, M. P.

1954. The character and significance of sound production among
fishes of the western North Atlantic. Bull. Bingham Oceanogr.
Coll. 14(3):1-109.
FISH, M. P., A. S. KELSEY, JR., and W. H. MOWBRAY.

1952. Studies on the production of underwater sound by North At-
lantic coastal fishes. J. Mar. Res. 11:180-193.
FITZINGER, L. J.

1873. Versuch einer natiirlichen Classification der Fische. Sit-
zungsber. Akad. Wiss. (Wien), Math.-Naturwiss. Kl. 67:5-58.
FORD, E.

1937. Vertebral variation in teleostean fishes. J. Mar. Biol. Assoc.
U.K. 22:1-60.
FRASER-BRUNNER, A.

1935a. Notes on the plectognath fishes. 1. A synopsis of the genera
of the family Balistidae. Ann. Mag. Nat. Hist., Ser. 10, 15:658-
663.

1935b. Notes on the plectognath fishes. II. A synopsis of the genera
of the family Ostraciontidae. Ann. Mag. Nat. Hist., Ser. 10, 16:
313-320.

1940a. Notes on the plectognath fishes. IH. On Monacanthus setifer
Bennett and related species, with a key to the genus Stephanoiepis
and descriptions of four new species. Ann. Mag. Nat. Hist., Ser.
11, 5:518-535.

1940b. Notes on the plectognath fishes. IV. Sexual dimorphism in
the family Ostraciontidae. Ann. Mag. Nat. Hist., Ser. 11, 6:390-
392.

1940c. The fishes of the genus Pseudomonacanthus, with descrip-
tions of two new species. Bull. Raffles Mus. 16:62-67.

1941a. Notes on the plectognath fishes. V. The families of tria-
canthiform fishes, with a synopsis of the genera and a description
of a new species. Ann. Mag. Nat. Hist., Ser. 11, 7:420-430.

1941b. Notes on the plectognath fishes. VI. A synopsis of the genera
of the family Aluteridae, and descriptions of seven new species.
Ann. Mag. Nat. Hist., Ser. 11, 8:176-199.

1941c. Notes on the plectognath fishes. VII. The Aracanidae, a dis-
tinct family of ostraciontid fishes, with descriptions of two new
species. Ann. Mag. Nat. Hist., Ser. U, 8:306-313.

1943. Notes on the plectognath fishes. VHI. The classification of
the suborder Tetraodontoidea, with a synopsis of the genera.
Ann. Mag. Nat. Hist., Ser. 11, 10:1-18.

1950. Studies in plectognath fishes from the "Dana"-Expeditions.
I. An interesting new genus of triacanthodid fishes from the Celebes
Sea. Dana-Rep. Carlsberg. Found. 35:1-8.

1951. The ocean sunfishes (family Molidae). Bull. Br. Mus. (Nat.
Hist.) Zool. 1:89-121.

FROST, C. A.

1930. A comparative study of the otoliths of neopterygian fishes
(concluded). Ann. Mag. Nat. Hist., Ser. 10, 5:621-627.
CARMAN, S.

1891. On Batistes uetula Linne. Bull. Essex Inst. 22:53-55.
GARNAUD, J.

1956. Note sur le squelette cephalique de Batistes capriscus L.
Bull. Inst. Oceanogr. 1081:1-12.
GEOFFROY SAINT-HILAIRE, I.

1827. Histoire naturelle des poissons du Nil. In Description de
I'Egypte, . . . Histoire Naturelle (Paris), 1:141-579.
GHIGI, A.

1905a. Ricerche suUa dentatura dei teleostei. Arch. Zool. (Unione

Zool. Ital., Napoli) 2:439-462.
1905b. Dimostrazione de preparati sullo sviluppo dei denti del

Batistes capriscus. C. R. VI Cong. Int. Zool. (Berne), p. 327.
1921. Sulla dentatura di alcumi plettognati iOstracion e Tetrodon).
Rend. Sess. Accad. Sci. 1st. Bologna, CI. Sci. Fish., New Ser., 15:
138-140.
GILL, T.

1872. Arrangement of the families of fishes, or classes Pisces, Marsi-
pobranchii, and Leptocardii. Smithson. Misc. Collect. 11(247):
1-49.
1885. Synopsis of the plectognath fishes. Proc. U.S. Natl. Mus.

7:411-427.
1888a. Some extinct scleroderms. Am. Nat. 22:446-448.
1888b. The extinct scleroderms. Am. Nat. 22:828-830.
1889. Note on the genus Spheroides. Proc. U.S. Natl. Mus. 11:

1892a. Notes on the Tetraodontoidea. Proc. U.S. Natl. Mus. 14:

705-720.
1892b. Note on the genus Chonerhinus or Xenoptcru*. Proc. U.S.

Natl. Mus. 14:697-699.
1897. The distinctive characters of the Molinae and Ranzaniinae.

Science, New Ser., 6:966-967.
GMEUN, J. F.

1788. Caroli a Linni . . . Systems Naturae. Tomus I, pars HI, p.

1033-1616. Leipzig.
GOELDI, E. A.

1884. Kopfskelett und Schultergflrtel von Loricaria cataphracta.

Batistes capriscus und Accipenser ruthenus. Vergleichend-ana-

tomische und entwicklungs-geschichtliche Studien zur Deck-

knockenfrage. Jenaische Z. Naturwiss., New Ser. 10, 17:401-

447.

GOEPPERT, E.

1895. Untersuchungen zur Morphologie der Fischrippen. Mor-
phol. Jahrb. (Leipzig) 23:145-217.
GOETTE, A.

1879. Beitrage zur vergleichenden Morphologie des Skeletsystems
der Wirbelthiere. II. Die Wirbelsaule und ihre Anhange. 5. Die
Teleostier. Arch. Mikrosk. Anat. (Bonn) 16:117-142.

GOODE, G. B.

1880. A study of the trunk-fishes (Ostraciontidae), with notes upon
the American species of the family. Proc. U.S. Natl. Mus. 2:261-
283.

GOODRICH, E. S.

1909. Part IX. Vertebrata Craniata. (First fascicle: Cyclostomes
and fishes). In Lankester, R. (editor), A treatise on zoology, p.
1-518. Adam and Charles Black, London.
1930. Studies on the structure and development of vertebrates.
837 p. London.
GOODSm, J.

1841 . On certain peculiarities in the structure of the short sun-fish
(Orthagoriscus mola). Edinburgh New Philos. J. 30:188-194.
GOSLINE, W. A.

1961. The perciform caudal skeleton. Copeia 1961:265-270.
1968. The suborders of perciform fishes. Proc. U.S. Natl. Mus.

124(36471:1-77.
1971. Functional morphology and classification of teleostean fishes.
Univ. Press. Hawaii, Honolulu, 208 p.
GOOAN, a.

1770. Historia piscium (Histoire des poissons). Amand K5nig,
Strasbourg, 252 p.
GREEN, E. H.

1901. The chemical composition of the sub-dermal connective tis-
sue of the ocean sunfish. Bull. U.S. Fish Comm. 19(for 1899):321-
324.

GREEN, E. H., and R. W. TOWER.

1902. The organic constituents of the scales of fish. Bull. U.S.
Fish Comm. 21(for 19011:97-102.

GREENWOOD, P. H., D. E. ROSEN, S. H. WETTZMAN, and G. S.
MYERS.

1966. Phyletic studies of teleostean fishes, with a provisional classi-
fication of living forms. Bull. Am. Mus. Nat. Hist. 131:339-454.
GREGORY, W. K.

1907 . The orders of teleostomous fishes. A preliminary review of the
broader features of their evolution and taxonomy. Aim. New
York Acad. Sci. 17:437-508.

1933. Fish skulls: A study of the evolution of natural mechanisms.
Trans. Am. Philos. Soc. 23:75-481.

GREGORY. W. K., and H. C. RAVEN.

1934. Notes on the anatomy and relationships of the ocean sunfish
(Mola mola). Copeia 1934:145-151.

GRENHOLM, A.

1923. Studien uber die Flossenmuskulatur der Teleostier. Ph.D.
thesis, Univ. Uppsala, 296 p.
GRIEB, A. W.

1935. Besonderheiten im Bau des Pharyngealapparates bei Fischen
aus Ordnung Plectognathi. [In Russ., with Ger. summ.) Tr.
Biol. Inst. Peterhof (Trav. Inst. Geol. Peterhof) 13-14:88-95.

GRONOVIUS, L. T.

1754. Museum ichthyologicum Vol.1, 70 p. Lugduni.

1763. Zoophylacii Gronoviani .... 136 p. Lugduni.
GUDGER, E. W.

1935. Some undescribed young of the pointed-tailed ocean sun-
fish, Masturus lanceotatus. Copeia 1935:35-38.

1936. From atom to colossus. Growth of the pointed-tailed ocean
sunfish from one-tenth inch to ten feet — an increase of 60,000,000
times its original weight. Nat. Hist. 38:26-30.

1937a. The natural history and geographical distribution of the
pointed-tailed ocean sunfish {Masturus tanceolatus), with notes on
the shape of the tail. Proc. Zool. Soc. London, Ser. A, p. 363-

1937b. The structure and development of the pointed tail of the
ocean sunfish, Masturus tanceolatus. Ann. Mag. Nat. Hist., Ser.
10, 19:1-46.

1939. Three six-inch pointed-tailed ocean sunfish, Masturus lance-

olatus, the largest post-larva on record. J. Elisha Mitchell Sci.
Soc. 55:305-313.
GUDGER, E. W.. and S. M. MacDONALD.

1935. The rarest of the ocean sunfishes. Sci. Mon. (New York)
41:396-408.

GONTHER. A.

1870. Catalogue of the fishes in the British Museum. Vol. 8, 549 p.
London.

1880. An introduction to the study of fishes. 720 p. Edinburgh.
HAINES, R. W.

1937. The posterior end of Meckel's cartilage and related ossifica-
tions in bony fishes. Q. J. Microsc. Sci. (Lend.), New Ser., 80:1-
38.
HARTING, P.

1865. Notices zoologiques, anatomiques et histiologiques, sur 1'-
Orthagoriscus ozodura; suivies de considerations sur I'osteg^nise
des teltostiens en g^n^ral. Verb. K. Akad. Wet. (Amsterdam)
ll(2):l-48.
HEER. 0.

1865. Die Urwelt der Schweiz. 622 p. Zurich.
1876. The primaeval world of Switzerland. Vol. 1,393 p. London.
HERTWIG, O.

1881. Uber das Hautskelet der Fische. Dritte Abtheilung. Mor-
phol. Jahrb. (Leipzig) 7:1-42.

HILGENDORF, F. M.

1886. Ueber Schliffen sogenannter Zusammengesetzter FischzShne.
Sitzungsber. Gesamten Naturforscher. Freunde Berl., 1886:87-
94.

1893. Ueber die Bezahnung der Gattung Mola (Orthagoriscus) .
Sitzungsber. Gesamten Naturforscher. Freunde Berl., 1893:3-4.
HOLLARD, H.

1853. Monographic de la famille des Balistides. Suite 1. Introduc-
tion et premiere partie. Ann. Sci. Nat. (Paris), Zool., Ser. 3, 20:
71-114.

1854a. Monographic de la famille des Balistides. Suite 2. Deuxiime
partie. Ann. Sci. Nat. (Pans), Zool., Ser. 4, 1:39-72, 303-339.

1854b. Monographie de la famille des Balistides. Suite 3. Ann.
Sci. Nat. (Paris), Zool., Set. 4, 2:321-366.

1855. Monographie de la famille des Balistides. Suite 4. Ann. Sci.
Nat. (Pans), Zool., Ser. 4, 4:5-27.

1857a. Monographie de la famille des Ostracionides. Ann. Sci.
Nat. (Paris), Zool.. Ser. 4, 7:121-170.

1857b. Etudes sur les Gymnodontes et en particulier sur leur os-
teologie et sur les indications qu'elle peut foumir pour leur classi-
fication. Ann. Sci. Nat. (Paris), Zool., Ser. 4, 8:275-328.

1860. Mimoire sur le squelette des poissons Plectognathes, etudi^
au point de vue des caractferes qu'il peut foumir pour la classifica-
tion. Ann. Sci. Nat. (Paris), Zool., Ser. 4, 13:4-46.
HOLMGREN, N., and E. STENSIO.

1936. Kranium and Visceralskelett der Akranier, Cyclostomen und
Fische. In Bolk, L., E. Goppert, E. Kallius, and W. Lubosch
(editors), Handbuch der vergleichenden Anatomic der Wirbeltiere.
Vol. 4, p. 233-500. Berlin.

HORA, S. L.

1924. Notes on fishes in the Indian Museum. VII. On a new genus
of "globe-fishes" (Fam. Tetraodontidae). Rec. Indian Mus. 26:
579-582.

1925. Notes on fishes in the Indian Museum. XIV. On a new species
of the genus Kanduka Hora. Rec. Indian Mus. 27:468-469.

HORSHELMANN, C.

1866. Ein Beitrag zur Anatomie der Zunge der Fische. Ph.D.
thesis, Univ. Dorpat, 41 p.
HUBBS, C. L.

1919. A comparative study of the bones forming the opercular series
of fishes. J. Morphol. 33:61-71.
INGER, R. F.

1953. A new fish from north Borneo, genus Tetraodon. Fieldiana,
Zool. 34:149-152.
ISOKAWA, S.

1955. Morphological studies of teeth of fishes. IV. Teeth of filefishes.
Zool. Mag. (Tokyo) 64:194-197.
IWAI, T.

1964. A comparative study of the taste buds in gill rakers and gill

arches of teleostean fishes. Bull. Misaki Mar. Biol. Inst., Kyoto
Univ. 7:19-34.
JACOBS, W.

1935. Ueber laut-erzeugung bei Fischen. Natur Volk 65:155-166.
JENYNS, L.

1842. The zoology of the voyage of H.M.S. Beagle. Part IV. Fish.
172 p. London.
JORDAN, D. S.

1905. A guide to the study of fishes. 2 vols. Henry Holt and Co.,

New York. [Vol. 1, 624 p.; Vol. 2, 599 p.)
1923. A classification of fishes, including families and genera as
far as known. Stanford Univ. Publ., Univ. Ser., Biol. Sci. 3:77-
243.
JORDAN, D. S., and C. L. EDWARDS.

1887. A review of the American species of Tetraodontidae. Proc.
U.S. Natl. Mus. 9:230-247.
JORDAN, D. S., and B. W. EVERMANN.

1898. The fishes of North and Middle America. Part n. Bull. U.S.
Natl. Mus. 47:1241-2183.
JORDAN, D. S., and H. W. FOWLER.

1903. A review of the trigger-fishes, file-fishes, and trunk-fishes of
Japan. Proc. U.S. Natl. Mus. 25:251-286.
JORDAN, D. S., and J. O. SNYDER.

1901. A review of the gymnodont fishes of Japan. Proc. U.S. Natl.
Mus. 24:229-264.
KASCHKAROFF, D. N.

1914a. Materialien zur vergleichenden Morphologic der Fische.
Vergleichendes Studium der Organisation von Plecjognathi. Bull.
Soc. Imp. Nat. (Moscow), New Ser., 27:263-370.
1914b. Zur Kenntnis des feineren Baues und der Entwickelung des
Knochens bei Teleostiem. I. Die Knochenentwickelung bei Ortha-
goriscus mola. Anat. Anz. 47:113-138.
1916. La structure et le developpement de I'os d' Orthagoriscus

mola. Zool. Zh. (Rev. Zool. Russe, Moscow) 1:136-138.
1925. Materials to the knowledge of bone structure in fishes. |In
Russ.] Bull. Soc. Nat. (Moscow), New Ser., 34:207-251.
KAUP, J. J.

1855. Uebersicht uber die Species einiger Familien der Scleroder-
men. Arch. Naturgesch. (Beri.) 21:215-233.
KLAUSEWITZ, W.

1957a. Tetraodon somphongsi n. sp., ein weiterer neuer Kugelfisch
aus Thailand (Pisces, Tetraodontidae). Senckenb. Biol. 38:205-
208.
1957b. Eine neue Kugelfisch-Art aus Siam: Tetraodon somphongsi.

Aquar.-u. Terrar.-z. (Stuttgart) 10:115-116.
1964. Uber den Maskenkugelfisch,Aro£hrondiadematiiS. Aquar.-
u. Terrar.-z. (Stuttgart;) 17:333-334.
KLEIN, E. E.

1872. Beitrage zur Osteologie des Genus Batistes Cuv. (Scleroder-

mi). Jahresh. Ver. Vateri. Naturkd. Wurttemberg 28:262-300.
1881. Beitrage zur Osteologie der Fische. Jahresh. Ver. Vateri.
Naturkd. Wurttemberg 37:325-360.

1884. Beitrage zur Bildung des Schadels der Knochenfishe. I.
Jahresh. Ver. Vateri. Naturkd. Wurttemberg 40:129-257.

1885. Beitrage zur Bildung des Schadels der Knochenfishe. n.
Jahresh. Ver. Vateri. Naturkd. Wurttemberg 41:107-261.

1886. Beitrage zur Bildung des Schadels der Knochenfishe. HI.
Jahresh. Ver. Vateri. Naturkd. Wurttemberg 42:205-300.

KLEIN, J. T.

1742. Historiae piscium naturalis. Missus tertius. 48 p. Gedani.
KOLUKER, R. A.

1859. On the different types in the microscopic structure of the
skeleton of osseous fishes. Ann. Mag. Nat. Hist., Ser. 3, 4:67-
77.

1860. Ueber die Knochen von Orthagoriscus. Sitzungber. Phys.-
Med. Ges. (Wurtzburg), 1859:38.

KURONUMA, K.

1943. Osteological study on Japanese fish. 1. Comparative oste-
ology, especially of cranial characters, in tetraodontid genera
Liosaccus, Lagocephalus and Sphoeroides. J. Sigen Kagaku
Kenkyusyo (Res. Inst. Nat. Resour., Tokyo) 1:119-130.
LACEPEDE, B. G. E.

1798-1803. Histoire naturelle des poissons. 5 vols. Paris. [Vol.

I (1798), 532 p.; Vol. D (1800), 632 p.; Vol. ffl (1802), 558 p.; Vol.

IV (1802), 728 p.; Vol. V (1803), 803 p.)
LATREILLE, P. A.

1825. Families naturelles du ligae animal, exposies succinctement

et dans un ordre analytique, avec I'indication de leurs genres. 570

p. Paris.
LE DANOIS, Y.

1954. Sur le dimorphisme sexuel des poissons de la famille des Di-
odontides. C. R. Acad. Sci. (Paris) 238:2354-2356.

1955. Sur le remaniement du sous-ordre des poissons Plectognaths
et la definition d'un nouveau sous-ordre: les Orbiculates. C. R.
Acad. Sci. (Paris) 240:1933-1934.

1956. A propos du syst^me muco-sensoriel des poissons Orbicu-
lates. C. R. Acad. Sci. (Paris) 242:684-686.

1959. Etude osteologique, myologique et systematique des poissons

du sous-ordre des Orbiculates. Ann. Inst. Oc^anog. 36:1-273.
1961a. Remarques sur les poissons Orbiculates du sous-ordre des

Ostracioniformes. Mem. Mus. Natl. Hist. Nat. (Paris), New

Ser. (A, zool.), 19:207-338.
1961b. Transitions morphologiques et osttologiques entre les

especes du genre Dichotomycterus (Tetraodontidae). Bull. Soc.

Zool. Fr. 86:694-700.
LEIS, J. M.

1977. Development of the eggs and larvae of the slender mola, Ran-
zania laeuis (Pisces, Molidae). Bull. Mar. Sci. 27:448-466.

1978. Systematics and zoogeography of the porcupinefishes
{Diodon, Diodontidae, Tetraodontiformes), with comments
on egg and larval development. Fish. Bull., U.S. 76:535-567.

LERICHE, M.

1905. Les poissons fiocines de la Belgique. M^m. Mus. R. Hist.
Nat. Belg. 3:49-228.

1906. Contribution a 1 'etude des poissons fossiles du nord de la
France et des regions voisines. Mem. Soc. Gtol. Nord (Lille)
5:1-430.

1907. Revision de la faune ichthyologique des terrains niogtaes du
bassin du Rhone. Assoc. Fr. Av. Sci., C. R. 35" Sess., (Lyon,
Atti Quarts Riun. Sci. Ital.. Padova. 1842:244-245.

1919. Sur des poissons fossiles de la region c6ti6re du Congo et sur
la presence de I'Eocene dans cette region. C. R. Hebd. Seanc.
Acad. Sci. (Paris) 169:479-481.

1920. Note preliminaire sur des poissons nouveaux du Paleoctoe
et de I'Eocene de la region cotifere du Congo. Rev. Zool. Afr.
Suppl. Hot. (Brussels) 8:80-86.

1926. Les poissons neogenes de la Belgique. M6m. Mus. R. Hist.
Nat. Belg. 32:365-472.
LE'YDIG, F.

1857. Lehrbuch der Histologic des Menschen und der Thiere. 551
p. Frankfurt.
LINNAEUS, C.

1758. Systema naturae Editio Decima, Tomus I, 823 p.

Holmiae.

1766. Systema naturae Editio Duodecima, Tomus I, 1327

p. Holmiae.
MASSALONGO, A. B.

1857. Vorlaufige Nachricht uber die neueren palfiontologischen
Entdeckungen am Monte Bolca. Neues Jahrb. Miner., Geogr.,
Geol. Petrefakten-Kunde (Stuttgart), 1857:775-778.
MA'reR, F. J. C.

1864. Ueber den Bau des Gehims der Fische in Beziehung aug eine

daraug Gegriindete eintheilung dieser Thierklasse. Nov. Act.

Acad. Caes. Leopold. -Carol. German. Nat. Curiosorum (Verb. K.

Leopold. -Carol. Dtsch. Akad. Naturforsch., Dresden) 30:1-40.

MCALLISTER, D. E.

1968. Evolution of branchiostegals and the classification of teleo-
stome fishes, living and fossil. Nat. Mus. Canada, Bull. 221,
239 p.
McCULLOCH. A. R., and E. R. WAITE.

A revision of the
S. Aust. 39:477-49

and iUs allies. Trans. R.

MEEK, A
1904.

Notes on the auditory organ and the orbit of Orthagoriscus

mola. Anat. Anz. 25:217-:

MOBIUS, k. a.

1889. Batistes aculeatus. ein trommelnder Fisch. Sitzungsber.
[K. Preuss.) Akad. Wiss. Berl., Phys.-Math. Kl., 1889, Erst.
Halb., p. 999-10006.
MOHR, E.

1928. Zur Mechanik der vorderen Ruckenstacheln bei Batistes und
Zeus. Zool. Anz. 75:49-53.
MONOD, T.

1950. Sur trois dispositifs de "veirouillage" osseux chez des pois-
sons. C. R. Prem. Conf Int. Afr. Quest, Inst. Fr. Aft. Noire (Da-
kar) 1:211-215.

1959a. Les nageoires ventrales du Batistes fordpatus Gmelin 1789.
Bull. Inst. Fr. Afr. Noire (Dakar), Ser. A, 21:695-709.

1959b. Notes sur I'epine latero-caudale et la queue de \' Acanthurus
monroviae. Bull. Inst. Fr. Afr. Noire (Dakar), Ser. A, 21:710-
734.

1959c. Verrouillage osseux a trois positions chez Triacanthus brevi-
rostris Temminck et Schlegel (Poissons, Balistiformes). Bull.
Mus. Natl. Hist. Nat. (Paris). Ser. 2, 30:498-501.

1960. Sur le systeme de "verrouillage" de la premiere ^pine dorsale
des Balistidae. Proc. XV Int. Cong. Zool., 1958:409-412.

1968. Le complexe urophore des poissons tiltostiens. Mem. Inst.
Fond. Afr. Noire 81:1-705.

MOULTON, J. M.

1958. The acoustical behavior of some fishes in the Bimini area.
Biol. Bull. (Woods Hole) 114:357-374.
MOLLER, H.

1844. Oeber den Bau und die Grenzen der Ganoiden und iiber das
naturliche System der Fische. Abb. Akad. Wiss. (Berl.), 1844:
117-216.
MYERS, G. S.

1934. Reports on the collections obtained by the first Johnson-
Smithsonian deep-sea expedition to the Puerto Rican Deep. Three
new deep-water fishes from the West Indies. Smithson. Misc.
Coll. 91(9):1-12.
1958. Trends in the evolution of teleostean fishes. Stanford

Ichthyol. Bull. 7:27-30.
1970. A monograph on plectognath fishes of the Superfamily Tria-
canthoidea. Copeia 1970:204-205.
NARDO, G.

1842. Considerazione sopra alcune nuove famiglie de'Syngnathi e
del Plectognathi, e sui caratteri anatomici che le distinguono.
Atti Quarta Riun. Sci. Ital., Padova, 1842:244-245.
NELSON, G. J.

1969. Gill arches and the phylogeny of fishes, with notes on the
classification of vertebrates. Bull. Am. Mus. Nat. Hist. 141:475-
552.

NIEUHOF, J.

1682. Gedenkwaerdige zee en lantreize door de voomaemste lan-

schappen van West en Oostindien. 2 parts in 1, 240 p. and 308 p.

Amsterdam.
NOWIKOFF, M.

1910. Ueber den Bau des Knochens von Orthagoriscus mola.

Anat. Anz. 37:97-106.
OKADA. Y. K.

1950. A coffer-fish with double caudal fins. Zool. Mag. (Dobutsu

Zassi, Tokyo) 59:8.
OKEN, L.

1816. Lehrbuch der Naturgeschichte. 6 vols.

Zool., Abth. 2, Fleischthiere, 1270 p. Jena.
OWEN, R.

1839. Sur la structure microscopique

dents des poissons Gymnodontes. Ann. Sci. Nat., Zool., Ser. 2,

12:347-352.
1840 & 1845. Odontography; or, a treatise on the comparative anato-
my of the teeth. 2 vols. London. [Vol 1 (1840), 655 p.; Vol 2

(1845), 37 p. -1-150 plates.)
1846. Lectures on the comparative anatomy and physiology of the

vertebrate animals. Part I. Fishes. 308 p. London.
1853. Descriptive catalogue of the osteological series contained in

the Museum of the Royal College of Surgeons of England. Vol. I:

Pisces, Reptilia, Aves, Marsupialia. 350 p. London.

3 parts: part ;

le developpement des

Fishes and reptile

1866. On the anatomy of vertebrates. Vol.
Longmans, Green and Co.. Lond., 650 p.
PARR, A. E.

1927. On the functions and morphology of the postclavicular ap-
paratus in Spheroides and Chilomycterus. Zoologica 9:245-269.
PATERSON, W.

1786. An account of a new electrical fish. Philos. Trans. R. See.
Lond. 76:382-383.
PATTERSON, C.

1964. A review of Mesozoic acanthopterygian fishes, with special
reference to those of the English Chalk. Philos. Trans. R. Soc.
Lend., Ser. B, Biol. Sci. 247:213-482.
PELLEGRIN, J.

1912. Sur la presence d'un banc de Ranzania truncate Retzius A la
Martinique. Bull. Soc. Zool. Fr. 37:228-231.
PFLUGFELDER, 0.

1930. Das Gebiss der Gymnodonten. Ein Beitrag zur Histogenese
des Dentins. Z. Anat. Entwicklungsgesch. (Berl.) 93:543-566.
PINTO, S. Y.

1959. Estudos morphologicos. I— Sobre as maxilas de Colomesus
psittacus (Bloch & Schneider, 1801) Gill, 1884 (Actinopterygii,
Tetrodontiformes, Tetrodontidae). Bol. Mus. Nac. Rio de J.
Zool., New Ser., No. 203, U p.
POPE, E. C.

1945. A fishy monstrosity. Aust. Mus. Mag. 8:383-384.
PUTNAM, F. W.

1871. On the young of Orthagoriscus moia. Am. Nat. 4:629-633
QUEKETT, J. J.

1850-1855. Descriptive and illustrated catalogue of the histological
series contained in the Museum of the Royal College of Surgeons
of England. 2 vols. London. [Not seen.)
RAFINESQUE, C. S.

1810. Indice d'ittiologica Siciliana. 70 p. Messina.
RANDALL. J. E.

1964. A revision of the filefish genera Amanses and Cantherhines.
Copeia 1964:331-361.
RATH, J. J. G. VOM.

1859. Beitrag zur Kenntniss der fossilen Fische des Plattenberges
im Canton Glarus. Z. Dtsch. Geol. Ges, (Berl.) 11:108-132.
RAUTHER, M.

1919. Notiz uber das Integument von Batistes. Anat. Anz. (Jena)

52:214-218.
1927a. Integument. In Bronn's Klassen und Ordnungcn des Tier-

Reichs. Vol. 6, p. 2-184. Leipzig.
1927b. Skelett. In Bronn's Klassen und Ordnungen des Tier-
Reichs. Vol. 6, p. 185-493. Leipzig.
RAVEN, H. C.

1939a. On the anatomy and evolution of the locomotor apparatus
of the nipple-tailed ocean sunfish {Masturus lanceolatus). Bull.
Am. Mus. Nat. Hist. 76:143-150.
1939b. Notes on the anatomy of Romania truncata. a plectognath
fish. Am. Mus. Novit. 1038:1-7.
REGAN, C. T.

1903a. On the classification of the fishes of the suborder Plectog-

nathi; with notes and descriptions of new species from specimens

in the British Museum collection. Proc. Zool. Soc. Lond. 1902(2):

284-303.

1903b. A revision of the fishes of the genus Triacanthus. Proc.

Zool. Soc. Lond. 1903(1): 180-185.
1910. The caudal fin of the Elopidae and of some other teleostean

fishes. Ann. Mag. Nat. Hist., Ser. 8, 5:354-358.
1929. Fishes. 14th ed. Encycl. Br. 9:305-329. London, New
York.
REUVENS, C. L.

1894. Remarks on the genus Orthagoriscus. Notes Leyden Mus.
16:128-130.
RICHARDSON, J.

1844-1848. Ichthyology of the voyage of H.M.S. Erebus and Terror,
under the command of Capt. Sir James Clark Ross, R. N., F.R.S.
Ichthyology. Edward Newman, Lond., 139 p., 60 pis. [Cited
page 52, published in 1845.]
1856. Ichthyology. 8th ed. Encycl. Br. 12:204-331. London.

RISHI, K. K.

1969. The head skeleton of Tetrodon patoca Ham. J. Asiat. Soc.
(Calcutta) 9:193-197.
RISSO, A.

1826. Histoire naturelle .... Vol 3. 480 p. Paris.
ROON, J. M. VAN.

1942. Some additional notes on external features and on the jaw
muscles of Orthagoriscus mola (L.). Zool. Meded. (Leiden) 23:
313-317.
ROON, J. M. VAN, and J. J. TER PELKWIJK.

1939. Mechanism of the jaw and body muscles of Orthagoriscus
mola L. Zool. Meded. (Leiden) 22:66-75.
ROSEN. N.

1912. Studies on the plectognaths. 1. The blood-vascular system.

Ark. Zool. 7(25):l-24.
1913a. Studies on the plectognaths. 2. The air-sac, with notes on

other parts of the intestines. Ark. Zool. 7(30):l-23.
1913b. Studies on the plectognaths. 3. The integument. Ark.

Zool. 8(10):l-29.
1913c. Studies on the plectognaths. 4. The body-muscles. Ark.

Zool. 8(18):1-14.
1916a. Studies on the plectognaths. 5. The skeleton. Ark. Zool.

10(8):l-28.
1916b. Ueber die Homologie der Fischschuppen. Ark. Zool. 10(7):
1-36.
ROSENTHAL, F.

1839. Ichthyotomische Tafeln. 2d ed. Unnumbered explanatory
pages for 27 plates. Berlin. [Exact copy of 1st ed., 1812-25.)
R'iT)ER, J. A.

1886. On the origin of heterocercy and the evolution of the fins and
fin-rays of fishes. Rep. U.S. Comm. Fish Fish., 1884:981-1107.
SALMON, M., H. E. WINN, and N. SORGENTE.

1968. Sound production and associated behavior in triggerfishes.
Pac. Sci. 22:11-20.

SANZO, L.

1930. Ricerche biologische su materiali raccolti dal Prof. L. Sanzo
nella Campagna Idrografica nel Mar Rosso della R.N. Ammiraglio
Magnaghi 1923-1924. VII. Plectognathi. R. &)m. Talassogr. Ital.
(Venezia), Mem. 167:1-111.

SARKAR, S. P.

1960. A study of functional morphology of the head of an Indian
puffer fish, Sphaeroides oblongus (Bloch). Ph.D. thesis, Univ.
Leiden, 119 p.

SCHLEGEL, H.

1850. Pisces (last part). In Siebold, P. (editor). Fauna Japonica,
p. 270-324. Leiden.
SCHNEIDER, H.

1961. Berichte neuere Ergebnisse der Lautforschung bei Fischen.
Naturwissenschaften 48:513-518.

SCOTT, E. 0. G.

1969. Description of Brachaluteres wolfei, sp. nov. (Aluteridae) and
first Tasmanian record of Urolophus paucimaculatus Dixon, 1969
(Urolophidae). Pap. Proc. R. Soc. Tasmania 103:35-45.

SHAW, G.

1798. The naturalists' miscellany, or colored figures of natural ob-
jects .... Vol. 9, plate 338. London.

1804. General zoology, or systematic natural history .... Vol. 5,
463 p. London.
SHEN, S.-C, and P.-C. LIM.

1973. Ecological and morphological study on fish-fauna from
the waters round Taiwan and it« adjacent islands. 6. Study
on the plectognath fishes.— a. The family of ostraciontoid fishes,
Ostraciontidae. Acta Oceanogr. Taiwan, Sci. Rep. Natl. Taiwan
Univ. 3:'245-268.

1974a. Study on the plectognath fishes— B. The family Canthi-
gasteridae. Bull. Inst. Zool., Acad. Sin. (Taipei) 13:15-34.

1974b. Ecological and morphological study on fish-fauna from
the waters around Taiwan and its adjacent islands. 9. Study
on the plectognath fishes— d. The family Balistidae. Acta
Oceanogr. Taiwan, Sci. Rep. Natl. Taiwan Univ. 4:191-224.
SHEN, S.-C, P.-C. LIM, and W.-H. TING.

1975. Ecological and morphological study on fish-fauna from the

waters around Taiwan and its adjacent islands. 8. Study on the
plectognath fishes— C. The family Tetraodontidae. Acta Ocean-
ogr. Taiwan., Sci. Rep. Natl. Taiwan Univ. 5, p. 152-178.
SHIPP, R. L,

1970. The pufferfishea (Osteichthyes: Tetraodontidae) of the At-
lantic Ocean and adjacent waters. Ph.D. thesis, Florida State
Univ., Tallahassee, 381 p.

1972a. Three new pufferfishes (Tetraodontidae: Sphoeroides) from
the southern Caribbean. Proc. Acad. Nat. Sci. Phila. 124:129-
134.

1972b. Competition between Atlantic and Pacific pufferfishes with
possible extirpation of species along some Atlantic shorelines.
Bull. Assoc. Southeastern Biol. (Phila.) 19:99.

1974. The pufferfishes (Tetraodontidae) of the Atlantic Ocean.
Publ. Gulf Coast Res. Lab. Mus. (Ocean Springs, Miss.) 4:1-
162.
SHIPP, R. L., and R. W. YERGER.

1969a. Status, characters, and distribution of the northern and
southern puffers of the genus Sphoeroides. Copeia 1969:425-
433.

1969b. A new puffer fish, Sphoeroides parvus, from the western
Gulf of Mexico, with a key to species of Sphoeroides from the At-
lantic and Gulf coasts of the United States. Proc. Biol. Soc.
Wash. 82:477-488.
SHUFELDT, R. W.

1917. Observations on the osteology of the porcupine tish {Diodon
hystrix). Anat. Rec. 11:405-407.

1926. Observa?6es sobre certos peixes e mammiferos do Brasil e
mais particularmente sobre sua osteologia. Rev. Mus. Paulista
(Sao Paulo) 14:503-561. [Followed by an English translation,
p. 563-614.1
SIEBENROCK, F.

1901. Ueber die Verbindungsweise des Schultergiirtels mit dem
Schadel bei den Teleosteern. Ann. K. Naturhist. Hofmus. (Wie-
ner Mus., Wein) 16:104-141.
SMITH, J. L. B.

1935. The South African species of the family Aluteridae. Rec.
Albany Mus. (Grahamstown) 4:358-364.

1949a. The sea fishes of southern Africa. 550 p. Cape Town.

1949b. A new aracanid fish from South Africa. Ann. Mag. Nat.
Hist., Ser. 12, 2:354-359.
SORBINI, L.

1968. Contributo alia sedimentologia della "Pesciara" di Bolca.
Mem. Mus. Civ. Stor. Nat. Verona 15:213-221.
S!?RENSEN, W.

1883. Om oppustningssaekken hos Tetrodon og aandedraettet hos
Clarias. Naturhist. Tidsskr. (Copenhagen) 13(tred. raek., trett.
bind):379-414.

1884. Om lydorganer hos fiske. En physiologisk og comparativ-ana-
tomisk unders^gelse. Ph.D. thesis, Univ. Copenhagen. 245 p.

1897. Some remarks on Dr. Thilo's memoir on "Die Umbildungen

an den Gliedmassen der Fische." Morphol. Jahrb. (Leipzig)

25:170-189.
SOULE, J. D.

1969a. Tooth attachment by means of a periodontium in the trig-
ger-fish (Balistidae). J. Morphol. (Boston) 127:1-3.
1969b. Amelogenesis in the trigger-fish Balistidae. Bull. S. Calif

Acad. Sci. 68:64-71.
1970. The organic matrix in developing dental enamel and dentin

studied by scanning electron microscopy. Bull. S. Calif Acad.

Sci. 69:104-111.
STANNIUS, F. H.

1854. Handbuch der Anatomie der Wirbelthiere. Erstes Buch.

Die Fische. 279 p. Berlin.
STARKS, E. C.

1901. Synonomy of the fish skeleton. Proc. Wash. (D.C.) Acad.

Sci. 3:507-539.
1905. Some notes on the myodome of the fish cranium. Science,

New Ser., 21:754-755.
1907. On the relationship of the fishes of the family Siganidae.

Biol. Bull. (Woods Hole) 13:211-218.
1916. The sesamoid articular, a bone in the mandible of fishes.

Stanford Univ. Publ., Univ. Ser., 40 p.

1926. Bones of the ethmoid region of the fish skull. Stanford

Univ. Publ., Univ. Ser., Biol. Sci. 4:139-338.
1930. The primary shoulder girdle of the bony fishes. Stanford

Univ. Publ., Univ. Ser., Biol. Sci. 6:149-239.
STEENSTRUP. J., and C. LOTKEN.

1898. Spolia Atlantica. Bidrag til kundskab om klump - eller

maanefiskene (Molidae). Dan. Vidensk. Selsk. Skr.. 6te Raek..

Natur\idensk. Math. Afd. (Copenhagen) 9:1-102.
STEPHAN, P.

1900. Flecherches histologiques sur la structure du tissus osseux des

poissons. Bull. Biol. Sci. Fr. Belg. (Paris), Ser. 5, 2 ( = 33 of com-

ple

STUDNICKA, F. K.

1916. Ueber den knochen von Or(/iagoracu5. Anat. Anz. 49: 151-
169, 177-194.
SUPINO, F.

1904. Contributo alio studio del tessuto osseo dell'Orthagoriscus.
Atti R. Accad. Lincei, Rend., CI. Sci. Fis., Mat. Nat., Ser. 5,
13:118-121.

1905. Morfologio del cranio dei Teleostei. Parte seconda. Plectog-
nathi. 16 p. Rome.

1907. n crano dei pesci. 100 p. Rome.
SWAINSON, W.

1838-1839. The natural history of fishes, amphibians, and reptiles,
or monocardian animals. 2 vols. London. (Vol. 1 (1838).
368 p.: Vol. 2 (18391, 448 p.]
SWINNERTON, H. H.

1902. A contribution to the morphology of the teleostean head
skeleton, based upon a study of the developing skull of the three-
spined stickleback {Gasterosteus aculeatus). Q. J. Microsc. Sci.
(Lond.), New Ser., 46:503-593.
TAVANI, G.

1955. Osservagioni su alcuni Plectognathi (Gymnodonti). Atti
Soc. Toscana Sci. Nat. (Pisa), Mem., (A) 62:176-200.
THILO, O.

1879. Die Sperrgelenke an den Stacheln einiger Welse, des Stich-
linges und des Einhomes. Ph.D. thesis, Univ. Dorpat, 21 p.

1896a. Die Gliedmassen der Fische. Korresp. Naturforsch. Ver.
Riga 39:96-106.

1896b. Die Umbildungen an den Gliedmassen der Fische. Mor-
phol. Jahrb. (Leipzig) 24:287-355.

1898. Erganzungen zu meiner Abhandlung "Die Umbildungen an
den Gliedmassen der Fische." Morphol. Jahrb. (Leipzig) 26:81-
90.

1899a. Sperrvorrichtungen im Tierreiche. Biol. Zentralbl. 19:
504-517.

1899b. DieEntstehungderLuftsackebeidenKugelfischen. Anat.
Anz. 16:73-87.

1900. Erganzungen zu meiner Abhtmdlung "Sperrvorrichtungen
im Tierreiche." Biol. Zentralbl. 20:452-461.

1902. Maschine und Tierkbrper. Verb. V Int. Zool.-Congr. (Berl.),
Sekt. 2 (Experimentalzool.):501-503.

1914. Die Vorfahren der Kugelfische. Biol. Zentralbl. 34:523-545.

1920. Das Maulspitzen der Fische. Das Entstehen und Vergehen
seiner Mechanik. Biol. Zentralbl. 40:216-238.
THOMPSON, DA. W.

1888. On the auditory labyrinth of Orthagoriscus mola L. Anat.
Anz. 3:93-96.
TRETJAKOFF, D. K.

1924a. Die Haul der Plectognathen. 1. Die Schuppenhaut von Ba-
tistes. Nauchno-lssled. Kafedry (Ber. Wiss. Fors. Inst., Odessa)
1:163-171.

1924b. Die Haut der Plectognathen. U. Der Panzer des Koffer-
fisches. Nauchno-lssled. Kafedry (Ber. Wiss. Fors. Inst., Odes-
sa) 1:172-181.

1925. Die Mikrostruktur der Flossenstrahlen bei den Pleuronectiden
und den Plectognathen. Anat. Anz. 60:182-191.

1926a. Die Haut der Plectognathen. ID. Die Schuppen von Ortha-
goriscus. Nauchno-lssled. Kafedry (Ber. Wiss. Fors. Inst.. Odes-
sal 2:85-87.

1926b. Die Haut der Plectognathen. IV. Die Dome von Diodon und
von Tetrodon. Nauchno-lssled. Kafedry (Ber. Wiss. Fors. Inst.,
Odessa) 2:88-99.

1926c. DieZahnederPlectognathen. Z. Wiss. Zool. (Leipzig) 127:

619-644.
TROIS, E. F.

1883-1884. Ricerche suUa struttura della flanzania fruncata. Atti

R. 1st. Veneto Sci. Lett. Arti (Venice), Ser. 6, 2[1883):1269-1306;

Ser. 6, 2[ 1884): 1543-1560.
TURNER, W.

1862. On tlie structure and composition of the integument of the

Orthagonscus mola. Nat. Hist. Rev. (Lend.), New Ser., 2:185-

188.
■nXER, J. c.

1962a. Triodon bursarius, a plectognath fish connecting the Sclero-

dermi and Gymnodontes. Copeia 1962:793-801.
1962b. The pelvis and pelvic fm of plectognath fishes; a study in

reduction. Proc. Acad. Nat. Sci. Phila. 114:207-250.
1963a. The apparent reduction in number of precaudal vertebrae in

trunkfishes (Ostraciontoidea, Plectognathi). Proc. Acad. Nat.

Sci. Phila. 115:153-190.
1963b. A critique of Y. Le Danois' worli on the classification of the

fishes of the Order Plectognathi. Copeia 1963:203-206.
1964. A diagnosis of the two species of South American puffer

fishes (Tetraodontidae, Plectognathi) of the genus Colomesus.

Proc. Acad. Nat. Sci. Phila. 116:119-148.
1965a. The trunkfish genus Acanthostracion (Ostraciontidae, Plec-
tognathi) in the western Atlantic: two species rather than one.

Proc. Acad. Nat. Sci. Phila. 117:1-18.
1965b. A synopsis of the four species of cowfishes (Acanthostracion,

Plectognathi) in the Atlantic Ocean. Proc. Acad. Nat. Sci. Phila.

117:261-287.
1965c. Poisonous and worthless fishes? Frontiers (Acad. Nat. Sci

Phila.) 30:36-45.

1966. Mimicry between the plectognath fishes Canthigaster valen-
tint (Canthigasteridae) and Paraluteres prionurus (Aluteridae)
Not. Nat. (Phila.) 386:1-13.

1967. A redescription of Triodon macropterus Lesson, a phyleti-
cally important plectognath fish. Proc. K. Ned. Akad. Wet
(Amsterdam). Ser. C, 70:84-96.

1968. A monograph on plectognath fishes of the Superfamily Tria-
canthoidea. Monogr. Acad. Nat. Sci. Phila., Vol. 16, 364 p.

1970a. The dorsal and anal spine-locking apparatus of surgeon
fishes (Acanthuridae). Proc. Calif. Acad. Sci. 38:391-409.

1970b. The progressive reduction in number of elements support-
ing the caudal fin of fishes of the order Plectognathi. Proc. Acad.
Nat. Sci. Phila. 122:1-85.

1970c. Osteological aspects of interrelationships of surgeon fish
genera (Acanthuridae). Proc. Acad. Nat. Sci. Phila. 122:87-
124.

1970d. Abnormal fin and vertebral growth structures in plectog-
nath fishes. Proc. Acad. Nat. Sci. Phila. 122:249-271.

1970e. An especially small, sexually dimorphic new species of file-
fish (Monacanthidae) from Australasian reefs. Proc. Acad. Nat.
Sci. Phila. 122:273-290.

1970f. New records of triacanthoid plectognath fishes. Not. Nat.
(Phila.) 435:1-7.

1973a. A new species of boxfish from the Eocene of Monte Bolca,
Italy, the first unquestionable fossil record of the Ostraciontidae.
Mem. Mus. Civ. Stor. Nat. Verona 2:103-127.

1973b. A new species of triacanthodid fish (Plectognathi) from the
Eocene of Monte Bolca, Italy, representing a new subfamily an-
cestral to the Triodontidae and the other gymnodonts. Mem.
Mus. Civ. Stor. Nat. Verona 2:128-156.

1974. Tetraodontiformes. 15th ed. Encycl. Br. T:162-164.

1978. Der Rotaugen-Kammkugelfisch, Carinotetraodon lorteti,
aus Siidostasien. Aquar.-u. Terrar.-z. (Stuttgart) 31:118-120.
VAN DOBBEN, W. H.

1935. Ober den Kiefermechanismus der Knochenfische. Arch.
Neerl. Zool. 2:1-72.
\TNCENT, F.

1963. Note preliminaire sur des emissions acoustiques chez Balistes
capriscus L. Cent. Etud. Rech. Sci. (Biarritz) 4:307-316.
VOGT, C.

1845. Quelques observations sur les caracteres qui servent 4 la clas-

sification des poissons ganoides. Ann. Sci. Nat. (Paris), Zool.
Ser. 3, 4:53-68.
VOLTA, G. S.

1796. Ittiolitologia Veronese del Museo Bozziano. 2 parts. Ver-
ona. [Part l:I-Ln; Part 2:I-CCCXXin.l
WAGNER, R.

1841. Icones zootomicae. Handatlas zur vergleichenden Anatomie.

45 p. Leipzig.
1845. Elements of the comparative anatomy of the vertebrate ani-
mals; designed especially for the use of students. J. S. Redfield,
Clinton Hall, New York, 264 p.
WAHLGREN, F.

1867. Nagra anteckningar om en stor klumpfish, Mola nasus (Raf.).
Acta Univ. Lundensis (Lund; Lunds Univ. Ars-skr.) 4:1-18.
WELLENBERGH, P. H.

1840. Observationes anatomicae de Orthagoriscus mola. Ph.D.
thesis, Univ. Lugduni Batavorum (Leiden), 32 p.
WEITZMAN, S. H.

1967. The origin of the stomiatoid fishes with comments on the
classification of salmoniform fishes. Copeia 1967:507-540.
WETTSTEIN, A.

1886. Ueber die Fischfauna des Tertiaeren Gitimerschiefers. Abh.
Schweiz. Palaontol. Ges. (Mem. Soc. Palfentol. Suisse, Basel) 13:
1-103.
WHITEHOUSE, R. H.

1910. The caudal fin of the Teleostomi. Proc. Zool. Soc. Lond.
1910(2):90-627.
WIEDERSHEIM, R.

1887a. Das Geruchsorgan der Tetrodonten nebst bemerkungen
uber die Hautmuskulatur derselben. In Fest-schrift Albert von
KoUiker zur feier seines siebenzigsten Geburtstages gewidmet von
seinen Schiilem, p. 73-84. Leipzig.
1887b. Ueber rudimentare Fischnasen. Anat. Anz. 2:652-657.
1888. On the degeneration of the olfactory organ of certain fishes.
Rep. 57th Meet. Br. Assoc. Adv. Sci., Manchester, 1887, Sect. D -
Biol., p. 736-738.
WILLEM, V.

1941. Le mode respiratoire de Balistes et d'Ostracion. Acad. R.
Belg., Bull. CI. Sci. (Brussels), Ser. 5, 27:288-294.

1942. Contribution a I'etude des organes respiratories chez les
teleosteens Plectognathes. K^ Partie: Les Balistes. Bull. Mus.
R. Hist. Nat. Belg. (Brussels) 18(35):l-24.

1944. Contribution a I'etude des organes respiratories chez les
teleosteens Plectognathes. 2^ Partie: Les Chaetodontiformes.
Bull. Mus. R. Hist. Nat. Belg. (Brussels) 20(6):1-21.

1945. Contribution a I'etude des organes respiratories chez les
teleosteens Plectognathes. 3^ Partie: Les Ostracionidis. 4e Partie:
Le role de I'interoperculaire dans la manoeuvre respiratoire chez
les teleosteens. Bull. Mus. R. Hist. Nat. Belg. (Brussels) 21(23):
1-16.

1947. Contribution a I'etude des organes respiratories ches les
teleosteens Plectognathes. 5^ Partie: Tetrodontes et Dmdon.
Bull. Mus. R. Hist. Nat. Belg. (Brussels) 23(17):1-17.
1949. Contribution a I'etude des organes respiratories chez les
teleosteens Plectognathes. 6« Partie: Mola. Bull. Mus. R. Hist.
Nat. Belg. (Brussels) 25(24):l-4.
WILLL\MSON, W. C.

1851. Investigations into the structure and development of the
scales and bones of fishes. Philos. Trans. R. Soc. Lond., 1851
(pt. n):643-702.
WILLUGHBY, F.

1686. Historia piscium libri quatuor, Oxford. 343 p.
WINTERBOTTOM, R.

1970. Evolution of the pelvic fin ray musculature of some triacan-
thoid fishes (Plectognathi). Copeia 1970:453-456.

1971. A reinterpretation of the mechanism of extension of the tetra-
odontid pufferfish postcleithral apparatus, with notes on the
oblique eye muscles of the diodontids (Plectognathi). Not. Nat.
(Phila.) 440:1-7.

1974. The familial phytogeny of the Tetraodontiformes (Acan-
thopterygii: Pisces) as evidenced by their comparative myology.
Smithson. Contrib. Zool. 155:1-201.

V\TNTHEk I. .

1877. Fiskene^ a.ioitri; en comparativ-anatomisk under80gelse.
Tredie afsnii. Xaturh. Tids. (Copenhagen), Ser. 3, 11:267-478.
WOODWARD, A. S.

1901. Catalogue of the fossil fishes in the British Museum (Na-
tural History). Part IV. Br. Mus. Nat. Hist., 636 p.
WORTMAN, J. L.

1886. The comparative anatomy of the teeth of the Vertebrata.
In Litch, W. F. (editor), The American system of dentistry, p.
351-504. Lea Brothers & Co., Philadelphia.
ZIGNO, A. DE.

1874a. Annotazioni paleontologiche. Pesci fossili nuovi del calcare

Eoceno dei Monti Bolca
Lett. Arti (Venice) 18:287

! Postale. Mem. R. 1st. Veneto
301.

1874b. Catalogo ragionato dei pesci fossili del calcare Eoceno di
M. Bolca et M. Postale disposti secondo il metodo dell' Agassiz
modificato dal Pictet (continua). Atti R. 1st. Veneto Sci., Lett.
Arti (Venice), Ser. 4, 3(Nov. 1873-Oct. 1874):1897-1929.

1887a. Due nuovi pesci fossili della famiglia dei Balistini scoperti
nel terreno Eoceno del Veronese. Soc. Ital. Sci. |Soc. XL (dei
quaranta)], Mem. Mat. Fis. (Venice), Ser. 3, 6:1-8.

1887b. Nuove aggiunte alia ittiofauna dell'epoca Eocena. Mem.
R. 1st. Veneto Sci., Lett. Arti (Venice) 23:9-33.

I I I I I

I I nil f-l i

I I I illi i

£ E s e e

I I J,

£ -s £ -s £

"' -s - s s ^ '

D. a a q.'m B. B. a. a.

5i^

s ° s

I I

I I

Js J ££ ££^£££ SSS-cS^

I 08 -e P -S £ £ £ ^ I £ £ £ £

"2

1 11 11

2 2 2 2 2

^« M^m®«« OD

.2 I S "; .2

si s

2 ^ Z: ^ -c 2 ■

O O O X P

5 Q (S

Table 2.— Total number of vertebrae and vertebral formulas of Recent plectognath flshes examined
radio^aphed and as cleared and stained specimens.

16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Triacanthodidae
Hollardiinae

HollardiagosUnei 8+12 1

Hollardia hoUardi 8+ 12 20

HoUardia meadi 8+ 12 3

Parahollardia lineata 8+12 17

ParahollardiaschmidtiS+ 12 17

Triacanthodinae

Atrophacanthus japonicus S+12 38

Bathyphyiax bom bifrons 8+12 1

Bathyphylax omen 8+ 12 1

Halimochirurgus alcocki &+12 8

Halimochirurgus centriscoides 8+ 12 7

Johnsonina eriomma 8+12 15

Macrorhamphosodes platycheilus 8+12 13

Macrorhamphosodes uradoi 8+ 12 6

Mephisto fraserbrunneri 8+ 12 1

Paratriacanthodes herrei 8+ 12 3

Paratriacanthodes retrospinis 8+ 12 6

Tnacantfwdesanomalus 8+ 12 19

Triacanthodes ethiops 8+ 12 7

Tydemania navigatoris 8+12 16

Triacanthidae

Pseudotriacanthus strigUifer 8+ 12 21

Triacanthus btaculeatus 8+12 31

Triacanthusmeuhofi 8+ 12 2

Tripodichthys angustifrons 8+12 1

Tripodichthys blochi 8+ 12 36

Tripodichthys oxycephalus 8+ 12 2

7>iiip/jic/i«hysu)e<)?ri8+12 8

Balistidae

Abalistesstellatus 1+11 1

Balistapus undulatus 1 +11 2

Batistes capriscus 7+11 32

Batistes forcipatus 7+11 1

Batistes polytepis 7+ U 30

Batistes uetuta 1+11 6

Batistoides conspicitium 7+11 2

Balistoides viridescens 7+11 1

Canthidermismaculatusl+ 11 10

Hemibalistes bursal + 11 2

Hemibalistes chrysopterus 1 + 11 1

Me/ichfhysmger 7+11 in 3 '1 3

7+10 in 1

Melichthysvidual+U 1

Odonus niger 1 + 11 2

Pseudobalistes flavimarginata 7+11 1

Rhinecanthusacuteatusl + 11 4

Rhinecanthus rectangulus 7+11 in 5 5 1

7+12 in 1

Rhinecanthus verrucosus 7+11 l

Sufflamenfrenatusl + 11 1

Sufflamen verres 7+11 2

Xanthichthys lineopunctatus 7+11 1

Xant/iic/if/iysringefw7+ll 4

Monacanthidae

Acanthaluteres spilomelanurus 6+ 14 1

Atuteraheudetotiil +13 8

v4/u(erQ monoceros 7+16 1

i4;u£cra sc^cp/i 7+16 in 11 1 11

7+15 in 1

Table 2.— Total number of vertebrae and vertebral formulas of Recent plectognath fishes examined as
radiographed and as cleared and stained specimens.— Continued.

16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Alutera scripta 7+ liin 3 3 1

7+16 in 1
Amansesscopas 1+12 ■*

Brachaluteres trossulus 7 + 13 22

Brachaluteres uivarum 7 + 13 2

Cantherhinespardalisl+ 12 11

CantherhinespuUusl+ 12 H

Cantherhines sandwichiensis 7 + 12 8

Cantherhines trachylepis 7 + 13 2

Chaetoderma spinosissimus 7+ 13 2

Laputa cingalensis 7+12 1

Monacanthus chinensis 7 + 12 3

Monacanthus ciliatus 6+ 13 in 49 49 1

6+14 in 1
Monacanthus mylii 7+12 1

Monacanthus tuckeri &+!% 6

Navodon setosus 7+13 1

Oiymonacant/ius longirostris S+ 17 in 6 16

8+ 16 in 1
Paraluteres prionurus 1 + IZ in 1 1 1

7 + 14 in 1
Paramonacanthus bamardi 7+12 1

Paramonacanthus cryptodon 7+ 12 in 3 3 1

7+ 13 in 1
Paramonacanthus curtorhynchus 7+12 3

Paramonacanthus oblongus 7+ 12 2

Pervagor melanocephalus 7+ 12 in 1 2

6+13 inl
Pervagor spilosomus 7 + 12 4

Peroagortomentosus 7+13 in 4 2 4

7 + 12 in 2
Pseudaluteres nasicomis 8+18

Psilocephalus barbatus 1+22 in 3 ^ ^

7+23 inl
8+22 inl
Rudarius ercodes 7+13 in 17 17 2

7 + 14in2
fludanasminutiis7+12 4

Stephanolepis auratus 7+12 1

Step/iano/epis cirrhi/er 7+ 13 in 1 1 1

7+ 12 inl
Stephanolepis hispidus 1 + 12 in 33 1 33 1

7 + 13 inl
7 + Uinl
Stephanolepis setiferl + 12 14

Aracanidae

Aracanaaunta (2 fused) 9+9 2

/Iracana ornafa (2 fused) 10+8 2

Caprichthys gymnura (2 fused) 10+8 1

Capropygia unistriata (2 fused) 10+8 1

Kentrocapros aculeatus (2 fused) 9+9 2

Strophiurichthys robustus (2 fused) 9+9 1

Ostraciidae
Ostraciinae

Lactoria comuta (4 fused) 9+9 5

Lactona/omasmii (5 fused?) 9+9 5

Ostracion lentiginosum (2+2 or 4 fused) 9+9 4

Ostracion tuberculatus (2 + 2 fused) 9+9 4

Rhynchostracion rhinorhynchus (4 fused) 9+9 1

Tetrosomusconcatenatus {4 (used) 9+9 1

Tetrosomusgibbosus (4 (used) 9+9 1

415

Table 2.— Total number of vertebrae and vertebral formulas of Recent plectognath fishes examined as
radiographed and as cleared and stained specimens.— Continued.

Lactophrysinae

Acanthostracion guineensis (5 fused) 9+9
Acanthostracion notacanthus (5 fused) 9+9
Acanthostracion polygonius (5 fused) 9+9
Acanthostracion quadricornis (5 fused) 9+10
Lactophrys trigonus (5 fused) 9+9
Rhinesomus bicaudalis (5 fused) 9+9
Rhinesomus triqueter (5 fused) 9+9
Triodontidae

Triodon macropterus 9+11
Tetraodontidae: figures followed by (A) are simpli-
fied from Abe's ( 1942) extensive analysis of
vertebral variation in Japanese puffers
Tetraodontinae
Amblyrhynchotes honckenii 8+11 in 5
8+ 12 in 3
Amblyrhynchotes hypselogenion 8+11 in 5
8+ 12 in 2
7+12 in 1
Amblyrhynchotes piosae 8+11
Arothron armilla 8+11
Arothron hispidus 8+ 10 in 12
8+9 in 2
9+ 10 in 1
Arothron immaculatus 8+10 in 59
8+9 in!
8 + 11 in 1
Arothron meleagris 8+10 in 2
9+ 10 in 5
9+11 in 1
Arothron nigropunctatus 9+9 in 6
8+9 in 1
8+ 10 in 1
Arothron reticulatus 8+10 (A)
Arothron setosus 9+ 10 in 6
9+9 in 1
8+9 in 1
Arothron stellatus 8+10
Carinotetraodon lorteti 7+10 in 2

7+9 in 1
Chelonodon fluuiatilis 8+ 10 in 38
8+llin9
8+9 in 4
Chelonodon patoca 8+11 in 25

8+ 10 in 1
Chonerhinos modestus 10+ 16 in 6
10+ 15 in 3
10+ 17 in 1
9+ 16 in 1
9+ 17 in 1
Colomesus asellus 8+ 11 in 8
8+ 12 in 1
8+ 10 in 1
Colomesus psittacus 8+ 11 in 5
8+ 10 in 1
Ephippion guttifer 8+ 12
Fugu chrysops 9+13 in 11
9+ 12 in 6
Fugu niphobles 8+ 13 in '
8+ 12 in i
8+11 in
8+14 in

Table 2.— Total number of vertebrae and vertebral formulas of Recent plectognath
radiographed and as cleared and stained specimens.— Continued.

8+ 12 in 1
9+13 inl
Fugu rubnpes 9+ 12 in 2
9+ 13 in 3
9+14inl
Fugu stictonotus 8+14 in 65
8+ 15 in 12
8+ 13 in 1
Fugu vermiculans 8+ 14 in 225 \
8+ 13 in 15 /
8+ 15 in 14 ( .
9+13in 1 } "^
9+14 in 1 i
7+15in 1 /
Fugu xantkopterus 8+13 (A)
Guentheridia formosa 8+9
Lagocephalus inermis 8+ 10
Lagocephalus laevigatus 8+11 in 10
8+ 10 in 2
Lagocephalus lagocephalus 8+ 10
Lagocephalus lunaris 8+9 in 9
8+11 inl
Lagocephalus scleratus 8+9 in 8
8+llin2
8+ 12 in 2
7 + 12inl
Lagocephalus spadiceus 8+9
Monotreta cutcutia 8+10
Monotreta gularis 8+10
Monotreta leiurus 9+11 in 2
10+11 in 2
10+ 10 inl
Monotreta palembangensis 10+11
Sphoeroides angusticeps 8+9
Sphoeroides annulatus 8+10 in 19
8+9 in 2
9+ 10 inl
Sphoeroides dorsalis 8+9 in 20
8+8 inl

Sphoeroides lobatus 8+9

5 143 173 21 1

2 10

2

1 20
7 5
2

417

Table 2.— Total number of vertebrae and vertebral formulas of Recent plectognath fishes examined as
radiographed and as cleared and stained specimens.— Continued^

Sphoeroides maculatus 8+11 in 26
8+ 10 in 8
Sphoeroides marmoratus 8+ 12 in 7
8+11 in 2
Sphoeroides nephelus 8+10 in 32
8+llin2
8+9 in 1
7+ 10 in 2
9+10 in 1
Sphoeroides pachygaster 8+10
Sphoeroides sechurae 8+9 in 35
8+ 10 in 1
Sphoeroides spengleri 8+9 in 34
8+10 in 1
Sphoeroides testudineus 8+ 10 in 74
8+9 in 32
8+8 in 1
8+11 in 1
7+9 in 2
Sphoeroides tnchocephalus 8+ 9 in 5
8+ 10 in 2
Tetraodon Uneatus 8+10 in 1
8+11 in 1
Tetraodon mbu 8+9
Tetraodon miurus 8+11
Torquigener hamiltoni 8+ U in 2
9+11 in 1
9+12 in 2
Torquigener pleurogramma 8+ 12
Torquigener pleurostictus 8+ 11
Xenopterus naritus 11 + 18 in 1
11 + 19 in 1
10+ 19 in 1
Canthigasterinae
Canthigaster amboinensis 8+9
Canthigaster bennetti 8+9 in 4
9+9 in 1
Canthigaster compressa 8+9 in 30
8+ 10 in 1
Canthigaster coronata 8+9
Canthigaster jactator 8+9 in 8
8+10 in 1
Canthigaster janthinoptera 8+9
Canthigaster margaritata 8+9
Canthigaster papua 8+9
Canthigaster punctatisstma 8+9 in 23

8+8 in 1
Canthigaster rivulata 8+9
Canthigaster rostrata 8+9
Canthigaster solandri 8+9 in 19
8+ 10 in 2
Canthigaster ualentini 8+9
Diodontidae

Chilomycterus affinis 13+10
Chilomycterus antennatus 12+8
Chilomycterus antillarum 10+9
Chilomycterus atinga 13 + 8
Chilomycterus mauretamcus 12+8
Chilomycterus orbicularis 11 + 8
Chilomycterus reticulatus 12+10
Chilomycterus schoepfi 11+8 in 3
12+8 in 2
10+8 in 1

-Total number of vertebrae and vertebral formulas of Recent plectognath fishes examined
radio^aphed and as cleared and stained specimens. — Continued.

Chilomycterusspinosus 10+S 1

Chilomycterus tigrinus 12+ 10

Dicotylichthys nicthemerus 11 + 9 1

Dicotylichthys punctulatus 12+9 1

Diodon holocanthus 12+9 10

Diodonhystrixn+9in2 3 1

11+10 inl

12+8 in 1
Diodon jaculif eras 10+9 1

Molidae

Masturus lanceolatus 8+8 2

Moia mola 8+9 2

Ranzanialaevis%+W 4

'A vertebra with bifid neural and haemal spines but centrum of normal si:

419

Table 3.— Numbers of teeth in the upper and lower jawg of ostracioids.

12 13 14 15 16 17

ARACANIDAE
Aracana aurita

Aracana flavigaster

Caprichthys

gymnura

Capropygia

Kentrocapros

aculeatus
Strophiurichthys

inermis
Strophiurichthys

robustus
OSTRACIIDAE
Acanthostracion

guineensis
Acanthostracion

notacanthus
Acanthostracion

polygonius
Acanthostracion

quadricornis
Lactophrys

tngonus

cornuta
Lactona

diaphana
Lactoria

fornasinii
Ostracion

cubicus
Ostracion

lentiginosum
Ostracion

tuberculatus
Rhinesomus

bicaudalis
Rhinesomus

triqueter
Rhyncho

Rhynchostracion
rhinorhynchus

Tetrosomus
concatenatus

Tetrosomus
gibbosus

upper
lower
upper
lower
pper

upper
lower
upper
lower
pper
lower

pper
lower
upper

upper
ower

ower

ower
upper
ower
upper

5 7 15
9 7

4 1 25

3 4 10 6

1

4 10

3

13

9 1

2

"

3 10

3 2

17

3

5

2

2
3

2 3
2

Index of Discussions of Genera

Abalistes 123, 128, 179

Acanthaluteres 146, 154, 167, 173-174, 176, 178,

180, 182

Acanthopleurus 4, 79, 88, 96-98, 118, 134

Acanthostacion . . 184, 218, 220-223, 228-229, 236-238,
239-240, 241, 242

Acreichthys 180

A later ta .

. .123, 136, 146, 153, 154, 167, 174, 178,
180-183, 344

Amanses 143, 146, 154, 167, 172, 173, 176,

178, 180, 182, 183
Amblyrhynchotes .... 290-291, 297-298, 329, 331-334,

336, 343

Anoplocapros 195, 198, 199, 205-206

Aracana 195-199, 203, 205, 206, 243

Arothron 290, 291, 297, 334-337, 340

Arotrolepis 146, 154, 175, 183

Atinga 358

Atrophacanthus 56, 66, 68, 70

Balistapus 109, 110, 120, 122-123, 128, 179

Balistoides HO, 123, 128, 179

Batistes 120, 123, 128, 130, 179

Balistomorphus 98, 110, 118, 120-121

Bathyphylax 56, 66-68, 70, 73, 344

Blandou'skius 154, 183

Brachaluteres 136, 144, 146, 153, 154, 167,

173, 174, 178, 181, 182, 183

Cantherhines 136, 146, 154, 167, 176, 178, 180,

182-183

Canthidermis 77, 99, 101, 110, 120, 122-123,

128, 129, 130, 133

Canthigaster 144, 153, 181, 290, 298-299,

314-316, 326, 340, 343

Caprichthys 195, 198-199, 204, 205-206, 243

Capropygia 195, 198-199, 204, 205-206, 243

Carinotetraodon 4, 144, 290, 297, 298, 314-316,

326, 334-336, 340, 342, 343

Chaetoderma . . . 136, 144, 146, 153, 154, 167, 174-175,

176, 178, 180-181, 183

Chelonodon 290, 291, 297, 315, 334-338,

340, 342, 343
Chilomycterus .... 344, 355, 357-359, 360-361, 364-370

Chonerhinos 265, 287, 290, 298, 337, 342-343

Colomesus . . 290-291, 297, 298, 330-331, 334, 343, 353

Cryptobalistes 78, 79, 98, 132, 133, 134

Dicotylichthys 358, 360, 361, 364, 367, 368, 369

Diodon 353, 354, 357, 358-360, 364, 368, 369

Eodiodon 357

Eolactoria 5, 207, 222, 223, 240, 242-243

Eoplectus .4, 7, 56, 64-65, 74-76, 77, 131, 134, 206, 261

Eotetraodon 287

Ephippion 77, 265, 290, 297, 298, 334, 335,

337-338, 342, 343

Eubalichthys 154, 178, 183

Fugu 290-291, 297-298, 329, 331-334,

335, 336, 343
Guenthendia . . . .290-291, 297, 298, 320, 322, 330, 331

Halimochirurgus
Hemibalistes . . .

Holtardia

Johnsonina ....

56, 66-68, 70, 344

128

. . . .56, 57, 66, 69
.... 56, 66-69, 70

Kentrocapros 193, 195, 198-199, 203-204, 205

Kyrtogymnodon 353

Lactophrys 220, 222, 223, 228, 229, 236,

237, 238, 239-240, 241, 242

Lactoria 222-223, 228, 229, 236-238, 240-243

Lagocephalus 277, 282, 290-291, 297, 298, 299,

316-317, 322-329, 331, 332, 334, 342, 343

Laputa 146, 154, 167, 176-178, 180, 181, 183

Leprogaster 173

Liosaccus 297

Macrorhamphosodes 56, 66-68, 70, 342

Marosichthys 79, 88, 98, 123, 134

Masturus 77, 377, 379, 381, 386, 392, 393-394

Melichthys 101, 110, 120-121, 123, 128-129, 179

Mephisto 56, 66-68, 70

Meuschenia 154, 178, 182

Mola 77, 123, 379, 381, 386, 392, 393-395

Moiacanthus 377

Monacanthus 143, 153, 154, 167, 172, 173,

174, 176, 178-181, 182, 183
Monotreta 290, 291, 297, 298, 314, 315, 334,

335-336, 338-340, 342, 343

Navodon 136, 154, 167, 176, 178, 180, 182, 183

Nematobalistes 123

Odonus 110, 120, 122, 123, 128, 133, 179

Oligobalistes 120, 121

Oligodiodon 355

Omegophora 335

Ostracion 221, 222, 223, 228-229, 236,

237, 240-241, 242, 243

Oxvmonacanthus 154, 167, 173-174, 176, 178,

181-182, 183, 342

ParahoUardia 56, 66, 69

Paraluteres 99, 146, 153, 154, 167, 173,

174-175, 176, 181, 182, 183
Paramonacanthus 153, 154, 167, 174, 176,

178-179, 180-181, 182, 183

Paratriacanthodes 56, 66-68, 70

Peruagor . . . 154, 167, 176, 178, 179-180, 181, 182, 183

Proaracana 5, 185, 195, 198, 204-205, 206

Prodiodon 260, 343, 352-355, 357

Progymnodon 260, 357

Protacanthodes . . .4, 75, 76, 77, 78, 79, 88, 93, 95, 98,
131-132, 134, 207, 210

Protobalistum 56, 57, 59, 63-64, 76, 206

Protriacanthus 98

Pseudaluteres 100, 144, 146, 154, 167, 172,

173-174, 176, 181, 182, 183, 344

Pseudobalistes 110, 123

Pseudomola 377

Pseudomonacanthus 88, 92-93, 97, 98

PsUocephalus 100, 135, 144, 146-147, 153, 154,

167, 172-173, 174, 175, 176, 178, 181, 182 183, 344

421

Ramania 77, 123, 262, 375, 377, 379,

384, 390-392

Rhinecanthus 101, 110, 120, 121, 122, 123, 128,

167, 179

Hhtne.somu.s 220, 222-223, 228, 229, 236,

237, 239-240, 241, 242

Rhvnchostracion 222, 223, 228, 229, 236,

237, 240-241, 242, 243

Rudarius 136, 143, 154, 167, 172, 173, 174,

176, 181, 182-183

Scobinichthys 154, 182

Sphoeroides 290, 291, 297, 298, 316-322,

324, 327, 328-333, 335, 343

Spinacanthus 56, 63-64, 76, 206

Stephanolepis 153, 154, 167, 176, 178-179,

180, 181, 182, 183
Strophiurichthys .... 195, 198, 199, 203, 204, 205, 206

Sufflamen 123, 128, 130, 179

Tetraodon 286-287, 290, 291, 297, 298, 314,

324, 335, 336, 337-338, 340, 342, 343

Tetro.somus 220, 222, 223, 228, 229, 236,

237, 238, 240-242, 243

Torquigener 265, 290-291, 297, 298, 329,

331-334, 336, 343

Triacanthodes 56, 66-70

Triacanthus 57, 88, 92, 93, 95, 98

Triodon 6, 65, 74-75, 167, 210, 243, 244,

253, 254, 255, 260, 261-263, 273, 323,
326, 344, 348, 349, 361, 394

Tripodichthys 88, 92, 93, 98

Tnxiphicht'hys 78, 79, 82, 88, 92-93, 98, 134

Tydemania 56, 66, 68, 70, 344

Verrunculus 123

Xanthichthys\Q\, 110, 120-121, 122, 123, 128, 133, 179

Xenopterus 265, 287, 290, 298, 337, 342-343

Zignoichthys 56, 65, 74, 77, 261, 344

WHSE

NOAA TECHMC AL RP:P()RTS
NMFS CIRCULAR AND SPECIAL SCIENTIFIC REPORT
GUIDELINES FOR CONTRIBUTORS

FISHERIES

( ONTKNTS OF MAM SCRIPT

lir^lpagc. (live the title (as I
ami I he author's name, and k
mailing address, and ZIP code.

else as possible) ol the paper
lote the author's alliliation,

CoMlents. Contains the text headings and abbreviated figure
1( ;;rii,l-, and table headings. Dots should follow each entry and
p.iL;r numbers should be omitted.

Abstract. Not to exceed one double-spaced page. Footnotes
and literature citations do not belong in the abstract.

■|'(\l See also Form of the Manuscript below. Follow the t'.S
(,' ■ nunent Printing Office Style Manual. 1973 edition. Fish
n.imt-. follow the American Fisheries Society Special Publica-
tion No. 6. A List of Common and Scientific Names of Fishes
from the United States and Canada, third edition, 1970. I'se
short, brief, informative headings in place of "Materials and
Methods."

Text footnotes. Type on a separate sheet from the text. For
unpublished or some processed material, give author, year, title
of manuscript, number of pages, and where it is filed— agency
and its location.

Personal communications. Cite name in text and footnote.
Cite in tootnote: -lohn -1. -Jones, Fishery Biologist, Scripps Insti-
tution ol Oceanography, La JoUa, CA 92037, pers. commun. 21

May 1977.

Figures. Should be self-explanatory, not requiring reference
to the text. All figures should be cited consecutively in the text
and their placement indicated in the left-hand margin ol the
manuscript. Photographs and line drawings should be ol
"professional" quality— clear and balanced, and can be re-
duced to 6'/.' inches |40 picas! for page width or to 3/- inches ( 19
picas) for single-column width, but no more than 9 inches (54
picas) high. Photos should be printed on glossy paper— sharply
focussed, good contrast. Label each figure. List, and typed dou-
ble spaced, each figure legend. DO NOT SEND original figures
to the Scientific Editor; NMFS Scientific Publications Office
will request these if they are needed.

Tables. Each table should start on a separate page and should
be self-explanatory, not requiring reference to the text.
Headings should be short but amply descriptive. Use only
horizontal rules. Number table footnotes consecutively across
the page from left to right in Arabic numerals; and to avoid con-
fusion with powers, place them to the left of the numerals. If the
original tables are typed in our format and are clean and leg-
ible, these tables will be reproduced as they are. In the text all
tables should be cited consecutively and their placement indi-
cated in the left-hand margin of the manuscript.

Acknowledgments. Place at the end of text. Give credit only
to those who gave exceptional contributions and not to those
whose contributions are part of their normal duties.

I.iteralure cited. In text as: Smith and -Jones (1977) or (Smith
and -lones 1977); if more than one author, list according to years
leg.. Smith I9:?6; -lones et al. 1975; Doe 1977). All papers re-
ferred to in the text should be listed alphabetically by the senior
author's surname under the heading "Literature Cited"; only
the author's surname and initials are required in the author line.
The author is responsible for the accuracy of the literature cita-
tions. Abbreviations of names of periodicals and serials should
conform to Biological A bstracts List of Serials uith Title A bbre-
iiations. Format, see recent SSRF or Circular.

.Xbbreviations and symbols. Common ones, such as mm, m,
g, ml, mg, °C (for Celsius). ' . , 'L, etc., should be used. Abbrevi-
ate units of measures only when used with numerals; periods are
rarely used in these abbreviations. But periods are used in et al.,
vs.. e.g., i.e.. Wash. (W'A is used only with ZIP code), etc.
Abbreviations are acceptable in tables and figures where there is
lack of space.

Measu

equiva

[•mcnts. Should be given in metric
nt units may be given in parentheses.

Other

lOK.M OK THE MANUSCRIPT

Original of the manuscript should be typed double-spaced on
white bond paper. Triple space above headings. Send good
duplicated copies of manuscript rather than carbon copies. The
sequence of the material should be:

FIRST PAGE

CONTENTS

ABSTRACT

TEXT

LITERATURE CITED

TEXT FOOTNOTES

APPENDIX

TABLES (eacii table should be numbered with an Arabic
numeral and heading provided)

LIST OF FIGIRE LEGENDS (Entire figure legends, includ-
ing "Figure" before each number)

FKIURES

ADDITIONAL INFORMATION

Send ribbon copy and two duplicated copies of the manuscript

Dr. -Jay C. Quast, Scientific Editor

Northwest and Alaska Fisheries Center

Auke Bay Laboratory

National Marine Fisheries Service, NO.^.A

P.O. Box 1.55

Auke Bay, AK 99821

Copies. Fifty copies will be supplied to the senior author and
100 to his organization free of charge.

UNITED STATES
DEPARTMENT OF COMMERCE

NATIONAL OCEANIC AND ATMOSPHERIC ADMINISTRATION

NATIONAL MARINE FISHERIES SERVICE

SCIENTIFIC PUBLICATIONS STAFF

1700 WESTLAKE AVE. N.. ROOM 336

SEATTLE. WA 98109

OFFICIAL BUSINESS

NOAA SCIENTIFIC AND TECHNICAL PUBLICATIONS

! ' rjii.iii ivas ciablished as pan of the Deparlment of

ics ot NOAA ;irc Iti assc
MiJ til iiKinilur and prcdicl

The Natitfiuil Oceanic iind A
Commerce on October 3. 1970. Tl
of natural and Icchnoiogical changt.
Ihe oceans and their iivinc resourci

■ss tne socioeconomic impact
I the state of the solid Earth,
of the Earth

The major compitnenis of NO.AA regula
tion in the following kinds of publications:

PROFESSION \I PAPERS — Important dcfir
research results, m.ijor techniques. ,ind special ii
ligations.

CONTRACT AND GRANT REPORTS — Re;
prepared by contractors or grantees under N(
sponsorship.

ATLAS — Presentation of analyzed data gene
in (he form of maps showing distribution of rail
chemical and physical conditions of oceans ani
mosphcre, distribution of fishes and marine n
mals. ionospheric conditions, etc.

pes of .

ntific and technical informa-

TECHNICAI. SERMCE PUBLICATIONS — Re-
ports conlaining data, observations, instructions, etc.
A partial listing includes data serials; prediction and
outlook periodicals; technical manuals, training pa-
pers, planning reports, and information serials; and
miscellaneous technical publications.

TECHNICAL REPORTS — Journal quality with
extensive details, mathematical developments, or data
listings.

TECHNICAL MEMORANDUMS — Reports of
preliminary, partial, or negative research or technol-
ogy results, interim instructions, and the like.

ol NOAA pubficotio

be obtained from;

ENVIRONMENTAL SCIENCE INFORMATION CENTER (D822)

ENVIRONMENTAL DATA AND INFORMATION SERVICE

NATIONAL OCEANIC AND ATMOSPHERIC ADMINISTRATION

U.S. DEPARTMENT OF COMMERCE

6009 Executive Boulevard
Rockville, MO 20852