on eae . aaa eT oe ti Oe PO PT ay) ye TR es Sh. AO Ry . — lie a) | ats a an | i ny er las rn : a a a) 1 PVG AM a 4S) a : ia.’ u ‘yy ve : io ol 4 vf r ’ , a t ’ u ay 9 ' 17 a4 : : -_ a »S i om , i” 7 14 s c y oa - 7 7 hy it) Aye r , ry, ti a , ee) : aa L = Sy ih a 7 } , Cie 3 yi 1 ? 7 : : : ; - h >. ee "i q _ 7 * 7 vs ! - >) Py 7 uM 7 ry _ 7 i ban = ny : 7 n a _ 7 . 7 ia 7 i a. i kot iam es 7 : oe ae : ? Mb ay , a ~W : ; ae ' : : A : 7 '"@ , _ : 7 i a ae A home

‘ a i‘. : Vee = : t a yy a 7 - a! 7 7 - u : a an aes ya ee - 7 4 a i | : fi athe 7 7 ona 4 or he ; i 7 : 7 ran a a “4m a a - 7 7 i : - a - r in _ «\ wees . o rs a, " oe 7 no - ‘ ay ( on - , ’ w , 7 2 : if ‘ = ~4 wr a ne ae iP a a i 0 = i) - i >" Pi . - a - es - a oe © oy 7 dy : ,) ; - ; De) oF a Are _ 7 a - oa : - . , > a Lite ~ a) 7 : bald “y ‘ ys 5 nit - 7 ee te ay — : m y i _ : 7 7 7 7 7 7 a : A 4 4° “ t vi yt ; > : ii : ; a i] : oY i, 7 ; 7 2m i - = mo y 7 z ’ y' : - iv 7 7 ; = mad 2 ‘ a : n>) _ ™~—) 7 a. - . 7 wi. 7 we nl - -_ 7 ; yy at, oe ls > _ . ie ; o : as 7 a ’ 7 AY TY © Pie wee i : a 7 : > = : - : as : : aa a F A i. 7 - : -~ 7 : = J 7 - a : : ; ae a ‘Un, ae _ ; : ay. > 7 & 7 U 7 ut 7 ) : 7 i. een i) - a ; 7 a: i : ie Dee ‘ 7 a : o < » | . : - - 4 . iN a) ‘ ti : J a? - 7 7 ae, : ; a au - i ’ 7 rh : a 1% br : a ; - _ "ee : =a 7 uy wo : {i Ly q ” ia ; i <] a, in - f oe 7 7 ° : "i , ae 7 ur Hh? yo i - : - ’ a 7 D 7 - a : 1h . : : a a 7 i : a :- i a i ry : r - ' ‘Se 4 ay hi’ an j : 7 - - he On i” an , ' , ee) o F - if fn) - - J ny , | ' oe , 7 _ i r oy \ _ 7 AP s On| - d ; a ; : a a a i ’ 7 ie 7 vs 4 > : ieee. ee 2 a | : : i b , a : i 7 oe i Lon 7 it 7 ’ htm it . : i v¥ i - 7 - a 7 7 <0 a a a , = : : Cy 7 ah ; nt? ‘ 1) vt 7 ‘ : ‘ 7 - - a nl > i a : ee. y i AD eu , s.¥ 7 i i ° _ - /a ; : » 7 a = ive ? a ° J : « - a i _ _ _—_ i. aL - 1 . - 12 > i 0 - ; : 7” — a > } : 7 p is r Pe: rl “iy v1 , a a ‘ : J wy, os 1 2 7 ' . Cae , i 7 i ti ; 1 a ar ey: . : i 7 : 7 : : ! : - - io 7 - Cee _ 7 ‘ : » i. ev Ss , = i ) _ an mt Bu 4, Org 7 S80 | i nn aa : 7 wa) 7 a 7 2a . 7 : _— ' oN in ay $4 . - 7 7 if A 7 7 =i » : 7 ' a x 7 7 - ' : Df i _ oF , im 7 = 7 7 7 rT _ a = \b ti +9 an 5 al) a I ; , fa ime a out : n ‘ _ oF fi 7 7 in ‘ oa 1s : a - a : ‘ty A : : 1% y 7 os a1 - xs : a ~ "os i j ae 7 i : - m ‘ vi a a - a j yi : > Gere" om 5 ; j : : - uy 0 | _ : mi p mT , on] aa) i ay ‘gs ‘ ; ; s ~ ' ’ x A ‘ m is i — x ; i iv ag a bat i i oe 9 ov) \a oa. J De ee a, FT eed . iy ms ft ane wy Ae aes ri le a, 1 Pale hae id " et Me a ee Meee | < 7) 77 ws 7 " : a" : a ee uN , a fh 9 a a a : ; P: Mt 7 ae 5 a 7 : : = Bp Hi i , ; ‘ - , 7 ; i ey ip 7 i VE) ry i a , 7 : : i 1 ; As Tey 7 Ng a ot ss ; Th rf he rr a mh i" i ‘Suen se 1 RA io en : i ates : i con Jee | » tee a t rt oak fi » y _ a i f a oo ied iy, Se "ay ‘te 6 Ae Ly) oa ‘o n° weriS ct ui ; : a ee a F i nee ie 7” i ; aa a 7 ’ 7 ; ut ‘ Pata wie : 14 a “IYe A 4) 7 t et 7 * Ran 7 a va queer aig : a f ne i 7 ey) 7 it y Ae a - P ete _ ; : : - f 5 Wy my oa oO hen oo in aig 1 j i an - v0 Sis, a ‘aan ein Pan : ; ; ; re _ =) eae if aye ee =o en ane : ; “NC are WN dhs? oe i) 7 o] 7 : i a oe i an : NM le) ae “4 na 4 os ahi 7 7 : 7 7 \Seae he todas We 4 - r : a hey oO ' Prey fi 1 : rie | 7 - 7 ina : nae eu — rg ay in . S i 4 ' - ? 7 f okie ; mee nf oe i ae — . a vai 7 wal 7 a Wy i a ae 7 7 i a! : : ; 7 , 7 iy ) ‘yo, ae 4 a yl : : : ae : a pay 7 aE D a? : rn i 7, ' = : : ee i 7 : ; i p j ny > : a : aTA oe : ane rey ke | tl oe a a, P : ey a 7 Mh . " fas on ane ees 9 " ha ee. ek Oe Lg |) he. “eee 7 Me an i uh Ai, a Me. aN ry wer va _— 7 Py an ' tae i. un y -F r i 7 ss 7 ; : mw > ; ; F 0 - t P ay - 7 if J 7 } aig Pie é ¥ } 7 : i) : — : * a. 7 oy a G, nV ed ai 7 vy, it 2 ee i = : y - / i ie iY ee “at a) a ae, i‘ y a AS ane a: ; ma ae ae : - nor - 4 : ’ Ye oe I nl a Zz - ; - : ae Vas sat " ay : 7 ‘S ae 7 : @ F i ri an oe a rf me a). ? or peek | * is arn fs a ; keer va oe Sean _ ; ' : ir a) : rina wy 7 hi 7] i v) ; | : A ae i, cu : io sey |. ah , ; ri ji , ie olen ; a - = wr : uf : ae int me) ; x av 7 vs - nas ci Ae 3 an i k va > a, f i. ee : +7 A | y. 1 Tan y nD i ; i : om + Mi i Q : We | ce ee 7 ee ke iy een 7 a | “) . | Ae i ; — u/ : Te ’ peat We ad Ot Oe : aan 7 a if AY a Anh Vs ‘ I : - tale ine ! : or ie « Iau ’ ie ’ Zz .' - ne. re, oe Pt | a re Te i Eten. t : aN mee * Care ia ® = OO i are ern i i ! 7? {5 ae 1; ; Df | f ae Lac) = 7 ‘aoe | a ae a abi! 0 nee ef : et ed - Ww ion Ave 7 ae avy uo Ma 7 — : it AS! Pa i, _ i f 7 rh (i a ‘ - j is i /} " ‘i : diss ae be ! - oe Dae eae Na Ws , a) ; feed) ae ) a mK m= PD ons : ! oe : jon ; : i a ‘ : ; u t.¥ ‘i i oA ; | tod ai | a a re : . ; Lavi : «4 : - l es 7 _ vie : " a fy t i) pal! o a in a Aan Pre on ce, | Tae rh ce \) a. i hy ae 4 , nt : - Best ay 7 . Wea ion * P in By)" 2 a - : 7 Vy > P] | a % y i) 1" - 7 ye ’ SF ; , : if — Ae -_ i Ly ie — : : - ne ~~ ‘ a ; Ma ae u mul A a ; : nt ) } a iy M in - ng | ; " an on Min a a rh i : i nw 7 Pa 7 : : ; ; i 4 ' a “s one ee vi : J 7 any 7 1%. ey Yi" Bint | a | a ane : Se | i Os ae ft - oi a) re F ’ 7 ite ul ' q ‘(an ¥ ra ; He : , af ; i 7 7 f «ae : oH) “nu jie : i 7 7 ow ‘Y * 7 a Al : 1 Wt “and [pan j a@ ; yy rae | an - » os he 7 ; i iy ' Tuan 5 7 i 4 a a if = i i a & a Te , ‘e } an 2 o hy a ae in 4 , it > oP oa ea a ¥ by , ag 2 ar ' oy — 7 b Wh a Wie ve RS " Ane >. rh 7 iy >, ; : { eee } an a a 7 ee Ak: 7 a a ~ ié 7 ae : ball Vs, a 7 } : i ni ay, as Ak ‘or i] fo, Pr >. | 7 ty oF i. Gee Mee i a A ; re ; b lit | ven : D' . ‘ | | ne my es (an ay rita | dat. ay ry Dy ; re . ), thy f es =; Po ed an q ; ied t wa or 7 4 ' mi yr a i ie - i ita. 7 v ven aN one ie vl ‘Wi % ne : ae q 7 ule o 7 ¢ i 7 = ne ‘spas on ‘th 7 ee wana a i - a | + fl ec a oy - ay eee ele 1g | i: a ‘ar ny . ia a Cy Like 0 te : oe) =a met, ih ib) 7 : ae Loe , - : Ts. a : f ny an ; > . 7 f UNIVERSITY SERIES: INHERITANCE OF SILKworMS, I. Vernon L. Kellogg, Professor of Ento- mology. 89 pp., 4 plates. 1908. Price, $1.00. THE OPISTHOBRANCHIATE MOLLUSCA OF THE BRANNER-AGASSIZ EXPEDI- TION TO Brazit. Frank Mace MacFarland, Professor of Histology 105 pp., 19 plates. 1909. Price, $1.00. A Srupy or THE NORMAL CONSTITUENTS OF THE POTABLE WATER OF THE SAN FRANCISCO PENINSULA. John Pearce Mitchell, Assistant Professor of Chemistry. 70 pp., 1 map. 1910. Price, 50c. SYNOPSIS OF THE TRUE CRABS (BRACHYURA) OF MONTEREY Bay, CALI- FORNIA. Frank Walter Weymouth. 64 pp., 14 plates. 1910. Price, 50c: . THE OSTEOLOGY OF CERTAIN ScCOMBROID FisHEes. Edwin Chapin Starks, Assistant Professor of Zoology. 49 pp., 2 plates, 1 text figure. 1911. Price, 50c. : A PuysicaAL THEORY OF ELECTRIFICATION. Fernando Sanford, Professor of Physics. 69 pp., 2 plates. 1911. Price, 50c. THE Matzke MerEmortAL VOLUME. Papers by John Ernst Matzke, late Professor of Romanic Languages, and Thirteen Colleagues. 162 pp. 1911. Price, $1.00. : : Das HistorisCHE PRASENS IN DER ALTEREN DEUTSCHEN SPRACHE. Bruno Boezinger, Assistant Professor of Germanic Languages. 91 pp. 1912. — Price; 50¢. ! Tue Errecr oF A Strictly VEGETABLE Dimer oN THE SPONTANEOUS ACTIVITY, THE RATE OF GROWTH, AND THE LONGEVITY OF THE ALBINO Rar. James Rollin Slonaker, Assistant Professor of Physi- ology. 36 pp., 1 plate, 15 text figures. 1912. Price, 50c. CATALOGUE DE Tous Lrs Livres DE Feu M. CHAPELAIN. (Bibliothéque Nationale, Fonds Francais, Nouv. Acq., No. 318.) Colbert Searles, Associate Professor of Romanic Languages. 119 pp., 2 plates. 1912. Price, -75c. THE DupLey MEMorIAL VoLuME. Papers by William Russel Dudley, late Professor of Botany, and Several Colleagues. 137 pp., 12 text figures, 9 plates. 1913. Price, $1.00. LELAND STANFORD JUNIOR UNIVERSITY PUBLICATIONS UNIVERSITY SERIES THE FISHES OF THE Stanford Expedition to Brazil BY EDWIN CHAPIN STARKS ASSISTANT PROFESSOR OF ZOOLOGY WITH FIFTEEN PLATES (ISSUED MARCH 17, 1913) STANFORD UNIVERSITY, CALIFORNIA PUBLISHED BY THE UNIVERSITY 1913 Ye ley ol Ra aes u Veet a i HAHA LIBAN et Myth Hive SLD On St RMIT TMVIMINER eT aL MONON Hae Rta wih aN i i NA y en # Ap Ky i Mp ayy i) AAR i iG d ie Ayo ny ag Dedhiten Q D LRA ORR ORO MR REN iy At MT A ‘ fi, einai eae ti Lee ' , bile , AREY Oy On AO Vy N i ae ; hat ; Abt ent } ‘ ‘ an Biv ) f aan ; By sb i t My f, q] Mh NE f d Wy Ki aia Ry Bath) , h mah AN) Say Nester a ey wt A it Pe AL wad ye 4h i f DSc a A Ma hat a f Yi neh ean SARL SN A iy, : ear At } nf Rey Ber a ni ; ¢ \ With i if nee eu Ai ey ire lant ) aay Ava aya RAPA) | aa HK ‘ i } Re Beatty PGs Oe Hara) LNA ea Aye iit near aN in ets if ry Ph Ni daa oe Mi IN (et I v ay rae Ht 1 I, tly " in j bt TABLE OF CONTENTS hs} Si ay ANG? ; » VA NrtrOductlomen ae. esers oO yh och Savste oes 3 ldemmanlhy, (Galleon een oaebad Ss Gon sdoe + idaranilhys Sol shana ye nga n Goo a cuore 5 Bamily wR binobatideey ).. 2.2.0... - 5) Mairyad liv aeleri GGG aoiar.,2/ss ors ice eel pectic 6 Metanilyae asyatichde sa. sa as ieicrleieie aie ate 6 amailvaselopiciass ashi. 6 v4y.04 eatery 8 HamanlyaeAllipilidaesys titers cee 8 iamuliva Glupeidae 2. 2a accies ascot 8 Hamily Eneraulididé 2.055... 9 Bamuly Osteoglossid@....2.5.6.008. 12 Family Symbranchide........../.. 12 amily Ophichthyides....i0....).. 12 RamaihyaiMurcenidce ss 358 ic sncesiisie s « 12 Bamuahyn@Characinida). s,s hncec- sees: 13 BRamiulya Gymnotide:.:. oss. acces -- 21 Family Electrophoride............ 25 Family Bunocephalide............. 25 Ramnalye suluridee yi) 65 3 oc. iele seieis 2 0 25 Family Hypopthalmide............ 34 Bamuily (oricatidae: 0.42554 5..4.c 0% 34 Banuly (Callichthyide...... 0.3.5. 39 Family Synodontide............... 39 Hamiuliya Poecilivdae® 2s). sans anaes: 39 Becsiy Belonidee ss sooo iis ale cles es 40 Family Hemiramphide............ 40 amily xoccetidces sso 50 Woy sc. 40 Hamilysyngnathide 2 0.030. . 40 Hanmiihy (Atherinide. 3) 4. heu.s ie so 41 Pamiily Mine ibides l3)s ec yee ie bk 41 Bamily Sphyraenide.. 0.60.5... 41 BRamily: Holocentrida) 0. 00.405... 4] Ramis scombridcey../ja.)0me et ooo 42 Banaihyeninichiunidae. ol. ae). oo: 42 Family Carangida: 0.5.2 0.4..05-% 43 Family Apogonichthyide.......... 44 Family \Centropomidz +... <.'!). 45 BRamilyasernanid ces Vacs sais 45 BRamahyetttanidice ys sais ye eis 46 . Ramiulynevesnatliclaenys sc sv ecpectss ale sy: 46 Mannilys Spariddes'. scien denen cei 47 Ram#ilynGerridcsey ys vac haces crs: 47 Pamily;,Polynemidzewi ys toa sae 49 Bamathys Scicemid eyes ae)s- aes seers 49 Bamily.Cichiltdaeny vous eee ese acs 54 Family Pomacentride...2...5..:.- 58 BRamilys wabridaeceaeccsscaon cee 59 RamilysscanGererecec aces ocean 61 Ramily Ephippidces. ace.cesciect- 61 Family Chetodontide............- 61 Ramily vAcanthunidee saa. sich cl 62 Bamailya Balistideeten sss seeie- cee 63 Family Monacanthide...........-. 63 amily, Ostractidaes 2.34. o--c0 == 63 Ramily, Detraodontide:.. <<... 2. - 63 Hamily, Diodontide.s. 3.2. .2a-.---- 64 Kamiya SCOnpenidaeycq secs esas 64 Hamulys Gobtidasecri serie cere see 64 Family Echeneidide ............-- 69 Family Malacanthide............. 69 Hamily, Dactyloscopida nyse 424. 70 Family’ Batrachoidide.. 0.3.) .25 71 BamilyGobiesocidaeasa.aieeerees 73 RamiulynBlennitdee:|yyasnectnca ce ele 74 Family, Pleuronectidzw 43.2... 4.4: 76 amily Solerdaayy.js esr ecialcieiietaes 76 RamilyeAntennceridce ses a. sees iM Family Ogcocephalide............. 77 Sanity ly Hy Mi vf ay if ah nM a i) r Hs i) 1 nth is GAY, i Weal het id Un bp ‘ ND hs iy, UN i ae De TU ; ha } oa yy Me i ny f ue } vay } ‘ shay Ni Ne A ny Re feo te FISHES FROM BRAZIL COLLECTED BY THE STANFORD EXPEDITION OF 1911 HE following pages contain an account of the fishes collected by the author as a member of the Stanford Expedition to Brazil in the summer of 1911. Several weeks were spent at Natal, in the State of Rio Grande do Norte, where the rock pools, though containing a rather small fauna, proved to be the most fruitful collecting ground in the vicinity. Seining was done in the harbor and as far up the estuary as Carnahubinha. ‘Two trips were made in a trawling boat, equipped with an otter trawl, that a local company had imported from England, but on account of the great quantities of seaweed that filled and wrecked the nets few fishes were secured. ‘The fish market was extemely poor, and contributed but very little to the collection. Seldom Over six or seven of the commonest species were represented at one time, and the prices that they commanded proved the supply to be far short of the demand. A trip was made to Lake Extremoz, nine or ten miles in a north- westerly direction from Natal. After several trials it was found impossible to do any seining in the lake owing to the great quantities of half dis- integrated vegetable matter that is held in suspension in the water, and that quickly filled the net and made it too heavy to land, so that most of the specimens that were taken were secured by fishermen with cast nets.* In the same direction as Lake Extremoz and about nine miles farther is the town of Ceara Mirim, where several days were profitably spent in collecting in the small creek and in several little muddy ponds that were disconnected from each other, but are doubtless connected during high water. Some seining was done, but more dependence was placed upon the very efficient boys, who waded about with dip-nets and caught fishes under the overhanging muddy banks. * The fish fauna of Lake Extremoz is made up of the usual fresh water forms of the region and, in addition, many typical salt water ones, such as Centropomus, Mugil and Gerres, showing that the lake has been connected with the sea in recent years, though it is said to be cut off by sand hills at the present time. + FISHES FROM BRAZIL Several days were spent at Lake Papary, about twenty-four miles south of Natal. The lake is large and shallow, with marshy banks and muddy estuaries choked with water-hyacinth, so that seining here, too, was im- possible. On our account the fishermen were given permission to practice a certain method of destructive fishing that is ordinarily prohibited by law. The fishes were surrounded by a great gill-net, into which canoes were pad- dled and the fishes taken in extraordinarily large cast-nets or driven into the meshes of the gill-net. By this and the more usual methods of fishing a rather large collection was taken from the lake. A few days were spent at Fortaleza, in the state of Ceara, with little results from an ichthyological standpoint, and a couple of weeks at Para, where collecting was confined to the excellent fish-market and a very good collection secured. Several specimens were taken by Dr. Fred Baker and Mr. W. M. Mann in the Madeira River during a trip taken after the conclusion of the main part of the expedition. I wish here to especially thank Dr. J. C. Branner, chief of our ex- pedition, not only for the opportunity of making this collection but also to thank him, and my colleagues as well, for much help of various sorts in the making of it. To Dr. Bashford Dean I am indebted for help without which my participation in the expedition would have been impossible. In this small paper it is scarcely possible to attempt to acknowledge much of the assistance that I at various times received, but I can not forbear the pleasure of here thanking Mr. José Joaquim de Carvalhoe e Araujo, whose hospitality and influence made our stay at Papary so pleasant and profitable. This collection, including the types of the new species, is deposited among the collections of Stanford University. A set of duplicates has been sent to the American Museum in New York. The accompanying plates, illustrating the new species, are from draw- ings made by Chloe Lesley Starks. Famity GALEID. 1. Carcharhinus platyodon (Poey). A specimen 29 inches in length, secured at Para, seems to be referable to this species. The mouth is twice as broad as it is long and the preoral part of the snout is contained 114 times in the space transversely between the corners of the mouth. The front of the head is semi-circular in outline. The fins are all more or less concave behind. The pectorals when folded EDWIN CHAPIN STARKS 5 back almost reach to opposite the posterior end of the dorsal base. The length of the caudal from the pit on the upper part of its base to its tip is equal to the space between the front of the head and the dorsal fin. The second dorsal is a little in front of the anal, and the base slightly exceeds that of the latter in length. The anal is unlike the second dorsal in form, being deeply notched behind with its lobes almost equal, while the dorsal is concave behind, with its inner lobe reaching far behind its outer. The color is slaty blue above and pure white below. The dorsal and caudal are outlined in black, especially the caudal, which has a broad, black posterior margin. The tips of the other fins are dark. 2. Hyproprion brevirostris Poey. A specimen, a couple of feet long, taken at Para. Its teeth are not mature enough to note their final character. They are abruptly widened at the base and very slightly serrate. In general proportions of fins and body this specimen agrees very well with current descriptions. FamMity SPHYRNID. 8. Sphyrna tiburo (Linnzus). A small specimen collected at Natal. FaMILy RHINOBATID-. 4, Rhinobatus percellens (Walbaum). Two fine specimens were taken in a trawl in deep water by a fishing company at Natal, that has imported a trawler from England. The greatest width of the disk is equal to the space from the tip of the snout to the middle of the eyes. The preoral part of the snout, from between the teeth, is contained 24 times in the length to the vent. The front of the vent is midway between the tip of the snout and the base of the caudal. The width of the mouth is contained 234 in the preoral part of the snout. The lower part of the tail has a projecting flap that is wider than the usual keel. Two papille are on the margin of the spiracle; the outer one much the larger. The distance between the dorsals is equal to the width of the tail behind the base of the ventrals. The ground color is white or light sienna, with round brown spots a little smaller than the eye. Between them are smaller spots, similar in color but less regular in outline. The spots are scattered over the upper parts so closely that the ground color shows only as reticulations. There are 6 FISHES FROM BRAZIL also a few small round, light spots scattered sparsely over the middle of the back. The dark spots are somewhat larger and more conspicuous on the tail and run back on the caudal fin. On one specimen there are two pairs of dark spots on the rostral ridges. The under part of the snout has a black blotch. Famity PRISTID-. 5. Pristis perrotteti Valenciennes. A large “saw,” 4 feet in length, was obtained in Natal. It has 18 teeth on each side and is much more tapering than in Pristis pectinatus. 6. Pristis pectinatus Latham. One small specimen was taken at Natal. FamMILy DASYATID. 7. Dasyatis gymnura (Miiller). A specimen taken at Para. 8. Dasyatis say (Le Sueur). Two specimens with the disk 11 inches in length, caught in a trawl in deep water at Natal, are referred to this species, though there are some dis- crepancies between them and current descriptions. ‘The disk is no longer than broad, and is nowhere concave. ‘The front of the disk is nearly straight on each side, meeting at an angle at the snout. The greatest width is opposite the gill openings. ‘The caudal is no longer than the disk, and has a broad fold above and below. ‘There are 6 or 7 tubercles on the median line of the back in front of the shoulder girdle, and one some distance behind this, in one of the specimens. A group of 2 or 3 very small spines is on the shoulder. The skin is otherwise perfectly smooth. The color in one of them in life was light dusky green, growing red- dish toward the edges of the disk. Posteriorly it was bordered with dusky and narrowly edged with white. The caudal was white, with the dorsal and ventral folds black. The other specimen is very much darker in alcohol, but the posterior white border of the disk is evident. Both of the specimens have large irregular black blotches below. 9. Potamotrygon hystrix (Miller and Henle). The species of this genus are very much in need of revision. Doctor Eigenmann has included this and the next with several others under one EDWIN CHAPIN STARKS ri species (The Freshwater Fishes of British Guiana). Material at hand from Para represents two distinct forms, which seem to be referable to this and the following species. Seven specimens were collected in the market at Para, and one near the mouth of the Madeira River (by Mr. Mann and Dr. Baker). ‘They are from 10 to 20 inches in length, or measuring the length of the disk only, from 5 to 11 inches. In these the preoral part of the snout is contained in the length to the front of the vent from 4 to 4% times. The width of the mandibular tooth-patch is contained from 13 to 13% times in the same length. The interspiracular width is contained from 14% to 1% times in the length of the snout from the eyes. The teeth are stained brown in all of the large specimens, and usually in the small ones. Counting the longest rows of teeth in the mandible from the outer side obliquely to the median line they number from 12 to 14. The largest papillae behind the teeth are about 3 times longer than broad, and nearly or quite half as long as the eye. The color is variable. All of them are light dusky brown with lighter spots scattered over the back of cream or light sienna color. ‘The spots are with blended edges shading into the ground color. Surrounding the spots at some distance from their edges are usually elongate, irregular, black blotches forming more or less regular broken rings of greater diameter than the eye. In one or two of the smaller specimens the rings are nearly com- plete, but in one they are entirely absent. Usually the black spots forming them are sharp-edged and very definite, but occasionally they are soft and blended. Sometimes they are so separated that they scarcely suggest rings, but run irregularly in various directions. The side of the tail is very dis- tinctly barred with alternate light and dark spots. , 10. Potamotrygon motoro (Miiller and Henle). Three specimens from 11 to 22 inches in length (or measuring the disk only, from 6 to 12 inches) were collected in the market at Para. The preoral part of the snout is longer than in P. Aystrix, being con- tained in the length to the front of the vent from 3% to 3% times. The width of the mandibular tooth-patch is contairled 10% to 11 times in the same length. The interspiracular width is contained 134 times in the snout from the eyes. The teeth are not stained brown and the longest rows from the outer side to the middle of the mandibular patch number 19 or 20. The longest papilla in the mouth are scarcely longer than broad, and do not exceed the length of the pupil. The disk is more thickly set with fine spinules than in P. hystrix, so that it is more like ordinary shagreen. 8 FISHES FROM BRAZIL In life the disk is dark brownish drab, growing more drab toward the edges. Scattered everywhere are small orange spots as large as the pupil and sharply defined by rings of color similar to the ground color but of a darker shade. ‘These are more crowded at the middle of the back and at the edges of the disk. On the tail they are less distinct and without much orange color. The side of the tail is mottled rather than barred. FamMILy ELOPIDZ. 11. HElops saurus Linneus. A common species on the Brazilian Coast. A few specimens were col- lected in Lake Papary. The elongate, transparent, larval forms of this or Albula were frequently taken in the seine. 12. Tarpon atlanticus (Cuvier and Valenciennes). One specimen was obtained at Lake Papary. Famity ALBULIDZA. 13. Albula vulpes (Linnzus). A single small specimen from Natal. Famity CLUPEID~. 14. Sardinella sardina (Poey). Specimens obtained in great abundance at Natal are placed here with some doubt. They have a distinct lateral band, which may be due to the action of formalin in destroying the overlying silvery pigment. ‘The scales are thin and entire edged, not laciniate as in Sardinella macrophthalmus, which they resemble in form. ‘The scales seem to be, moreover, more firmly adherent than is usual in S. sardina. In the lateral band they are very similar to the Pacific Coast species, S. stolifera, but the eye is considerably larger. 15. Opisthonema oglinum (Le Sueur). Several specimens were collected at Natal. The variation of the depth is remarkable in this species as it is in the Pacific species, Ofisthonema libertate. It varies from 234 to 334 times in the length to the caudal base. 16. TIlisha flavippinnis (Valenciennes). A few specimens collected in the market at Para. EDWIN CHAPIN STARKS 9 17. Pristigaster cayanus Cuvier. A couple of fine specimens were collected by Dr. Baker and Mr. Mann in the Madeira River about 400 miles above its mouth. FAMILY ENGRAULIDID. 18. Anchovia clupeoides (Swainson). Several specimens from 5 to 8 inches in length were taken in Lake Papary. The anal rays vary fom 28 to 32 (not counting those in front of the first long one), but are usually 30, as described by Dr. Steindachner. His specimens were from 3% to 5 inches (‘9 to 12 cm.”) in length. The smallest specimens at hand agree very well with his description, but the large ones are much deeper, the depth being much greater than the length of the head, and is contained from 3 to 33% times in the length of the body, while the head is contained from 334 to 4 times. 19. Anchovia januaria (Steindachner). Specimens that answer very well to the description of this species, which has hitherto been known only from the harbor of Rio Janeiro, were taken in abundance in the harbor of Natal. 20. Anchovia pallida Starks, new species. Plate 1. The head and body are deep and compressed. The depth is equal to the length of the head and is contained 3% times in the length of the body to the caudal base. The snout is shorter than the eye and projects almost its full length beyond the tip of the mandible. The maxillary is broad and rounded behind, and scarcely reaches to the posterior end of the mandible. Small even teeth are present in both jaws. The diameter of the eye is contained 41% times in the head, and 234 times in the postorbital part of the head. The interorbital width slightly exceeds the length of the snout. The cheek is long and triangular, and measuring from the eye to the lower angle of the preopercular ridge, its length is twice that of the eye. The greatest width of the opercle is no greater than that of the eye. Forty slender gill-rakers, barely as long as the eye, are on the lower limb of the first gill arch. The front of the anal is under the middle of the dorsal base, and the front of the dorsal is midway between the base of the caudal and the front margin of the eye. There are 14 dorsal rays, counting 2 rudimentary ones 10 FISHES FROM BRAZIL in front of the first long one, and 21 anal rays with the 2 short anterior ones. The base of the anal slightly exceeds the length of the head. The length of the pectoral is two-thirds that of the head, and it reaches a little past the front of the ventral. There are 37 cross series of scales and 7 longitudinal series, counting under the front of the dorsal. No color whatever is present in the alcoholic specimen, except a few fine, scattered dark points on the back. The type and sole specimen is 4 inches in length, and was collected in the market at Para. In the shape of the head and body as well as in various other char- acters this species is very much like Cetengraulis edentulus, but there is no trace of membrane connecting the branchiostegal membranes that charac- terizes the genus Cetengraulis. Should future specimens, however, prove it to belong to that genus it may be known from C. edentulus by the much shorter operculum as compared with the long oblique cheek, longer maxil- laries and pectorals, the anal under the middle of the dorsal base, and several other minor differences. C. juruwensis differs in the same characters, and in addition is much more slender. In the genus to which it is here referred it seems to be closest to Anchovia vaillanti (Steindachner), but it has twice as many gill-rakers and lacks a lateral stripe. 21. Anchovia brownii (Gmelin). A few specimens seined at Natal from 234 to 334 inches in length. It was not taken so commonly as A. januaria or Lycengraulis grossidens. These specimens were compared with some from Jamaica and agree in all respects, but they show some discrepancies with current descriptions. The depth is contained from 5 to 5% times in the length to the caudal base. The eye is contained from 334 to 4 times in the head; the snout from 44% to 4% times. The anal rays, counting back from the first long ray, number 18 or 19. The front of the anal is under the posterior fourth of the dorsal base. The front of the dorsal is midway between the base of the caudal and the front of the eye. 22. Pterengraulis atherinoides (Linnzus). This species is very common in the market at Para, where many speci- mens were secured. In the following notes specimens from 6 to 10 inches in length are considered. The head is contained from 4 to 4% times in the length to the base of the caudal; the depth from 3% to 4 times. The eye is much longer than EDWIN CHAPIN STARKS 11 the snout and is contained in the head from 5 to 5% times. The maxillary has a bluntly rounded end which is even with the posterior end of the mandible. 23. lLycengraulis grossidens (Cuvier). A few specimens were seined in the harbor of Natal, the largest 61% inches in length. The head is contained from 4 to 4% times in the length to the caudal base; the depth from 4% to 434 times. The eye is longer than the snout and is contained from 2 times in the postorbital part of the head (in speci- mens 3% inches long) to 24 times (in specimens from 5 to 6 inches long). The number of teeth in the mandible is very variable, as might be expected from their uneven size and position. The maxillary teeth are more even in size and smaller. Those on the anterior part of the bone are directed downward or slightly backward, while those on the posterior part are di- rected slightly forward. The mandible curves upward and is sharp at the tip. The maxillary is slender, slightly widened, lance-head-like, posteriorly, and ends in a slender point just behind the articulation of the mandible. The head is rather broadly rounded behind at the gill cover, and not particularly elongated obliquely. The insertion of the dorsal is midway between the base of the caudal and a point varying from the middle to the posterior edge of the eye. The front of the anal is under the middle of the dorsal. In the large specimens the pectorals reach to, or nearly to, the ventrals. The head and particularly an area just behind the upper half of the shoulder girdle is richly supplhed with ramifying canals. On the cheek and on an area behind and above the eye the canals form a net-work, in- closing small spots that appear, at first sight, like well separated, imbedded scales. 24. Lycengraulis batesii (Giinther). A single specimen, 8 inches long, was collected in the market at Para. In this specimen the teeth in the mandible are fewer and a little farther apart than in Lycengraulis grossidens, though more were described for the type. Their number is doubtless variable. This species differs in being much more slender, in having the front of the anal only slightly behind that of the dorsal, in having the head shorter as compared with the entire length, and in having the gill-rakers shorter. The last are low on the side of the arch and are about as long as the mandibular teeth, though they are not ‘‘like tubercles,” as originally described. 12 FISHES FROM BRAZIL FAMILY OSTEOGLOSSID. 25. Osteoglossum bicirrhosum Vandelli. A specimen was collected in the market at Para. FamMiILty SYMBRANCHID-. 26. Symbranchus marmoratus Bloch. Numerous specimens were taken at Lake Papary and in the muddy little lagoons about Ceara Mirim. BRAMILY, OPHICH TH VIDAs: 27. Myrichthys oculatus (Kaup). A single specimen was seined in the harbor at Natal. Famity MURAENIDA. 28. Rabula megalops Starks, new species. Plate 2. The head is contained 224 times in the trunk to the vent, and 5 times in the entire length. The body is not much compressed, and the head and neck not at all. The depth of the body is a third of the length of the head. The gape does not quite extend to the posterior part of the eye, and is a fourth of the length of the head. The jaws are straight and capable of being completely closed. The teeth are sharp and easily detached. ‘They are biserial in the upper jaw, with a group of enlarged curved teeth in front. There are no teeth on the shaft of the vomer running back between the lateral teeth. In the lower jaw the teeth are uniserial on the side and biserial toward the front. The eye is contained 14% times in the length of the snout, and 8 times in the head. The snout is broadly rounded as viewed from above. The anterior nostril is in a tube, and the posterior one is in the upper lip, opening downward, and not visible in a side view. The dorsal fin begins a distance behind the gill opening equal to the length of the snout. The longest rays are toward the posterior end of the dorsal, where their length is contained 224 times in the length of the head. The ground color is light, but made almost uniform dull brown by dark points which more or less run together. On the back and side this is scarcely appreciable except with the aid of a lens, appearing to the unaided eye uniform brown, but on the belly the points are better separated, EDWIN CHAPIN STARKS 13 and also on the tail a few light spots of the ground color show. ‘The fins are uniform dusky brown like the body, but posteriorly they grow slightly darker toward the margin, and are narrowly edged with white. The type and only specimen is 5 inches in length, and was collected in the rock pools at Natal. This species differs from Rabula panamensis (Steindachner) in having the jaws straight so that the mouth may be closed, and in having the dorsal beginning behind the gill opening. 29. lLycodontis moringa (Cuvier). Several small specimens were taken in the rock pools at Natal. On some of them the dark spots have so run together as to leave none of the usual ground color, or just a trace of it showing as a few indefinite, ir- regular markings. The light border to the anal and posterior part of the dorsal seems to be constant. In life this species is sienna yellow, with slaty brown spots scattered irregularly over the head and body. ‘The anal is dusky, and narrowly and sharply edged with milk white. ‘The white border is continued around the caudal and onto the dorsal, where it becomes narrow anteriorly and dis- appears. The dorsal has a dark border at the edge of the fin anteriorly and next to the white edge posteriorly. ‘The base of the dorsal is spotted like the body. 30. Lycodontis funebris (Ranzani). A single specimen a couple of feet in length was taken at Natal. It is uniform in color and shows no longitudinal lines on the fins. Famity CHARACINIDA. 31. Curimatus leucostictus Higenmann and Higenmann. One specimen was secured by Mr. Mann and Dr. Baker in the Madeira River about 400 miles from its mouth. ‘There are traces of a light longi- tudinal bar on the posterior part of the side, its upper edge at the lateral line. 32. Curimatus schomburgki Giinther. Several specimens collected at Para are entirely similar to specimens from British Guiana collected by Dr. Eigenmann. It is probable that these should all be referred to Curimatus ,4§prinoides (Linneus) . The dorsal reaches variably to the adipose dorsal or to as far back as the base of the caudal. 14 FISHES FROM BRAZIL 33. Curimatus elegans Steindachner. Numerous specimens were taken at Lake .Kaxtremoz, Lake Papary and at Ceara Mirim. The largest ones are almost 6 inches in length. Small ones, 2% inches long, have a black lateral band which tapers to a point and dis- appears anteriorly, or it may be continued as a narrow line to the operculum. The dorsal has a black spot at the middle of its base. In the large speci- mens there is a rather broad diffused dark lateral band most evident poste- riorly where it follows the lateral line, and anteriorly.arching just above the lateral line. The posterior part of the lateral line has a small black spot above and below the pore on each scale. The dorsal spot may be pale but it is usually evident. 34. Gasterotomus latior (Spix). A specimen, 8 inches in length, was collected by Mr. Mann and Dr. Baker in the Madeira River 400 miles from its mouth. The head is contained 334 times in the length to the caudal base, and the depth 244 times, being considerably deeper than the picture published by Spix (Pisc. Bras. Pl. 41). The front of the dorsal is midway between the tip of the snout and the posterior part of the base of the adipose dorsal. The pectoral is contained 134 times in the head, and is not so long as the ventral. It does not reach to the ventral, and the ventral reaches a little over half way from its base to the anal. The dorsal contains 11 rays, and the anal 14. The base of the latter is equal to the combined length of the snout and the eye, and the tip of the adipose dorsal is above the base of the last ray. The caudal fin is broken. There are 90 series of scales, all of which are roughly denticulated, but to the touch scarcely feel rough. The ventral median line from the pectorals to the anal is trenchant. The scales scarcely meet on the median line in front of the dorsal, leaving an extremely narrow naked line. 35. Hemiodus microlepis Kner. Four specimens were taken in the Rio Madeira about 400 miles above its mouth by Mr. Mann and Dr. Baker. Mr. Fowler (Proc. Phil. Acad. Sci. 1906, p. 319) found the ‘‘Lower lobe of caudal much longer than upper, just the reverse of that indicated by Kner.” In the specimens at hand two of them have the upper lobe a trifle longer, one has them about equal, and the other has the lower lobe a little longer. Except that these specimens are considerably more slender (the depth is contained from 4%4 to 434 times in the length) they agree very well with Mr. Fowler’s description. EDWIN CHAPIN STARKS 15 36. Anisitsia notata (Schomburgk). Five specimens were collected at Para. 37. Prochilodus migricans Agassiz. Several specimens were collected at Para, and a couple by Dr. Baker and Mr. Mann in the Madeira River. ‘The latter differ in having the stripes, which follow the rows of scales on the back, and the dusky cross bars much more conspicuous. In the Para specimens these markings are more or less obscure, but usually evident. 38. Schizodon elongatus Steindachner. A specimen taken by Dr. Baker and Mr. Mann in the Rio Madeira about 400 miles from its mouth. 39. Schizodon fasciatus (Spix). A couple of specimens collected by Mr. Mann and Dr. Baker at the mouth of the Madeira River. 40. Leporinus friderici (Bloch). One specimen was taken at Para. 41. Leporinus pachyurus Cuvier and Valenciennes. A single specimen was found in the market at Para. 42. Leporinus affinis Giinther. A specimen taken at Para. 43. lLeporinus fasciatus Bloch. Several specimens were collected at Para. 44, Leporinus maculatus Miiller and Troschel. One specimen was taken at Lake Extremoz. 45. Cheirodon insignis Steindachner. A few specimens taken in the little ponds about Ceara Mirim. In the male the lower edge of the caudal peduncle is produced and armed with from 7 to 9 spines along its edge. The caudal spot is inconspicuous in the male and the first ray of the anal is longer, making the posterior margin of the fin more concave. There is considerable variation among the females in the concavity of the anal and in the depth. 16 FISHES FROM BRAZIL 46. Astyanax bimaculatus Linneus. Many specimens were taken at Para, Lake Extremoz, Lake Papary and Ceara Mirim. Specimens of 4 inches in length and over from Para have the caudal spot and lateral band almost entirely absent. British Guiana specimens of all sizes (collected by Dr. Eigenmann) have the spot and band conspicuous. Comparing small specimens the difference is not so great. ‘The anal has 31 or 32 rays, and there are from 30 to 33 series of scales. The specimens from Lake Extremoz, Lake Papary and Ceara Mirim have the spot and band as in the British Guiana specimens or even darker, and the anal has from 25 to 27 rays. These are perhaps referable to Astyanax bimaculatus nove Eigenmann, though the lateral band is not so definite as in the picture of the original specimen. Among the smaller specimens of from 2 to 2% inches in length are many variations of the humeral spot. Sometimes it is elongate horizontally ; sometimes it is round; sometimes the faint streak extends up and down from it; and sometimes the streak becomes so pronounced that the spot is elongate vertically. The larger specimens do not vary in this respect, the spot being always elongate vertically. There are breeding females among these only 2 inches in length. In life the Lake Papary specimens were silvery with golden yellow tinges, which were more conspicuous over the faint lateral band. The pectoral and caudal were yellow, the latter much darker, and the anal and ventrals bright red. I fail to appreciate the naked predorsal line in either these or the British Guiana specimens, though the latter have been referred by Dr. Eigen- mann to the genus Poecilurichthys, which he differentiates by this character in his key. 47. Chalcinus angulatus curtus Garman. Four specimens were collected by Mr. Mann and Dr. Baker in the Rio Madeira about 400 miles from its mouth. ‘The depth is contained in the length to the caudal from 234 to 2°/+ times. 48. Chalcinus angulatus fuscus Garman. A single specimen taken at the mouth of the Rio Madeira by Mr. Mann and Dr. Baker is evidently referable to the subspecies. The back is dark and has longitudinal dusky stripes following the rows of scales. The lips and barbels are dusky, and the latter are as long as the eye. All of the fins are dusky. EDWIN CHAPIN STARKS il 49, Chalcinus rotundus (Schomburgk). One specimen was taken by Dr. Baker and Mr. Mann at the mouth of the Rio Madeira. Comparing it with a specimen collected by Dr. Eigen- mann at British Guiana it differs only in having no dark pigment anywhere on the fins or body. 50. Chalcinus elongatus Giinther. A number of small specimens up to 8 inches in length were taken at Para. The length of the pectoral is usually equal to half of the distance between the snout and the dorsal, but it varies from this to half of the distance between the snout and the last dorsal ray. It reaches to the middle of the ventral, which reaches half way from its base to the anal. The front of the anal is under the last dorsal ray or a little behind. ‘The pectorals are sometimes colorless or sometimes dark with fine black points. The adipose dorsal is over the last anal ray and coterminous with it, or often the latter projects a trifle farther back. 51. Piabucus dentatus Koelreuter. Three specimens, the largest 6 inches in length, were obtained at Para. They agree very well with the description published by Dr. Eigenmann (F. W. Fish. of Brit. Guiana, p. 316) except in depth of body. It is there stated to be 3.6 of the length. Giinther (Cat. V., p. 343) gives the depth as one-fifth or one-sixth of the total length, including the caudal, and the picture published by Bloch (Ausl. Fische, p. 382) agrees with this. In my specimens the depth is from 4% to 434 in the length to the caudal base, or 5% to 534 with the caudal. It is possible that Dr. Eigenmann’s descrip- tion should read 4.6 rather than 3.6. 52. Pygopristis gibbosus Starks, new species. Plate 3. Though this species was found to be one of the most common of the forms related to Serrasalmo in the market at Para it seems to have remained undescribed, for it has been referred to Pygopristis denticulatus. The type of Pygopristis denticulatus came from British Guiana, and as the specimen with which I compare this species was collected by Dr. Eigenmann at Lama Stop-off, British Guiana, there is every reason to sup- pose that it is a representative of that species. The depth in front of the dorsal is contained in the length 1% times, or sometimes a trifle less. In P. denticulatus it is 134 times (though Dr. Eigenmann reports 1.66). The outline of the body is much more angu- 18 FISHES FROM BRAZIL lated and the curve from the dorsal to the snout much more pronounced than in the Guiana species, being more convex at the nape and concave above the eyes. The head is contained from 3% to 336 times in the length. The eye is contained from 3% to 334 times in the head. Five trilobed teeth are at each side of the upper jaw, and 7 at each side of the lower. They are about as distinct, and sharp, and as well developed posteriorly on the mandible as anteriorly, while in P. denticulatus they are lower, blunter, and less projecting posteriorly. ‘The mandible is rather slender; its depth in front is less than half of the diameter of the eye, and is contained 3 times in its own length. In the other species the mandible is very heavy, its depth being equal to three-quarters of the eye, or half of its own length. There are 33 ventral scutes besides a paired one in front of the vent and another behind it, but there are none at the side of the.vent as in P. denticulatus, where there are four at each side posterior to the median ones. There are 15 or 16 dorsal rays, and 32 or 33 anal rays. In other species there are 19 dorsal rays and 35 anal rays. The distance from the dorsal to the adipose dorsal is scarcely shorter than the base of the dorsal, while in P. denticulatus it is not over half as long as the dorsal base. ‘The first anal rays are a little produced beyond the others. The length of the pectoral is contained 1% times in the head; that of the ventral 2 times. The pectoral reaches to above the ventral base, and the ventral to the last unpaired ventral scute. The scales are larger than in the other species, there being 71 tubes in the lateral line, 82 cross series, and about 32 verti- cally from the lateral line to the front of the dorsal. ‘The scales are rather uneven and difficult to count crosswise, but the difference in size when the scales above the lateral line are directly compared is very striking between the two species, being much finer in P. denticulatus. The specimen at hand of P. denticulatus, and those described by Dr. Eigenmann, are uniform in color, without spots or a dark caudal margin. In alcohol this species is slate-blue above, silvery below and on the side, and the fins yellow at the base. Over the back and upper part of the side are scattered dark blue spots as large as the pupil and smaller. A dusky blotch is just below the anterior part of the lateral line. The caudal is edged with black. The type is 6 inches in length, and several cotypes are as large and others slightly smaller. 53. Pygocentrus piraya (Cuvier). One small specimen was taken at Para. The side is closely covered with dusky spots and the posterior edges of the vertical fins are dark. EDWIN CHAPIN STARKS 19 54. Serrasalmo spilopleura Kner, A fine specimen of this species was taken by Mr. Mann and Dr. Baker at the mouth of the Madeira River. 55. Metynnis hypsauchen (Miiller and Troschel). A single small specimen was taken at Para. 56. Metynnis maculatus (Kner). One specimen was taken at Lake Papary and two at Lake Extremoz. One of the specimens from the last locality has the anal slightly convex and angulated in front, as in the picture accompanying the original descrip- tion. The other has it more rounded in front and produced at a rounded angle at the anterior third; probably the mark of the male. It also has the spots larger and more definite than in the others. A much darker spot is present in all of them above the anterior end of the lateral line. 57. Myleus parma (Giinther). Four specimens 6 inches long and one 11 inches, taken in the market at Para. The large one has the middle anal rays produced (male), and the procumbent predorsal spine is scarcely evident. The depth (in the smaller ones) is contained 1%4 times in the length, and the head 4 times. The back is highest at the front of the dorsal, and the ventral outline deepest at the anal, so that the cross axis of the body is oblique. The anal lobe is a trifle longer than the head, and longer than the anterior dorsal rays. The pectoral does not reach to opposite the ventral, which reaches half way, or a little more, from its base to the anal. There are 36 to 38 ventral scutes. The front lobe of the anal is black, and all of the ventral fins are margined with black behind. 58. Mylosoma aureum (Spix). Three specimens were taken at the mouth of the Madeira River by Dr. Baker and Mr. Mann. ‘The depth is from 1% to 1% in the length to the caudal base. The anal is angulated behind the posterior third, as in the drawing published by Spix. There are 43 or 44 ventral scutes, with 4 or 5 paired ones surrounding the vent. One specimen has 49 scutes, all of them simple and running at one side of the vent. In this it is apparently. ab- normal. 59. Mylosoma albiscopus (Cope). Two specimens taken at Para seem to be referable to this species. The anal is less angulated than in Wylosoma aureum but more evenly rounded, 20 FISHES FROM BRAZIL as in Dr. Cope’s drawing of the type. There are from 47 to 50 simple scutes along the median ventral line, and from 7 to 9 pairs are crowded in around the vent. 60. Mylosoma herniarius (Cope). Two specimens were collected by Mr. Mann and Dr. Baker just below the mouth of the Madeira River in the Amazon. They are slightly less deep than the picture published by Mr. Fowler (Proc. Phil. Acad. Sci. 1906, p. 477, fig. 56), which has the depth equal to the distance from the snout to the adipose dorsal. In these specimens the depth is equal to the same distance, less 34 the diameter of the eye. The sigmoid curve of the ventral outline is scarcely so pronounced. ‘The anal is covered with scales (M/ylo- soma) and on one of the specimens it is angulated posteriorly, as it usually is in M/. aureum, while in the other it is more evenly rounded as in Mr. Fowler’s drawing. ‘The scutes along the ventral ridge number 40, and there are none surrounding the vent. 61. Raphiodon vulpinus Spix. Several specimens, a foot in length, were collected at Para. The depth of the body is equal to the length of the head, or is contained from 5'%4 to 5% times in the length to the caudal. The eye is contained from 4% to 5 times in the head, and the maxillary 1% times. The cheek is almost en- tirely covered by the expanded suborbitals. The adipose eyelid has a ver- tical elongate opening, as in some of the herring-like fishes. The pectoral is a trifle shorter than the anal base, or one-fourth of the length to the caudal. The front of the dorsal is a little behind that of the anal. The caudal is not rounded, as in the picture published by Spix (Pisc. Bras. tab. 26), but is angulated at the tips of the outer rays and double lunate behind, with the middle one or two rays produced considerably beyond the rest of the fin and carrying the lateral line to their tips. The ventrals are in front of the anal a distance equal to the combined length of the snout and the eye. 62. Hydrolycus pectoralis (Giinther). A couple of specimens, 8 inches long, were taken by Dr. Baker and Mr. Mann in the Amazon River near the mouth of the Madeira River. The body is much deeper than the length of the head, and is con- tained 31% times in the length to the caudal; the head, without the pro- jecting jaw, 4% times. The pectoral is scarcely as long as described for the type. In one specimen it scarcely extends to the front of the dorsal; in the other it extends just to, though its length as contained in the entire EDWIN CHAPIN STARKS 21 length is the same (about 3 times). The anal has from 45 to 47 rays, and its base is contained from 334 to 4 times in the length. ‘The caudal is rounded and the middle rays are not produced with the lateral line. The ventrals are a little more than half as long as the head. ‘The anal and caudal are very closely covered with scales. The scales on the body are slightly ctenoid. A large black spot is present on the shoulder girdle above, a small one on the base of the lower pectoral ray, and one on the adipose dorsal. 63. Hydrocynus cuvieri (Agassiz). This species is not uncommon in the market at Para, where a number of specimens were collected. As our steamer entered the Rio Para, numbers of what appeared to be this form could be seen darting swiftly over the surface of the stream, with the greater part of their bodies out of the water. 64. Hoplias malabaricus Bloch. This speciés was taken at Cedro Dam, near the town of Quixada, at Ceara Mirim, at Lake Extremoz and at Lake Papary. 65. Hoplerythrinus uniteniatus (Spix). One specimen was found in the market at Para. I may here call atten- tion to an evident mistake in Dr. Eigenmann’s key to the family Characide (Freshwater Fishes of British Guiana, p. 258). For the above genus and for Erythrinus the walls of the anterior portion of the air-bladder are said to be cellular. In our representatives of this genera it is the second chamber of the air-bladder, the part behind the pneumatic duct, that is cellular. The walls of the anterior chamber and the posterior part of the posterior chamber are simple in structure. FamMity GYMNOTID. 66. Sternarchus macrolepis Steindachner. A few specimens were taken at Para. The teeth in the upper jaw are extremely easily lost, and in about half of the specimens at hand they are not to be detected. There is considerable variation in the length and slenderness of the tail and in the position of the vent. 67. Sternarchus albifrons (Linnzus). A single specimen was found in the market at Para. It has 15 scales between the lateral line and the median line of the back. The head is contained 6% times in the length, and 1% times in the greatest depth. 22. FISHES FROM BRAZIL The angle of the mouth is under the posterior edge of the eye. The snout is contained 3 times in the length of the head, and is equal in length to the longest anal rays. The anal has 152 rays, and its posterior end is dis- tant from the caudal a distance equal to the length of the caudal. The head is slate-color and darker than the rest of the body. On its upper surface is a white, rectangular, longitudinal spot, and from it a thin, light, broken line runs posteriorly along the back. An abrupt white band crosses the body above the posterior part of the anal fin, involving the latter, and a second one crosses the base of the caudal fin and caudal peduncle. The tip of the caudal fin is white. The fins are otherwise jet black. 68. Sternarchella sima Starks, new species. Plate 4. The ventral outline of the head and body is more strongly curved than the dorsal. The greatest depth of the body is 7 in the length. The head is small, with its dorsal contour abruptly rounded at the snout, and its ventral contour slightly and evenly curved to the tip of the mandible; its length is contained from 8 to 8% times in the length of the body. The mouth is very small; its gape is more transverse than lateral, nearly straight in front and abruptly curved down at the end of the maxillary on the side. The maxillary reaches to under the very inconspicuous posterior nostril. Small, sharp, movable teeth are present in several series on both jaws. ‘The eyes are small and the thick skin that covers them is continuous with the side of the head, so that their extent is difficult to appreciate. Their diameter is contained 8 times in the length of the head, 2 times in the snout, and 2 times in the interorbital space. At the isthmus is a deep longitudinal groove, at the middle of which the vent is placed a little nearer to a vertical line from the preopercle than one from the eye. The scales are rather large and thin. There are from 75 to 80 tubes in the lateral line. Over an area extending above the base of the anal rays, and about equal in width to the length of the rays (the area of the inter- haemal rays) the scales are abruptly smaller. There are 8 or 9 scales between the lateral line and the scaleless- area of the back at the widest part of the body. The anal begins below the base of the pectoral or very slightly in front of it, and ends a distance from the caudal base nearly equal to the length of the head. It contains 193 rays, the longest of which are a trifle under half of the length of the head. The pectoral is contained 1%% in the head, and has 14 or 15 rays. The length of the caudal is equal to the diameter of the eye. EDWIN CHAPIN STARKS 23 In life the culor was flesh-color, with the back slightly darker with slaty brown. In alcohol the only pigment is a dusky brown narrow band along the back. This species may be known from other members of its genus by its having a blunt, rounded snout overhanging a small mouth. Four specimens were taken at Para, all of them about 6 inches in length. 69. Sternarchogiton natterei (Steindachner). A single small specimen was collected by Dr. Baker and Mr. Mann at the mouth of the Madeira River. 70. Sternarchorhamphus mulleri (Steindachner). Many specimens of this species were taken at the market at Para. Among them there is a very great variation in the caudal and fin. In many of them the peduncle is nearly twice as long as the head and tapers almost to a point. Its narrowest part is scarcely wider than the very small eye, and the caudal fin is only a trifle wider. In others it is less than the length of the head and tapers more rapidly to a width similar to the first ones. In one specimen it is no longer than the snout and scarcely tapers at all, being three or four times as wide as the narrowest ones, or equal to one- fourth of the length of the snout, while the fin is correspondingly large. One specimen has the peduncle very wide and short, and the fin very diminu- tive as usual. These specimens do not otherwise differ from each other except in the position of the vent, and this variation has no correlation with the caudal variation. The vent may vary from a position a trifle behind the eye to one considerably in front of it. These specimens are all colorless in alcohol except a slight tinge of orange-yellow on the fins and a slight dusky tinge on the back. One speci- men, however, differs in being a rather dark slaty-brown, almost uniform, but slightly darker above; about the opercular region are blue shades; the fins are bright orange, growing yellow toward the tips of the rays, with the tips dusky, or on the pectoral, nearly black. 71. Sternarchorhynchus curvirostris (Boulenger). This species is common at Para, though it has not been before recorded from the lower Amazon. 72. Sternarchorhynchus oxyrhynchus (Miiller and Troschel). A small specimen was taken by Dr. Baker and Mr. Mann about 400 miles above the mouth of the Rio Madeira. 24 FISHES FROM BRAZIL 73. Rhamphichthys rostratus (Linnezus). One specimen was collected at Para. The postorbital part of the head is contained 1% times in the length of the snout. 74. Rhamphichthys reinhardtii Kaup. Many specimens of this common species were seen in the market, though on account of their large size only 5 were collected. I follow Doctor Eigenmann in recognizing this species, while believing with him that it is probably a long-snouted variation of &. rostratus. Though there is little difference between these species, except the rather extraordinary one of the length of the snout, it seems better to keep them separate until intermediate variations are reported. In the specimens before me the body is more slender and more gradually tapering backward than in the single specimen of Ahamphichthys rostratus, but as this character shows considerable variation it is less remarkable. The postorbital part of the head, measured obliquely to the middle of the gill opening, is contained in the length of the snout from 1% to 134 times. 75. Steatogenes elegans (Steindachner). Though this well-marked form was found to be rather common in the market at Para it has not been reported except at the mouth of the Rio Negro and in the Rio Jurua. It agrees very well with the description and picture of the type, though it differs from the latter in having the tail a little more slender and tapering to a finer point, and in the anal not stop- ping so abruptly posteriorly but gradually growing lower until it disappears. It has the same number of anal rays. 76. Higenmannia virescens (Valenciennes). This species was found to be common in the market at Para. Most of the specimens have the ventral outline more strongly arched than the dorsal, as alleged for Ezgenmannia humboldti, and specimens of LE. virescens men- tioned by Dr. Eigenmann from Rio Grande do Sul. Others have both out- lines equally arched. ‘The character is very evidently due to preservation and cannot be used in classification. In the thick head and other characters these specimens agree very well with current descriptions. 77. Sternopygus macrurus Bloch and Schneider. Several specimens of this form were taken at Para. The life color is dark olive-green on the back, shading downward to bluish-slate color on the lower parts. There is a dark bluish spot above > EDWIN CHAPIN STARKS 25 gill opening, a light stripe longitudinal along the posterior half of the side, which may be absent, and the fins are light straw-color. 78. Gymnotus carapo Linnzus. Specimens were taken at Para and at Lake Papary. The following color note was made from a fresh specimen from the former locality. ‘The back is a very dark dusky-greenish-brown, shading through soiled sienna on the side to dusky-bluish below. The cross bars are black and the anal and pectoral are dusky. Famity ELECTROPHORID. 79. EHlectrophorus electricus (Linnzus). Two specimens were taken at Para. ‘They were dusky-yellowish-green above, growing yellow on the sides and bright yellow on the lower part of the head, while the lower part of the body was drab, growing lighter anteriorly and merging with the yellow of the head. Blended light spots, as big as the eye, were scattered over the body, and the margin of the ver- tical fin was light yellow. Famity BUNOCEPHALID. 80. Platystacus cotylephorus Bloch. A single specimen was taken at Para. 81. Aspredo aspredo (Linnzus). Three specimens of this species were collected at Para. FAMILY SILURIDA. 82. Felichthys bagre (Linneus). Three specimens were collected at Para. There is considerable varia- tion among specimens of the same length in the length of the barbels. In two of them, 15 inches long, the maxillary barbel reaches to the end of the ventral in one, and to the end of the anal in the other. The mental bar- bel reaches nearly to the pectoral base in one, and scarcely past the eye in the other. The pectoral filament reaches past the tips of the middle caudal rays in one, and not to the caudal in the other. The pectoral filament reaches past the anterior fourth of the anal in one, and to the posterior fourth in the other. The interorbital bone cannot be described as flat in these specimens, for it is considerably convex. 26 FISHES FROM BRAZIL 83. Sciadeichthys proops (Cuvier and Valenciennes). Two specimens were collected in the Para market, and one at Ceara. 84, Selenaspis herzbergii (Bloch). This was by far the most abundant of the subfamily Tachisurine in the market at Para during our stay. The patch of teeth that develops be- hind the palatine patch and forms the “backward projecting angle” may be found in all stages of development. Sometimes it is not present at all; sometimes it forms a small disconnected patch; sometimes it is very large, connected with the palatine patch, and extends well back in the roof of the mouth. All intergradiations are present. It is not altogether a develop- ment of age or size as has been supposed, for it may be better developed in a specimen 8 inches long than in one a foot long, while other specimens of similar sizes, differing from these in no other manner, may or may not have it developed. However, it is usually best developed in large specimens, while specimens 7 or 8 inches long only occasionally have it. There is considerable variation in the roughness of the top of the head, and of the sharpness of the median ridge on the occipital process. One or two specimens were seen with a narrow but very evident groove running back from the frontal but not reaching to the occipital plate. 85. Tachysurus nuchalis (Giinther). A few specimens were taken at Para which differ from the description of the type in having the maxillary barbels extend scarcely to the middle of the pectoral spine instead of ‘nearly to the end.” The tooth-patches on the palatines are the shape and size figured by Dr. Giinther, and the palatine teeth are coarse. The inner mandibular teeth are coarse and all of the teeth are brown-tipped. The gill membrane forms a distinct free fold across the isthmus, thus differing from Tachysurus spixt. 86. Pimelodina flavipinnis Steindachner. Three specimens, from 12 to 15 inches in length, were taken at Para. The maxillary barbel in the large specimen reaches only to the end of the adipose dorsal, thus agreeing with Pimelodina nasus Eigenmann. In the other two it reaches to or a little beyond the caudal base, as described for Pimelodina flavipinnis. The length of the adipose fin is from 2% to 2% in the length, thus coming about between these two species. ‘The inter- orbital space is convex, and the head behind the eyes is not compressed as in Pimelodina nasus. The eye is contained about twice in the interorbital, or nearly midway between the condition described for these two species, as EDWIN CHAPIN STARKS 27 is also the length of the pectoral as compared with the dorsal. In other respects (except “Die kopfbreite zwischen den kiemendeckeln erreicht 1% der kopflange,”” which is a mistake, as shown in Dr. Steindachner’s picture of the type) these specimens agree with Pimelodina flavipinnis. The eye in the head, the depth above the ventrals, the net-work of sensory tubes over the head, the distance of the adipose dorsal from the dorsal, and the color, with several rows of brown spots on the upper half of the body, all agree with that species. 87. Pinirampus pinirampu (Spix). Four specimens of this well-marked species were taken at Para. 88. Rhamdia sebe (Cuvier and Valenciennes). A number of specimens were collected at Para. Comparing a specimen 6 inches in length collected in British Guiana by Dr. Eigenmann with these of equal size the eye is a little smaller, the body and caudal peduncle not so deep or so compressed, and the fontanel a trifle wider. 89. Rhamdia quelen (Quoy and Gaimard). A couple of specimens were taken in the ponds at Ceara Mirim. The maxillary barbels do not reach to the middle of the adipose dorsal, but otherwise these differ only in minor characters from Rhamdia sebe and these are such characters that might be expected to intergrade in a sufficiently large series. It is probable that they represent only one species. 90. Pimelodus altipinnis Steindachner. A common species in the market at Para, where several specimens were taken. 91. Pimelodus clarias (Bloch). A couple of specimens collected by Dr. Baker and Mr. .Mann in the Rio Madeira about 400 miles from its mouth. 92. Brachyplatystoma filamentosum Lichtenstein. A number of specimens of this species were collected at Para. 93. Brachyplatystoma vaillanti (Cuvier and Valenciennes). This species is a common one in the market at Para, where a number of specimens were collected. 94. Brachyplatystoma rousseauxii (Castelnau). A half dozen specimens, from 11 to 14 inches in length without the caudal filament, were collected at Para. The following notes show some 28 FISHES FROM BRAZIL slight differences from the description published by Dr. Eigenmann (South American Nematognathi, p. 198), and add a few variations. The occipital crest reaches at least 34 of the distance from its base to to the dorsal spine. The eye is contained from 6% to 7% times in the snout; 34% to 4 times in the interorbital space; and 14 to 16 times in the length of the head. The maxillary barbel often reaches to the tip of the pectoral, and the postmental is usually coterminous with it, or sometimes is a trifle shorter. The upper lobe of the caudal bears a long filament, variable in length, but often as long as the entire body. The lower lobe is in one or two case filamentous and twice the length of the head, though usually it is but little produced or half that length. The vent is only 6 diameters of the eye from the anal. The entire side of the body is not reticulated and porous, but it bears a lateral band which is about half as wide as the interorbital space under the dorsal, grows a little wider anteriorly and tapers to a point posteriorly. ‘The head is contained from 3% to 3% times in the length. In life it is a strikingly beautiful fish, with metallic golden red, cop- pery and silvery reflections. 95. Tenionema platynema (Boulenger). A single specimen, 22 inches in length, was collected at Para. It is referred to this species, though in many of its characters it resembles as well T@nionema steeret EKigenmann and Bean, making it appear probable that the two are identical, as suggested in the original description of the latter species. The eye is contained 4 times in the interorbital space, 14 times in the snout, 20 times in the head, the adipose dorsal is 3 times as long as deep, and the ventral and pectoral are equal in length. All of these char- acters are as in 7. platynema. ‘The length of the head is 3 times as long as deep, and the length of the first dorsal ray is contained 114 times in the head as in 7. steerei. The depth of the body is 6% times in the length, and . the depth of the caudal peduncle is 2 times in its length. The fins and the sides are bright yellow in alcohol; its colors in life were not recorded. Platypogon, new genus. This genus is related to Brachyplatystoma, but has the barbels with their inner edges bearing a thin dermal membrane, and has not the char- acteristic small eye and flat, depressed snout. The teeth are freely movable and in broad bands. On the premaxillary, and to a less degree on the mandible, they grow longer posteriorly and point backward. ‘The palatine and vomerine teeth form a continuous band. The fontanel does not extend EDWIN CHAPIN STARKS 29 back as a groove, and the top of the head is covered with a moderately thick skin. The supraoccipital process barely reaches to the dorsal plate. The adipose dorsal is situated over the anal and is about equal to it in length. The anal is emarginate and the caudal is forked. 96. Platypogon cerulorostris Starks, new species. Plate 5. The head is conical, as deep as wide opposite the posterior edge of the operculum, and equal to two-thirds of its length. It is contained 4 times in the length of the body to the caudal base. The depth of the body slightly exceeds its width, and is contained from 5% to 5% in the length. The depth of the narrowest part of the caudal peduncle is con- tained from 34% to 3% times in the head. The skin forms a moderately thick covering to the bones of the top of the head, forming a smooth surface with fine striations barely indicated. ‘The fontanel does not extend behind the eye, and a very short groove, not longer than the pupil, extends back and terminates abruptly. This is only made evident by removing the skin. The occipital process tapers back to a rather narrow point at the dorsal plate, being scarcely, or barely, in contact with it and not at all con- nected. The eye is large and round. Its diameter is contained 2% times in the interorbital space, 6 times in the head, and 24 times in the snout. The premaxillary band of teeth is widest at the middle, where it is equal to two-thirds of the diameter of the eye. The teeth grow long posteriorly and lie flat, pointing straight backward. The vomerine and palatine patches of teeth form a continuous band, a little narrower than the premaxillary band, and slightly narrower in the middle than at either side. All of the teeth are freely movable. The snout is convex above, and moderately broadly rounded in front. Its length is contained 3 times in the head. The rictus reaches to under the anterior margin of the eye. The barbels have a membraneous posterior margin, especially tle mentals and _post- mentals, where the membrane is as wide, or a little wider, than the fleshy part of the barbel. The maxillary barbels reach to the base of the ventrals, the postmental barbels a little past the middle of the pectorals, and the mental barbels to the base of the pectorals. The dorsal spine is filamentous, a third longer than the head, and reaches to the adipose dorsal. ‘The dorsal has 6 rays, and its base is con- tained 2%4 times in the space between it and the adipose dorsal. The base of the adipose dorsal is almost as long as that of the anal, and is situated directly opposite to it. Its vertical height is half of the length of its base. 30 FISHES FROM BRAZIL The anal has 14 rays, counting two in front of the longest ray. It is deeply emarginate, and its longest anterior rays reach past the tips of the last ones. The ventrals are situated a little less than the length of the head in front of the anal, and reach three-fourths of the distance to the anal. ‘The caudal is widely forked, and its lower lobe is a trifle the longer. Its longest rays slightly exceed the length of the head. The ground color is slaty-brown on the back, grading downward to soiled white on the lower parts. On the upper part of the side two more _ or less definite rows of dark brown spots with blended edges, their diameter about half that of the eye. The top of the head is brown, with a conspicuous white spot over the fontanel. ‘The tip of the snout has a deep blue spot almost the exact color of modern blue-black writing ink. ‘The fins and humeral processes are deep orange-yellow, more intense on the caudal. Counting from the middle of the caudal fin downward the sixth to the eighth rays are abruptly black. Three specimens were collected at Para, from 9 to 12 inches in entire length. 97. Sorubim lima (Bloch and Schneider). A single specimen, 10 inches long, was collected by Dr. Baker and Mr. Mann in the Madeira River about 400 miles above its mouth. The plates on the head are finely striate, granulate, and separated by well-marked sutures. The anterior half of the occipital plate is bounded laterally by parietal and temporal plates. Its granulated surface does not reach to the dorsal plate, but below the thin membrane that covers it it extends to, and joins, the dorsal plate. The dorsal plate is spearhead- shaped, with its point forward. The fontanel is open from the middle of the eye to within a diameter of the eye of the occipital plate, and anteriorly is represented as a groove to in front of the eye. A membraneous groove in the middle of the occipital plate appears like a second fontanel, but it does not pierce the bone. 98. Platystomatichthys sturio Kner. Several specimens were collected at Para, the longest 14 inches in length. They show some variation from Dr. Eigenmann’s description. The projection of the snout beyond the lower jaw is contained from 2% to 2% in the length of the head. The eye is contained from 10% to 12% times in the same distance. In none of them are the palatine teeth as far remote from the vomerine, as described and figured by Dr. Eigenmann. Usually the palatine teeth are almost in contact with the vomerine teeth, there being EDWIN CHAPIN STARKS Sill only a fine line of naked skin between. Occasionally the space is a little wider, but in none is it as wide as in the picture. 99. Doras dorsalis Cuvier and Valenciennes. Many specimens of this common species were collected at Para. In most of them the stomach was tightly filled with a small univalve shell. The dorsal median plates and the ventral plates on the caudal peduncle are apparently never present in specimens as small as 5 inches in length, but in those 6 inches long some of them are present at least as rudiments. Though they are usually the best developed in the largest specimens they may be almost entirely absent. There is also much variation in the rough- ness of the head. In one specimen the bones of the head are so finely granular as to appear almost perfectly smooth. These specimens show some slight variations from Dr. Eigenmann’s description (S. Am. Nematog. Cal. Acad. Sci.,1890). The caudal peduncle measured at the bases of the spines is as wide as deep. Usually a series of slight broken ridges extends from the frontal to the dorsal plate, and a well-developed groove on the dorsal plate extends to its tip, but only oc- casionally ‘“‘a marked groove extends form it [the frontal] to the tip of the dorsal plate.”” The pectoral spine is as long as the dorsal spine or often a trifle longer, and in the young of 6 inches in length it is equal in length to the head, or slightly shorter, but in specimens 10 or 11 inches in length it is considerably longer than the head. 100. Centromochlus heckelii (Filippi). Several specimens of this peculiar little species were taken at Para, the largest 4 inches in length. 101. Trachycorystes galeatus (Linnzus). Several specimens were collected in the market at Para. The follow- ing color notes were made from fresh specimens. The ground color is opaque, soiled light yellow, usually with black blotches scattered over the head, body and fins. These may be absent or often take the shape of hori- zontal elongate spots, or the form of irregular lines suggesting penciled lines. The under parts are milk-white, often more or less soiled with groups of small dark dots. 102. Trachycorystes striatulus Steindachner. A few specimens were collected in the little ponds about Ceara Mirim. These much resemble Trachycorystes galeatus, but the head’ is not so rough 32 FISHES FROM BRAZIL or naked; the humeral process is shorter and does not project so much later- ally at the base over the base of the pectoral spine; no pectoral pore is present; the top of the head between the eyes in the adult is a little flatter ; and in the specimens at hand the color is darker and the under parts dusky with fine points. These differ from the original description in having 2 or 3 fewer anal rays. 103. Pseudauchenipterus nodosus (Bloch). A number of specimens from 6 to 9 inches in length were taken at Para, which I refer, with considerable hesitation, to this species, though they agree very well with the picture* and short description of Pseudau- chenipterus guppy: Regan (Proc. Zool. Soc. 1906, Pl. XXIII). The head cannot be described as having the “upper surface, excepting the snout, rugose, not covered by skin; frontal bones not swollen.” On the contrary, the top of the head is as described by Dr. Eigenmann (S. Am. Nematog- nathi, p. 290) for P. nodosus, with the frontals swollen and of distinctly “honey-comb structure.” It is covered with thin skin. ‘These specimens have the color markings as described and figured for P. guppyi, with the same light lateral band and vertical row of lhght spots, and with or without the black border to the caudal. These markings are occasionally obscure, though they are usually very conspicuous. They differ from Dr. Eigen- mann’s description of P. nodosus as follows: ‘The pectoral spine is usually sharp-pointed and does not end in a filament. Its length is from 4 to 434 (not from 224 to 34%), thus agreeing with the pictures published by Bloch and by Bleeker. The humeral process usually extends to the middle of the pectoral spine or a trifle beyond. 104. Auchenipterus nuchalis (Spix). Very common at Para, where a large number were collected. There is a remarkable variation in the maxillary barbel. In most of them the barbel extends straight back under the eye. It is very slender, its anterior end ossified, and it fits into a shallow groove immediately below the eye. In the opposite extreme the base of the barbel is a wide curved bone, much wider than the pectoral spine, and its curve forms at least a fifth of a complete circle (though it is not an even curve, being more abrupt below the eye). It extends upward from the eye, and ends above the level of the eye a little behind the head. It does not taper much posteriorly, but ends * The title of the picture and the reference in the text are transposed with those of P. pasee. EDWIN CHAPIN STARKS 33 abruptly at the beginning of the fine soft part of the barbel. The groove that it occupies encroaches somewhat upon the eyeball below. ‘There is only about one of these specimens to a dozen of the others. Between these extremes are nearly all of the intermediate stages, and coérdinating with it are the size and shape of the first anal rays. These are enlarged and form a process in the specimens with the large, bony, curved maxillary barbel, but not at all differentiated from the other rays in those with the straight, fine, scarcely ossified barbel, while they are slightly enlarged in those with the barbel slightly enlarged, curved and ossified. The enlarged anal process suggests a copulatory organ. It is as long as the pectoral spine, or two-fifths longer than the rays behind it, and is formed of two rays, with a third slightly enlarged and closely attached to it. These appear to be the first rays, but upon dissecting away the integu- ment two slender shorter rays are found in front of them. The enlarged rays, though they become stiff and spine-like, do not lose their articulations. The tip of the process is strongly hooked up and back. In a specimen with the ossified portion of the barbel no wider than the pectoral spine, and the ossified portion shorter and passing into the fleshy portion imperceptibly, the anal process is enlarged but not hard and spine-like at the tip, and not hooked up. In those where the condition of the barbel is about midway between the extremes, that of the anal process is also about midway between, perceptibly enlarged, but not longer than the other rays. Whether these characters go with sex I am unable to ascertain with the specimens at hand. Some specimens have the abdominal cavity filled with fat, and the bases of the fins thickened with fat. This gives to the fins of preserved specimens an orange-yellow color. In other specimens this condition is not apparent. It cannot be correlated with the condition of the barbel or anal process. 105. Ageneiosus ucayalensis Castelnau. Many specimens were taken in the market at Para, where it seems to be as abundant as Ageneiosus dentatus, which it much resembles. From the latter it may be known by the slightly more slender body, the longer snout as compared with the width of the head at the eyes (as compared with the length of the head or the body the difference does not appear), the smaller eyes, the wider band of teeth on the premaxillary, the longer and more strongly curved mouth, the more strongly projecting upper jaw, and the slenderer spines. The color is usually more bluish, less spotted above, be- coming more abruptly white on the sides, and seldom with a black or dusky 34 FISHES FROM BRAZIL spot on the base of the caudal rays above and below. ‘The difference in the width of the band of teeth as given by Dr. Eigenmann, ‘depth of inter- maxillary band of teeth in front equals a diameter of the eye” in this species, and “depth of intermaxillary band of teeth scarcely more than half as wide as eye’ in A. dentatus is not such a difference as might appear, for it is usually considerably more than half as wide as the eye in the latter species, and the difference between them is as much a variation in the size of the eye as of the teeth. When the bands of teeth are directly compared with each other it is only a little wider in A. ucayalensis. 106. Ageneiosus dentatus Kner. Common in the market at Para, where several specimens were taken. On the back are indications of cross bands or mottled spots, especially in the young, and the color is made up of rather coarse dots. FAMILY HYPOPHATHALMIDA. 107. Hypophthalmus edentatus Spix. This species is very common in the market at Para, where a series of specimens from 4 to 16 inches in length was collected. ‘These show very well the branching tubes of the lateral line system developing with size, and also shortening of the barbels, as described by Dr. Eigenmann. 108. Cetopsis cecutiens Lichtenstein. A single specimen from Para, and two from the Rio Madeira, the latter collected about 400 miles above the mouth of the river by Dr. Baker and Mr. Mann. In the largest one, 10%4 inches in length, the skin has so grown over the eye as to almost completely hide it. FamMity LORICARID. 109. Farlowella hargreavesi Higenmann, This species was described as new in my manuscript, but it has recently appeared under the above name in Doctor Eigenmann’s Freshwater Fishes of British Guiana (Mem. Carnegie Mus. Vol. V, p. 252). I leave my description and comparisons stand as I have written them, however, as they show some slight variation from, and add a little to, the description of the type. A single specimen, 8 inches in length, was collected in the market at Para. EDWIN CHAPIN STARKS 35 ‘The head, measured to the end of the temporal plate, is contained 334 times in the length to the caudal. It is wider than in related species, and rapidly tapers, as viewed from above, to a point half way between the eyes and the tip of the snout, thence the taper is scarcely perceptible to the slightly expanded tip of the snout. The length of the snout is contained 456 times in the entire length to the caudal. Opposite the eyes the depth of the head is contained 14% times in its width. The surface of the head and snout are everywhere finely granular, with the granulations not bristle-like or otherwise differentiated at the side. The teeth of the upper jaw are long and slender, hooked inwards, their points flattened and unevenly divided into two or three unequal cusps. There are at least 24 of them on each side of the jaw. The diameter of the eye is contained 14 times in the length of the head; 3 times in the interorbital; once in the space to the anterior nostril; and 114 times in the space to the upper end of the gill opening. : The breast has a median series of plates; the one between the ventrals and just in front of the vent is enlarged. At the vent a lateral plate extends entirely across the body and is pierced by the vent. Directly behind it is the last median ventral plate. Seven plates are in the median series between the occipital plate and the dorsal. The same number of lower lateral plates are between the pectoral and ventral. The pectoral is contained 2% times in the head, and reaches a diameter of the eye past the front of the ventral. The ventral is contained 3% times in the head, and does not reach to the anal. The anal has 6 rays, is inserted behind the front of the dorsal and reaches a trifle past it. Its first ray is contained 134 times in the head. The caudal lobes are produced. A conspicuous dark band, about twice the diameter of the eye at its widest part, runs from the tip of the snout to the caudal. The entire snout is dark and the color parts into the lateral band of each side where the snout grows broad. The band thence curves upward, surrounding the eye, and arches backward, following the contour of the back. It follows along the side of the caudal peduncle with a round white spot at the posterior edge of each plate, and divides on the caudal fin in a black stripe along each lobe near the edge. The other fins are colorless, and except the ventral, have round dusky spots on the first two or three rays. This species is related to Farlowella gladius and F. gracilis, but differ as shown in the appended table. Of the only other two species with a median series of ventral plates F. kneri has a much blunter, shorter, and wider snout, and a much larger eye, while F. oxyrhynchus has the post- dorsal part of the body much shorter as compared with the predorsal part. 36 FISHES FROM BRAZIL TABLE SHOWING DIFFERENCES BETWEEN FARLOWELLA HARGREAVESI, F. GLADIUS AND F. GRACILIS. For Ff. gladius the measurements are taken both from the original description and from a description published by Regan (Trans. Zool. Soc. Lond. XVII, 1904, p. 303). hargreavest gladius | gracilis | Length of head in length of body.......... 38 33@ 32 Width of head in length of head........... 3 a + | Dai uot avers 6 Ae spn ats are eh ey han byte! galps Ata Les 14 18 @ 123 ID VemmmterOr Oita emir ele ereree mite ctr te 3 oe 3@ Interorbital in head@.n. cue eee eseles oy 5 58 Distance from supraoccipital to first dorsal ray in length of body 22.4...) 64 6 6} Postorbital part of head in snout from nakedonraltancae ree eee ee ener Zs Die 33@ Number of lateral iscutesas: 0. eee 35 33 33 Length of 2nd and 3rd dorsal scutes in their Width nila ie cis ottsweneam meee ey ait 2 3 © 22@ Pectoral ending in relation to front of ven- Graal atte sok ots We orci tee seer past not to just to @ ene thyotmMorsal Tayi se louse sieresteriere ater 13 2 21@ Wenethiot frst analgray esc + s)pe ere eerie ts Ds 21@ Wiblackilateralibande en esis ene etocnteed: present absent absent @According to Boulenger. According to Regan and the picture it is 33. @According to Regan. According to the picture it is 163. @According to Regan. The picture shows it to be 23. @According to the picture. These characters not described by Regan, who states ‘‘other characters similar to the preceding species [F. gladius].”’ @According to picture. Boulenger describes them ‘‘as more than twice as long asibLroade 110. Loricaria cataphracta Linneus. Three specimens from Para. 111. Plecostomus emarginatus (Cuvier and Valenciennes). A few specimens were collected by Dr. Baker and Mr. Mann in the Madeira River about 400 miles from its mouth. 112. Plecostomus pusarum Starks, new species. Plate 6. In the following description only specimens between 6 and 8 inches in length are considered, though smaller ones were taken. The head to the temporal plate is contained 3 times in the length. The temporal and occipital ridges are only very slightly indicated, not nearly so well developed as in Plecostomus plecostomus. The width of the EDWIN CHAPIN STARKS Sil head is from 1 to 1% diameters of the eye less than the length of the head. The depth of the head at the point of the supraoccipital is equal to the length of the snout. ‘The interorbital space is contained 2% times in the head, and the length of the snout from 134 to 14% times. - The distance from the anterior nostril to the tip of the snout is equal to the interorbital space. The middle of the interorbital space is only a trifle higher than the supraorbital rim, while the intermediate space between these two points is somewhat sunken on each side. ‘The eye, inside of the eyelids, is contained from 9 to 10 times in the head, and 3% to 4 times in the interorbital space. The band of teeth on each side of the mandible is contained 31% times in the interorbital space, and the width of the lower lip 2% times. The barbel does not reach to opposite the posterior edge of the lip, and is equal to the space between the eye and the posterior nostril. The distance from the tip of the snout to the first dorsal spine is con- tained from 2% to 2%4 times in the length of the body. The vent is mid- way between the base of the caudal and the pectoral spine. The depth of the body in front of the dorsal spine is contained from 4 to 4% times in the length, and the greatest width from 3 to 3% times. The caudal peduncle in front of the adipose dorsal is as wide as it is deep. The first dorsal spine is a trifle less than the length of the head, and when the fin is reclined its tip does not reach to the middle of the last ray. The last ray is equal to the length of the snout, and for a short distance near its base it is adnate to the back. The tips of the last rays broadly overlie the adipose dorsal, reaching at least to its middle. On the posterior margin the dorsal fin is convex. The length of the pectoral spine is contained from 3 to 3% times in the length of the body, and almost a third of its length extends past the base of the ventral. The ventrals reach well past the posterior end of the anal base, or half way or more from their base to the caudal fin. The anal has a spine and 4 soft rays, and the tips of its rays do not reach so far back as the dorsal rays, but reach half way, or nearly, from their base to the caudal. The lower caudal lobe is longer than the upper, and the lowest branched ray reaches nearly to the tip of the lower simple ray, which is scarcely filamentous at the tip. The lower simple ray is contained from 21% to 2% times in the body, the upper ray from 27% to 234 times. The supraoccipital is bounded behind by a single plate. Two upper rows of plates are slightly carinate, and four plates extending back from the pectoral are bluntly carinate; the carinations are scarcely evident elsewhere. The longitudinal spinules grow coarser on the caudal peduncle, and are coarser and less numerous than in P. plecostomus. ‘There are 25 or 26 38 FISHES FROM BRAZIL lateral plates; 3 between the occipital and dorsal spine, the last very narrow medially ; 7 between the dorsals; 4 between the adipose dorsal spine and the caudal; 12 between the anal and the caudal. ‘The cross lines on the belly divide the granulations into much larger plates than in P. plecostomus, and there is a larger naked area at the bases of the ventral fins. Dark spots cover the body, on the ventral as well as the dorsal surface, and the fins. Each dorsal membrane bears two rows of them. This species is most closely related to P. flecostomus, but differs in having a smaller eye, the occipital and temporal ridges scarcely developed, the dorsal fin overlapping the adipose dorsal, the plates on the belly coarser, the spinules on the lateral scales coarser, and the whole posterior part of the body shorter, so that the anal and ventrals and vent are more posteriorly placed in relation to the caudal base. Numerous specimens from 2 to 8 inches in length were collected in the little disconnected ponds and in the muddy stream at Ceara Mirim, the boys catching them under the overhanging grassy banks in their hands and in dip-nets. The type is 8 inches in entire length. (Pusarum, named in memory of the boys to whom I owe the collection of fishes at Ceara Mirim— the best collectors we found in Brazil.) 113. Plecostomus plecostomus (Linnzus). Three specimens were taken at Para, the largest 8 inches in length. This species resembles Plesoctomus verres very closely. It differs not only in having the supraoccipital bounded behind by a single plate, but the dorsal is less angulated, less oblique and straight on its posterior edge, and its outline more rounded. ‘The scales on the breast and belly are more segre- gated into little square areas formed by diagonal cross lines. 114. Plecostomus verres (Cuvier and Valenciennes). Several specimens were collected at Para, the largest a foot in length. They agree very well with current descriptions except that “scutes carinate except on caudal peduncle” (Regan Trans. Zool. Soc. Lond. XVII, p. 209) does not adequately describe these, for the scales on the caudal peduncle have low keels, not so sharp and high as they are anteriorly, but still very evident. 115. Acanthicus hystrix Spix. A specimen of this rare species, 14 inches long without the caudal filaments, was collected in the market at Para. It has 8 soft rays in the dorsal. The barbels, unlike the specimen described by Dr. Eigenmann, scarcely reach as far posteriorly as the edge of the lower lip. EDWIN CHAPIN STARKS 39 Famity CALLICHTHYIDA:, 116. Callichthys callichthys Linnzus. Three specimens were taken at Para. 117. Hoplosternum thoracatum (Cuvier and Valenciennes). Several specimens were taken at Lake Papary and two at Para. FamMity SYNODONTID~. 118. Synodus feetens (Linnzus). Two small specimens taken at Natal. FamMILty PQ:CILLID. 119. Anableps anableps (Linnzus). A single specimen from Para. 120. Peecilia vivipara Bloch and Schneider, A large number of specimens from 1 to 234 inches in length were col- lected in the little ponds at Ceara Mirim, and in a pond in a deserted part of a public park at Ceara. They are identical with specimens collected in British Guiana by Dr. Eigenmann. The head is contained from 3% to 4 times in the length to the caudal base; the depth from 3 to 3% times. The eye is contained from 3% to 4 times in the head, or 2 times in the interorbital space. The width of the mouth is 1% times in the interorbital space. The front of the dorsal is midway between the base of the caudal and the opercle, or varying to slightly in front of the last point. The length of the pectoral is equal to the com- bined length of the eye and the postorbital part of the head, or to the depth of the caudal peduncle. There are 6 or 7 rays in the dorsal, and the same number in the anal, though the base of the latter is much shorter. The cross series of scales number from 26 to 28. The males do not exceed a length of 2 inches. They have from 4 to 6 cross bars, usually a little narrower than the interspaces but not always, behind the front of the dorsal. Just above the point of the pectoral on the upper part of the side is a brown spot as large as the eye. It is conspicuous in specimens up to 1% inches in length, but in large ones it becomes length- ened into an anterior cross bar. The females have no conspicuous cross bars except a slight trace of them in the younger specimens. They may retain the dark side spot until they are 2%4 inches in length, but usually it dis- 40 FISHES FROM BRAZIL appears by the time they are 2 inches long. On both sexes the scales are bordered with brown, and the upper and lower rays of the caudal some- times bear a spot irregular in size, and varying in color from black to its complete absence. Famity BELONIDA. 121. Tylosurus almeida (Quoy and Gaimard). Specimens were secured at Lake Papary, Natal and Para. FamMity HEMIRAMPHID#. Mw 122. Hyporhymphus unifasciatus (Ranzani). Numerous specimens from 6 to 10 inches in length were taken at Natal. None of them has the mandible, from the tip of the upper jaw, shorter than the rest of the head, as stated by Jordan and Evermann (Fishes of N. and Mid. Am. p. 720), nor are the ventrals always inserted midway between the eye and the base of the caudal, but their insertion varies from that point to midway between the base of the caudal and the gill opening as in Hyfo- rhamphus rose. These specimens were compared with some from Jamaica and found to agree in all of their characters. Famity EXOCC@:TIDA. 123. Cypselurus rubescens (Rafinesque). A large specimen of this fish flew on board of our steamer while en route between Ceara and Natal. ‘The light area on the pectoral is not nearly as large or distinct as usual. FamMity SYNGNATHID. 124. Hippocampus punctulatus Guichenot, A few specimens of this form were collected in the harbor at Natal. Two distinct color phases are represented, but no other differences ap- pear. The following colors were taken from live female examples. The only male collected is of the color of the second one here described. (1) The lower parts of the head, belly and tail are bright canary yel- low, and the upper parts are yellowish green. Two dark brown cross bars occur between the head and the dorsal fin. ‘The second one is more con- spicuous, covering two rings, and covered with irregular, longitudinal, white lines. A third cross bar is on the tail four rings behind the dorsal. Be- EDWIN CHAPIN STARKS 41 hind this are some scarcely distinguishable bars on top of the tail. Very fine white points are scattered over the body and head. (2) The belly and under parts of the head are light gray, almost white, while the upper parts grow gradually darker to almost black. There is no yellow anywhere. The head and body are nearly everywhere covered with dark spots as large as the pupil and smaller, with numerous fine white points in between, some of which are arranged as irregular light lines. The body color is very much the same as the cross bars on the yellow specimens. A few light lines radiate from the eye. No cross bars are present. An- other specimen of this type of coloration has the white dots run together into large, numerous solid lines, giving a much lighter general effect. Famity ATHERINID-. 125. Menidia brasiliensis (Quoy and Gaimard). Many specimens were seined in the harbor at Natal and collected by the fishermen in Lake Papary. These specimens have 3 or 4 dorsal spines, not 4 or 5 as described by Doctor Giinther (Cat. III, p. 404). Famity MUGILIDA. 126. Mugil brasiliensis Agassiz. This species was very abundant at Lake Papary and Lake Extremoz. In the former lake specimens were taken nearly three feet in length. 127. Mugil curema Cuvier and Valenciennes. Specimens were taken at Para, Natal and Lake Extremoz. 128. Mugil trichodon Poey. This was the commonest mullet in the harbor and rock pools at’ Natal, where several specimens were taken. FamMILty SPHYRAENIDA. 129. Sphyrena barracuda (Walbaum). A few small specimens taken at Natal. FamMILyY HOLOCENTRID£. 130. Holocentrus ascensionis (Osbeck). A few specimens of this brilliant fish were taken in the tide-pools at Natal. In life it is very resplendent with metallic golden-red and coppery 42 FISHES FROM BRAZIL color having brassy reflections. The under parts are clear white, and the side is crossed with longitudinal light pearly stripes. A white stripe runs back from the maxillary to below the base of the preopercular spine, in- volving the upper half of the maxillary. The vertical limb of the pre- opercle is edged with white. The dorsal spines are light yellow, and the membrane between them is pink. The other fins are pinkish, and the iris is dark red. FAMILY SCOMBRID. 131. Scomberomorus cavallo (Cuvier and Valenciennes). This species is common in the market at Para, where several specimens about a foot in length were obtained. These specimens do not have fewer dorsal spines than S. maculatus, as alleged by Week and Newland (Proc. Acad. Nat. Sci. Phila. p. 233, 1884), nor is the spinous dorsal without a black blotch anteriorly, as stated by Jor- dan and Evermann (Fishes of N. and Mid. Am., p. 875). Compared with S. maculatus the teeth are much more compressed and wider at the base in the mandible and not so numerous, there being only 8 or 10 on each side. The species may be most readily recognized by the very short, scarcely developed, gill-rakers which are less than half the pupil in length. 132. Scomberomorus maculatus (Mitchill). A single large specimen was taken in Natal. FamMity TRICHIURIDA. 133. Trichiurus lepturus Linneus. Only a single specimen, a couple of feet in length, was taken at Para. but it is an extremely common species, especially at Ceara. Great quantities of them are dried and they may be seen in the stores at Ceara and Natal in big bales corded together like so much firewood. The bones are subject to hyperostosis, or a deposition of bone cells on the surface of the normal bones until they are several times their ordinary size. It appears in the form of tubercles, or abrupt enlargements of irregular size scattered over the interspinous bones at irregular intervals, and not con- fined to certain bones as appears to be the case among the species of the family Carangide. The specimen at hand resembles Evoxymetopon teniatus Poey in color but in no other respect. There are conspicuous orange-colored longitudinal EDWIN CHAPIN STARKS 43 lines on the body; one at the base of the dorsal; two narrow pale ones above the middle of the side; one wide, dark and well-marked one parallel to the lateral line; the fifth one following the lateral line; and two narrow ones on the lower part of the side. Famity CARANGID-. 134. Oligoplites saurus (Bloch and Schneider). A few small specimens were seined at Natal. 135. Oligoplites saliens (Bloch). Several specimens of this species were taken at Para. The following key to the species of Oligoplites is founded upon speci- mens in Stanford University. All of the species are represented by an abundant number of specimens except Oligoplites altus and Oligoplites re- fulgens, of which only a single well-preserved specimen of each is available. Key to the Species of Oligoplites. a Top of head everywhere densely covered with pores opening into short canals ramifying beneath the skin. b Depth from 234 to 3 in length; body angulated at front of soft dorsal and anal; outer mandibular teeth conical and no more crowded or movable than other teeth; the band of mandibular teeth not so wide; maxillary reaching considerably past the eye. mundus. bb Depth 3% in length; body scarcely angulated; outer mandibular teeth crowded, slender and movable, projecting above other teeth and laterally flattened; the band wider; maxillary reaching but slightly past eye. altus. aa Top of head without pores or only a few at the side running forward from above opercle, but never any medially on top of head between eyes or on snout. ce Maxillary reaching to middle of eye; depth equal to length of head ; side abruptly silvery below a double dark stripe on back. refulgens. cc Maxillary reaching to posterior border of eye or beyond; depth greater than length of head, the dark color of back more gradually merging into the silvery color of side. +4 FISHES FROM BRAZIL d