Skip to main content

Full text of "Bulletin"

See other formats


V.A.,x_-tC^ \ 



SMITHSONIAN INSTITUTION 

BUREAU OF AMERICAN ETHNOLOGY 

BULLETIN 167 



ARCHEOLOGICAL INVESTIGATIONS 
AT THE MOUTH OF THE AMAZON 



By 
BETTY J. MEGGERS and CLIFFORD EVANS 




*jf^<»^^l, lUOiy^s 



\ 



i 



UNITED STATES 

GOVERNMENT PRINTING OFFICE 

WASHINGTON : 1957 



For sale by the Superintendent of Documents, U. S. Government Printing 0£Bce 
Washington 25. D. C. 



LETTER OF TRANSMITTAL 



Smithsonian Institution, 
Bureau of American Ethnology, 

Washington, D. C, June BO, 1956. 
Sm : I have the honor to transmit herewith a manuscript entitled 
"Archeological Investigations at the Mouth of the Amazon," by 
Betty J. Meggers and Clifford Evans, and to recommend that it be 
published as a bulletin of the Bureau of American Ethnology. 
Very respectfully yours, 

M. W. Stirling, Director. 
Dr. Leonard Carmichael, 

Secretary, Smithsonian Institution. 



^SOrt^ 



"» '?i"0. 






Ci^o^ 



May this re fort be one of many memorials to 

WENDELL C. BENNETT 

by those like us whom he with friendly 
guidance helped along the road to anthro- 
pological careers 



in 



CONTENTS 



PAGE 

Preface xx v 

Introduction 1 

Background of the Lower Amazon Archeological Expedition 1 

Itinerary 3 

Problems and compromises in field technique 6 

Organization of the report 12 

Tropical Forest Culture 17 

Ethnographic definition of Tropical Forest culture 17 

Culture traits discernible archeologically 24 

Environmental limitations on culture in the Tropical Forest 26 

The Territory op AmapX 33 

Geographical description 33 

The Arua Phase 37 

Description of sites and excavations 37 

Site A-5— Cafezal 37 

Site A-8— Aurora 38 

Site A-23 — Ilha da Fortaleza, Conceicao 40 

Data from other investigations 41 

Igarap6 dos Macacos 41 

Rio Novo 41 

Jos6 Antonio 42 

Villa Calgoene 42 

Teso da Mina 42 

Sucurijii 42 

Lago dos Patos 42 

Cachoeirinha 42 

Agahyzal 43 

Analysis of material 43 

Ceramic history 44 

Diagnostic features of the Arua Phase 44 

The Mazagao Phase 44 

Description of sites and excavations 44 

Site A-1 — Pigacd Occupation 45 

Site A-2 — Lauro 45 

Site A-3 — Pigacd Cemetery 48 

Burial Group 1 51 

Burial 2 56 

Burial3 56 

Burial 4 57 

Burial 5 57 

Site A-4— Valentim 58 

Burial 1 60 

Burial Group 2 60 

Burial 3 62 

V 



VI BUREAU OF AMERICAN ETHNOLOGY [bull, lez 

The Territory of AmapA — Continued 
The Mazagao Phase — Continued 

Description of sites and excavations — Continued page 

Site A-5 — Cafezal 65 

Site A-6 — Ilha das Igagabas, Igarap6 do Lago 66 

Data from other investigations 67 

Rio Iratapurii Sites 68 

Rio Vilanova Sites 69 

Ilha do Pard Site 70 

Rio Mazagao Sites 73 

Igarape do Urubii Sites 75 

Rio Maracd Sites 75 

Analysis of materials of the Mazagao Phase 78 

Pottery type descriptions 78 

Anauerapucu Incised 79 

Camaipi Plain 81 

Jarf Scraped 83 

Mazagao Plain 85 

Pigacd Incised 87 

Uxy Incised 89 

Vilanova Plain 92 

Unclassified Decorated 94 

Pottery artifacts 94 

Nonceramic artifacts 95 

Ceramic history 95 

Diagnostic features of the Mazagao Phase 102 

The Arista Phase 103 

Description of sites and excavations 103 

Site A-7— Amapd City 104 

Site A-8— Aurora 106 

Site A-9— Rel6gio 107 

Site A-10 — Montanha da Pluma 107 

Site A-11 — Montanha de Aristd 108 

Site A-12— Cruzeiro 116 

Site A-13— Matapi 118 

Site A-14^-Macapd 118 

Site A-15— Vila Velha 119 

Site A-16— Ilhas do Campo 120 

Site A-18— Maica 121 

Site A-19— Renovado 122 

Site A-20— Vila Cunanf 122 

Site A-21— Pracuiiba 122 

Site A-22 — Conceigao 123 

Data from other investigations 123 

Rio Oiapoque site 123 

Rio Uagd, sites 125 

Rio Aracaud sites 125 

Cunanf sites 126 

Monte May6 sites 128 

Ilha do Carao site 130 

Agahyzal site 131 

Igarap^ Tartarugalzinho sites 131 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON VH 

BVANSJ 

The Territory of AmapX — Continued 

The Arist6 Phase — Continued page 

Analysis of materials of the Arista Phase 132 

Pottery type descriptions 132 

Arista Plain 132 

Ariste Painted 135 

Davi Incised 137 

Flexal Scraped 140 

Serra Plain 143 

Serra Painted 145 

Uagd Incised 148 

Unclassified Decorated 150 

Pottery artifacts 150 

Nonceramic artifacts 150 

Ceramic history 151 

Diagnostic features of the Ariste Phase 156 

Conclusions and interpretations 158 

Maraj6 Island 168 

Geographical description 168 

Tropical Forest Phases 174 

The Ananatuba Phase 174 

Description of sites and excavations 174 

Site J-7— Sip6 174 

Site J-8— Maguari 177 

Site J-9 — Ananatuba 177 

Site J-10— Sororoco 178 

Data from other investigations 179 

Site J-19 179 

Site J-20 179 

Analysis of materials of the Ananatuba Phase 179 

Pottery type descriptions 179 

Ananatuba Painted 180 

Ananatuba Plain 181 

Carmo Brushed 184 

Sip6 Incised 185 

Sororoco Plain 187 

Unclassified Decorated 188 

Pottery artifacts 189 

Nonceramic artifacts 189 

Ceramic history 189 

Diagnostic features of the Ananatuba Phase 193 

The Mangueiras Phase 194 

Description of sites and excavations 194 

Site J-5— Croari 194 

Site J-7— Sipo 198 

Site J-13— Bacurf 199 

Site J-16— Canivete 200 

Site C-3— Porto Real 201 

Data from other investigations 203 

Site J-17— Flor do Anajds 203 



Vin BUREAU OF AMERICAN ETHNOLOGY [bull, ibt 

Maraj6 Island — Continued 

Tropical Forest Phases — Continued 

The Mangueiras Phase — Continued page 

Analysis of materials of the Mangueiras Phase 203 

Pottery type descriptions 203 

Anjos Plain 203 

Bacurf Brushed 207 

Croarf Brushed 207 

Esperanga Red 208 

Mangueiras Plain 210 

Pocoat6 Scraped 213 

Pseudo-Sip6 Incised — Mangueiras Phase Variety __ 215 

Unclassified Decorated 216 

Pottery artifacts 217 

Nonceramic artifacts 217 

Ceramic history 217 

Diagnostic features of the Mangueiras Phase 221 

The Formiga Phase 222 

Description of sites and excavations 222 

Site J-4— Mucajd 222 

Site J-6 — Formiga 224 

Data from other excavations 226 

Site J-18— Coroca 226 

Analysis of materials of the Formiga Phase 227 

Pottery type descriptions 227 

Catarina Plain 227 

Coroca Plain 228 

Embauba Plain 230 

FormigaPlain 232 

Mucajd Corrugated 234 

Pseudo-Sip6 Incised — Formiga Phase Variety 237 

Sauba Brushed 238 

Unclassified Decorated 238 

Pottery artifacts 239 

Nonceramic artifacts 239 

Ceramic history 239 

Diagnostic features of the Formiga Phase 24 1 

The Arua Phase 242 

Description of sites and excavations 242 

Site J-2/3— Chaves Airport 242 

SiteJ-11— Carmo 244 

Data from other investigations 245 

Analysis of material 245 

Ceramic history 245 

Diagnostic features of the Arua Phase 245 

Comparison and interpretation of the Tropical Forest Phases, 

with a method for computing village duration 245 

Characterization of the Tropical Forest Phases 257 

The Marajoara Phase 259 

Description of sites and excavations 259 

Site J-14— Monte Carmelo 259 

Mound 1, Guajard, 259 

Mound 2, Monte Carmelo 278 

Mound 3, Bacatal 279 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON IX 

EVANS] 

Maraj6 Island — Continued 

The Marajoara Phase — Continued 

Description of sites and excavations — Continued page 

Site J-15— Os Camutins 279 

Mound 1, Camutins 281 

Mound 2 286 

Mounds 287 

Mound 4, Sacrario 288 

Mound 5, Sacacao 288 

Mound 6 288 

Mound 7, Sao Bento 289 

Mound 8 289 

Mound 9 289 

Mound 10 289 

Mound 11 289 

Mound 12, Carmo 289 

Mound 13 290 

Mound 14, Inajasal 290 

Mound 15 292 

Mound 16, Tesinho 293 

Mound 17, Bel^m 293 

Mound 18, Arraial 295 

Mound 19 295 

Mound 20 295 

Data from other investigations 295 

BacurlAlto 296 

Cajueiros 296 

Camutins 297 

Caratat6ua 301 

Cuieiras 301 

Curuxys 301 

Desterro 302 

Diamantina 302 

Fortaleza 302 

Guajard 306 

IlhadosBichos 306 

Ilha dos Marcos 307 

Laranjeiras 307 

Macacao 308 

Matinados 308 

Menino Deus 308 

Monte Carmelo 308 

Nazareth 309 

Pacoval 309 

Pacoval do Cururii 315 

Pacoval dos Mello 316 

Panellas 317 

Sanharao 317 

Santa Brigida 317 

Santa Izabel 317 

Santo Andre 318 

Serra 318 

Tap^ra 318 



X BUREAU OF AMERICAN ETHNOLOGY [boll. 167 

Makaj6 Island — Continued 

The Marajoara Phase — Continued 

Data from other investigations — Continued page 

Tesodas Iga^abas 318 

Teso de Severino 318 

Tesodos China 319 

TesodoGentil 322 

Teso dos Gentios 323 

Unnamed mounds 323 

Conclusions 324 

Analysis of materials of the Marajoara Phase 324 

Pottery type descriptions 324 

Anajds Double-slipped Incised 326 

Anajds Plain Incised 328 

Anajds Red Incised 331 

Anajds White Incised 332 

Ararl Double-shpped Excised 336 

Arari Plain Excised 339 

Ararl Red Excised 341 

Arari Red Excised, White-retouched 344 

Arari White Excised 347 

Camutins Plain 348 

Carmelo Red 353 

Goiapf Scraped 355 

Guajard Incised 356 

Inajd Plain 358 

Joanes Painted 359 

Pacoval Incised 366 

Unclassified Decorated 370 

Ceramic and non ceramic artifacts 371 

Axes 372 

Beads 374 

Earplugs 374 

Figurines 375 

Labrets 377 

Spindle whorls 378 

Spoons 380 

Stools 381 

Tangas 382 

Whistles 384 

Miscellaneous 384 

Ceramic history 385 

Diagnostic features of the Marajoara Phase 398 

Conclusions and interpretations 404 

The Islands of Mexiana and Caviana 425 

Geographical descriptions 425 

Mexiana 425 

Caviana , 427 

The Acauan Phase 429 

Description of sites and excavations 430 

Site M-1 — Jacareuba 430 

Site M-3— Acauan 430 

Site J-12 — Jurupucu 436 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON XI 

BVANSJ 

The Islands of Mexiana and Caviana — Continued 

The Acauan Phase — Continued page 

Data from other investigations 439 

Analysis of materials of the Acauan Phase 439 

Pottery type descriptions 439 

Acauan Excised 439 

Carobal Incised 442 

Floripes Corrugated 444 

Paciencia Scraped 446 

Piryzal Plain 448 

Vergal Incised 451 

Unclassified Decorated 452 

Pottery artifacts 453 

Nonceramic artifacts 453 

Ceramic history 453 

Diagnostic features of the Acauan Phase 456 

The Arua Phase 457 

Description of sites and excavations 457 

Habitation sites on Mexiana and Caviana 457 

Mexiana 457 

M-2 — Papa Cachorro 457 

M-7— Aberta 457 

Caviana 458 

C-5— Morera 458 

C-6 — Croatasal 459 

C-7— Sao Domingo 460 

C-8— Pacajd 461 

C-10— Sao Bento 461 

C-13— Alta Piratuba 463 

C-14 — Limaozinho 464 

C-15— Patahua 464 

Cemetery sites 465 

Mexiana 465 

M-4 — Fundo das Panellas 465 

M-5— Mulatinho 481 

Caviana 493 

C-1 — Teso das Igagabas 493 

C-4 — Teso dos Indios 495 

C-6— Croatasal, Section A 499 

C-9— Frei Joao 502 

C-11 — Vaquejador de Sao Sebasteao 509 

C-12— Condino 515 

Data from other investigations * 520 

Mexiana 520 

M-6— Recreo 520 

M-8 — Limao da Fora 520 

M-9— Chapeu 521 

Caviana 521 

Campo Redondo 522 

Sao Domingo 522 

Teso da Samahuma 522 

Esperanga 523 

Teso dos Indios - 523 



XII BUREAU OF AMERICAN ETHNOLOGY [bdll. ler 

The Islands of Mexiana and Caviana — Continued 
The Arua Phase — Continued 

Data from other investigations — Continued 

Caviana — Continued page 

Bacabal I 523 

Pesqueiro 524 

Prainha 524 

Rebordello 524 

Analysis of materials of the Arua Phase 525 

Pottery type descriptions 525 

Aberta Incised 525 

Nazare Brushed 525 

Piratuba Plain 526 

Unclassified Decorated 532 

Pottery artifacts 534 

Nonceramic artifacts 534 

Ceramic history 534 

Diagnostic features of the Arua Phase 538 

Conclusions and interpretations 539 

The Historical Aftermath 556 

Chronology of European contact 556 

Ethnohistorical information 566 

Territory of Amapd 566 

Tribes 566 

Population 569 

Culture 569 

Subsistence 570 

Settlement pattern 571 

Transportation 573 

Manufactures 573 

Dress and ornament 573 

Social organization 574 

Recreation 576 

Life cycle 576 

Religion 578 

Warfare . 578 

Lore and learning 679 

The Islands 579 

Tribes 579 

Population 581 

Culture 581 

Settlement pattern 581 

Watercraft 582 

Manufactures 582 

Social organization 582 

Life cycle 582 

Warfare 583 

Ethnohistorical-archeological correlation 583 

Implications of the Cultural Sequence at the Mouth of the Amazon. 589 

Literature Cited 608 

Appendix (tables 1-52) 619 



^I?S^^^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON XIII 

EVANS J 

ILLUSTRATIONS 

PLATES 
(All plates follow page 664) 

1. Views of the Rio Araguari above its junction with the Rio Amaparf, Territory 

of Amapd. 

2. Arua Phase stone alineraent at A-8 — Aurora in the central part of the 

Territory of Araapd. 

3. Mazagao Phase sites in the southern part of the Territory of Amapd. 

4. Ariste Phase habitation site of A-9 — Rel6gio and its environment in the 

central part of the Territory of Amapd,. 

5. Arista Phase burial site of A-10 — Montanha da Pluma in the northern 

part of the Territory of Amapd. 

6. Arista Phase sites in the northern part of the Territory of Amapd. 

7. Mazagao Phase vessels from A-3 — Pigacd Cemetery. 

8. Camaipf Plain vessels collected by Lima Guedes from the Rio Vilanova 

and now in the Museu Paraense Emilio Goeldi. 

9. Mazagao Phase vessels collected by Lima Guedes from the Rio Vilanova and 

now in the Museu Paraense Emilio Goeldi. 

10. Mazagao Plain sherds showing coarse quartz temper and typical applique 

decoration, 

11. Type sherds of Anauerapucii Incised, Mazagao Phase. 

12. Type sherds of Pigacd Incised, Mazagao Phase. 

13. Type sherds of Uxy Incised with rectilinear motifs, Mazagao Phase. 

14. Type sherds of Uxy Incised with curvilinear motifs, Mazagao Phase. 

15. Decorated pottery types of the Mazagao Phase. 

16. Unclassified Decorated sherds from the Mazagao Phase. 

17. Zoomorphic burial urns of the Maracd Phase from Ilha do Pard. 

18. Anthropomorphic burial urns of the Maracd Phase, collected by Lima 

Guedes from the Rio Maracd and now in the Museu Paraense Emilio 
Goeldi. 

19. Lid heads of Maracd Phase anthropomorphic burial urns in the collection 

of the Museu Paraense Emilio Goeldi. 

20. Arist6 Plain burial vessels from A-11 — Montanha de Arist6. 

21. Decorated pottery types of the Arist6 Phase. 

22. Fragmentary vessels of Flexal Scraped from the Arista Phase cemetery of 

A-11 — Montanha de Arista, Cave 2. 

23. Type sherds and a complete vessel of Serra Plain from Arista Phase sites. 

24. Vessels of Serra Painted from cemetery sites of the Arist6 Phase. 

25. Serra Painted jar from A-15 — Vila Velha and a sample of the glass trade 

beads found inside. 

26. Pottery types from the Arista Phase. 

27. Aerial views of Marajo Island (courtesy of the United States Army Air 

Force) . 

28. Aerial views of Maraj6 Island (courtesy of the United States Army Air 

Force) . 

29. Typical environment on northern Maraj6. 

30. Habitation sites of the Tropical Forest archeological Phases on Maraj6 

Island. 

31. Formiga Phase sites in the campo of northern Maraj6. 

32. Marajoara Phase habitation mounds on the Igarap6 Camutins, central 

Mara j 6. 



JAM 1 1958 



XIV BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

[^33. Marajoara Phase cemetery mounds on the Igarape Camutins, central 

'0.-- Maraj6. 

J 34. Marajoara Phase cemetery mounds on the upper Rio Anajds, central 

f- Marajd. 

35. Type sherds of Ananatuba Plain showing rim variation and handle con- 

struction, Ananatuba Phase. 

36. Type sherds of Carmo Brushed, Ananatuba Phase. 

37. Type sherds of Sip6 Incised, Design Type 1: row of scallops, Ananatuba 

Phase. 

38. Type sherds of Sip6 Incised, Design Type 2: zoned, fine cross-hatch, Anana- 

tuba Phase. 

39. Type sherds of Sip6 Incised, Ananatuba Phase. 

40. Type sherds of Sip6 Incised, Design Type 5 : zoned, parallel lines, Ananatuba 

Phase. 

41. Type sherds of Sip6 Incised, Ananatuba Phase, Design Types 6 and 7. 

42. Miscellaneous pottery from Ananatuba Phase sites. 

43. Type sherds of Bacurl Brushed, Mangueiras Phase. 

44. Type sherds and miniature vessel of Croari Brushed, Mangueiras Phase. 

45. Rim sherds from vessels of Mangueiras Plain, showing rim form and occa- 

sional notched or lobed decoration. 

46. Type sherds of Pocoat6 Scraped, Mangueiras Phase. 

47. Decorated sherds from the Mangueiras Phase. 

48. Decorated pottery types of the Formiga Phase. 

49. Trade sherds of the Marajoara Phase excavated at the Formiga Phase site 

of J-6 — Formiga. 

50. Type sherds of Anajds Double-slipped Incised, Marajoara Phase. (American 

Museum of Natural History.) 

51. Vessels of Anajds Plain Incised, Marajoara Phase. 

52. Vessels of Anajds Plain Incised, Marajoara Phase. (American Museum of 

Natural History.) 

53. Type sherds and vessels of Anajds Red Incised, Marajoara Phase. 

54. Type sherds of Anajds White Incised, Marajoara Phase. (American 

Museum of Natural History.) 

55. Anajds White Incised vessels, Marajoara Phase. 

56. Type sherds of Arari Double-slipped Excised, Marajoara Phase. 

57. Arari Plain Excised vessels. 

58. Type sherds of Ararf Plain Excised, Marajoara Phase. 

59. Type sherds of Arari Red Excised, Marajoara Phase. 

60. Arari Red Excised vessels, Marajoara Phase. 

61. Arari Red Excised vessels, Marajoara Phase. 

62. Arari Red Excised vessels, Marajoara Phase. 

63. Type sherds of Arari Red Excised, White-retouched, Marajoara Phase. 

(American Museum of Natural History.) 

64. Rim sherds of Camutins Plain and Inajd Plain bowls and jars from Marajo- 

ara Phase habitation mounds. 

65. Rim sherds of Camutins Plain and Inajd Plain bowls and jars from Marajoara 

Phase cemetery mounds. 

66. Inajd and Camutins Plain vessels associated with Marajoara Phase burials. 

67. Vessels from Marajoara Phase cemeteries. 

68. Large rim adornos from Camutins Plain and Inajd Plain Vessel Shape 4, 

Marajoara Phase. 

69. Rim adornos from Camutins Plain and Inajd Plain Vessel Shape 4, Mara- 

joara Phase. 



01 



^^:^.?^?^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON XV 

EVANS J 

70. Type sherds of Goiapf Scraped, Marajoara Phase. 

71. Sherds and vessels of GuajarA Incised, Marajoara Phase. 

72. Type sherds from Joanes Painted bowls with red-on-white designs, Mara- 

joara Phase. (American Museum of Natural History.) 

73. Vessels of Joanes Painted, Marajoara Phase. 

74. Neck sherds from large Joanes Painted burial urns, Polychrome Type B — 

red and black on white; J-15, Mound 1, surface. 

75. Vessels of Joanes Painted, Polychrome Type B; Marajoara Phase. 

76. Joanes Painted burial urn, Marajoara Phase. Two views of polychrome an- 

thropomorphic jar L, J-14, Mound 1, cut 1; now in the Museu Paraense 
Emilio Goeldi, Belem. 

77. Type sherds of Pacoval Incised, Marajoara Phase. (American Museum of 

Natural History.) 

78. Pacoval Incised vessels, Marajoara Phase. 

79. Pottery figurines from Marajoara Phase cemetery sites. 

80. Miscellaneous pottery objects from Marajoara Phase cemetery sites. 

81. Pottery spoons from Marajoara Phase sites showing range in size and posi- 

tion of "spout." 

82. Pottery tangas from Marajoara Phase burial urns. 

83. Pottery stools and stool fragments from Marajoara Phase sites. 

84. Painted pottery stools. 

85. Small pottery vessels with large, grotesque, "wing" adornos. 

86. Sherds with incised and excised designs from Arauquin, Venezuela. (Uni- 

versity Museum, Philadelphia.) 

87. Vessels with incised and excised decoration from Colombia. 

88. Decorated sherds from the Lower Amazon. 

89. Acauan Phase site of M-3 — Acauan. 

90. Type sherds of Acauan Excised with rectilinear motifs, Acauan Phase. 

91. Type sherds of Acauan Excised with curvilinear motifs, Acauan Phase. 

92. Acauan Phase pottery. 

93. Type sherds of Floripes Corrugated, coarse variety, Acauan Phase. 

94. Type sherds of Floripes Corrugated, fine variety, Acauan Phase. 

95. Ornamental rims of Piryzal Plain, Acauan Phase. 

96. Decorated pottery types of the Acauan Phase. 

97. Decorated sherds from the Acauan Phase site of J-12 — Jurupucii. 

98. Arua Phase habitation sites on Caviana Island. 

99. Arua Phase cemetery of M-4 — Fundo das Panellas. 

100. Arua Phase cemeteries. 

101. Arua Phase sites. 

102. Type sherds of Aberta Incised, Arua Phase. 

103. Type sherds of Nazar6 Brushed, Arua Phase. 

104. Rim sherds of Piratuba Plain, Arua Phase, showing typical thickening and 

folded-over treatment. 

105. Small Piratuba Plain vessels associated with Arua Phase burial jars. 

106. Piratuba Plain vessels from M-5 — Mulatinho, Mexiana, Arua Phase. 

107. Sherds from Piratuba Plain platters or griddles with punctate decoration 

around the rim, Arua Phase. 

108. Piratuba Plain sherds with impressed ring decoration or applique ribs, 

Arua Phase. 

109. Piratuba Plain sherds with impressed ring decoration from M-4 — Fundo das 

Panellas burial jars. 

110. Piratuba Plain sherds with applique decoration, Arua Phase. 

111. Fragmentary vessels of Piratuba Plain, Arua Phase. 

112. Miscellaneous sherds from Arua Phase habitation sites. 



XVI BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

FIGURES 

PAGE 

1. The Territory of Amapd, showing geographical features and location 

of archeological sites 34 

2. Ground plan of A-8 — Aurora, a stone alinement of the Arua Phase. _ 39 

3. Ground plan of A-23 — Ilha da Fortaleza, Conceigao, a stone alinement 

of the Arua Phase 40 

4. Ground plan of A-1 — Pigacd and A-3 — Pigacd Cemetery, Mazagao 

Phase 46 

5. Ground plan of A-2 — Lauro, a habitation site of the Mazagao Phase_ 47 

6. Foot of a zoomorphic (turtle) urn from A-3 — Pigacd Cemetery, 

Mazagao Phase 49 

7. Glass trade beads from A-3 — Pigacd Cemetery, Mazagao Phase 50 

8. Jar C (Mazagao Plain), Burial Group 1, A-3 — Pigacd Cemetery, 

Mazagao Phase 52 

9. Vessel E (Uxy Incised), Burial Group 1, A-3 — Plgacd Cemetery, 

Mazagao Phase 54 

10. Stone artifacts from A-3 — Pigacd. Cemetery, Mazagao Phase 55 

11. Glass trade beads from A-4 — Valentim, Mazagao Phase 59 

12. Reconstruction of jar A, Burial Group 2, A-4 — Valentim, Mazagao 

Phase 61 

13. Pottery vessels from Burial 3, A-4 — Valentim, Mazagao Phase 63 

14. Anauerapucu Incised lid from Burial 3, A-4 — Valentim, Mazagao 

Phase 65 

15. Ground plan of Site A-6 — Ilha das Igagabas, a habitation site of the 

Mazagao Phase 66 

16. Rim profiles and vessel shapes of Anauerapucu Incised, Mazagao 

Phase (Appendix, table 2) 79 

17. Rim profiles and vessel shapes of Camaipi Plain, Mazagao Phase 

(Appendix, table 3) 82 

18. Rim profiles and vessel shapes of Jari Scraped, Mazagao Phase (Ap- 

pendix, table 4) 84 

19. Rim profiles and vessel shapes of Mazagao Plain, Mazagao Phase 

(Appendix, table 5) 86 

20. Rim profiles and vessel shapes of Pigacd Incised, Mazagao Phase 

(Appendix, table 6) 88 

21. Rim profiles and vessel shapes of Uxy Incised, Mazagao Phase (Ap- 

pendix, table 7) 90 

22. Rim profiles and vessel shapes of Vilanova Plain, Mazagao Phase 

(Appendix, table 8) 93 

23. Seriation of Mazagao Phase sites on the basis of pottery type frequency 

(Appendix, table 1) 96 

24. Trends in the popularity of common vessel shapes of the Mazagao 

Phase (Appendix, table 10) 102 

25. Stone artifacts from A-7 — Amapd, Arista Phase 105 

26. Ground plan of A-9 — Rel6gio, a habitation site of the Arista Phase. _ 106 

27. Worked sherd scraper from the Ariste Phase cemetery of A-11 — 

Montanha de Arista, Cave 1 109 

28. Ground plan of A-11— Montanha de Arista, Cave 2, Arista Phase 110 

29. Small stone chisel from A-11— Montanha de Arista, Cave 2, Arist6 

Phase 111 

30. Pottery figurine (Ariste Plain paste) from A-11 — Montanha de Arista, 

Cave 2, Arista Phase 112 



l^f?^^^^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON XVH 

EVANS J 

PAGE 

31. Flexal Scraped Bowl (vessel E) from A-11 — Montanha de Ariste, 

Cave 2, Arista Phase 113 

32. Flexal Scraped Jar (vessel G) from A-11 — Montanha de Arista, Cave 

2, Arist6 Phase 114 

33. Serra Painted design on the shoulder of vessel P from A-1 1 — Montanha 

de Ariste, Cave 2, Arist6 Phase 116 

34. Reconstruction of Ariste Painted vessel with anthropomorphic face 

from A-11 — Montanha de Arista, Cave 3, Arist6 Phase 117 

35. Detail of the anthropomorphic face on the Arista Painted vessel 

shown in figure 34 117 

36. Burial urn and lid of the Arist6 Phase found by Hamy (1897) on the 

Rio Oiapoque 124 

37. Profile and top view of shaft-burial at the Cunani Site, Arist6 Phase 

(After Goeldi, 1900.) 127 

38. Rim profiles and vessel shapes of Ariste Plain, Arista Phase (Ap- 

pendix, table 12) 133 

39. Rim profiles and vessel shapes of Arista Painted, Arista Phase 

(Appendix, table 13) 136 

40. Arista Painted sherds from A-11 — Montanha de Arista, Cave 2, 

Arista Phase 138 

41. Rim profiles and vessel shapes of Davl Incised, Arist6 Phase (Ap- 

pendix, table 14) 139 

42. Rim profiles and vessel shapes of Flexal Scraped, Arista Phase 

(Appendix, table 15) 141 

43. Rim profiles and vessel shapes of Serra Plain, Ariste Phase (Appendix, 

table 16) 144 

44. Rim profiles and vessel shapes of Serra Painted, Arista Phase (Ap- 

pendix, table 17) 146 

45. Rim profiles and vessel shapes of Uagd Incised, Ariste Phase (Ap- 

pendix, table 18) 149 

46. Seriation of Ariste Phase sites on the basis of pottery type frequency 

(Appendix, table 11) 152 

47. Maraj6 Island, showing major streams, vegetation pattern and the 

location of archeological sites 168 

48. North coast of Maraj6 Island, showing vegetation pattern and the 

location of sites J-6 through J-13 174 

49. Plan of J-7 — Sipo, a habitation site of the Ananatuba Phase 175 

50. Partially drilled sherd from J-7 — Sipo, Ananatuba Phase 176 

51. Cylindrical pottery objects from J-9 — Ananatuba, Ananatuba Phase. 178 

52. Rim profiles and vessel shapes of Ananatuba Painted, Ananatuba 

Phase (Appendix, table 24) 180 

53. Rim profiles and vessel shapes of Ananatuba Plain and Sororoco 

Plain, Ananatuba Phase (Appendix, tables 23 and 26) 182 

54. Rim profiles and vessel shapes of Carmo Brushed, Ananatuba Phase 

(Appendix, table 24) 184 

55. Rim profiles and vessel shapes of Sip6 Incised, Ananatuba Phase 

(Appendix, table 25) 186 

56. Seriation of Ananatuba Phase sites on the basis of pottery type 

frequency (Appendix, tables 21 and 22) 190 

57. Plan of J-5 — Croari, a habitation site of the Mangueiras Phase 195 

58. Pottery tubular pipes from Mangueiras Phase sites 197 

59. Figurine parts from J-5 — Croari, Mangueiras Phase 197 

391329^57 2 



XVin BUREAU OF AMERICAN ETHNOLOGY [bull, wt 

PAGE 

60. Lab ret (?) fragment from J-5 — Croari, Mangueiras Phase 198 

61. Plan of J-13 — Bacurl, a habitation site of the Mangueiras Phase 199 

62. Plan of J-16 — Canivete, a habitation site of the Mangueiras Phase. _ 200 

63. Labrets and bicouical object of pottery from C-3 — Porto Real, 

Mangueiras Phase 202 

64. Rim profiles and vessel shapes of Anjos Plain, Mangueiras Phase 

(Appendix, table 29) 205 

65. Rim profiles and vessel shapes of Bacuri Brushed, Mangueiras Phase 

(Appendix, table 30) 206 

66. Rim profiles and vessel shapes of Croari Brushed, Mangueiras Phase 

(Appendix, table 30) 208 

67. Rim profiles and vessel shapes of Esperanga Red, Mangueiras Phase 

(Appendix, table 32) 209 

68. Rim profiles and vessel shapes of Mangueiras Plain jars, Mangueiras 

Phase (Appendix, table 31) 212 

69. Rim profiles and vessel shapes of Mangueiras Plain bowls, Mangueiras 

Phase (Appendix, table 31) 213 

70. Rim profiles and vessel shapes of Pocoat6 Scraped, Mangueiras Phase 

(Appendix, table 32) 214 

71. Rim profiles and vessel shapes of Pseudo-Sip6 Incised, Mangueiras 

Phase Variety (Appendix, table 32) 216 

72. Seriation of Mangueiras Phase sites on the basis of pottery type 

frequency (Appendix, table 28) 218 

73. Stratigraphic evidence for the origin by acculturation of Mangueiras 

Plain Vessel Shape 4 (Mangueiras Phase) from the Ananatuba 
Phase (Appendix, tables 27 and 33) 220 

74. Stratigraphic evidence for the origin by acculturation of Man- 

gueiras Plain Vessel Shape 3 (Mangueiras Phase) from the Anana- 
tuba Phase (Appendix, tables 27 and 33) 221 

75. Plan of J-4 — Mucajd, a habitation site of the Formiga Phase 223 

76. Plan of J-6 — Formiga, a habitation site of the Formiga Phase 225 

77. Drilled sherds from J-6 — Formiga, Formiga Phase 226 

78. Rim profiles and vessel shapes of Catarina Plain, Formiga Phase 

(Appendix, table 35) 229 

79. Rim profiles and vessel shapes of Coroca Plain, Formiga Phase 

(Appendix, table 35) 231 

80. Rim profiles and vessel shapes of Embauba Plain, Formiga Phase 

(Appendix, table 36) 233 

81. Rim profiles and vessel shapes of Formiga Plain, Formiga Phase 

(Appendix, table 37) 235 

82. Rim profiles and vessel shapes of Mucajd Corrugated, Formiga Phase 

(Appendix, table 38) 236 

83. Rim profiles and vessel shapes of Pseudo-Sip6 Incised, Formiga 

Phase Variety (Appendix, table 38) 237 

84. Rim profiles and vessel shapes of Sauba Brushed, Formiga Phase 

(Appendix, table 38) 238 

85. Seriation of Formiga Phase sites on the basis of pottery type fre- 

quency (Appendix, table 34) 240 

86. North coast of Maraj6 Island in the vicinity of Chaves, showing the 

locations of J-2/3 — Chaves Airport and J-4 — Mucajd, 243 

87. Stone ax from J-2/3 — Chaves Airport, a habitation site of the Arua 

Phase 244 



EVANsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON XIX 

PAGE 

88. Plan of J-14, Mounds 1, 2, and 3, a mound group of the Marajoara 

Phase 260 

89. Burial stratigraphy of J-14, Mound 1, cut 1, Marajoara Phase 261 

90. J-14, Mound 1, cut 1, vessels A, B, C, and D, Marajoara Phase 262 



91. J-14, Mound 1 

92. J-14, Mound 1 

93. J-14, Mound 1 

94. J-14, Mound 1 

95. J-14, Mound 1 

96. J-14, Mound 1 

97. J-14, Mound 1 

98. J-14, Mound 1 

99. J-14, Mound 1 
100. J-14, Mound 1 



cut 1, jar E, Marajoara Phase 264 

cut 1, jar F, Marajoara Phase 265 

cut 1, jar H, Marajoara Phase 266 

cut 1, jar I, Marajoara Phase 268 

cut 1, jar J, Marajoara Phase 269 

cut 1, jar K, Marajoara Phase 270 

cut 1, jar L, Marajoara Phase 272 

cut 1, jar M, Marajoara Phase 274 

cut 1, jar N, Marajoara Phase 276 

cut 1, jar O, Marajoara Phase 277 

101. Plan of mounds composing J-15 — Camutins, a mound group of the 

Marajoara Phase 280 

102. Plan of J-15, Mound 1, Marajoara Phase, showing the location of 

excavations 282 

103. Artifacts from J-15, Mound 1, cut 2, Marajoara Phase 284 

104. Detailed plans of J-15, Mounds 3, 4, 6, 15, 16, habitation mounds of 

the Marajoara Phase 287 

105. Detailed plan of J-15, Mound 4 — Inajasal, a habitation mound of 

the Marajoara Phase, showing the location of cut 1 290 

106. Profile of west face of cut 1, J-15, Mound 14, Marajoara Phase 291 

107. Detailed plan of J-15, Mound 17, a Marajoara Phase cemetery, 

showing location of excavations 293 

108. Map of Maraj6 Island, showing the location of Marajoara Phase 

cemetery sites 296 

109. Plan of the Fortaleza Mound Group of the Marajoara Phase on the 

Rio Goiapi 303 

110. Detailed plan of Mound 7 of the Fortaleza Group showing the extent 

of Farabee's excavations in this Marajoara Phase cemetery site 305 

111. Plan of the Marajoara Phase site of Pacoval made by Lange in 1913__ 312 

112. Plan of the Marajoara Phase site of Pacoval made by Hilbert in 1951__ 313 

113. Plan of Teso dos China mound group of the Marajoara Phase 321 

114. Rim profiles and vessel shapes of Anajds Double-slipped Incised of 

the Marajoara Phase 327 

115. Rim profiles and vessel shapes of Anajds Plain Incised, Marajoara 

Phase 329 

116. Rim profiles and vessel shapes of Anajd,s Red Incised, Marajoara 

Phase 331 

117. Rim profiles and vessel shapes of Anajds White Incised, Marajoara 

Phase 334 

118. Rim profiles and vessel shapes of Ararl Double-slipped Excised and 

Arari Red Excised, Marajoara Phase 337 

119. Rim profiles and vessel shapes of Ararf Plain Excised, Marajoara 

Phase 339 

120. Rim profiles and vessel shapes of Ararf Red Excised, White-retouched, 

Marajoara Phase 345 

121. Rim profiles and vessel shapes of Camutins Plain and Inajd Plain 

bowls, Marajoara Phase (Appendix, tables 45 and 46) 350 



XX BUREAU OF AMERICAN ETHNOLOGY [boll. 167 

PAGE 

122. Rim profiles and vessel shapes of Camutins Plain and Inajd Plain 

jars, Marajoara Phase (Appendix, tables 45 and 46) 351 

123. Less common rim profiles and vessel shapes of Camutins Plain and 

Inajd Plain, Marajoara Phase 352 

124. Rim profiles and vessel shapes of Carmelo Red, Marajoara Phase 354 

125. Rim profiles and vessel shapes of Goiapf Scraped, Marajoara Phase.. 355 

126. Rim profiles and vessel shapes of Guajard Incised, Marajoara Phase. _ 357 

127. Rim profiles and vessel shapes of Joanes Painted, Marajoara Phase 

(Appendix, table 47) 360 

128. Joanes Painted, Polychrome Type A and B, Marajoara Phase 364 

129. Joanes Painted, Polychrome Type C, Marajoara Phase 365 

130. Rim profiles and vessel shapes of Pacoval Incised bowls, Marajoara 

Phase 367 

131. Rim profiles and vessel shapes of Pacoval Incised jars, Marajoara 

Phase 368 

132. Marajoara Phase miniature axes 373 

133. Marajoara Phase stone ax from J-14, Mound 1, cut 1 374 

134. Marajoara Phase pottery ear plugs 375 

135. Marajoara Phase pottery labrets from J-15, Mound 1, surface. Re- 

construction is based on a complete stone specimen from Panellas.. 377 

136. Marajoara Phase pottery spindle whorls 379 

137. Standardized measurements on tangas of the Marajoara Phase 383 

138. Marajoara Phase grooved polishing stone from j^ J-15, Mound 1, 

surface 385 

139. Ceramic stratigraphy of J-15, Mound 14, cut 1, showing trends in 

the Marajoara Phase plain wares (Appendix, table 39) 386 

140. Ceramic sedation of the Marajoara Phase sites of J-15, Mounds 1 

and 17, and J-14, Mound 1 (Appendix, table 40) 386 

141. Seriation of Marajoara Phase cemetery sites based on relative fre- 

quency of Inajd Plain and Camutins Plain (Appendix, table 41) . _ 388 

142. Seriation of J-15 habitation mounds based on the relative frequency 

of Inajd Plain and Camutins Plain (Appendix, table 42) 390 

143. Trends in Marajoara Phase decorated types revealed by the seria- 

tion of the 4 Marajoara Phase cemetery sites of Pacoval, Fortaleza, 
Camutins, and Guajard (Appendix, table 43) 394 

144. Trends in unusual pottery artifacts and adornos of the Marajoara 

Phase shown by comparing the cemetery sites of Pacoval, Fortaleza, 
Camutins, and Guajard (Appendix, table 44) 396 

145. Distribution of the various archeological Phases on Maraj6 Island.. 406 

146. Archeological distribution of distinctive traits of the Marajoara Phase. 412 

147. Tassels shown as ear ornaments on Marajoara Phase anthropomorphic 

jars , 416 

148. Ethnographic distribution of distinctive traits of the Marajoara 

Phase 416 

149. Sequence and relative duration of the prehistoric cultures on Maraj6 

Island 422 

150. Mexiana Island, showing major streams, vegetation pattern and loca- 

tion of archeological sites 426 

151. Caviana Island, showing major streams, vegetation pattern and loca- 

tion of archeological sites 428 

152. Plan of M-3 — Acauan, a habitation site of the Acauan Phase 431 

153. Acauan Phase anthropomorphic or zoomorphic rim adornos on Piryzal 

Plain from M-3 — Acauan. 434 



EVANsf^ *^° ARCHEOLOGY AT MOUTH OF AMAZON XXI 

PAQB 

1 54. Acauan Phase spindle whorl from M-3 — Acauan 435 

155. Acauan Phase figurine head from M-3 — Acauan 436 

156. Acauan Phase stone flake from M-3 — Acauan 437 

157. Acauan Phase pottery stamp from M-3 — Acauan 438 

158. Acauan Phase figurine from J-12 — Jurupucu 438 

159. Rim profiles and vessel shapes of Acauan Excised, Acauan Phase 

(Appendix, table 49) 440 

160. Rim profiles and vessel shapes of Carobal Incised, Acauan Phase 

(Appendix, table 49) 443 

161. Rim profiles and vessel shapes of Floripes Corrugated, Acauan Phase.- 445 

162. Rim profiles and vessel shapes of Paciencia Scraped, Acauan Phase 

(Appendix, table 49) 447 

163. Rim profiles and vessel shapes of Piryzal Plain, Acauan Phase 

(Appendix, table 50) 449 

164. Plan of M-2 — Papa Cachorro, a habitation site of the Arua Phase. _ 458 

165. Arari Excised vessel with adorno, a Marajoara Phase trade sherd 

found at the Arua Phase site M-2 — Papa Cachorro 459 

166. Location and environmental situation of C-5 — Morera and C-6 — 

Croatasal, habitation sites of the Arua Phase 460 

167. Arua Phase pottery animal feet 461 

168. Plan of C-8 — Pacajd,, a habitation site of the Arua Phase 462 

169. Ajua Phase stone axes 463 

170. Plan of M-4 — Fundo das Panellas, an Arua Phase cemetery site, 

showing the positions of the burial jars 466 

171. Reconstructed burial jars from the Arua Phase site M-4 — Fundo das 

Panellas 467 

172. Painted decoration on Piratuba Plain sherds from Arua Phase sites. _ 474 

173. Plan of M-5 — Mulatinho, a cemetery site of the Arua Phase, showing 

the position of the burial jars and other objects 482 

174. Arua Phase anthropomorphic vessel 4 from M-5 — Mulatinho 485 

175. Arua Phase zoomorphic adorno from jar 24, M-5 — Mulatinho 489 

176. Arua Phase nephrite pendants from M-5 — Mulatinho 492 

177. Detail of modeling on jar 7 from the Arua Phase Site C-1 — Teso das 

Igagabas and a reconstruction of the vessel to which it belonged 494 

178. Plan of C-4 — Teso dos Indies, a cemetery site of the Arua Phase, 

showing the location of the burial jars 496 

179. Reconstructed jars from Arua Phase cemeteries 497 

180. Detailed plan of Section A, the burial area of C-6 — Croatasal 499 

181. Base shapes of Piratuba Plain jars, Arua Phase 500 

182. Arua Phase anthropomorphic burial jar from Section A of C-6 — 

Croatasal 501 

1 83. Arua Phase pottery figurines 502 

184. Arua Phase stone axes from C-9 — Frei Joao 504 

185. Arua Phase stone artifacts from C-9 — Frei Joao 505 

186. Arua Phase decorated sherds from C-9 — Frei Joao 506 

187. Arua Phase zoomorphic rim adorno in the form of a bird from C-9 — 

Frei Joao 507 

188. Arua Phase nephrite pendants and beads from C-9 — Frei Joao 508 

189. Plan of C-11 — Vaquejador de Sao Sebasteao, an Arua Phase cemetery 

site 510 

190. Piratuba'^Plain bowl associated with Burial jar 4 from the Arua Phase 

site C-11 — Vaquejador de Sao Sebasteao 511 



XXn BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

PAGE 

191. Tiered jar of Piratuba Plain associated with Burial jar 12 from the 

Arua Phase site of C-11 — Vaquejador de Sao Sebasteao 512 

192. Arua Phase pottery beads associated with burial jar 4, C-11 — 

Vaquejador de Sao Sebasteao 513 

193. Stone axes of the Arua Phase 514 

194. Plan of C-12 — Condino, a cemetery site of the Arua Phase, showing 

the location of the burial jars 516 

195. Piratuba Plain platter from the Arua Phase site, C-12 — Condino 518 

196. Rim profiles and vessel shapes of Piratuba Plain bowls, Arua Phase 

(Appendix, table 52) 528 

197. Rim profiles and vessel shapes of Piratuba Plain jars, Arua Phase 

(Appendix, table 52) 529 

198. Piratuba Plain sherds with modeled decoration from various Arua 

Phase sites 530 

199. Piratuba Plain sherds with modeled decoration from various Arua 

Phase sites 531 

200. Piratuba Plain sherds with modeled decoration from various Arua 

Phase sites 532 

201. Seriation of Arua Phase sites on the basis of vessel shape frequency of 

Piratuba Plain (Appendix, table 52) 636 

202. Designs made by potterj^ stamps 547 

203. European forts and other settlements at the mouth of the Amazon 

prior to the middle of the 18th century 560 

204. Early map of the tribes inhabiting the Guiana coast published by 

William Delisle in 1703 568 

205. The cultural sequence at the mouth of the Amazon 590 

206. Routes of migration and diffusion in northern South America, recon- 

structed from the evidence of the affiliations of the archeological 

Phases at the mouth of the Amazon 600 

TEXT TABLES 

A. Glass beads from A-3 — Pigacd Cemetery 51 

B. Glass beads from A-4 — Valentim 68 

C. Glass beads from A-15— Vila Velha 120 

D. Relative frequency of Ananatuba Phase and Mangueiras Phase wares 

at J-7— Sip6 192 

E. Duration of Ananatuba Phase village sites 253 

F. Duration of Mangueiras Phase village sites 263 

G. Duration of Arua Phase village sites 254 

H. Duration of Formiga Phase village sites 254 

I. Differential results of two methods of calculating rate of village refuse 

accumulation 255 

J. Wai Wai village duration 257 

K. Frequency of Inajd and Camutins Plain wares at Teso dos China 322 

L. Temporal differences in Anajd,s White Incised decoration 336 

M. Temporal differences in Arari Plain Excised decoration 341 

N. Temporal differences in Arari Red Excised decoration 344 

O. Temporal differences in Arari White Excised decoration 348 

P. Differences in wear on tanga fragments 382 

Q. Standardized measurements on tangas 383 

R. Duration of Acauan Phase village Site M-3 — Acauan 456 

S. Glass beads from M-5 — Mulatinho 492 



EVANsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON XXIH 

PAGE 

T. Base type and dimensions of burial jars from C-6 — Croatasal, Section A. 50 1 
U. Chronological sequence of European exploration and settlement 557 

APPENDIX TABLES 

1. Frequency of pottery types in surface collections and stratigraphic excavations 

at sites of the Mazagao Phase. 

2. Frequency of rim and vessel shapes of Anauerapucu Incised in sites of the 

Mazagao Phase. 

3. Frequency of rim and vessel shapes of Camaipl Plain in sites of the Mazagao 

Phase. 

4. Frequency of rim and vessel shapes of Jari Scraped in sites of the Mazagao 

Phase. 

5. Frequency of rim and vessel shapes of Mazagao Plain in sites of the Mazagao 

Phase. 

6. Frequency of rim and vessel shapes of Pigacd Incised in sites of the Mazagao 

Phase. 

7. Frequency of rim and vessel shapes of Uxy Incised in sites of the Mazagao 

Phase. 

8. Frequency of rim and vessel shapes of Vilanova Plain in sites of the Mazagao 

Phase. 

9. Reduction of the individual rim and vessel shapes of Mazagao Phase pottery 

types to common vessel shapes. 

10. Frequency of common rim and vessel shapes irrespective of pottery type in 

sites of the Mazagao Phase. 

11. Frequency of pottery types in sites of the Arista Phase. 

12. Frequency of rim and vessel shapes of Arista Plain in sites of the Arista 

Phase. 

13. Frequency of rim and vessel shapes of Arist6 Painted in sites of the Arista 

Phase. 

14. Frequency of rim and vessel shapes of Davf Incised in sites of the Arista 

Phase. 

15. Frequency of rim and vessel shapes of Flexal Scraped in sites of the Arista 

Phase. 

16. Frequency of rim and vessel shapes of Serra Plain in sites of the Arista Phase. 

17. Frequency of rim and vessel shapes of Serra Painted in sites of the Arista 

Phase. 

18. Frequency of rim and vessel shapes of Uagd Incised in sites of the Arista 

Phase. 

19. Reduction of the individual rim and vessel shapes of Arista Phase pottery 

types to common vessel shapes. 

20. Frequency of common rim and vessel shapes irrespective of pottery type in 

sites of the Ariste Phase. 

21. Frequency of pottery types in stratigraphic excavations at site J-7 of the 

Ananatuba Phase. 

22. Frequency of pottery types in stratigraphic excavations at sites J-8, J-9, 

and J-10 of the Ananatuba Phase. 

23. Frequency of rim and vessel shapes of Ananatuba Plain in sites of the Ana- 

natuba Phase. 

24. Frequency of rim and vessel shapes of Ananatuba Painted and Carmo Brushed 

in sites of the Ananatuba Phase. 

25. Frequency of rim and vessel shapes of Sip6 Incised in sites of the Ananatuba 

Phase. 



XXIV BUREAU OF AMERICAN ETHNOLOGY [bull, le? 

26. Frequency of rim and vessel shapes of Sororoco Plain in sites of the Ananatuba 

Phase. 

27. Frequency of 3 jar shapes in Ananatuba Phase plain wares that influenced 

the pottery of the Mangueiras Phase. 

28. Frequency of pottery types in surface collections and stratigraphic excava- 

tions at sites of the Mangueiras Phase. 

29. Frequency of rim and vessel shapes of Anjos Plain in sites of the Mangueiras 

Phase. 

30. Frequency of rim and vessel shapes of Bacurl Brushed and Croarf Brushed 

in sites of the Mangueiras Phase. 

31. Frequency of rim and vessel shapes of Mangueiras Plain in sites of the Man- 

gueiras Phase. 

32. Frequency of rim and vessel shapes of Esperanga Red, Pocoat6 Scraped and 

Pseudo-Sip6 Incised in sites of the Mangueiras Phase. 

33. Frequency of 3 jar shapes in Mangueiras Phase plain wares showing in- 

fluence from the Ananatuba Phase. 

34. Frequency of pottery types in surface collections and stratigraphic excava- 

tions at sites of the Formiga Phase. 

35. Frequency of rim and vessel shapes of Catarina Plain and Coroca Plain in 

sites of the Formiga Phase. 

36. Frequency of rim and vessel shapes of Embauba Plain in sites of the Formiga 

Phase. 

37. Frequency of rim and vessel shapes of Formiga Plain in sites of the Formiga 

Phase. 

38. Frequency of rim and vessel shapes of Mucajd Corrugated, Pseudo-Sip6 

Incised and Sauba Brushed in sites of the Formiga Phase. 

39. Frequency of pottery types in J-15, Mound 14, Cut 1, a habitation site of 

the Mara j oar a Phase. 

40. Frequency of pottery types in stratigraphic excavations at J-14, Mound 1, 

J-15, Mound 1, and J-15, Mound 17, cemetery sites of the Marajoara 
Phase. 

41. Adjusted frequency of plain pottery types from 17 Marajoara Phase cemetery 

mounds. 

42. Frequency of the pottery types in surface collections from J-15 habitation 

mounds, Marajoara Phase. 

43. Frequency of decorated pottery types at 4 Marajoara Phase cemetery mounds. 

44. Frequency of unusual pottery artifacts and adornos at 4 Marajoara Phase 

cemetery sites. 

45. Frequency of rim and vessel shapes of Camutins Plain in sites of the Mara- 

joara Phase. 

46. Frequency of rim and vessel shapes of Inajd Plain in sites of the Marajoara 

Phase. 

47. Frequency of rim and vessel shapes of Joanes Painted in sites of the Mara- 

joara Phase. 

48. Frequency of pottery types in surface collections and stratigraphic excava- 

tions at site M-3 of the Acauan Phase. 

49. Frequency of rim and vessel shapes of Acauan Excised, Carobal Incised and 

Paciencia Scraped of the Acauan Phase. 
60. Frequency of rim and vessel shapes of Piryzal Plain of the Acauan Phase. 

51, Frequency of pottery types at sites of the Arua Phase. 

52. Frequency of rim and vessel shapes of Piratuba Plain in sites of the Arua 

Phase. 



PREFACE 

Any field project undertaken in the interior of a country away from 
the settled metropolitan centers of the world depends so heavily on 
personal contacts, friendliness, hospitality, and cooperation for its 
successful completion that it is impossible to single out each individual 
to whom special mention is due. However, there are a number of 
people whose contributions were instrumental in making our work 
possible, and we wish to take this opportunity to express our sincere 
appreciation to them for their generous assistance and cooperation. 

For his enthusiastic interest in, and active support of our proposal 
to do archeological research at the mouth of the Amazon, we are 
particularly indebted to Dr. William Duncan Strong, Loubat pro- 
fessor of anthropology and at that time chairman of the Department 
of Anthropology of Columbia University. It is no exaggeration to 
say that without his aid, our plans would never have been realized. 
Dr. Charles Wagley and Dr. Julian H. Steward, also of the Columbia 
faculty, aided us in details of the planning and together with Dr. 
Strong read and criticized the portions of this report that were sub- 
mitted as doctoral dissertations to Columbia University. We are 
indebted to Dr. Wendell C. Bennett for both encouragement and 
advice when our plans were in a formative stage, and for friendly and 
continuing interest in our work and its results. We deeply regret 
that he never saw the completed monograph, but hope that in dedi- 
cating it to him we are able to convey an impression of what his 
friendly guidance meant to those of us who never had the privilege 
of being his students. For showing an interest in supporting the first 
intensive archeological field work in the Amazon Valley we wish to 
record our gratitude to the Wenner-Gren Foundation for Anthropo- 
logical Research, Inc. (at that time called The Viking Fund, Inc.) 
and to the William Bayard Cutting Traveling Fellowship Fund of 
Columbia University. 

The successful launching of the expedition in Brazil was due to the 
cooperation of Sra. Heloisa Alberto Torres, then director of the Museu 
Nacional in Rio de Janeiro, who offered us permission to work under 
the auspices of that museum. For her scientific and oflEicial assistance 
as well as her friendly interest, aid, and advice from the beginning to 
the end of the trip, we offer om- warmest thanks. Sra. Maria Alberto 
Torres was also extremely helpful in guiding us through the intricacies 
of ofiicial details in a foreign country. Dr. and Mrs. Charles Wagley, 

XXV 



XXVI BUREAU OF AMERICAN ETHNOLOGY [boll, m 

who preceded us to Rio de Janeiro and Belem, did a great deal to pave 
the way for our reception. 

In Belem, we are particularly obligated to Sr. Machado Coelho, at 
that time director of the Museu Paraense Emilio Goeldi, for his coop- 
eration and for placing at our disposal a house on the museum grounds 
in which we lived while classifying and analyzing our excavated 
materials; to Sr. Frederico Barata, who not only allowed us to examine 
his collection of archeological specimens but was ready to aid us in 
any other way that was within his capacity; to Sr. Eurico deMelo 
Cardoso Fernandes, whose interest in anthropology was primarily 
responsible for our being invited to work in the Territory of Amapa; 
to Sr. Jose Ambrosio de Miranda Pombo, who was instrumental in 
obtaining permission for us to work on the Island of Mexiana and in 
the Municipio of Chaves, Maraj6 Island; to Jose F. Cottim, who 
volunteered his services as our interpreter in official situations before 
our laiowledge of Portuguese was adequate; to Benjamin Pinto y 
Sousa, who patiently washed and numbered all our sherds and per- 
formed numerous other indispensable services both on Caviana and in 
Belem; to Dr. Caspar Cesar de Andrade, director of the Servigo 
Especial de Saiide Publica in Belem, and his staff for certain medical 
supplies and advice ; to Dr. Acylino de Leao, delegate to the Conselho 
de Fiscalizagao de Expedigoes Artisticas e Cientificas, for his under- 
standing and cooperation in official matters regarding the archeological 
collections ; to Sr. Fritz Ackermann and Sr. Felisberto de Camargo for 
permitting us to study their private archeological collections. At 
various intervals while in Belem we spent many enjoyable hours in 
the homes of Mr. and Mrs. George T. Colman, Mr. and Mrs. Gordon 
Pickerell, Mr. and Mrs. M. L. Albuquerque, and Sr. and Sra. Philippe 
Farah. Their hospitality did much to make our months in Belem 
pleasant memories. 

While in the Territory of Amap^ we became indebted foremost to 
Governor Janary Gentil Nunes, who, because of his interest in the 
Territory of Amapa and his eagerness to develop all its potentialities, 
gave us overwhelming cooperation and placed at our disposal every- 
thing from motorboats to airplanes, maps, and workmen in order to 
expedite our research before the rainy season reached its height. 
Thanks to him, we were able to accomplish here in 1 month almost as 
much as we had been able to do in 4 months on the islands. Sr. Newton 
Wilson Cardoso, director of the newly formed Museu Territorial, ac- 
companied us on all our trips here to learn the technique of archeologi- 
cal field survey and excavation.^^ He '-proved 'so apt a pupil that we 
were able to include in our archeological analysis of the region addi- 
tional sites and materials he later collected by himself. We wish to 
thank Sr. Fritz Ackermann, for making available to us the information 



EVANsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON XXVII 

he had recorded about archeological sites during geological surveys. 
For their friendliness, cooperation, and hospitality while we were in 
the Territory of Amapa, we are also indebted to Sr. and Sra. Janary 
Gentil Nunes, Sr. and Sra. Jose Ferreira Teixeira, Sr. Raul Montero 
Valdez, Sr. Benedito Jose Carneiro de Amorim, Sr. Hermogenes da 
Lima Filho, Sr. Atahualpa Maranhao, and Prof. Jose Tostes. 

The carrying out of extensive archeological survey in north Brazil 
involves the granting of permission by land owners to trespass, ex- 
cavate and in many cases also to use the ranch house for headquarters. 
For granting their permission to work on their property, for placing 
at oiu- disposal the facilities of the fazendas, and for assistance in nu- 
merous other ways, we wish to thank the following persons: On 
Maraj6 Island — Sr. and Sra. Dionysio Bentes de Carvalho, Sr. 
Rodolfo Chermont, Sr., Sr. Rodolfo Chermont, Jr., Sr. Armando 
Teixeira, Sr. Lauro de Miranda Lobato, Sr. Raul Bittencourt, Capt. 
Am^ncio Antonio dos Santos, Sr. Raimundo Brito, Sr. Noe Xavier 
de Andrade, Sr. Fernando Teixeira; on Mexiana Island — the members 
of the Mexiana Cooperativa, especially Sr. Edgar Guama whose 
scientific interests permitted us to enjoy many evenings of pleasant 
conversation and whose cooperation was absolutely overwhelming, 
and Sra. Lelio Lobato for her friendly hospitality while at the main 
fazenda house; on Caviana Island — Sr. Dionysio Bentes de Carvalho, 
Coronel Lusignan Dias, Dr. Claudio Dias, Dr. Armando Morelli, Sr. 
Mario Lobato, Sr. Angelino Lobato, Sr. Antero da Silva Melo Filho, 
Sr. Benjamin Pinto e Sousa, Sr. and Sra. Tiburcio da Silva Melo, Sr, 
Nadir Pinto e Sousa, Sr. and Sra. Manoel Alves da Silva. 

The most extensive archeological investigations in the Amazon 
area are those made by Nimuendajii during the 1920's under the aus- 
pices of the Ethnographical Museum, Goteborg, Sweden. The un- 
pubhshed notes have been edited by Stig Ryden, supplemented with 
detailed descriptions of the specimens in Goteborg Museum collections. 
Ryden has generously made available to us this manuscript (Ryden, 
MS.) and photographs of the plates, and has granted permission to 
quote passages in this report. His cooperation has added considerable 
comparative information to the sections on Maraj6, Caviana, and the 
Territory of Amapd. It is a pleasure to acknowledge our indebted- 
ness to him. 

In the technical analysis of certain specimens, we wish to thank 
the following scientists for their cooperation, information, and expert 
opinions: Marshall T. Newman, Division of Physical Anthropology, 
United States National Museum, for the study of extremely fragmen- 
tary human bone material; Doris M. Cochi-an, Division of Reptiles 
and Amphibians, United States National Museum, for identification 
of reptile bones; Junius B. Bird, Department of Anthropology, 



XXVin BUREAU OF AMERICAN ETHNOLOGY [bull, ler 

American Museum of Natural History, for comments on clay frag- 
ments with cord impressions; Arthm* Woodward, Glenn A. Black, 
and Kenneth Kidd for analysis of the glass trade beads. 

There are two people whose exceptional contribution to the success 
of our work requires special acknowledgement. Mr. George T. Col- 
man, United States Consul at Belem, helped us immeasurably by 
seeing om' equipment through the customs and acting as intermediary 
in similar official situations at no little saving to us in time and mental 
anguish. The interest he and Mrs. Colman showed in us and om- work 
is typical of their enthusiasm for all things pertaining to the culture 
and people of Brazil. Peter Paul Hilbert, ethnologist of the Museu 
Paraense Emilio Goeldi in Belem, accompanied us on our second trip 
to Maraj6 Island, took his indoctrination into jQeld archeology with 
aplomb, and has since returned to the Island and done further work. 
His generosity in putting his data at our disposal is evident in the 
frequency with which he has been quoted in the Marajo section of this 
report, and it is difficult to express accurately the extent of our debt 
to him for this contribution. 

Finally, we would like to record our warm thanks to the caboclos who 
were our guides, our workmen, our companions, and frequently our 
hosts. They and their fellow Brazilians in all walks of life made our 
stay in Brazil so pleasant, as well as scientifically rewarding, that we 
are eager to return. 

The third and fourth sections of this report formed the nuclei of 
dissertations submitted to the Faculty of Political Science of Colum- 
bia University in partial satisfaction of the requirements for the de- 
gree of doctor of philosophy. Clifford Evans presented "The 
Archeology of the Territory of Amapd, Brazil (Brazilian Guiana)" 
in March 1950, and Betty J. Meggers presented "The Archeological 
Sequence on Maraj6 Island, Brazil" in February 1952. Now, both 
of these sections have been reorganized and partly rewritten for in- 
clusion in the final monograph, and all statements, conclusions, and 
interpretations included herein supersede any previously made which 
may slightly differ. 

B. J. M. 

C. E. 
Division of Archeology, 

United States National Museum, 
Smithsonian Institution, 
Washington, D. C, May 21, 1964 



ARCHEOLOGICAL INVESTIGATIONS AT THE 
MOUTH OF THE AMAZON 



By Betty J. Meggers and Clifford Evans 



INTRODUCTION 

BACKGROUND OF THE LOWER AMAZON ARCHEOLOGICAL 

EXPEDITION 

Prior to the introduction of extensive survey and stratigraphic 
excavation by trained archeologists, the interpretation of the arche- 
ology of an area must be based on the more elaborate and decorative 
pieces of pottery that have found their way into museums and on 
vague comments recorded by travelers in pursuit of adventure or by 
scientists after other kinds of information. This situation applied to 
the mouth of the Amazon prior to 1948. In the 19th century, Marajo 
Island in particular exerted a great fascination on numerous scientists 
as well as laymen. The Marajoara mounds were first recorded in the 
18th century by an anonymous visitor who was impressed by the 
weU-made vessels they contained. When the Amazon was undergoing 
exploration during the latter part of the 19th century by geologists, 
botanists, general naturalists, and laymen, these sites were frequently 
visited and examined. Among those who wrote detailed accounts of 
their activities and impressions are Jose Vieira Couto de Magalhaes, 
Domingo Soares Ferreira Penna, Joseph B. Steere, Orville A, Derby, 
Charles F. Hartt, and Ladislau Netto. In 1895 and 1896 Emilio 
Goeldi and Aureliano Lima Guedes conducted survey and excavation 
in the Territory of Amapa and reported the now well-loiown sites at 
Cunani and Maraca. Most of these men made collections of the more 
elaborate types of pottery and these were sent to museums in North 
America and Brazil. 

In the early decades of the 20th century, the mounds of Maraj6 
continued to be visited and excavated. Those who conducted the 
most extensive explorations represented two new categories of pro- 
fessional allegiance: JournaHsts, like Algot Lange who dug in Pacoval 
in 1913 and Desmond Holdridge who examined several mounds east 
of Lago Ai-ari in 1930; and anthropologists, including W. C. Farabee 
who made extensive excavations in 1914 at Fortaleza and in 1916 at 



2 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

the Camutins, Curt Nimuendajii who tested sites in the Cabo Maguari 
area in 1922, Heloisa Alberto Torres who visited Pacoval do Cururu 
in 1930, and Antonio Mordini who excavated at Teso dos Gentios in 
1926 and Panellas in 1928. It is unfortunate that none of these 
individuals has written a detailed account of his findings except Lange 
(1914), whose excavation technique is unreliable. Farabee left 
detailed field notes on some of his work, but they are largely rendered 
useless by the loss of the pottery identifications. However, he de- 
posited a large collection of complete vessels and a sample of sherds 
at the University Museum in Philadelphia. Thus, in spite of a 
long sequence of articles and numerous visits of inspection and even 
excavation, the descriptions of Marajoara Phase remains are so incom- 
plete and indefinite that they serve more to tantalize than to inform. 

Our interest in the archeological situation at the mouth of the 
Amazon dates from 1943, when Meggers began an analysis and inter- 
pretation of a small collection from Marajo Island made in 1871 by 
J. B. Steere, and deposited at the University Museums in Ann Arbor, 
Mich. (Meggers, 1947). This study revealed the meagerness and 
indefiniteness of the information on the Marajoara Phase sites and 
their contents, in spite of the relatively voluminous literature, and 
indicated that no reliable conclusions could be drawn without strati- 
graphic excavation. This conclusion was strengthened after a de- 
tailed examination was made of the larger and more representative 
Lange collection from Pacoval at the American Aluseum of Natural 
History in New York. A classification of the sherds by surface 
treatment (plain, slipped, double slipped) and decoration (incision, 
excision, painting) revealed a variety of types, some simple and others 
complex. The probable selectivity of the collection, coupled with 
Lange's method of excavation (p. 312), indicated any efforts to 
deduce temporal significance from the differences in decorative styles 
would be purely speculative. The further the study of these museum 
collections proceeded, the more obvious became the need for scientific 
fieldwork as a basis for the interpretation of the archeological remains 
at the mouth of the Amazon- 

With all these factors in mind the authors, then graduate students 
in the Department of Anthropology at Columbia University, drafted 
a program for Lower Amazon archeological exploration. The research 
problems were discussed in detail with Drs. Wm. Duncan Strong, 
Julian H. Steward, Charles Wagley, Wendell C. Bennett, and Gordon 
K. Willey, and the interest with which these individuals received the 
project led to correspondence with ofliicials in Brazil to investigate 
the possibility of their cooperation. Although the details were not 
fully completed by correspondence, sufficient encouragement was 
received from Dr. Heloisa Alberto Torres, then director of the Museu 



EVA^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 3 

Nacional in Eio de Janeiro, Brazil, to warrant proceeding with a 
formal application for research funds and with preparations for the 
expedition. Through the generosity of the Wenner-Gren Foundation 
for Anthropological Research, Inc. (at that time known as The 
Viking Fund, Inc.) of New York a joint research grant was obtained 
for a year of "Archaeological Study in the Lower Amazon, Brazil" 
from July 1, 1948 to July 1, 1949. Dr. Wm. Duncan Strong, then 
Chairman of the Department of Anthropology of Columbia University 
presented the research project to the authorities of the University 
with the result that further financial assistance was received in the 
form of a Wniiam Bayard Cutting Traveling Fellowship. Through 
Dr. Strong's cooperation we were permitted to work as representatives 
of the Department of Anthropology of Columbia University, an ar- 
rangement which aided our work immensely and facilitated our re- 
lations with Brazilian authorities. 

Beyond the procedures followed above to guarantee the financial 
and scientific support of the project, a number of difficulties in planning 
were encountered. In spite of the fact that we talked to several 
people who had spent considerable time in the tropical forest regions 
of South America, no one was able to ofi'er any concrete advice on 
such specific things as field equipment needs, or on the problems and 
possibilities of labor, transportation, etc., beyond the hmits of the 
main course of the Amazon or its major tributaries. Relying on 
Evans' prior experience in Peru, we consequently sent down many 
items that had seemed to be necessary field equipment, but which 
later proved completely useless or impractical in the lowland tropics 
and were shipped back to a central base at the first opportunity. 

ITINERARY 

We left Miami, Fla., on July 1, 1948, flying directly to Rio de 
Janeiro to complete official negotiations with the Brazilian authorities. 
Our arrival was preceded by that of Dr. and Mrs. Charles Wagley who 
had come to Brazil under UNESCO auspices to conduct research for 
the International Hylean Amazon Institute (Wagley, 1953). Not 
only did the Wagleys pave our way in Rio de Janeiro, but they 
preceded us to Belem, Para, where their announcement of our arrival 
made om* reception more cordial than it otherwise would have been. 
In Rio de Janeiro, Dr. Heloisa Alberto Torres, director of the Museu 
Nacional, offered us her complete scientific cooperation and allowed 
us to work under the auspices of the museum. In addition, her 
friendly aid, hospitality, and assistance expedited the acquisition 
of the necessary official papers with the result that on July 17 we 
flew north to Belem, the capital of the State of Para and the gateway 
to the Amazon. In spite of a handful of letters of introduction pro- 



4 BUREAU OF AMERICAN ETHNOLOGY [bull, lei 

vided by Dr. Torres, the contacts the Wagleys had made for us, 
and the cooperation of Dr. Machado Coelho, director of the Museu 
Goeldi, we were delayed for 6 weeks in town by problems surrounding 
the negotiation with officials and landowners for permission to under- 
take archeological investigations on their property. Since Brazilian 
federal law regarding antiquities or subsurface rights does not auto- 
matically grant permission to trespass nor does it actually protect 
the archeological sites, our itinerary was controlled by the willingness 
of the owners to cooperate and their preference as to when we should 
visit their property. 

A workable schedule was finally achieved and we left Belem 
August 20, 1948, by wood- burning steamer through the inland route 
to Chaves on the north coast of the Island of Maraj6. Chaves 
served as our first base of operations, from which excavations were 
conducted at Sites J-1 through J-5. On September 2, we moved 
eastward to Fazenda Santa Catarina, where a new base was estab- 
lished. From there and the various outstations of the Fazenda we 
worked imtil September 23, covering Sites J-6 through J-12. Since 
the owner-manager of Mexiana Island had requested that our visit 
coincide with his presence on the Island, we left Maraj6 Island at 
this time and established a new base of operations at Fazenda Nazar6 
on Mexiana. Sites M-1 through M-7 were studied while on Mexiana 
Island from September 24 to October 22 w^hen we moved to Caviana 
Island. A base of operations was estabhshed at Fazenda Sao Joao da 
Caridade and Sites C-1 through C-15 were excavated between 
October 22 and December 14. During this period w^e returned on 
November 19 to Chaves in order to examine further Site J-4. Before 
going back to Caviana Island on November 23 we went along the north 
coast of Maraj6 toward the east to examine Site J-1 3. Our baggage 
had become of some considerable size by this time, even though the 
Maraj<5 sherds had been shipped back to Belem earlier, and since we 
were unable to arrange successfully for direct transportation from 
Caviana to Macap^ in the Territory of Amapd, we hired a sailboat to 
bring all the specimens into Bel6m by way of the coastal route while we 
returned to Bel6m via the inland steamer which was to stop in Chaves 
on December 18 on its monthly trip. Ai'riving in Belem on December 
21, we made our headquarters in a house provided on the grounds of 
the Museu Paraense EmUio Goeldi, which was to serve not only as our 
living quarters but as an excellent laboratory space for storing and 
studying the numerous archeological specimens. After the baggage 
arrived from the islands, along with a general handyman and assistant, 
Benjamin Pinto e Sousa, the work was laid out in the laboratory 
so that he could wash and number all the specimens collected so far 



^yf^s]^ ^ND ARCHEOLOGY AT MOUTH OF AMAZON 5 

while we returned to the field for another month before the rainy 
season began. 

On January 2, 1949, we flew from Belem to Macapa, the capital of 
the Federal Territory of Amapa, where we had been invited to under- 
take archeological investigations by the Governor, Janary Gentil 
Nunes, Through his magnificent cooperation all the governmental 
facilities, including m.otor launches, trucks, airplanes, maps, archival 
records, and obscure reference books were put at our disposal so that 
our work in the Territory of Amapa would be facilitated as much as 
possible in the limited time available. We were accompanied on all 
our trips by Sr. Newton Wilson Cardoso, director of the newly form.ed 
Museu Territorial. As a result of some specimens brought back by a 
geologist, Fritz Ackermann, from the Rio Pigaca, we began with a 
survey of the Rio Vilanova and its tributaries. This resulted in data 
on Sites A-1 through A-6 and A-13. From January 15 to 21, we 
explored the Rio Araguari-Amapari without finding any sites. On 
January 22 we flew to Amap4 and worked at Sites A-7 through A-12 
until January 30, when we returned to Macapa. The remaining few 
days in Macapa were spent in getting data on Site A-14 and in 
photographing and taking notes on the various specimens that Sr. 
Cardoso had in his custody in the Museu Territorial. On February 4 
we returned to Belem by air and immediately moved into our house- 
laboratory on the Museu Goeldi grounds. 

The rainy season was now at its height and m^any of the specimen 
bags and labels showed such severe effects of mildew that some of 
the identifications were almost illegible. We began to work immedi- 
ately on the classification of the pottery while Sr. Benjamin Pinto e 
Sousa continued to wash and number the rest of the sherds. In 
addition to analyzing all our own sherd material and photographing 
all the complete specimens and representative samples of the pottery 
types, we also classified, described, and photographed all the speci- 
mens in the Museu Goeldi which had any sort of provenience data, 
as well as some specimens in the private collections of Sr. Frederico 
Barata and Sr. Fritz Ackermann. This work continued until May 5, 
when, although the rainy season lasted somewhat longer than was 
normal, we left for a final trip to the interior of Maraj6 Island to 
collect data on the elaborate Marajoara Phase burial mound complex. 

On May 6, accompanied by Peter Paul Hilbert, the ethnologist of 
the Aluseu Goeldi, we sailed from Belem to the center of Maraj6 
Island, making our headquarters at Fazenda Campo Limpo near the 
upper Rio Anajds. Sites J-14, J-lo (with 17 artificial mxounds), and 
J-16 were excavated in the area and we returned to Belem on May 23. 
The remaining time in Belem was spent in completing the analysis of 
the previously excavated materials as well as the nev/ly acquired 

391329—57 3 



6 BUREAU OF AMERICAN ETHNOLOGY [bdll. 167 

specimens from the Marajoara mound^cultures.". On June 23 we flew 
to Macapd for the day to deliver a talk on the results of our work in 
the Territory of Amapd. While there we found that Sr. Newton 
Wilson Cardoso had visited several more sites since working with us 
and had proved himself an apt pupil by taking accurate notes and 
keeping materials by site, as well as making some stratigraphic ex- 
cavations. He kindly allowed us to take this material back to Belem 
for classification and study. 

The last week or so in Belem was somewhat hectic. Not only did we 
have to pack our equipment, but the final details of the study of all 
the materials excavated had to be completed and a division had to be 
made into type collections to be left at the Museu Goeldi in Belem, 
the Museu Territorial in Macapd, the Museu Nacional in Rio de 
Janeiro, and smaller samples to be exported for distribution to muse- 
ums in the United States.^ On July 1, we flew from Belem to Belo 
Horizonte, Minas Gerais, where for several days we visited the caves 
of the Lagoa Santa region with Mr. H. V. Walter and Sr. Josephat 
Paula Penna (Evans, 1950). On July 4 we continued to Rio de Ja- 
neiro to close our ofiicial business with the Museu Nacional and to 
report to Dr. Heloisa Alberto Torres the progress of our year's field- 
work in the Amazon. After paying respects to the many friends we 
had made in both Rio de Janeu'o and in the north, we left Brazil by 
air arriving in New York on July 14, 1949. 

PROBLEMS AND COMPROMISES IN FIELD TECHNIQUE 

Archeology in the tropical forest of South America presents, in 
addition to the usual problems, many diflSculties that are not en- 
countered in the more arid or more accessible parts of the New World. 
Manuals of field procedure and precision methods of excavation 
technique frequently cannot be followed, and the field situation must 
be met with an understanding of what is pertinent and what is unprofit- 
able in order to gain the maximum of information in the shortest possi- 
ble time. Othermse, one could easily spend a full year in the field and 
have very little to show for it. This we learned, however, only by 
experience. For the benefit of those who may follow us, we will 
outline briefly some of the major problems and compromises. 

Evans, who had recently returned from 9 months of fieldwork on the 
coast of Peru, superintended the assembling of the field equipment. 
We included all those items that had been essential or helpful in that 
work, and some of these proved to be even more important in the 
tropical environment, particularly specimen bags of unbleached muslin, 

> These type collections have been deposited at the United States National Museum, the University 
Museum in Philadelphia, the American Museum of Natural History, the Peabody Museum of Yale Uni- 
versity, the Museum of Anthropology at the University of Michigan, and the Museum of Anthropology of 
the University of California. 



Ev^ANsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 7 

linen tags, and duplicate sets of field notes. In the hope that aerial 
photographs might reveal the location of the Marajoara mounds as 
they do ruins on the coast of Peru, we secured permission to examine 
those taken of Maraj6 Island by the United States Air Force, going to 
some difficulty since they were still classified as "confidential." The 
results were highly disappointing because the forest growth obliterated 
all but the most abrupt and extreme alterations in the terrain. Later, 
after experience on the ground and in low elevation flights over the 
savanna and jungle, we became fuUy convinced that aerial photog- 
raphy has nothing to offer as a means of locating archeological sites 
in the Amazon area. However, we derived one important benefit 
from the aerial survey of the Amazon. As a result of this work, the 
Aeronautical Chart Service of the United States Army Air Force has 
been able to revise and correct its World Aeronautical Charts to such 
a degree of accuracy that when on the ground we could follow each 
bend and curve of all but the smallest streams. From the standpoint 
of a more useful scale for groundwork, the Aeronautical Chart Service 
made available the Preliminary Work Sheets, Scale 1:500,000, from 
which the final copies of the World Aeronautical Charts, Scale 
1 :1, 000,000, are made. Not only did these maps save us considerable 
time and trouble, but they gave a degree of accuracy to the site loca- 
tions and the geographical features of the area that otherwise could 
not have been attained. 

In addition to these excellent maps, we took along surveying equip- 
ment, such as a plane table and tripod, alidade, and stadia rod, on the 
assumption these materials would be useful in the mapping of each 
archeological site. Two weeks in the field demonstrated that not only 
it was impossible to carry around this equipment, but also it was com- 
pletely nonfunctional for several reasons: (1) Generally, the sites were 
not large enough or with enough surface features to warrant the use 
of the alidade and plane table; (2) to sight a line through vegetation 
required a cutting operation that was not economically feasible or 
practical; (3) a sufficiently accurate map could be obtained with greater 
ease and in less time with grid paper, a compass, a tape, and a hand 
level. In other words, we made a compromise in technique here 
because if we had not done so we would have had to sacrifice results in 
terms of the number of sites we could examine and we are convinced 
that a site map so derived would show no more pertinent information 
than is now available on our various plans. 

Field technique must be adaptable to the situation so that the most 
scientific data can be obtained under the peculiar local circumstances. 
To demonstrate the point, it is pertinent to mention why we used the 
system of sinking several smaU strata cuts, generally 1.5 by 1.5 meters, 
into various sections of the site instead of digging a long trench or a 



8 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

larger, single strata cut. Again, the site situation is the determining 
factor. In all the sites of the Tropical Forest level of culture the accu- 
mulation of refuse is too shallow to make it essential to dig a large cut 
to provide sufficient space to throw out the dirt as the cut increases 
in depth. The nature of the refuse makes it more functional to place 
several small cuts in various parts of a site in order to test more of 
the occupation zone. Trenches are not feasible because of the quan- 
tity of trees and their root systems that cover most sites. A series of 
small cuts can be finished in a limited time going from the surface to 
sterile, whereas a larger excavation might not reach completion in the 
time available at certain sites. The question has been raised as to 
why we did not trench the large Maraioara mounds in two directions. 
Our answer is threefold: (1) Property owners were extremely hesitant 
to allow any digging in theu* mounds because they offer the only high 
ground for their cattle during the wet season and also because they 
do not want their "treasure" disturbed. Permission to dig even small 
test holes was difficult to obtain and permission to trench the mounds 
would never have been granted. (2) Sufficient labor for extensive 
trenching activities was not available at any cost. (3) Scientifically 
speaking, it was far more important to test several sites than restrict 
ourselves by extensive excavation on one site. This approach has per- 
mitted us to evaluate and interpret more extensive digging conducted 
earlier by Farabee (MS., 1914) and others. 

For those who have not had the fortune or misfortune, depending on 
one's viewpoint, to work in Amazon archeology, the tremendous 
problem created by roots cannot be overemphasized. Although the 
cuts were laid out originally with square sides, the first layer of dig- 
ging always produced roots that often caused a slight modification of 
shape; however, the area covered in each cut was always well con- 
trolled. Not only was it impossible to polish the walls of strata cuts, 
but if they had been polished they would have shown no details. The 
intense rainfall, high humidity, and easily leached soil take out any 
materials that would make a clear-cut line of strata distinguishable on 
the walls of cuts. In the artificial mounds of the Marajoara Phase, 
soil conditions did vary and here it was possible to smooth the walls 
of the cut sufficiently to plot the various featiu-es. In all the other 
sites the excavation technique was careful and well controlled, but 
not carried to the point of diminishing returns by trying to follow out 
preconceived ideas that no strata cut is properly executed unless the 
ritual of polishing and smoothing is faithfully carried out. In other 
words, the entire excavation technique in Amazon archeology can 
be summarized in a few words: not once was technique abandoned 
because of a lack of interest, nor Vas it modified to the extent that 
the data obtained would be unreliable; but it was necessary at all 



f^?S^^^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 9 

times to be realistic about technique and to apply the method to the 
peculiar local situation that would bring results, rather than blindly- 
become a slave to technique irrespective of the total results. 

Kain and humidity create problems that can only be appreciated if 
one has tried to work in a tropical forest situation in part of the 
rainy season. Granted, we stayed in Belem classifying our materials 
at the height of the rainy season, but some of the fieldwork had to be 
conducted during this part of the year. Tarpaulins were used to 
cover the excavations to keep them from filling with water during a 
downpour, but even then we were digging in mud. Sm*vey trips in 
dugouts up streams and rivers, going from intense sun one hour to a 
heavy shower the next, made it difficult to keep notes and photo- 
graphic equipment dry. Cameras and film had to be kept in airtight 
cans dehydrated with silica gel. The intense rainfall in the Amazon 
not only creates physical hazards that restrict the work and actually 
makes it impossible to undertake archeological fieldwork during 
February, March, and April, but it reduces the archeological evidence 
to objects of stone or pottery. Postholes, matting, thatching, and 
other details of house construction are so quicldy destroyed by decay 
that unless the posts burned (apparently an extremely rare situation) 
there is absolutely no evidence of such features. Proof of this factor 
is easily obtained by digging on the site of a former caboclo house 
where the exact position is known. If over 5 years have passed, the 
area has passed into secondary growth, posts and postholes have 
disappeared, all decayed vegetable matter has been leached out or 
washed away and except for areas darkened with charcoal or ashes 
there is no sign of the occupation other than occasional broken arti- 
facts. Bone materials destroy rapidly even in secondary urn burials. 
Except in those urns in which the water supply was constant (in other 
words the jar broke and was filled by rain or by seepage and remained 
moist thi'oughout the dry season) or where the urn and its lid had kept 
the contents constantly dry, bone has turned to dust, usually dis- 
tinguishable only as fine, white flecks in light gray to black soil. The 
few bone scraps we were able to salvage were in extremely poor 
condition and had to be treated with a dehydrating agent mixed with 
a stabilizing cement, such as acetone and duco or acetone and ambroid. 

None of the problems and compromises mentioned so far have been 
dictated by another situation inherent in the Amazon area, one that 
has a decided effect on the method of carrying out the fieldwork — the 
lack of modern transportation facilities and the sparse population. 
In spite of all the modern mechanical aids to mankind, one is reduced 
to the necessity of utilizing the primitive, local means of transporta- 
tion. More than once after a slow and difficult dugout trip we wished 
for an outboard motor, but there were many other situations in 



10 BUREAU OF AMERICAN ETHNOLOGY [boll, ler 

which paddling in a dugout was 100 percent more practical than travel- 
ing by outboard motor. To use motors it is necessary to haul all the 
gasoline from a main base and establish caches of fuel. To do this 
would involve organization and planning of supplies that would be 
more time consuming and frustrating in the long run than the use of 
local transportation. By taking advantage of the larger sailboats, 
sailboats with auxiliary motors, or launches to traverse some of the 
longer distances, traveling light when going by dugout, and depositing 
our collections and main equipment at various bases, we were able to 
reach all areas fairly easily. Those who have never traveled in the 
interior of the Amazon, along the smaller streams where only a hunter, 
wood cutter, or rubber cutter might live, sometimes find it diflBcult 
to understand the importance of the dugout as a means of transporta- 
tion. Not only is it a sturdy craft, capable of taking a lot of punish- 
ment from submerged debris, but it is quite stable, easy to propel and 
will hold a fairly large load. Nevertheless, in archeology more than 
once we had to keep in mind the fact that, although we went "empty 
handed" to a site except for a few digging tools, specimen bags, and 
photographic equipment, we always returned laden down with sherds. 
This is not to say that we now feel that our final results have suffered 
as a result of the limitation of transportation (after completion of the 
study, we have only one site that we feel could be better interpreted 
with another day's work), but several times we had to take into con- 
sideration the fact that another bag of surface material or another 
test excavation would be out of the question because of the lack of 
hands to carry the resulting sherds back to the dugout, or because the 
waterline of the dugout would be lowered below the margin of safety. 
Archeology in the Amazon is not like that of areas of the world where 
one can drive to the site, load the car down, and then drive back again 
if necessary. 

Fortunately, the cultures are simple, the sites are small, and a maxi- 
mum of data can be secured with a minimum of digging. The prob- 
lem of labor in the Amazon is much more severe than in many other 
parts of the New World. Most of the people live by working on cattle 
ranches, cutting wood, gathering rubber, or by hunting. Miles and 
miles of rivers and streams can be traveled without seeing any human 
habitation. Local labor is consequently not available iu quantity. 
If one had the financial resources to buy a boat large enough to house 
a crew of men and to transport food for this crew, then labor could be 
brought from Belem to the interior. However, this is not practical 
for many reasons. The expense of such a project would be prohibitive 
on the budget of most New World archeological expeditions; the 
laborers would not be familiar with the local situation and it would 
still be necessary to hii-e local guides; permission from landowners to 



Jf,f,*\?fP ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 11 

trespass with such a large crew would not be easy to obtain ; transporta- 
tion of such a large crew to more remote sites would overtax the 
available facilities. As a result of this acute labor shortage and the 
necessity of constantly changing guides as we moved from one area to 
another, we found ourselves doing a larger part of the actual excava- 
tion work than would normally be expected. Only during the first 
month on Maraj6 did we have the same workmen long enough for us 
to train them to work in a strata cut. Otherwise the guide went 
hunting or dug in another part of the site for sherds to increase the 
sample from the site. In the long run, however, such a system means 
that one does not have to question the data when analysis might 
suggest inadequate or careless excavation technique that can so often 
be blamed on an inexperienced crew. 

Looking back on the Amazon situation and having the benefit of a 
second tropical forest expedition (Evans and Meggers, MS.) behind 
us before writing this introduction, we believe we have found the 
equipment best suited for South American tropical forest archeology, 
as well as developed the ability to travel light with a minimum of 
unessential equipment but with a maximum of protection for such 
things as cameras, exposed film, notes, etc. In spite of this we still 
have no general solution for the problems of transportation. Re- 
gardless of how much planning is done beforehand or how much 
money one has available, there is no way to avoid traveling by foot, 
by horse, by bullock, by dugout, and by sailboat, even though oc- 
casionally the airplane, jeep, truck, car, outboard motor or launch 
may be thrown in for the sake of variety. In other words, the local 
situation frequently cannot be predicted. One might carry an out- 
board motor and gasoline for weeks and then discover that the local 
conditions of a particular stream make use of the motor impossible; 
and paddling a dugout the only resort. 

There is one universal fact, however, and that is the contribution of 
the local guide to the success of South American tropical forest arche- 
ology. This guide is essential and invaluable not only because of 
his knowledge of the location of sites along a specific stream and in 
the adjoining area, because of his hospitality, his ability to obtain 
extra fish and game to supplement the food supply, and the use of 
his thatched shelter as a base, but also from the standpoint of his 
intimate knowledge of the local custom, the local problems, the 
local geographic features, and most of all for his ability to arrange 
for an extra helper, a dugout, an extra set of paddles or whatever else 
might be needed. A good guide can anticipate the archeologist's 
requirements and make archeology in the Amazon more than just 
hard work. 



12 BUREAU OF AMERICAN ETHNOLOGY [bum.. 167 

ORGANIZATION OF THE REPORT 

Archeologists attempting to organize the description of a variety 
of cultures occupj'ing a relatively large geographical area that is 
broken into several well-defined units are faced with various prob- 
lems in determining the best method of presentation. They must 
find a logical order that will meet the needs of the rare student or 
specialist who will read from cover to cover in the proper direction, 
and they must also anticipate and attempt to provide for the larger 
audience that will proceed in reverse order and frequently never 
probe deeper than the general conclusions. The organization of this 
report attempts to meet the requirements of both types of readers. 

Analysis of the archeological remains brought out the fact that the 
natural geographical divisions were correlated with clearly defined 
cultural boundaries. A twofold separation into the mainland (Ter- 
ritory of Amapa) and the islands (Marajo, Mexiana, and Caviana) 
was therefore not simply an arbitrary convenience but rather an aid to 
the understanding of the archeological sequences. Further subdivision 
of the islands into Maraj6 on the one hand and Mexiana and Cavi- 
ana on the other was warranted by the widely different roles played 
in local prehistory. Within each of these areal divisions the geo- 
graphical description is followed by the discussion of the archeological 
cultures in chronological order. A uniform outline was employed to 
assure equal coverage and to facilitate comparison. In addition to 
the details of the sites and their excavation and the analysis and 
description of the artifacts, this gives a summary of any information 
from other investigations and concludes with a condensation of the 
diagnostic features of the cultxu-e as represented by or deduced from 
the archeological remains. Each geographical unit closes with a 
detailed analysis of the evidence for the chronologica,l position of the 
cultures in the sequence and of their probable affiliations. 

One culture, the Arua, is found on all three of the islands and on 
the mainland, and after considerable deliberation v\^e decided to treat 
it in the following manner. The details of site description and ex- 
cavation, and the information from other investigations are given 
separately in each of the areas; data on Arua sites in the Territory 
of Amapa are included in that section and similar information on 
sites on Marajo are given under that island. Since the majority of 
the sites of this culture are located on Mexiana and Caviana, and 
since the Arua is the only pottery-making group to have dominated 
those islands, the analysis of the pottery and other artifacts, the sum- 
mary of diagnostic features, and the detailed interpretation of the 
culture are given in this part of the report. 

One of the disadvantages of archeology in the tropical forest is that 
the climate soon disposes of all but the most durable remains, in 



EVANsf^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 13 

this case objects of pottery and stone. As a result, any attempt at 
reconstruction of the cultural pattern must be based on knowledge of 
living cultures. Since we have made the effort to secure the maxi- 
mum amount of information from the archeological remains, and 
have drawn heavily on ethnographic clues for this purpose, the re- 
port begins with a summary of the Tropical Forest Pattern of cultm-e, 
emphasizing material traits and generalizing sociopolitical and re- 
ligious aspects. This is followed by a notation as to which of these 
traits might be discerned in the archeology. Finally, the significance 
of the environment in the formation and stabilization of this type of 
cultural adjustment is discussed. The evidence that environment 
has played an important part in producing the Tropical Forest Pat- 
tern justifies the rather detailed environmental descriptions that 
introduce each geographical section. 

The historical information, including location and description of the 
tribes at the mouth of the Amazon in the early postcontact period, 
has been placed after the archeological evidence for two reasons: (1) In 
this position it follows the general chronological order of the report, 
which is from early to late, and (2) it supplements the archeological 
remains but in turn is subject to verification or evaluation in terms 
of the archeological picture. This section includes the chronology 
of conquest and settlement, the information recorded about the 
aboriginal cultures, and an analysis of the amount of correspondence 
between this and the evidence from archeology. 

Since this is the first report of survey and excavation in the Tropical 
Forest area of South America, it has been necessary to describe in 
detail all of the sites and the cultural remains as a basis for future 
work. We have tried to reduce repetition to a minimum and to keep 
the detailed descriptions separate from the interpretations that are 
derived from them. This makes it possible, we hope, for any reader 
interested only in the major conclusions to satisfy himself with as 
much or as little specific information as he desu'es. Such a person 
can begin with the chapter on "Implications of the Cultural Sequence 
at the Mouth of the Amazon" and if he desires documentation he can 
turn to the conclusions and interpretations at the end of each of the 
geographical sections. If his interest is sufficiently stimulated, he can 
pursue the facts as far as he wishes. It must be emphasized, however, 
that the critic cannot fairly attack any theories or interpretations 
given in these chapters without delving deeper into the report and 
examining the supporting data on which they are based. 

The only term in the report that warrants some explanation is 
om* use of "Phase" when referring to our various archeological com- 
plexes. Phase has been used to designate distinct archeological 
cultures with a definite geographical distribution and persistence 



14 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

through time. Although this parallels, in a general way, the modified 
terminology of the Midwestern Taxonomic System (Cole and Deuel, 
1937), it is not an attempt to introduce this system to the Amazon 
region, where the archeological situation is not sufficiently well known 
as yet to warrant its use. The term "Phase" was selected instead 
of tribe, group, culture, complex, etc., because it carries absolutely 
no ethnological connotation. At present there is no way of deter- 
mining whether each of the archeological Phases corresponds to one 
tribe or several, or whether two Phases correspond to a single tribe.^ 

In addition to limiting the cultm'al reconstruction, the conditions 
of preservation in the tropical forest place difficulties in the way of 
arriving at temporal evaluations. Unfortunately, insufficient un- 
contaminated charcoal was found to make Carbon 14 techniques 
applicable. In an attempt to compensate for this, we have tried to 
establish a time sequence by developing formulas for calculating the 
rate of refuse accumulation in the archeological village sites (pp. 245 flP.). 
The results are admittedly tentative and before the system can be 
considered reliable there will have to be further check of the formulas 
in other South American Tropical Forest situations and particularly 
in ethnographic village sites. 

A few words should be said about the pottery type descriptions. 
We have not considered all the variations in the ceramic complex of a 
Phase as independent. Instead, we have recognized the plain wares as 
primary and the decorated types as the result of applying ornamenta- 
tion to the surface of a minor proportion of one or more of the plain 
wares. This approach is revealed in the pottery type descriptions 
by the absence of complete details on paste, temper, textiu-e, and 
surface treatment under each of the decorated types; instead, the 
reader is referred to the plain ware or wares on which the decoration 
was placed for these details. For example, Anauerapucii Incised 
designs always occur on Mazagao Plain paste. Since the details of 
paste and surface are the same in both these types, they are given in 
the plain type description only. This procedure was followed in the 
interest of emphasizing the interrelationships between the pottery 
types within a Phase. In the Marajoara Phase, a tabulation of the 
decorated sherds according to the plain ware on which the decoration 
was placed permitted the use of smaU, selected samples for seriation 
and made it possible to secm^e a relative date on sites that would be 
undatable otherwise (pp. 386-388). 

In naming the pottery types, a consistent method was followed, 
which deviates from that used in some other areas. The first term is 
a proper noun, either the name of a major site of the Phase or of some 

> In an article that appeared as this report was submitted for publication, Phillips and Willey (1953) 
recommend the use of the term " phase" in this manner' 



evTns?^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 15 

geographical feature or landmark in the region of distribution of the 
sites belonging to that Phase. The second word is descriptive and 
distinguishes decorated from undecorated surfaces. An undecorated 
surface is described as "plain," rather than as "orange," "white," 
"brown," or "gray" as is sometimes done. The use of a color term 
signifies a slip or paint, as in the case of "Carmelo Red," "Arari Red 
Excised," and "Anajas White Incised" of the Marajoara Phase. 

Occasional applique, modeling, or punctation has not been singled 
out for separate consideration as a decorated pottery type in any of the 
archeological Phases at the mouth of the Amazon because the occur- 
rence is too restricted or sporadic to be of temporal or cultm-al signifi- 
cance. Applique reaches an appreciable frequency only in the Arua 
Phase, but the fluctuation from site to site makes it of no value for 
seriation (see fig. 201). Although change in frequency through time 
was not the only criterion used in deciding whether or not a variant in 
the ceramic complex should be emphasized by making it a pottery 
type, this was an important consideration in doubtful cases. When a 
separate pottery type did not seem warranted, such specimens were 
described either as "Unclassified Decorated" or as occasional orna- 
mentation of the plain ware. 

The observant reader of the pottery type descriptions will notice 
a slight variation in the format of the vessel shape descriptions. This 
is the result of a friendly disagreement between the authors as to the 
most useful method of presenting the information and not of editorial 
oversight. The same reader will note that the drawings of the rim 
profiles have been rendered solid black for plain wares and in outline 
for decorated types to make them distinguishable at a glance. 

Each site is designated by a key letter and a number, in addition to 
the local name. The letter indicates the geographical region in which 
the site is located (A — Territory of Amapa; C — Caviana; J — Maraj6, 
formerly called Joanes; and M — Mexiana) and the number, the 
particular site. This system, which permits the addition of future 
sites in each area, has been followed in other parts of the New World. 
It is especially convenient for designating sites in foreign countries 
where the local names are often difficult for Americans to pronounce, 
much less remember. 

Throughout this report the authors have kept in mind that the ac- 
quisition of archeological data has one main purpose : to reconstruct the 
cultures of the past and their interrelations through history. For 
this reason, we have attempted to revitalize the dead fragments of the 
cultures we have found and to resurrect some semblance of their 
former, living condition. Some of our more conservative colleagues 
may^ object to our efforts to reconstruct the social organization, the 
evolutionary development or decline of certain cultures, or to see the 



16 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

various archeological Phases as expressions of different cultures 
adapting themselves in various ways to a tropical forest situation, but 
it is our sincere hope that these interpretations will not only make the 
report more useful to anthropologists as a whole, but will demonstrate 
that archeologists can do more than just accumulate bushels of 
potsherds. 



TROPICAL FOREST CULTURE 

ETHNOGRAPHIC DEFINITION OF TROPICAL FOREST CULTURE » 

To be in a position to evaluate properly the interpretations and 
conclusions reached about the archeological cultures of the mouth of 
the Amazon, it is necessary for the reader to be familiar with the evi- 
dence on which they are based. This includes not only the archeo- 
logical material but also the ethnographical details that are character- 
istic of the Tropical Forest Pattern, since these constitute one of the 
bases for the recognition of the Marajoara Phase as something unasual 
and distinct from the general uniformity of sequence from past to 
present. 

Since the Amazon forest has held similar potentialities and limita- 
tions for human adaptation as long as man has been a resident of the 
South American Continent, a basic general consistency of culture 
through time might be expected. Another reason for this belief is the 
uniformity in general features that is characteristic of cultures in the 
Tropical Forest today. This similarity results from the necessity 
for securing a living under similar conditions of food supply, natural 
resources, and other aspects of the environment that encroach upon 
men in their daily lives, and the ease with which useful mventions and 
discoveries may be swept along the innumerable waterways to be in- 
corporated into the cultures of distant tribes. Underlining the dom- 
inant role of the envu^onment in channeling the cultural adjustment is 
the characteristically wide variation between Tropical Forest cultures 
in traits of no survival significance, such as types of body adornment, 
methods of disposal of the dead, and observances surrounding birth, 
puberty, and death. 

Tropical Forest culture, as distinguished in the Handbook of South 
American Indians (Steward, editor, 1946-50), is both a cultural area 
and a level of cultural development. In the former capacity, it is a 
cultural complex based on "the cultivation of tropical root crops, 
especially bitter manioc; effective river craft; the use of hammocks as 
beds; and the manufacture of pottery" (Lowie, 1948, p. 1), which 
occupies the immense Amazon drainage bounded on the north by the 
Orinoco and its tributary the Guaviare, on the west by the Andean 
highlands, on the south by the Chaco and on the east by the Matto 

« Throughout this section, unless otherwise noted, the data presented are taken from the various articles in 
The Handbook of South American Indians, Volume 3: The Tropical Forest Tribes (Steward, editor, 194&- 
50). 

17 



18 BUREAU OF AMERICAN ETHNOLOGY [bull, m 

Grosso uplands and the Atlantic Ocean. A smaller concentration 
occurs in a strip along the Atlantic coast, south to the present bound- 
ary of Uruguay and inland as far as riverine and tropical forest condi- 
tions exist (see Steward, 1946-50, vol. 3, map 1). 

As a level of cultural development. Tropical Forest culture is inter- 
mediate between the Marginals, nomadic hunters and gatherers of 
wild foods, and the class-divided, occupationally specialized Circum- 
Caribbean and Andean peoples living by permanently productive 
agriculture. Less efficient Tropical Forest agriculture makes possible 
a semisedentary type of life, with concomitant possibilities for amassing 
material possessions, but is not profitable enough to remove the neces- 
sity for constant exploitation of the wild resources of the forest and 
the streams, or to permit the concentration of population and the 
occupational division of labor prerequisite to the development of a 
more formalized system of social and political control. The result is 
that, whereas the Tropical Forest Pattern verges toward the Circum- 
Caribbean in its material cultural inventory, it more closely resembles 
the Marginal Pattern in its social organization and religious develop- 
ment. 

Because archeological remains are slim, an understanding of the 
present cultural pattern is helpful as a guide to achieving the fullest 
interpretation of the clues from the past and to visualizing the range 
of adjustment probably characteristic in prehistoric as well as in his- 
toric times. In the brief description that follows, settlement pattern 
will be given more emphasis than usual because it is one of the few 
aspects of culture that can be described almost as fully for extinct as 
for living cultures. 

Agriculture. — A variety of plants is cultivated by living tribes in 
the Tropical Forest area, with some regional variation and other 
recent modifications brought about by the introduction of Old World 
crops like bananas and sugarcane. Of primary aboriginal significance 
were the root crops, with bitter and sweet manioc as staples and the 
sweetpotato, cara, and arrowroot also widely grown. Beans were 
raised in the western part of the area, but seem not to have been 
introduced into the Guianas until post-European times. Maize was 
everywhere of secondary importance. Palms and fruit trees, some- 
times planted but more often exploited in the wild, include papaya, 
guava, ingd, genipapo, avocado, castanha (Brazil nut), cupuassu, gua- 
rand, manga, assai, and other palms. Tobacco, calabashes, and urucu 
(for dye) were among the nonfood crops. 

Fields were located in the vicinity of the settlement or scattered in 
the sm-rounding forest where conditions of soil and drainage were 
suitable. Size is variable: a Ym-acare field was 10 by 500 meters 
(M6traux, 1948 c, p. 487), an Amanaye field, 912 by 1,188 meters 



^J.*l?o?^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 19 

EVANS J 

(Nimuendajii and M6traux, 1948, p. 200). The Tacanan clearings 
average 50 by 20 meters (Metraux, 1948b, p. 381), and those in the 
Guianas about 4,025 to 8,050 square meters (Gillin, 1948, p. 825). 
Before the introduction of iron axes, trees were cut by alternately 
charring the trunk with fire and cutting away the burned wood with 
stone axes. The Mojo either waited for a strong wind to topple the 
girdled trees or selected for felling those whose fall would carry ad- 
jacent ones with them. After drying, the brush was burned and the 
crops planted between the charred stumps and unconsumed trunks. 
Clearing a new field was often a collective undertaking, in which the 
owner rewarded his helpers with a feast on its completion. 

Because of rapid soil exhaustion, new fields were constantly being 
cut in the Guianas and were required everywhere each 2 to 3 years 
at the most. Tapirape fields were planted for 2 years, the second- 
year crop confined to manioc (Wagley and Galvao, 1948, p. 168) . The 
Cubeo situation is typical: 

The periodic exhaustion of the soil by manioc produces a seminomadic tribal 
life. But mobility is limited not only by tribal boundaries but by the necessity of 
maintaining contact with the gardens nearing exhaustion. To avoid abrupt 
transitions, the Indians select a new site not too far from the old one and begin 
to plant it many months before moving. They continue to harvest the abandoned 
gardens until the entire new crop has reached maturity 8 months to a year later. 
Abandoned fields are reputedly not replanted, although their owners may con- 
tinue to harvest the fruit trees for a considerable time. Presumably according 
to its quality, the soil is exhausted in 3 to 5 years. [Goldman, 1948, p. 770.] 

The yield of a typical garden may be judged by the fact that a Cubeo 
field of about an acre produces approximately 5 tons of manioc a 
year. This is harvested at an average rate of 25 pounds a day and 
converted into flat cakes, one of which lasts an adult 2 days if supple- 
mented with other food. Manioc gardening and the preparation of 
the plant for eating consume 75 percent of a woman's time (ibid.). 

Hunting. — The variety of bird and animal life made hunting impor- 
tant, but the paucity of large land mammals made it more time-con- 
suming than in forested areas in other parts of the world. Blowguns 
with poisoned darts, spears, and bows and arrows tipped with bamboo, 
hardwood, bone, or sting ray barbs were the major weapons, but 
traps, nets, and deception were also employed. The Indians of the 
Guianas, a typical example, 

manifest virtually all the tricks adaptable to their fauna. They imitate the call 
of the tapir, deer, monkeys, and birds to allay their suspicions; stalk deer; fire 
the savanna grass and encircle large game in communal drives; dig out armadillos 
from their burrows; or lie in ambush, screened by a shelter built on the ground 
or in a tree. [Lowie, 1948, p. 10.] 

Fishing. — The Amazon network of rivers and streams provides a 
constant and plentiful food supply that was thoroughly exploited by 
the Tropical Forest peoples, as well as by the Marginals. Numerous 



20 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

and varied techniques were used, of which drugging was perhaps the 
most productive. For this purpose, over 100 species of narcotic 
plants were used. Spearing, shooting with bow and arrow, and 
captui'ing in traps and weirs were also common methods. Not only 
fish, but turtles, caymans, frogs, manatees, and turtle and cayman 
eggs were utilized. 

Forest jproducts. — The forest was a source not only of food, but of 
most of the other adjuncts of life. Woods for stools and mortars; 
fibers for baskets, hammocks, mats and lashings; reeds for arrow 
shafts; materials for house manufacture, poisons, medicinal plants, 
oils, and resins were only a few of the products gathered. Often one 
plant yielded m.aterials for many uses, like the huriti palm, which was 
a kind of "country store" for the Warrau, providing 
leaves for roofing houses, fibers for thread, and rope used to make hammocks, 
edible pith, materials for sandals from the leaf sheath, conelike fruits regarded 
as a confection when soaked in water, sap for the manufacture of an alcoholic 
drink, and the edible larvae of a beetle. [Gilliu, 1948, p. 826.] 

Settlement pattern. — The riverine environment of the tropical forest 
presents two basic choices for village location: away from the river or 
along the shore. Considerations of defense, elevation, and proximity 
to food sources contribute to the selection of the site. The Caraja, 
for example, feel that it is preferable to be closer to fishing grounds than 
to gardens, and build on a high bank overlooking the river. This 
location is also chosen by the Mura, Apiaca, and tribes in the Uaupes 
and Montana regions. JIvaro settlements are on a steep hill at the 
head of a stream. 

Other tribes, who favor the depths of the forest, also do so for 
reasons of a subsistence nature. Gillin observed this in the case of 
the Barama River Caribs : 

Successful hunting requires a wide range of virgin forest on all sides, a territory 
in which the hunters are not handicapped by competition from neighboring vil- 
lagers or passersby on the river. Furthermore, it is the practice to locate cassava 
fields on hills or slopes in order to facilitate drainage of the soil. Suitable facili- 
ties for natural drainage are most often found at some distance from the river. 
[Gillin, 1936, p. 31.] 

Tribes who shun the river include the Encabello, whose villages are 
4 to 9 km. away, the Awishira, 9 to 18 km. away, and tribes of the 
Upper Xingu, 3 km. away. A nearby creek provides the domestic 
water supply and a path gives access to the river. Land above flood 
level is almost universally chosen, but the Omagua often settled on 
islands, beaches, or lowlands likely to be inundated, and in this 
respect they are more comparable to the pre-European inhabitants 
of the Island of Maraj6. 

Houses were of two fundamental types, communal and single-family, 
of which the communal type is predominant. It varies from the small 



™^lf^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 21 

structures (about 4 X 10 meters) housing 3 to 8 families, character- 
istic of the Omagua and Tapirape, through somewhat larger dwell- 
ings of the Aroana (18.2 X 6.1 meters), the Witoto (10 X 20 meters), 
the Parintintin (20 meters long), and the Jivaro (13 X 26 meters) 
to the immense structures of the Tupinambd (up to 150 meters long), 
the Awishira (22.5 X 90 meters), and the Apiacd, which sheltered sev- 
eral hundred people. Details of construction also have a wide range, 
including circular, rectangular, and elliptical floor plan; conical, 
gabled or arched roof; thatched or open sides; the interior unparti- 
tioned or divided by mats into family compartments. Individual 
family houses were characteristic of the Tupl-Cawahib (3.5-5.5 meters 
square), tribes of the Guianas and the Montana, and the Encabello 
(sometimes occupied by two families). Pile dwellings are built by 
the Warrau and Tucuna when on inundated sites; otherwise the floor 
is of packed earth. 

Village composition is variable and not coordinated with the nature 
of the house, except where a single communal house constitutes the 
village, as is characteristic in the upper Amazon. Otherwise both 
types may be arranged in a circle around a central plaza, in rows or 
haphazardly scattered in the clearing. The dimensions of the clear- 
ing are rarely recorded, but in one Barama River Carib village of 
half a dozen houses it measured 206 by 136 feet and was roughly 
elliptical in outline (GilUn, 1936, p. 101). 

In population, the Tropical Forest villages run the gamut from two 
or three families (Chimane) to more than a thousand individuals 
(Tiipinambd). The majority contain under 200 people, housed in 
one or more communal houses. The average population for villages 
with individual family houses is somewhat less. 

Village permanency. — Information is scarce on the length of time 
that villages continue to be occupied, but where this is mentioned 
it is invariably short. The Tupinambd move when the soil in the 
vicinity is exhausted or the thatch on the house begins to deteriorate, 
that is every 4 to 5 years, and the new village is near the old one. 
Montana villages move every 2 to 3 years, the Jivaro at least every 
6 years, the Cubeo every 3 to 5 years, the Tapirape every 4 to 5 
years. In addition to soil exhaustion, the decimation of game ani- 
mals or the destruction of nearby palm trees makes a change desir- 
able. Among some groups, abandonment is customary at the death 
of a member of the household. 

Furnishings. — Wherever they are mentioned, floors are described 
as of packed earth and neatly swept at all times. Furnishings are 
sparse, but usually include wooden stools, often carved in the shape 
of an animal, mats and hammocks or platform beds. Personal 

391329—57 4 



22 BUREAU OF AMERICAN ETHNOLOGY [boll. 167 

belongings, weapons, gourd bottles, baskets, etc. are often stored in 
the rafters. 

Dress and ornament. — In aboriginal times the vast majority of the 
women wore no clothing, and male covering was confined to a penis 
sheath. Women of some of the Montana tribes wore a pubic cover 
of a shell (Zapa) or a leaf (Z^paro), and in the Upper Xingii of a 
miniatiu-e straw triangle. On the Jurua-Purus a short, apronlike 
fringe of cotton was substituted. Depilation of all or part of the 
body hair was frequently produced with resin or latex. Body painting 
is widespread, employed particularly on festive occasions. A great 
variety of ornaments — beads, bracelets, anklets, earrings, labrets, 
diadems, ligatures — are created from the brilliant and profuse selection 
of materials made available by nature: wood, human and animal 
teeth, feathers, bone, shell, stone, beetle wings, fruit shells, seeds, 
jaguar claws, bird beaks, woven cloth, and bast fibers. 

Transportation. — The effective exploitation of the Tropical Forest 
environment requires dependable watercraft. The rivers are not only 
the avenues of transportation and communication, but also barriers 
to be crossed. Canoes are indispensable to many types of fishing. 
As a result, watercraft is one of the diagnostic traits of Tropical 
Forest culture. Their greater lightness makes bark canoes most 
useful in the upper reaches of streams or where rapids make frequent 
portages necessary. Elsewhere, dugouts are common. The Tupi- 
namba, who manufactured both types, had bark canoes 40 feet long, 
holding 25 to 30 persons, and dugouts manned by 60 men. Sails 
appear to have been aboriginally employed along the Guiana coast, 
but the more usual propulsion was with paddles, supplemented by 
poling in very shallow water. 

Manufactures. — Another diagnostic of Tropical Forest culture is 
the manufacture of ceramics. These are simple in shape and orna- 
mentation, in accord with their utilitarian function. Calabashes 
were everywhere important as containers, and were put to many 
uses elsewhere associated with pottery. 

Twilled basketry was widespread and employed for a great many 
articles in daily use. Among the Guiana Indians, where the art of 
basketry reached a high degree of proficiency, the products included — 

tubular manioc presses (tipitfs), cassava and farinha sifters, fire fans, plated 
rectangular boxes, wicker pot stands, sitting mats, carrying baskets, handbags, 
rectangular telescoping two-piece containers for household goods, trays for 
holding cotton . . . , rectangular and round hanging trays, deep bucket-shaped 
utility baskets, bottle-necked farinha baskets, fish traps, conical landing baskets 
for fish, hour-glass-shaped containers, rattles for babies, cover nets for the suspen- 
sion of pots, knapsack covers, and hollow-woven belts. [Gillin, 1948, p. 839.] 

Hammocks and other articles were woven from palm (acta, tucum 
or buriti) or cotton fibers. The wooden spindle had a whorl of wood, 



mvANsV ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 23 

turtle shell, a round wild seed, bone, calabash, clay or a sherd, and 
was usually discoid, 2.5 to 5.0 cm. in diameter. 

Among the typical musical instruments were hollow log drums, 
gourd rattles, and bark and clay trmnpets. 

Social and political organization. — A Tropical Forest village is 
typically composed of one or more kin groups tracing their relationship 
in the Guianas through the female line, elsewhere patrilineally. 
When the village consists of a single extended family or sib, local 
exogamy is observed. Marriage with cross-cousins is often preferred. 
There is no social stratification and no well-defined leadership. Al- 
though a headman is recognized, his main functions consist in organiz- 
ing fishing and hunting expeditions, supervising on ceremonial 
occasions, and arbitrating disputes. The advent of foreigners or of 
war, requiring consolidated action on the part of the group against 
the outside, however, could put greater authority in his hands (e. g., 
Apiac^). Polygyny was generally permitted, but common only 
among chiefs. There was little occupational division of labor within 
a tribe, except along sex lines, but certain tribes made products of 
recognized superiority that were sought in trade. The only person 
who possessed knowledge of an exclusive nature was the shaman, 
who was not a full-time specialist. He treated the sick by blowing 
and sucking the affected area, washing in herbal decoctions, and 
sweat baths, and also foretold the future. 

Life cycle. — Food taboos are often observed during pregnancy, 
especially by the mother. After birth, which usually takes place in 
seclusion, the couvade is widely practiced, though with various degrees 
of duration and intensity. At puberty both sexes frequently undergo 
ordeals in which flagellation, scarification, and exposure to biting ants 
are common components. In the Guianas, this ordeal was a pre- 
requisite for marriage. 

In contrast to the relative uniformity of other aspects of Tropical 
Forest culture, the methods of disposal of the dead are numerous, 
widely variant, and with no apparent correlation to geographical or 
linguistic and therefore presumably historical unity. The body is 
often buried beneath the floor of the dwelling along with ornaments 
and utensils, after which the building may be abandoned temporarily 
or permanently or not at all. Among the Mundurucii a male of high 
status is exhumed after the flesh has decayed, cremated, and the ashes 
are buried in a jar. The Tupinamba wrap the body in a hammock 
and squeeze it into a large jar, which they bury in the house floor or 
in the open, building a fire in the vicinity to keep evil spirits away. A 
Mura was buried with his possessions wherever he happened to die. 
The Omagua disinterred the body 3 months after burial, washed and 
painted the skeleton, and set it adrift in a vase. The Garaj^ exhumed 



24 BUREAU OF AMERICAN ETHNOLOGY [bdll. i67 

the body the next season and placed the remains in an ui-n, which 
was not reburied. Cremation was less common, but practiced by 
Rucuyen and Atorai, the former keeping the ashes in a jar, the latter 
bm'ying them. The Guaharibo "bm-n the bodies of their dead, collect 
the calcinated bones, and pound them in a mortar, and keep them in 
their houses in globular baskets of closcl}^ woven mamuri. When they 
move theu' residence or travel, they carry with them the bones of 
their ancestors" (Spruce, 1908, quoted by Metraux, 1948 e, p. 864). 
Other groups, among them the Tapajo, cremated the corpse or the 
exhumed bones and mixed the ashes with a beverage, which they 
drank. 

CULTURE TRAITS DISCERNIBLE ARCHEOLOGICALLY 

One of the most striking features of the Tropical Forest Pattern of 
culture is the extent to which the material culture is composed of 
traits of a perishable nature. This, coupled with the warm and 
humid environment, makes it almost intangible from an archeological 
point of view. Of the busy village, with its large, thatched houses, the 
variety of household utensils, the array of manufactured items, and 
the gaudy feather headdresses and other ornaments, all that remains 
is a scattering of potsherds, a few chips from cassava board graters, 
and perhaps a few stone axes. The cemeteries that provide informa- 
tion in other parts of the world are often absent, meaning that burial 
could have been by any of the varied methods practiced in the region 
today, few of which would leave any trace even if the spot could be 
found. A few traits can be deduced, but of those listed as basic 
diagnostics of the Tropical Forest Pattern — agriculture, watercraft, 
hammocks, and pottery — pottery alone remains to the archeologist. 

Pottery, then, is the key to more than the unraveling of the arche- 
ological sequence. It is the only link that exists between the arche- 
ological past and the ethnographic present. If we are to trace the 
Tropical Forest Pattern of culture backward through time, it has to 
be done through the medium of pottery. For this to be done ade- 
quately and accurately, it is necessary to understand the functional 
associations that pottery has, not only in terms of its method of 
manufacture and its use, but also in the broader perspective of its sig- 
nificance as indicative of the subsistence level and sociopolitical attain- 
ments of the culture. Such an understanding can only be reached by 
a study of the living cultures of the Tropical Forest Pattern. This 
approach is, of course, not essential if the main goal is to reconstruct 
the prehistoric sequence in a limited area. Archeologists should not 
be content with this, however. For their data to be of any value to 
others than themselves, they must make it possible to trace types of 
culture and not just types of pottery backward through time. Since 



"vTnsT^ ^'^^ ARCHEOLOGY AT MOUTH OF AMAZON 25 

most field workers do not publish all the details in their notes, they are 
the only ones in a position to know and evaluate all the facts. It is 
up to them therefore to make the cultural reconstructions that can 
be used by others in the analysis and interpretation of cultural prob- 
lems of wider significance. In this report, an effort is made to practice 
what we are preaching and to deduce the absent from the present so 
as to restore the dead cultures as much as possible to their living 
condition. 

The first step in making such a restoration is to establish a common 
denominator between the ethnographic and archeological horizons. 
The characterization of the extinct cultures begins with the material 
evidence in the form of pottery and the extent and composition of the 
site. These data can be compared with similar information from liv- 
ing cultures, and when the correspondence is good it can be assumed 
with considerable reliability that the sociopolitical, religious, and perish- 
able material aspects of the culture wiU also be comparable in general 
features. On the basis of this kind of analysis, all but one of the arche- 
ological Phases found on the Islands of Maraj6,Mexiana, and Caviana 
and in the Territory of Amapa can be identified as belonging to the 
Tropical Forest Pattern of culture. They represent semisedentary 
agriculturalists living in small communities, possessing the major 
technologies (except metallurgy) and a social organization character- 
ized by lack of differentiation whether in occupation, wealth, or 
social position. Only the Marajoara Phase exhibits more advanced 
characteristics. 

There are questions raised by the archeology that cannot be an- 
swered by existing ethnographic data. In a functioning culture, the 
small details of daily living that emerge as significant in archeological 
sites escape notice in the multitude of subsistence, technological, 
sociopolitical, religious, recreational, and psychological patterns that 
the ethnographer must record. There is a distinction in density 
and distribution of sherd refuse that suggests differences in house 
type and village pattern, but little or no information on refuse accu- 
mulation is available from living groups for comparison, and the 
rapid decomposition prevents the formation of post molds that would 
reveal size and shape of the houses. Knowing how many vessels are 
made and broken by a family during a measured period of time 
would aid in estimating the population of a village or the length of 
time it was in use. Even an indication of how broken pottery is 
disposed of might prove or refute what has been suggested as a pos- 
sible interpretation in this report (pp. 245 ff.). What is the area of a 
village? How much refuse has accumulated in the period of its 
habitation? Having no data with which to answer these and similar 
questions, the archeologist is forced to resort to logic to make inter- 



26 BUREAU OF AMERICAN ETHNOLOGY [bull, ler 

pretations from his meager data. Only the eventual help of the 
ethnographers wiU place these postulations on a firmer scientific basis. 

ENVIRONMENTAL LIMITATIONS ON CULTURE 
IN THE TROPICAL FOREST 

We are familiar with the pattern of culture characteristic of the 
living tribes of the lowland tropical forest, and have noted some of 
the ways in which this type of culture is an adaptation to the environ- 
ment in which it exists. In the present report, it will be shown 
that 7 of the 8 archeological Phases identified on Mexiana, Cavi- 
ana, and Maraj6 Islands and in the Territory of Amapd fall within 
the Tropical Forest Pattern. They differ from one another in details 
of pottery type and decoration, in village size and composition, and 
in burial customs, but all of these variations come within the range 
exhibited among living Tropical Forest tribes. 

The sites and ceramics of the eighth culture, the Marajoara Phase, 
are so outstanding that they previously completely overshadowed the 
less spectacular remains of the earlier archeological horizons on Maraj6 
Island. Their exploitation is so obviously profitable, even in the 
eyes of the caboclos, that it is only with difficulty and persistence that 
one is able to secure information on Ananatuba, Mangueiras, and 
Formiga Phase sites within the limits of the Marajoara Phase area of 
distribution. The high degree of technical and artistic competence 
attained by the Marajoara Phase ceramicists caused early writers to 
suggest that the makers must have been descended from, or at least 
have had contact with, Egyptian or Oriental civilizations (e. g.. Lisle 
du Dreneuc, 1889, p. 19). This evaluation cannot be given scientific 
credence today, but the observation on which it is based, namely, 
that Marajoara Phase culture is considerably more highly developed 
than other living or extinct cultures in the area, receives the support of 
modem archeological investigation. The quality and standardiza- 
tion of the ceramics, the differential elaborateness of the burials, and 
the large earthworks are material indications of a level of social and 
poHtical organization more comparable to that of Circum-Caribbean 
and Andean cultures than to Tropical Forest tribal society. 

The appearance of this advanced culture on Maraj6 Island in the 
midst of a succession of simpler ones throws the contrast between the 
two levels of development into high relief and raises questions that 
otherwise might not come to the attention of the archeologist. Why, 
for example, did none of the other archeological cultures attain, or 
even begin to reach such a high level of development? Why^did the 
Marajoara Phase undergo a cultural decline on Maraj6 Island? 
Could it have originated elsewhere in the Tropical Forest Area? 



Ev^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 27 

Seeking answers to these questions requires a study of anthropo- 
logical theory. Analysis of the forces contributing to the evolution 
of culture elsewhere has shown that agriculture exercises a dominant 
role (White, 1949; Childe, 1951). Wherever it has been introduced, 
there is an almost immediate and revolutionary change in the culture; 
where it has not penetrated, the culture never advances (except in 
special situations) beyond a nomadic hunting and gathering level, 
with undifferentiated social organization and simple technology. 
Agricultm-e is not a simple "open sesame" to the unlimited vistas of 
civilization, however. Its effectiveness as a subsistence base depends 
on two factors: the potentiality of the environment and the agricul- 
tural technology of the culture. The variant combinations of these 
extant in the world explain and in some cases determine the differ- 
ences in level of development that can be described (Meggers, 1954). 

The principle behind this conclusion can be summarized briefly. 
A food-gathering type of economy is undependable and time consum- 
ing. The return per man-hour of labor expended is small and suflSi- 
cient only to satisfy immediate needs. The supply of roots, fruits, 
and seeds is seasonal, and game is unconcentrated. In order to main- 
tain an adequate food supply, constant activity is required by all the 
able-bodied members of the community, which is limited to a small 
group typically composed of kin. This type of cultural adjustment, 
characterized by a minimum of material goods and a minimum of 
sociopolitical organization, was universal over the world until the 
commencement of the Neolithic, which is marked by the introduction 
of domesticated plants and animals. It has survived until the present 
in scattered environments where agriculture cannot be introduced. 

The adoption of agriculture as the basic food source meant that 
man was able for the first time to devote a good part of his time and 
attention to other things than the securing of food. As a result, the 
introduction of agriculture everywhere transformed the typically 
nomadic life of hunters and gatherers with remarkable rapidity into 
a new pattern characterized by settled villages and by the acquisition 
of the ceramic and textile arts. This initial revolution brought little 
alteration in the social organization — no strong chiefs, social classes, 
occupational specialization — or in religious concepts or practices. 
These advances came later and depended upon the increasing pro- 
ductivity of agriculture; in other words, on the deflection of larger 
amounts of time and effort from food production to be expended 
instead on cultm-e building. 

Where the techniques are absent or the environment prohibits 
their use and agriculture does not increase in productivity, the cul- 
ture is arrested temporarily or permanently after the consummation 
of the first stage of advance. In temperate regions like Europe and 



28 BUREAU OF AMERICAN ETHNOLOGY [bull, ist 

North America, the fertility of the soil can be permanently main- 
tained and the yield often increased by scientific crop rotation, fer- 
tilization, and similar means. In desert regions like Coastal Peru 
or in fertile river valleys like that of the Nile, the soil is almost un- 
limited in its ability to produce abundant crops year after year, 
which selective plant breeding can augment. But there is no evidence 
from geographers, soil experts, agronomists, or botanists that such 
a thing is possible where tropical forest conditions requne slash-and- 
burn agricultm-al exploitation, and anthropological data add con- 
firmation. No culture deriving its subsistence from slash-and-burn 
agriculture is able to maintain any of the traits of advanced agri- 
cultural societies, such as well-developed leadership, class distinction, 
occupational specialization, priests, temples or high gods, large and 
permanent cities and towns, and empires. There are only small, 
scattered and semipermanent villages and a relatively simple develop- 
ment of some of the basic technologies, like ceramics, textiles, wood- 
working and basketry.* 

Much speculation has surrounded the promising potentialities of the 
American Tropics as the garden spot of the world. Observers of the 
densely populated areas in equatorial Asia have been led to view the 
Amazon drainage as equally capable of intense exploitation, lacking 
only in sufficient advertising. Anthropologists, seeing that the Ama- 
zon lagged behind tropical regions nearly everywhere else in the 
world in the level of cultural development, have been inclined to 
invoke the late start of the American Indians compared with cultures 
in the Old World, the constant state of hostility and warfare between 
the Amazon tribes, or simply to leave the question unanswered. 

There is abundant evidence, both from geographers and ethnolo- 
gists, however, that the limited productivity of slash-and-burn agri- 
culture is the true cause. Robert Pendleton (1950, p. 115), a leading 
authority on tropical land use, has recently put the situation in deci- 
sive language: 

In higher latitudes, and particularly in the United States, a widespread opinion 
prevails that such humid regions as the enormous Amazon basin, now occupied 
by luxuriant and apparently limitless tropical high forests, must certainly have 
rich soils, and hence, great potentialities for the production of food, fiber, and 
other agricultural crops. ... It is true that certain regions such as those with 
recently active volcanoes, and those recent alluvial soils in humid equatorial low- 
lands which are not deeply flooded, do have great crop growing potentialities; 
they are producing and can continue to produce much from the soil. Never- 
theless, on the whole, the soils of the humid equatorial regions have distressingly 
limited possibilities for plant production. . . . This pessimistic attitude is no 
longer the result of mere opinion, for in a number of widely scattered regions in 

* Maya culture, which superficially looks like an exception, exhibits a history of decline very similar to 
that undergone by the Marajoara Phase (Meggers, 1954). 



E\^ANsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 29 

the humid low latitudes agricultural scientists have been and still are seriously 
at work. 

The reason for this seeming contradiction in plant productivity is in 
the differential ability of the crops to utilize the resources of the soil: 

The reason for the rapid decline in productivity is that practically all of the 
plant nutrients within reach of the roots of the forest trees have been taken up and 
are in the growing trees. Almost all the plant offal (dead leaves, twigs, fruit, 
fallen trees, etc.) which falls to the ground is quickly attacked by termites and 
decay organisms; as a consequence it rapidly disapi^ears. Organic matter cannot 
persist long on the soil; leaf mold as it is known in the north temperate U. S. does 
not develop. However, the heartwood logs of certain very durable sorts of trees 
will last a couple of years or more. The nutrients thus released and washed into 
the soil by the frequent drenching rains are quickly taken up by the tree roots 
lying in wait just under the soil surface. All the nutrients within reach of the 
tree roots are in the vegetation, and are being cycled. When the forest is cut and 
burned the cycle is broken, the plant nutrients being released in soluble form in 
the ash. The soil itself is extremelj' acid, often being pH4. The burning slightly 
reduces the acidity and supplies available nutrients for the crop plants which may 
be planted in the clearing. But before the annual or biennial crop plants can 
develop extensive root systems sufficient to absorb any considerable proportion of 
these liberated nutrients, most of the soluble materials will have been washed 
down deep into the subsoil by the almost daily rains — thus quite out of reach of 
the roots, [op. cit., p. 116.] ^ 

The effects of this leaching process are dramatically reflected in 
differences in yield from the same field in successive years. Wagley 
(1953, p. 67) reports that the second planting is only about half as 
productive as that of the first year after clearing. Re-use of the 
area before it has had sufficient time to return to tall secondary 
growth results in a less productive harvest than is achieved if the 
vegetation is allowed to reach this stage before another attempt at 
cultivation (op. cit., p. 68). Wasteful as it appears to be, slash-and- 
burn agriculture is the only method of exploitation that is adapted 
to the major portion of the Amazon area. The adverse conditions of 
high temperature and humidity, heavy rainfall, and low initial fer- 
tility of the soil make short intervals of cultivation separated by long 
periods of fallow and reforestation the only circumstance under 
which the long range pursuit of agricultural return is feasible, given 
the plants available aboriginally. 

In addition to the general poverty of the soil for agricultural pur- 
poses, there is a further factor that serves to reduce the utility of the 
land. This is its topography and elevation. Estimates of the possi- 
bilities of tropical agriculture often leave this out of consideration and 
as a result make the picture appear considerably brighter than it ac- 
tually is. Higbee (1948), for instance, has estimated that the land in a 



» Similar conclusions on the poverty of tropical forest agricultural resources have been reached by Stamp 
(1952, pp. 61-63) and Richards (1952, pp. 401-403). 



30 BUREAU OF AMERICAN ETHNOLOGY [bull, ler 

60-mile radius around the Maya site of Tikal could feed 500,000 people. 
This calculation is based on the observation that the production from 
1 acre of land will feed 1 person for 2 years. After clearing, 30 years of 
fallow are required for the return of fertility, before reclearing is 
profitable. Under these conditions, an allotment of 15 acres per person 
would insure a permanent food supply. Division of the area within 
a eO-mUe radius of Tikal into 15-acre plots gives Higbee his estimated 
population of 500,000. However, this method of calculating sub- 
sistence potential fails to make allowances for irregularities in the 
terrain. To be usable for agriculture, the land must be above flood 
level and have a minimum of slope. In a region where rivers rise from 
10 to 20 or more feet in the rainy season, a substantial part of the 
land is submerged for several months each year. Hills often have 
steep banks and summits too small for a field. Our own estimate of 
agriculturally usable land in British Guiana, Brazilian Guiana (Ter- 
ritory of Amapa), and on the Islands of Mexiana, Caviana, and 
Maraj6, is that it constitutes about one twenty-fifth of the total dry 
season extent. Since this is based on traveling over the countryside 
and along the rivers rather than on a specific survey, and in order to 
avoid an error on the conservative side, we increased this figure 2)^ 
times, bringing it to 10 percent of the total land area. 

Tropical areas with more favorable conditions for agriculture exist, 
but Maraj6 Island is not one of them. On the contrary, its poten- 
tiality is rather lower than average. The forested western part, 
poorly drained even during the dry months, is inundated during the 
rainy season. The campo dominating the eastern half is also hostile to 
agriculture. Unlike the fertUe plains of temperate regions, the 
tropical grasslands are even lower in agricultural potential than the 
forests. It is only with extensive preparation of the soil with fertilizer 
and by careful nurturing that the modern ranchers succeed in bringing 
a rare fruit tree to maturity (Lage, 1944, pp. 244-245; Pendleton, 
1950, pp. 119-120). Only in the limited area along the southeastern 
coast can the conditions be said to be at all favorable to cultivation 
(pi. 27, b.) Productivity can be judged on the basis of efforts to es- 
tablish agricultural colonies on the opposite side of the Baia de 
Marajo, where the land is part of the same formation as on the Island: 

The peasants who pioneered here soon found that while they could get a good 
crop of food the first year after cutting and burning the primeval forest and could 
get a following crop or two of mandioca, no further cropping was worthwhile for 
them, even though very little labor was needed to cut down and clear the second 
growth that came in after they abandoned their 2 or 3 years' cultivation of crops 
in the new clearing. [Pendleton, 1950, p. 116.] 

The inescapable effects of reliance on slash-and-burn agriculture 
have been recorded repeatedly by ethnographers: "The periodic ex- 
haustion of the soil by manioc produces a seminomadic tribal life" 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 31 

BVANSJ 

among the Cubeo (Goldman, 1948, p. 770) ; "The Jivaro community 
is . . . moved at least every 6 years as new farm land is needed" 
(Steward and M^traux, 1948, p. 621), etc. The pattern of "shifting 
cultivation" requires that the rest of the culture remain simple enough 
to retain its mobility, to be capable of ready transferral from place to 
place, or become extinct as local food resources give out.^ 

Examples of the degeneration or extinction of cultures that had 
become adjusted to permanently productive agriculture and were 
attracted or pushed into the tropical forest are also abundant. John- 
son (1948 b, p. 196) summarizes the Central American situation: 

The few colonies which the Meso-Americans sent into the Tropical Forest were 
mere outposts, some of which succumbed to the environment, while others, 
probably under environmental influence, adopted the indigenous culture. The 
colonies which retained their Meso-American features were evidently not estab- 
lished long enough before the Conquest for local environmental and cultural 
influences to have changed them. [Cf. Steward, 1949 c, pp. 759-760.] 

Students of the Andean cultures have commented that even the 
remarkably organized Inca system was unable to surmount the hmi- 
tations of the lowland tropics. Stirling notes that "archeological 
sites ... in the valleys of the Upano and Namangosa Rivers dem- 
onstrate that the material culture of the Jivaros in pre-Columbian 
times resembled that of the ancient cultures of the highlands much 
more closely than do present-day survivals" (1938, p. xi; also Steward, 
1948 a, pp. 13-14). 

This process of deculturation can be observed in progress m the 
changes that occurred in the culture of the Marajoara Phase during 
its habitation of Maraj6 Island. In this instance we have as complete 
possession of the facts as we are likely to have for the assessment of 
the causes of this decline. We have comparative material in the 
form of four other cultures of the Tropical Forest Pattern that occu- 
pied the same area at different times. These form a sharp contrast 
to the Marajoara Phase and emphasize its more advanced character, 
which can be paralleled only by cultures of the Circum-Caribbean 
and Sub-Andean levels of development. We have a detailed knowl- 
edge of the environment today, and the high probability that in the 
short time represented by the archeological sequence there was no 
notable ecological alteration. All of this evidence makes as clear a 
case as possible for the conclusion that this environment cannot 
support a culture more advanced than the Tropical Forest Pattern. 

If this is true, then the Tropical Forest Pattern represents the 
maximum development of culture that could have been attained in 
the area where agricultural exploitation is limited to slash-and-burn. 
This limitation is first and foremost an environmental one, which 

» For another discussion of the interrelationship between culture and agricultural potential, see Linton, 
1940. 



32 BUREAU OF AMERICAN ETHNOLOGY [bull, ist 

operates in terms of restricting the subsistence resources, both in 
quantity and permanence. Some variation exists within the region, 
and this is correlated with larger or smaller communities, ranging 
from two or three families to a thousand or more individuals (Tupi- 
namba). The upper limits of this range, however, resemble culturally 
the lower limits more closely than they do members of the more 
advanced Circum-Caribbean and Andean Areas. The cultural de- 
velopment of the Tropical Forest Area cannot be said to have been 
"arrested" by the advent of the Europeans as it might have been in 
other parts of the New World; it had already been arrested by the 
agricultural deficiencies of the environment in which it existed. 

An understanding of this situation permits a more realistic inter- 
pretation and evaluation of the past and present cultures at the mouth 
of the Amazon than would otherwise be possible. The similarities 
between the archeological Phases and their comparability to living 
Tropical Forest cultures become the expected components of a total 
pattern of adaptation to and limitation by a particular type of en- 
vironment. The deculturation suffered by the Marajoara Phase and 
its lack of influence on tribes in the nearby area become understand- 
able and explainable. Knowing the limitations of the tropical forest 
for the development of culture makes it possible to conclude that 
some other part of the South American continent with greater sub- 
sistence potential must hold the key to the origin of the Marajoara 
Phase, and this clue can be pursued and verified by use of the com- 
parative method (pp. 412-418). 

That the ecological situation in the Tropical Forest Area can be 
so sharply defined is a fortunate and unusual circumstance. In most 
other types of environment, the limitations and possibilities for 
cultural development are less readily delimited, and differences in 
technological acliievement, especially in the realm of agriculture, 
can play an important role in determining the productivity of the 
subsistence and tlirough it the level to which the culture can attain. 
Hence the approach employed here may not turn out to be particularly 
useful to archeologists woridng in other parts of the New World. 
This does not argue against making fullest use of it in the tropical 
forest, where the data recovered by archeology are so meager that 
all conceivable methods of analysis and interpretation must be 
explored. 



THE TERRITORY OF AMAPA 
GEOGRAPHICAL DESCRIPTION 

The Federal Territory of Amap§, was created in May 1944 in the 
area commonly known as Brazilian Guiana (fig. 1). The Rio Oia- 
poque separates it from French Guiana on the north and the Tumuc- 
Humac Range, extending westward from the headwaters of the Rio 
Oiapoque, is on the boundary with Dutch Guiana. The western and 
southern limits follow the meandering Rio Jari from its headwaters 
near the Serra Tumuc-Humac to its mouth, which opens into the 
lower Amazon just opposite the Ilha Grande de Gurupli. The 
mouth of the north channel of the Amazon (Rich, 1942, pi. 25) and 
the Atlantic Ocean combine to form the eastern boundary. This 
vast equatorial region extends from 49°52' to 54°50' West Longitude 
and from 4°25' North Latitude at the mouth of the Rio Oiapoque to 
1°20' South Latitude at the mouth of the Rio Jari. The area of 
137,419 km.^ given in the most recent Territorial report, is an approxi- 
mation based upon aerial photographs and incomplete ground surveys 
(Moreira, 1948, p. 1 ; Reis, 1949, pp. 7-11 ; World Aeronautical Charts, 
895, 946). 

The muddy waters of the Amazon discolor the Atlantic Ocean for a 
distance of 200 miles out from land and, in spite of the strong tide 
effects and ocean currents, none of the water along the southeastern 
shore of the Territory of Amapa is contaminated with salt. The 
coastline is constantly shifting, especially between the Cabo do Norte 
and the mouth of the Rio Oiapoque. Sand and mud bars running 
parallel to the coast are backed by a belt of marshy lagoons into 
which the water penetrates at high tides, and across which the rivers 
meander to empty into the sea through openings or channels in the 
bars. This section of the coast is frequently subject to heavy seas, 
high winds, and strong tides, presenting grave hazards to navigation. 

The topographical features of the Territory of Amapa are controlled 
in part by the fact that the north is composed of the same geological 
structure of crystalline hilly uplands found in Dutch, British, and 
French Guiana. This culminates on the western extreme in the 
Tumuc-Humac Range, which runs for 250 km. in an east-west direc- 
tion and attains an elevation of 916 meters. This continuous high 
range of igneous rock forms a line of division between the streams that 
flow north into the Atlantic (e. g. Rios Oiapoque, Cassipore) and those 

33 



34 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

that drain south into the Amazon (e. g. Rios Jari, Maraca). Several 
lesser mountains outcrop in the northern part of the Territory. The 
Serra Lombarda is the largest with its highest peak reaching an ele- 
vation of 500 meters. Numerous rock shelters and caves are found 
in the large, eroded, igneous outcrops scattered throughout the 
Territory, especially in the north. 

Geographically, the Territory of Amap§, presents a mixture of low- 
lands flooded during the rainy season, dry grasslands (savanna) with 
scattered trees, dense rain-forest vegetation, undulating uplands, 
and small mountains, AU the topographical and vegetational fea- 
tm-es resemble those of the other Guianas. The association of up- 
lands, ranging from between 15 and 100 meters in altitude, and low 
mountain ranges with flooded lowlands gives the whole area an unusual 
combination of topographical and vegetational features. Flooded 
lowlands occur along the coast from the mouth of the Rio Oiapoque 
to within a few kilometers of the city of Macapa, extending inland in 
a zone ranging from 10 to 100 km. in width. The most extensive 
unbroken lowland is the lake region between the mouth of the Rio 
Araguari and the city of Amapa. Marshes and hundreds of deep 
lakes ranging in size from small ponds (pi. 4, b) to Lago Novo, which 
is 40 km. long and 20 km. wide at the lowest water of the dry season, 
cover more area than the woodlands and rolling grassy meadows. 
During the rainy season, it is possible to travel by boat from one lake 
to another across the flooded campo ; even during the dry season most 
of the lakes are interconnected by smaU igarapes or streams. The 
Rio Flexal drains part of the lake region, offering an exit to the coast. 
Throughout the year these deep lakes are bountiful in fish and harbor 
a large number of waterfowl. 

Such a topography makes the Territory of Amapa relatively in- 
accessible by land and, now as in the past, the waterways are the 
main routes of transportation and communication. Although an 
abundance of rapids (pi. 1 , b) makes most of the rivers unsuitable for 
steamers or motorboats, they are navigable in sniaU canoes. The 
Rio Oiapoque, one of the largest, is unobstructed only as far as the 
modern town of Clevelandia, 85 km. above the mouth. Passing 
from here to the headwaters, 270 km. farther in a southwesterly direc- 
tion, requires portage around or paddling up and over 35 major rapids. 
The Oiapoque empties into a bay 20 km. wide and 45 km. long, formed 
by two peninsulas, one on the French Guiana side and the other, Cabo 
do Orange, on the Brazilian side. The Rio Ua9a, whose headwaters 
lie in the foothills of the Serra Lombarda, flows from the south into 
the same bay. 

Going from the Rio Oiapoque south to the Cabo do Norte and the 
Ilha Maraca, four principal rivers— Rios Cassipore, Cunani, Cal§oene, 



L 




1949 EXCAVATIONS 
NIMUENOAJU'S EXCAVATIONS 
LIMA OUEOES' EXCAVATIONS 
COUDREAU'S EXCAVATIONS 
VI.' 330 METER CONTOUR LINE 
^ SWAMPY LOWLAND 
ZI9 ELEVATION IN METERS 

AFTER US AIR FORCE WORLD AERO- 
NAUTICAL CHARTS S»4, ess, 946 



FlocME 1- — The Territory of \mM\t&. showing geogrsphieftl features ftnd I(i««tion of arrheologicAl siteo. 



"v^A^Nsf ^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 35 

and Amapa Grande — flow into the Atlantic Ocean. The longest and 
largest of these, the Rio Cassipore, stretches northward for 300 km. 
from its som^ce in the Serra Lombarda to the sea. It is obstructed 
by rapids and falls to such an extent that in early colonial times its 
headwaters were often reached by ascending the Rio Calgoene and 
crossing over by a small igarape joining the two rivers to avoid the 
hardships of portage. 

One of the major rivers of the Territory of Amapa, the Rio Araguari, 
empties into the Atlantic Ocean just south of the Cabo do Norte and 
divides the Territory into a northern and southern sector. It has 
two main forks, the Araguari proper, which flows due south from the 
foothills of the Serra Lombarda, and the Amapari draining from the 
eastern extreme of the Serra Tumuc-Humac. Although this river is 
1 to 3 km. in width along much of its lower course and has a deep 
channel and swift current, a stretch of rapids and falls between the 
towns of Ferreira Gomes and Porto Grande and a silted-up mouth 
has made the lower 150 km. of the river of only secondary importance 
from the standpoint of modern navigation. The land along the river 
is subject to flooding during the rainy season, except for occasional 
high bluffs and rounded hiUs, but the region is not geographically 
distinct from or more hostile than other parts of the Territory (pi. 1). 
However, the low banks may account for its position as an important 
cultural boundary in aboriginal times. 

South of the Rio Araguari-Amaparl all the rivers and igarapes drain 
into the north channel of the Lower Amazon instead of the Atlantic 
Ocean. There are no unusual features along the Rios Matapi, Vila- 
nova (formerly Anauerapucii), and Marac^; these drainages are similar 
to all others in the region with meandering courses, deep channels, 
greatly affected by tide action toward the mouth, broken by rapids 
toward the headwaters, crossing lowlands along the coastal fringe and 
lower reaches and cutting through uplands, higher hills, and moun- 
tains in the headwaters. A large number of tributary igarapes and 
rivers form a network of inland waterways for each river drainage. 

The longest and most sinuous river of the entu-e Territory of Amapa 
is the Rio Jari, which forms the southwestern boundary. Its course is 
roughly 800 km. long, with its headwaters and upper branches draining 
the south side of the Serra Tumuc-Humac, but only the lower 150 km. 
are free of rapids and falls. This river penetrates the thickest forest 
of the entire Teriitory, unbroken by savannas and uplands. The 
southern part of the Territory consists of undulating uplands with a 
heavy, equatorial rain-forest vegetation interspersed with occasional 
grassy savannas dotted with trees (pi. 3, a). The coastal fringe along 
the north bank of the Amazon is higher land, ranging from 5 to 20 



36 BUREAU OF AMERICAN ETHNOLOGY [bull, ler 

meters in elevation and covered with thick, high forest and dense 
underbrush. 

The chmatic features must be mentioned briefly. Although the 
equator passes through the southern part of the Territory, the climate 
can in no way be called disagreeable. The average monthly tempera- 
ture ranges from 24° to 28° C. (75.2° to 82.4° F.), giving a mean 
annual temperature of 26° C. (78.8° F.). The daily range is from a 
minimum of 20.5° C. (69° F.) at 5:30 a. m. to a maximum of 32.6° C. 
(91° F.) at 3 p. m. The nights are always cool and agreeable; con- 
tinuous and strong offshore breezes cool the coastal regions. The 
relative humidity of the free air during the dry season ranges from 
99 percent at the period of minimum temperature to 50 to 60 percent 
at the time of maximum temperature, rising to 70 to 99 percent during 
any time of the day during the wet season. Rainfall varies slightly 
by region but usually fluctuates in 13 to 15 year cycles from an annual 
rainfall of 203.2 cm, (80 inches) to 342.9 cm. (135 inches), usually with 
light, scattered daily showers during the dry season from late July to 
November (Pinto, 1930, pp. 30-42; Le Cointe, 1945, pp. 79-93; 
U. S. Air Force Meteorological Observations, personal communi- 
cation) . 

The popular conception of the Amazonian equatorial "jungle" as 
dense, monotonous, and impenetrable is not applicable here. The 
plant growth of the Territory of Amapa varies considerably from high, 
thick, virgin, equatorial rain-forest vegetation of large broadleafed 
trees with lim.ited undergrowth, to dense thickets of heavy under- 
growth and tangled vines along the banks of all waterways, to sprin- 
kled palms and other trees on the undulating grassy uplands, to the 
typical marsh and lowland growth of reeds, water lilies, and hyacinths. 
The variety of trees and shrubs is enormous. Some measured forest 
areas record up to 3,000 different species per square kilometer. AnimpJ 
life is less bountiful but includes the paca, agouti, peccary, coati, deer, 
tapu*, jaguar, ocelot, many species of monkeys, etc., in addition to 
reptiles. Bird life is profuse, but often withdrawn to the high fringes 
of the tall forest, making it difficult to observe. Without any doubt, 
the excellent fish found in abundance and in great variety in all 
streams, rivers, and lakes provided a major source of food in the past, 
as it does today. In relation to the potential food supply, it is perti- 
nent to mention the inherent sterility of the soil for the support of 
intensive cultivation. Heavy rains constantly percolating through 
the upper layers of the soil and dissolving the soluble minerals, plus 
the vigorous bacterial action under conditions of high temperature 
and humidity which quickly destroys any organic matter that falls to 
the ground, rid the soil of vital plant foods and humus. 



Kv^Nsf^ *^" ARCHEOLOGY AT MOUTH OF AMAZON 37 

The natural mineral wealth of the region is today of great economic 
significance (Ackermann, 1948), but only nugget gold, hematitic iron 
in a relatively pure state, white chalk, and yellow and red ochre are 
found in a free or accessible state. With the exception of the ochres 
and the chalk, none of these minerals appears to have been utilized 
by the Indians of the region. 

The Territory of Amapa is one of the regions of greatest variation of 
topographical and vegetational features in the Amazon drainage. 
This combination of flooded lowlands and lakes, undulating uplands, 
low mountain ranges, savanna and dense forest with a multitude of 
igarapes and rivers in part affected the living habits of the indigenous 
population but, in all probability, its influence on the first European 
explorers, who bent their efforts to the control and colonization of the 
region, was even greater. 

THE ARUA PHASE 

DESCRIPTION OF SITES AND EXCAVATIONS 

The only prehistoric cultural Phase in the Territory of Amapa with 
a distribution on both sides of the Rio Araguari-Amapari is the Arua. 
Its history on the islands of Marajo, Mexiana, and Caviana pro- 
vides the sequel to its occupation of this part of the mainland, 
from which it was apparently expelled by the peoples of the Mazagao 
and Ariste Phases. The Ai-ua Phase is of further significance be- 
cause it is the earliest ceramic-producing cultural Phase in the Terri- 
tory. Although it is represented by a limited number of excavated 
sites, these add important information to the Arua Phase as it is 
known from the islands. Reference to the sites, designated by their 
numbers, on the map (fig. 1) while reading the following descriptions 
will aid considerably in establishing a picture of the geographical 
distribution of each Phase. 

SITE A-6 — CAFEZAL 

The only habitation site in the Territory of Amapa showing occupa- 
tion by two distinct cultural Phases is Cafezal, located on the north- 
east side of the Rio Vilanova about 5 km. above the junction of the 
Rio Pigaca. The steep-sided, conical hill on which it is located is 25 
meters high and separated from the river bank by about 2 km. of 
low, poorly drained land. The entire region is covered with a dense 
forest containing an unusually large number of palm trees amid 
thick underbrush. A small rivulet drains along the foot of the hill 
and into the Rio Vilanova. The site itself is on the northwest side of 
the hill 15 meters from the top. Tests in many spots showed that the 
deposit covered an almost circular area 10 meters in diameter and did 

391329—57 5 



38 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

not exceed 10 cm. in depth. The potsherds were scattered thickly over 
the surface and the body of a huge jar was buried in their midst. A 
trench 2 by 1 meters was excavated near the center of the site to the 
depth of sterile soil and all specimens were cataloged as a unit. Pot- 
tery types belonging to both the Mazagao and the Arua Phases were 
found, but there was no stratigraphic separation because (1) the 
deposit was so shallow that there was natural intermixture and migra- 
tion of sherd material and (2) the area had been used, recently, as a 
manioc garden and hence had undergone extreme artificial mixture 
from cultivation activities. 

The large jar in the center of the site was uncovered and found to be 
decorated with a rib running around the shoulder bearing an irregular 
row of impressed rings (pi. 101, b). The rim was broken off 22 cm. 
below the present ground surface and just above the ring-impressed 
design. The vessel measured 78 cm. at its widest diameter, 61 cm. 
at the broken neck, with the existing fragment 61 cm. high. The 
flat base was 29 cm. in diameter. Large fragments of the heavy, ex- 
teriorly thickened rim found nearby establish the mouth diameter as 
36 cm. The body wall was from 1 .5 to 2.0 cm. thick and was unevenly 
smoothed, pitted, and crackled. The decorative rings average 1.1 
cm. in diameter and were punched to a depth of 3 to 5 mm. with the 
end of a hollow cane. Since the contents of the jar had been removed 
by curious caboclos, its function was not ascertainable. In the Arua 
Phase on the islands this type of jar was used for secondary burial. 
Of the 839 sherds cataloged from Cafezal, 230 or 27.5 percent represent 
the Arua Phase and the remainder are from the Mazagao Phase. 
These Arua sherds classify as 230 (100 percent) Piratuba Plain, 2 of 
which have punctate decoration. 

SITE A-8 — AURORA 

Site A-8 is a stone alinement on the east shore of the Rio Flexal, 
which drains part of the lake region north of the Rio Araguari- 
Amapari and south of the citj^ of Amapa. Unfortunately, caboclos in 
search of bm-ied treasure had disarranged the stones so that no idea 
could be gained of their original positions (fig. 2) . The}^ are situated 
on a rise in the savanna one-quarter kilometer from the river bank 
between two small lakes, one to the east and another to the northeast. 
This ground, rising 3.5 meters above the high-water level, is the 
highest land in the area and commands a magnificent view in all direc- 
tions over the sm-rounding high-grass savanna dotted with gi'oves of 
trees, the winding com'se of the Rio Flexal, and the lakes. All the 
stones are biotitic granite with a high percentage of quartz, of which 
the closest natural outcrop is about 5 km. downstream. At the time 
of our visit only two stones remained standing and these were leaning 



"vANsT^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 39 

badly (pi. 2). One extended 3 meters above ground and was roughly 
triangular in cross section, with the greatest width 30 cm.; it leaned 
45 degrees to the southeast. The other, 6 meters southwest of the 
first and leaning at an angle of 45 degrees to the northeast, measured 
2.10 meters above ground and was roughly triangular with 20 cm. 
to a side. Eighteen other granite stones, ranging in size from 75 by 
30 by 13 cm. to 3.00 by 0.30 by 0.14 meters, were scattered on the 
surface over an area 11 by 9 meters. All but six were under a meter 
long, but those six ranged from 1.75 to 3.00 meters in length. All of 
the rocks were roughly hewn, with no evidence of redressing. 



THICKNESS laM. 




THICKNESS .19 M 



9 10 1.0 H 



Figure 2. — Ground plan of A-8 — Aurora, a stone alinement of the Arua Phase. 

The caboclos said that no whole vessels had ever been found in their 
treasure digging and our tests in the area revealed no traces of bone 
material from burials and no concentration of potsherds or vessels. A 
few sherds \vere scattered in the native clay (tan to dark-brown 
flecked with orange) from the surface to a depth of 5 cm. near one of 
the standing stones. Only 78 sherds were collected, of which 36 were 
excellent representatives of the early varietj'' of Piratuba Plain wdth 
occasional punctate decoration and 30 were Aberta Incised (pi. 102, 
g-k). The majority of these sherds appear to be from only 2 or 3 
vessels. The remaining 12 sherds are ceramic types of the Arista 
Phase and also seem to represent only a few vessels (see pp. 106-107). 
This stone alinement resembles those found by Nimuendajti (see pp. 



40 BUREAU OF AMERICAN ETHNOLOGY [bull, m 

41-43) in the northern part of the Territory of Amapd in its general 
characteristics and in the fact that no burials and onlj a few scattered 
sherds are associated with it. Its location on a high point in the area 
with an unobstructed view of both sunrise and sunset, the laborious 
transportation of stones from a distance of at least 5 km., and their 
placement in some sort of definite pattern, seem to warrant the con- 
clusion that this site was used as a place of worship or ceremonial 
gathering. 

SITE A-23 — ILHA DA FORTALEZA, CONCEigAO 

Information on this stone alinement comes from Sr. Newton W. 
Cardoso, who visited it in March 1949. It is on a point of land between 
the lower Rio Flexal and the Canal de Carapaporis, which passes be- 
tween the mainland and the Ilha de Maracd. One and a half kilom- 
eters from the coastline are two small hiUs, one measuring 50 meters 
long by 25 meters wide and the other, 60 meters west of the first and 
roughly circular, measuring 35 by 45 meters and 5 meters high. Dur- 
ing the rainy season only these two small rises escape flooding. Pot- 
sherds and remnants of a stone alinement are found on the second hill 
only. The caboclos could remember when there were a large number 
of standing stones and many fallen ones arranged roughly in a large 
circle. Only 6 of these (largest 1.00 X 0.25 X 0.25 meters) remained 
at the tune of Sr. Cardoso's visit (fig. 3), the others having been car- 
ried off for modern building purposes. The nearest source of rock 
that he could ascertain is 10 km. away by dugout. Once again the 
small elevation commands a perfect view of the surrounding region. 



AFTER N.W. CARDOSO, FIELD 
SKETCH. 




FLOODED IN 

RAINY SEASON 



NATURAL ELEVATION 
IF ABOUT S M 




Figure 3. — Ground plan of A-23 — Ilha da Fortaleza, ConceiQao, a stone aline- 
ment of the Arua Phase. 



KvTwsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 41 

Potsherds were distributed sparsely from the surface to a depth of 
10 cm. or less in a roughly circular area with no concentration near 
the stones. Seventy-six sherds were collected from the surface and 288 
from subsurface testing. All of the sherds were good, typical, early 
style Piratuba Plain resembling the type from Cafezal (A-5) and Site 
M-2 on Mexiana. No bone fragments or whole vessels were found. 
The fact that the site has a more extensive refuse deposit than is gen- 
erally associated with these stone structures makes it possible that a 
small Arua village was located here prior to the erection of the stone 
alinement. 

DATA FROM OTHER INVESTIGATIONS 

Arua Phase habitation sites, typically small and shallow, appear to 
have escaped notice by the previous investigators in the Territory of 
Amapd. In any event, no one thought the pottery types sufficiently 
interesting to warrant transportation to a museum, even one so 
accessible as the Museu Goeldi in Belem. 

During his archeological explorations in the Territory of Amapd 
during parts of 1923 and 1925, Nimuendajii encountered numerous 
stone alinements similar to our Site A-8 — Aurora. The brief sum- 
mary that follows is taken in part from Linne's published accounts 
(1928 a, 1928 b) but principally from Ryden's (MS.) translation and 
study of the notes and materials collected by Nimuendajii and depos- 
ited at the Ethnographical Museum in Goteborg. Where it was possible 
to identify the cultural affiliation from a study of the photographs, 
drawings, or Ryden's descriptions of the artifacts, we have done so. 

IGARAPfi DOS MACACOS ' 

Several granite slabs were on a slight rise of land near the igarape 
One slab, 1 meter tall and 10 cm. thick ,was vertical with fragments of 
another scattered nearby. Fragments of a few vessels and several 
stone axes came from the site. 

RIO NOVO 

On a small hill about 6 meters from the river bank, there is a stone 
alinement consisting of three parts : (1) The eastern part composed of a 
vertical, granite slab 2.45 meters high, 1.15 meters wide and 10 cm. 
thick, with a smaller pillar leaning against it as support and several 
fragments scattered over the ground; (2) the central part with a ver- 
tical, granite slab and a looted "grave-shaft" nearby, which was cov- 
ered with a large flat stone; and (3) the western part 20 meters from 
the central group where five granite slabs (largest one 1.60 meters tall) 
were placed irregularly over an area of 5 square meters. Except for 
a large stone ax, no artifacts were found. 



42 BUREAU OF AMERICAN ETHNOLOGY [bull, le? 

Although Nimuendajii did not visit them, he received information 
that further up the Rio Novo there were several other similar stone 
alinements. 

JOs£ ANTONIO 

One of the largest stone alinements in the Territory of AmapA 
extends about 100 meters along the Rio Calgoene. (A ground plan 
and photograph are given in Linne, 1928 b, fig. 4 and pi. I-l.) Large 
portions have been demolished by treasure hunters and people seeking 
stones for road paving, house foundations, and anchors, but about 
150 stones are still available on the surface. Nimuendajii divides the 
alinement into three parts. A, B, and C, each apparently distinct from 
the other. The granite slabs appear originally to have been vertical 
with smaller stones propping up the bases. Although a few scattered 
sherds were found around some of the stones there was no concentra- 
tion; a few complete axes came from the area. Traces of charcoal to 
a depth of 1 meter were perhaps produced by the original slash-bum 
clearing of the land. Nimuendajii was impressed by the fact that 
although this group of stones represented a tremendous amount of 
work including transporting them from some distance, pottery was 
exceedingly sparse. The descriptions suggest the pottery is typical 
Piratuba Plain. 

VILLA CALgOENE 

Three vertical stone slabs erected in a triangle are said to have once 
been standing a short distance from the old Villa Calgoene. 

TESO DA MINA 

A large stone alinement consisting of a larger and a smaller group 
was near the Amapa Grande, but an organized party of treasure 
hunters had so disarranged the numerous granite slabs that their 
original position was indeterminable. No pottery was found by 
Nimuendajii. 

SUCURUtJ 

Two small alinements on the Rio Sucurijii, a source of the Rio 
Mayacare, had been totally destroyed by treasure seekers. 

LAGO DOS PATOS 

On the Lago dos Patos of the Rio Sucurijii was a recently distiu-bed 
alinement of 12 stones. No artifacts were found. 

CACHOEIRINHA 

South of the town of Amapd on the Igarape da Serra there is a large 
stone alinement arranged in four separate groups with some stones 
still erect but with most of them scattered. Nimuendajii reports that 



"-?,<l?i=f^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 43 



EVANS] 

one large granite slab was 4.38 meters long. A few sherds were 
found at the base of some of the stones, Eyden's descriptive com- 
ments suggest they belong to the pottery type, Piratuba Plain. 

AgAHYZAL 

Along the Rio Frechal (today sometimes spelled Flexal) there is a 
large group of stones, only a few of which still stand because of 
disturbance. Contrary to the situation at other stone alinements, a 
great number of plain or incised sherds were found, which had a 
sandy paste different from the few sherds found at the other aline- 
ments. In his description, Ryden (MS.) comments that a red-brown 
paint was on the exterior surface of sevenil sherds but the majority 
were incised. Our identification of these sherds from Ryden's plate 
26 places the majority of them in the pottery types known as Uaga 
Incised and Davi Incised. From these observations and our own 
investigations at Site A-8 — Aurora, it is clear that both the quantity 
and quality of most of these sherds are not the same as usually found 
by the alinements. They are pottery types representative of the 
Ariste Phase and must have been deposited some time posterior to 
the construction and use of the alinement by the Arua. 

ANALYSIS OF MATERIALS 

Since the Arua made only one kind of undecorated pottery, Piratuba 
Plain, a light-tan to orange-surfaced, sherd-tempered ware, and only a 
fraction of a percent of the sherds are decorated, the sedation of the 
sites cannot be based on percentage analysis of pottery types alone. 
Careful study and comparison of the sherds from 7 cemeteries and 15 
habitation sites belonging to the Arua Phase on the Islands of Marajd, 
Mexiana, and Caviana revealed certain pronounced differences in 
vessel shape, decorative style and general quality of ceramic that 
seemed indicative of time lapse. Glass beads found at two sites 
establish them as late and provide a terminal point along with his- 
torical records of Arua occupation. The seriation based on vessel 
shape is characterized by the disappearance of ring-impressed decora- 
tion and by improved control of the ceramic medium, shown in thinner 
walls and more regular surfaces, and more ingenious vessel shapes. 

The Piratuba Plain and the few decorated substyles of this type from 
Sites A-5, A-8, and A-23 are of the cruder variety of Arua pottery. 
The jar from Site A-5 is ornamented with a ring-impressed, applique 
band, and a few sherds with irregular incised lines (Aberta Incised) 
were found at Site A-8. These characteristics place the Arua Phase 
sites in the Territory of Amapd at the beginning of the Arua sequence, 
and the absence of contact materials adds confirmation to this seriation. 



44 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

CERAMIC HISTORY 

The development of the various styles and substyles of pottery from 
the Arua Phase of the Territory of Amapd can only be discussed 
intelligently when the Arua materials from Mexiana, Caviana, and 
Maraj6 have been studied. The analysis is therefore postponed 
until a later section of the report (see pp. 245, 525-537 for details). 

DIAGNOSTIC FEATURES OF THE ARUA PHASE 

The evidence from the habitation sites indicates that the pattern of 
small villages occupied for a short time, characteristic of the Arua 
Phase on the islands, was the same on the mainland. The crudely 
made and predominantly undecorated pottery, Piratuba Plain, fits into 
the interpretation of a relatively low cultural level. Although 
secondary burial in large jars placed on the surface in remote parts of 
the forest is characteristic of the Arua Phase on the islands, no such 
sites have, as yet, been reported in the Territory of Amapd. Stone 
alinements are associated with the Arua Phase in the Territory of 
Amapd and the lack of similar structures on the islands can be explained 
by the fact that no stone was available. Although their function is 
problematical, they were always constructed on a high place com- 
manding the best unobstructed view of the surrounding area, even 
if such construction meant the transportation of the stones by dugout 
from as far as 10 km. away. Large-scale disturbance by treasure 
seekers makes it impossible to reconstruct the original position of the 
stones in many cases, but in others the arrangement varies from a 
single row of stones to crude circles and triangles. The presence of a 
burial shaft at Rio Novo and sherds of Ariste Phase pottery types 
at Aurora and Agahyzal reflect an occasional usage of the high 
areas with peculiar stone alinements by the later peoples of the 
Ariste Phase. Arua burials have not been found in the vicinity of 
the structures. The scattered sherds from occasional vessels do not 
suggest any extensive offertory practice utilizing pottery vessels; 
nevertheless, it seems most likely that these structures had some 
ceremonial function in the Arua culture. 

THE MAZAGAO phase 
DESCRIPTION OF SITES AND EXCAVATIONS 

The geographical description of the Territory of Amapa indicated 
that the Rio Araguari-Amapari divides the area into northern and 
southern regions (fig. 1). This geographical barrier seems to have 
been significant as a cultural boundary between two contemporaneous 
cultural groups, the Ariste Phase to the north and the Mazagao Phase 
to the south. The following description of the sites, excavations, and 



BV^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 45 

materials of the Mazagao Phase will demonstrate the geographical 
limitation of this cultural group. 

SITE A-1 — PigAcX OCCUPATION 

Fifteen kilometers up the Rio Pigaca from its confluence with the 
Rio Vilanova, a large area had been cleared on the northeast bank for 
a manioc garden, A cemetery (Site A-3) was in the midst of the 
garden. Our exploration of the vicinity revealed a large habitation 
site (fig. 4) on a slight hill 25 meters to the south of the cemetery site. 
The occupation site, A-1, covers a large part of this hill, with the 
forest growth of the area very dense, undoubtedly enriched from the 
large amount of ash in the refuse ; however, all the trees are secondary 
growth. In the area of the site the steep bank rises 16 meters above 
the Rio Pigaca and 4 meters above the level of the bank just to the 
north in the region of the cemetery (A-3). The hill would have 
provided a complete command of movements up or down the stream, 
as well as an excellent defense position. Sherds were scattered over 
an area roughly conforming to the hilltop, measuring 110 meters in a 
north-south direction, and 60 meters in an east-west direction. The 
area was tested intermittently with 10 small test pits to determine the 
extent and depth of the deposits. The black, sandy-loam refuse layer 
with scattered sherds varied in thickness from the surface only, to a 
depth of 20 cm., averaging 10 to 15 cm. Beneath the refuse, the light- 
orange, sterile clay was tested to a depth of 1.15 meters. Owing to the 
extreme unevenness and shallowness of the refuse, stratigraphic work 
was not feasible. Instead, a test pit 2 by 2 meters was dug in the 
northern part of the site, in what appeared to be the region with the 
thickest concentration of sherds, and the materials cataloged as a 
unit. In this deposit, sterile clay was encountered at a depth of 12 to 
15 cm. below the surface. 

Besides 518 sherds, the following nonceramic objects were found: 
1 large piece of yellow ochre (5.0X5.5X1-5 cm.) with one surface 
flattened from polishing, another slightly depressed from use scratches, 
with the remaining surfaces irregular; 2 scraps of Jutahi resin (one 
5.0X3.5 cm.; the other 4X2 cm.); 1 coarse-grained, granite ham- 
merstone fragment roughly rectangular in cross section (3.5-4.5 cm. 
wide, 2.0 cm. thick, 7.0 cm. long) with the edges slightly rounded but 
very little reshaping, one end slightly battered; 3 fragments of fire- 
burnt clay, and 21 fragments of quartz, granite, and indurated sand- 
stone conglomerate of which 11 were fire burnt. 

SITE A-2 — LAURO 

Two kilometers downstream from Site A-1, on the opposite (north- 
west) side of the river (fig. 5), the flood plain extends about 30 meters 
back from the edge of the clear, fast-running Rio Pigaca. The bank 



46 



BUREAU OF AMERICAN ETHNOLOGY 



[BDLL. 167 





■^'■■- 




.•*' 


•.^■•.■'•■> '.•■■:-\ 


Vv ■•:•■■'• ■•.'■.'/.':■'. 


^•••■/^ 


^. ■ 




. N 


sivj^^ 


^-i<^--: :•■■■:■• 


':0-r 








^^^^^1 


^^^^'^^^O^'' 


: lx\.'. ■ ■ 




X ■.'••.■ 




::; x^^ 


^si^^ 1 


W/y: 




■ . ' Vi .■ '. 


: > ■ 


:-:'-';^>''--i^i\Sj' 


^ApS^^i:] 


' 


* • ■ "\ . • 




". -■ '.*" ■ 




■.':.':'■■':■'■'.: :..'':y>\ 


■ \\\ \ ■ 




•^ • \ •■ 


.'x-f^'. 


^r.; 




■ '.'■' >-■..•■■.•■..■. 


\ \ VT'v 




• "X* •" ."v 








i- "..'■'.'' ^ ■■■■'■■•' ' 


■ v ' \" ' • ' 


..'■,-.■■ 




X X 








.■x..-'..°v.:-- ■.".■■• 


\ v 






* . ■ '• " ^s* .* 








.••■■■.«»■.■. 


• * \' ' \ • 






;".;:;• -/s. 








•V; ;••.".'■ ■.■/■"; •.•:••■;■ 


■ "^tv- 


W:-. 








S. ! -y^ '. ' 


•1 

1 

. '1 


i^^M.^ 


1 








^4 

< 




' . ' ' ■ * . • 










1 








t^ 








i 




1 

1 








■■ 








■1 
1 




1 

1 








O 








1 




1 








« 






'. • . ■ .**. 


•1 
:.\ 
.1 




1 










/^ 


- ^ 








• 1 




«- 










• • • , ' 'm ' 






■I 
1 

.1 
.1 


2 / 

o 

s 


3 
O 

< 










2 






•I 

• \ 
■ 1 


8 ./V 


■ J 


















• 1 


5 i« V 


y\ 




















2 -• ' 


— 1 












r 

\ 






1 


< 


1 
\ 

1 












9 \ 






-1 


3 5 














.• '^ 


» ■ • • • 




■ 1 


X 


1 










• 


ty 






. 1 


is 

a. 


1 

1 










Jo 






1 
•1 


< 


1 
1 








■ 



Ph 

o 

153 



o 



(Ih 



o 

o 



^,f.1?f?^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 



47 




Figure 5. — Ground plan of A-2 — Lauro, a habitation site of the Mazagao Phase. 

rises sharply for about 7 meters, levels off a little, and then rises more 
gradually to a flattened summit 14 meters above the flood plain. The 
dense forest and heavy undergrowth had been cleared from the slope 
as well as the adjacent summit for the planting of a manioc garden. 
After the brush burning, the owner noticed sherds scattered on the 
ground in the northwest corner of his garden. Our sampling showed 
that this area had been a large habitation site, 83 meters in length, 
parallel with the river, and 52 meters in width. Throughout this 
extent the soil was gray-black, sandy loam and the undergrowth thick 
in the uncleared areas; beyond the site the soil was light brown with 
sparse undergroAvth. Site A-2 — Lauro was visited during the rainy 
season when, in spite of good drainage, the ground was extremely wet. 

Stratigraphic excavation was attempted first, in the hope that the 
refuse might be deep enough to provide evidence of a ceramic change 
through time. Two cuts were made, the first outside and the second 
inside the zone of cultivation (fig. 5). The sherd sample was then 
increased by a surface collection and two test excavations, which were 
2 by 2 meters upon completion. 

Strata cut 1, 1.5 by 1.5 meters, controlled in 15 cm. levels, was 
excavated in the center of the east quarter of the site inside the 
undisturbed limits of the forest. The black, sandy loam of the refuse 
layers contained many small quartz and granite pebbles intermixed 
sparsely v/ith small sherds. Many of the stones were fire burnt, but 
most of them were rock fragments natiu-al to the soils of this part of 
the Territory of Amapa. Strata lines are not visible in this type of 



48 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

soil or refuse deposit due to the leaching effect of high rainfall in a 
tropical climate. At a depth of 35 to 38 cm. the soil changed to 
reddish-brown or light-brown clay without any mixture of sherds. 
There was no soil change in the native sterile clay, tested to a depth 
of 1.00 to 1.50 meters. The juncture of the refuse strata with the 
natural soil was irregular, conforming to the unevenness of the original 
ground surface. Level to 15 cm. produced 216 sherds, 1 burnt-clay 
fragment, and 18 rocks; level 15 to 30 cm., 86 sherds, 2 burnt-clay 
fragments, 1 waterworn, oval pebble probably used as a pottery 
smoother (2.7X1.8X1.4 cm.); 1 grooved fragment of sandstone, proba- 
bly a "shaft smoother" (groove depth 5 mm.; groove width 5-8 mm.; 
fragment 5.0X4.0X2.2 cm.); and 7 rock fragments, none fire biu-nt. 
Level 30 to 38 cm. had 29 sherds, 1 yellow ochre fragment with use 
scratches on one surface and the other surface areas irregular (size, 
4.0X2.8X0.5 cm.), and 3 small rocks. 

Strata cut 2 was dug in the unplanted corner of the garden 20 meters 
south of the center of the site area, using the same dimensions and 
levels as in strata cut 1, The refuse features were identical to those 
of cut 1, with the sherds giving out at a depth of 25 to 28 cm. upon 
an irregular and uneven surface. Level to 15 cm. contained 221 
sherds, 1 broken, natural, waterworn pebble of fine-grained diorite 
with one end showing extensive battering and use as a hammerstone 
(fragment length 5.0 cm., width 6.5 cm., thickness 2.0 cm.), 1 unworked 
quartz flake, 1 burnt-clay fragment, and 13 rocks. Level 15 to 30 cm. 
had 77 sherds and 4 bm-nt-clay fragments. 

The two test excavations and surface collections added 873 sherds 
and the following nonceramic objects to the materials from Lauro: 
1 ax fragment of fine-grained diorite with the bit missing, butt- 
end flat with the surfaces well-polished (fragment length 8.5 cm., 
width 5.0 cm., thickness 3.5 cm. at the butt end tapering to 2.0 cm., 
with an oval cross section); 1 percussion flake of fine-grained diorite 
that could have been used as a scraper but shows no evidence of 
intentional or use retouch (roughly triangular, 5 cm. long, 4 cm. wide 
at the bulb of percussion, 8 mm. thick and 8 mm. wide at the point) ; 
4 burnt-clay fragments; and 43 rock fragments of which half are fire 
burnt and the others are a misceUaneous collection of quartz, iron 
concretions, granite, and indurated sandstone fragments, probably 
all natural inclusions in the soil. 

All the sherds were relatively small, due to the brittleness of the 
cariape and sand-tempered paste, with their surfaces badly eroded. 

SITE A-3 PigAcX CEMETERY 

Pigaca cemetery is 25 meters north of the occupation site, A-1, on 
a flat area at the edge of the steep river bank, 4 meters lower than the 



MEOGBBS AND 
BVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



49 



hill (fig. 4). A manioc garden covers the entire cemetery and much of 
the surrounding area; as a result, the majority of the vessels were 
badly broken from cultivation activities. However, since these 
vessels originally had been partially buried in the ground, a number 
of them were still intact. The amount of excavation that could be 
undertaken was restricted by the planted crop. Fortunately, there 
was a good-sized area near the bank not under cultivation and surface 
sherds indicated this to be the center of the site, which extended over 
an area roughly 30 meters in diameter. 

The light-brown sandy loam of the cemetery is not distinguishable 
from the soil of the surrounding area, indicating no use except as a 
depository for the burial jars. In addition to 12 complete or partially 
broken jars or bowls, 1,281 sherds were collected from the site. Most 
of these large fragments belong to only a few vessels. One of the 
most unusual pottery fragments was a large hollow foot (Mazagao 
Plain) with five toes probably representing a turtle, measuring 8 cm. 
high, with the diameter of the sole 14 cm. and of the leg 10 cm. (fig. 6). 
No other fragments of this urn were found nearby. It is undoubtedly 




Figure 6. — Foot of a zoomorphic (turtle) urn from A-3 — PiQacd Cemetery, 

Mazagao Phase. 



50 



BUREAU OF AMERICAN ETHNOLOGY 



[BDLL. 167 















O 
h 

© 



CM 




RED 



BLUE 



I I WH 



ITE 



GREEN 



FiGUKE 7. — Glass trade beads from A-3 — PiQacd Cemetery, Mazagao Phase. 



MEGGERS AND 
BVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



51 



from a zoomorphic jahoty urn of the type found by Lima Guedes and 
Farabee on Ilha do Pard (see pp. 71-73 and pi. 17), 

A few years previously Fritz Ackermann, a geologist of the Territory 
of Amapd., had excavated a jar in the same cemetery 7 meters north- 
west of our Burial Group 1 in which he discovered a large number of 
European glass trade beads. These specimens are now in the Museum 
in Macapa. Since the beads from the burial have not been kept as 
a unit, an exact count of each type is not possible; the following 
varieties are included : 



Table A. — Glass beads from A-3 — Pigacd Cemetery 



Color 


Description 


Count 


Clear glass with white 

stripes. 
Blue with white, red 

and blue overlay. 
Azurite blue with white 

stripes. 


Round with white lines inside running lengthwise; 
sometimes called "Gooseberry" (fig. 7, a). 

Round, 6 mm. diameter, with a colored, barber-pole 
overlay on the exterior (fig. 7, 6). 

Round with inlaid narrow white stripes running al- 
most from hole to hole (fig. 7, c). 

Round, 5-S mm. diameter (fig. 7, d) 


(?) 
(?) 
(?) 
Most common variety 




Round, 5 mm. diameter (fig. 7, e) - 


in sample examined. 
(?) 


"White, red, and green. . . 
Red and blue 


Layered glass with star-shaped cross section revealing 
an inner white layer, a red middle layer, and an 
outer layer of light green upon which there are darker 
green stripes. No terminal grinding. Sometimes 
called "Chevron or 12 Apostle beads" by bead ex- 
perts (fig. 7,/). 

Variety of "Chevron bead" with a red core, blue in- 
terior layer, and a solid blue exterior. Ends ground 
to expose the star-shaped red layer in contrast to the 
blue; barrel-shaped (fig. 7, g). 

Small, "Seed beads" varyhig In shape from round to 
barrel to disk-shaped, ranging from 1-2 mm. in di- 
ameter (fig. 7, h). 


(?) 
(?) 


Porcelain white, azurite 
blue or dark, opaque 
blue. 


(?) 



Burial Group 1. — Five burial jars were found together shghtly 
west of the center of the cemetery area (fig. 4). The base of jar A 
was 45 cm. below the present surface. This burial jar had been 
broken by the later bm-ial of jar C and further disturbed by a large 
root growing through it. The existing fragment of jar A, a small, 
fiat, pedestal base 12 cm. across, with curved sides rising 20 cm. to 
a maximum existing diameter of 32 cm., contained no bone frag- 
ments. The vessel was a good example of Mazagao Plain. 

The base of jar C was 42 cm. below the surface next to jar A. It 
had a cylindrical body 40 cm. tall with two applique nubbins sug- 
gesting breasts. The jar was 21 cm. in diameter at the mouth, 25 
cm. in the widest body diameter and had a short, pedestal base 18 cm. 
in diameter (fig. 8). It was plain except for 4 small holes 1 cm. 
below the rim edge, matching in spacing 4 corresponding holes at the 
lower edge of the lid. The face-lid (fig. 8), almost completely restored 
from fragments found inside and around the jar, was similar in gen- 
eral shape to the truncated-cone type used in the Kio Maracd area. 
It was 10.5 cm. high, 21.5 cm. in diameter at the rim that joined the 



52 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 




J I 



Figure 8. — Jar C (Maz 



3 6 CM 

Plain), Burial Group 1, A-3 — PiQacd Cemetery 
Mazagao Phase. 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 53 

ETANSJ 

jar, and 13 cm. in diameter at the flat top. The face was asymmetri- 
cally applied by incising eyes and eyebrows and adding appliques for 
nose and mouth. The surfaces of both the lid and jar were slightly 
uneven, but not rough, and both were good examples of Mazagao 
Plain. 

A mixture of sand, a few miscellaneous, scattered sherds and bone 
scraps filled jar C. Traces of bone appeared at a depth of 18 cm. 
from the rim of the jar, all in extremely poor condition due to the 
moisture that had collected in the jar after the lid had fallen in. 
Apparently this jar had been only partially filled with bones and 
sand, filling up completely when the lid broke and fell inward. At 
a depth of 20 cm. inside the jar, a small shallow bowl (pi. 7, c) was 
inverted over several scraps of the occiput. Just below it, right- 
side-up at a depth of 25 cm., was a miniature jar (pi. 7, e) containing 
three molar fragments and pure white sand. The bone fragments of 
the urn represent the secondary burial of a young adult, but no 
anthropometric details can be given because of the poor condition 
of the bone. Both small vessels are excellent examples of Mazagao 
Plain. The small, open bowl measures 11.6 cm. in diameter and 
4.5 cm. in height, with a flat base 5.0 cm. in diameter. The lip is 
uneven and modeled with two pairs of small, triangular, rim-adornos 
opposite each other. The miniature jar is much cruder and more 
asymmetrical, with a globular body 9 cm. in diameter, a flat base 
5 cm. in diameter, a short neck 5.5 cm. in diameter with a slanting 
rim. The total height varies from a maximum of 5.5 cm. on one 
side to a minimum of 4.5 cm. on the side opposite. 

Another tall, cylindrical vessel, jar B, was 19 cm. north of jar C 
with its base 31 cm. below the surface. Since the existing jar fragment 
is 33 cm. tall and the fragments of a possible plain, inverted bowl-Hd 
were inside, a portion of the jar and lid must have originally projected 
aboveground. The interior was filled with sandy loam; 15 cm. below 
the rim, bone scraps were mixed with sand. No teeth were found nor 
were any bone fragments large enough for identification. The jar was 
21 cm. at widest body diameter, with a mouth diameter of 18. cm. and 
with a short, pedestal base 1 cm. high and 12 cm. in diameter. On the 
back of the jar a dorsal ridge 5 mm. high extended vertically from the 
rim halfway down the body. One small nubbin on the opposite side 
just below the broken top probably represented a breast. The vessel 
was Mazagao Plain with a large amount of crushed white quartz and 
mica temper visible on the siu-face. The exterior surfaces were 
smoothed but irregular, with the body wall 1 cm. thick. Several 
fragments from inside the jar appear to be rims of both the jar and 
lid; they have small holes near the lips, similar to those of jar C. 

391329—57 6 



54 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 



A large, depressed globular jar, vessel D (Mazagao Plain), was 48 
cm. from jar C and 22 cm. to the east of jar B, with its base 39 cm. 
below the sm-face. The rim was missing, but the body measured 31 
cm. high, 43 cm. in maximum diameter, 24 cm. in mouth diameter, 
and had a flat, pedestal base 11 cm. in diameter. The interior of the 
jar was filled with very wet, sandy loam and traces of bone fragments. 
A small, broken bowl (vessel E) was inverted over skull fragments 
inside the jar, near its center and 20 cm. below the rim. Beneath 
these, a few scattered fragments of long bones were arranged parallel 
to each other alongside a scrap of the left side of the mandible, which 
contained several badly worn molars. The bones of this secondary 
burial were too fragmentary to permit a detailed analysis beyond the 




Figure 9. — Vessel E (Uxy Incised), Burial Group 1, A-3 — PiQacd Cemetery, 

Mazagao Phase. 

fact that they were of an adult human with complete tooth eruption 
and badly worn molars with caries. 

Vessel E was reconstructed (fig. 9). It is a poor, late example of 
Uxy Incised with crudely applied, light incisions, a triple nubbin 
applique around the waist and a thickened, slightly everted rim. The 
surfaces are light tan to dark gray, and the shape is very irregular and 
asymmetrical, measm'ing 22 cm. in mouth diameter, 12.4 cm. in height 
and 9 cm. in diameter of the flat, slight pedestal base. 

The fragments of jar Z, a cylindrical, anthropomorphic burial urn 
seated on a clay bench, were 42 cm. west of jar C and only 20 cm. below 
the surface. A small ax (fig, 10, a) of indurated sandstone, 6.2 cm. 
long, 4.9 cm. wide, and 2.7 cm. thick, was next to the left leg of the 
bench. This ax was only partially shaped beyond the natural form 
of the rock, ^vith the upper end rounded and used as a hammerstone 
and the bit well polished. A small, granite polishing stone (fig. 10, 
b), rounded on all surfaces and probably used in ceramic polishing, 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 55 

BVANSJ 




I I 

I CM 





Figure 10. — Stone artifacts associated with Jar Z from A-3 — PiQaed Cemetery, 
Mazagao Phase, a, Small ax. b, Polishing stone. 



56 BUREAU OF AMERICAN ETHNOLOGY [bull, ler 

was among the sherd fragments of the vessel. It was oval, measming 
4.8 cm. long, 2.3 to 2.7 cm. wide and 1.2 cm. thick. 

Although a complete reconstruction of jar Z was not possible owing 
to the eroded condition of this soft and unusually sandy variety of 
Mazagao Plain, sufficient fragments were recovered to establish the 
form as a local copy of the well-known Maraca seated anthropo- 
morphic urns. The reconstructed bench was 31.0 cm. long, 14.5 cm. 
wide and stood 4.5 cm. high. It had the tail (5.5 cm. long) of a cutia 
(agouti) at one end and a well-modeled cutia head at the other end 
(pi. 7, a). The head rose 8 cm. above the bench and was 4 cm. wide 
and 6 cm. long. A small, raised, punctate rib ornamented the bench 
top at each end. The oval, tubular body of the man seated upon the 
bench was 21 by 13 cm. in diameter, with the height unreconstructable 
from the fragmentary condition of the sherds. The legs with swollen 
calves, indicating use of ligatures, were attached to the body so that 
the feet were suspended in midair. The legs are fairly well modeled 
with prominent ankle bones, but each foot has 8 toes. Both arms are 
circular in cross section and the seven-fingered hands have cup- 
shaped palms to rest upon the knees. The head of the body is the lid 
of the burial urn but, although all the facial features of mouth, nose, 
eyebrows, and one ear were recovered, insufficient pieces prevent a 
complete reconstruction. The mouth was 6.2 cm. long and 1.0 cm. 
wide, with the teeth represented by a long, horizontally incised line 
intersected by numerous vertical lines. The eyes were made by 
apphques 3.8 cm. long, 1.8 cm. wide, and 4 mm. high with a long, deep, 
lengthwise incision. The eyebrow was formed by a slight rib. An 
applique 5.2 cm. long and 1.4 cm. wide formed the nose, which had 
two small holes at the base to represent the nares. 

From the fill 1 meter to the southeast of this jar and a few centi- 
meters below the surface came a piece of red ochre with facets worn 
by rubbing. 

Burial 2. — The broken upper edge of another vessel was level with 
the sm-face of the ground 1.60 meters south of jar D of Burial Group 1. 
Excavation revealed a tall Mazagao Plain jar having a cylindrical 
body 22 cm. in diameter with a bulbous expansion at the bottom 31 cm. 
in diameter and with a slightly concave base. The existing height was 
35 cm. Inside the jar, bone scraps were mixed with sandy loam from 
a depth of 15 cm. to the jar bottom, but the condition was too poor 
to permit any identification. No teeth were found. 

Burial 3. — A large, broken, globular-bodied jar was 1 meter south of 
jar D of Burial Group 1 and 50 cm. from Burial 2, with its base 50 cm. 
below the surface. A large root passed directly through the body and 
no bones remained inside. The vessel was typical Mazagao Plain 
with a very sandy paste and a brownish-red surface. The flat rim 



fvANsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 57 

was slightly thickened on the exterior, giving an exterior mouth 
diameter of 32 cm. The body diameter was 50 cm., diameter of the 
slight, pedestal base 20 cm., and the estimated body height 60 cm. 

Burial 4- — This taU, cylindrical jar of Mazagao Plain was 1.25 
meters north of jar D of Burial Group 1. The rim was broken off 
approximately 5 cm. below the surface and the base was 41 cm. in 
the ground. The burial urn had the same general shape as jar C of 
Burial Group 1, measm-ing 22 cm. in diameter at the broken upper 
edge, 28 cm. in maximum body diameter, and 36 cm. in existing height. 
The small, flat, pedestal base was 15 cm. in diameter. No lid fragment 
was recovered. The vessel was completely filled with moist dirt, 
with the lower third of the contents including coarse sand mixed with 
decomposed bone scraps. Two small vessels rested side by side in 
the bottom of the burial urn. One, an Anauerapucu Incised bowl 
(pi. 7, d) with an incised, in-sloping shoulder and a single, short, strap 
handle, contained pure, coarse sand, small bone fragments and a 
little dirt. It is an excellent example of the typical Anauerapucti 
Incised rectilinear, squared-spiral, incised design filled with white chalk. 
The bowl has a gray, fire-clouded, well-smoothed surface and good 
symmetry, and is by far the best-made vessel from the cemetery. 
The mouth diameter is 9.0 cm., body diameter 13.5 cm., and height 
5.6 cm. The small, strap, looped handle has a slight groove down the 
center and is 1.5 cm. wide. The base is slightly flattened. Its com- 
panion, a small Vilanova Plain jar (pi. 7, b) with two broken-off 
protrusions leaving holes low on one side, was filled with coarse sand, 
3 molars, and a few bone scraps. The neck is slightly constricted and 
the rim thickened on the exterior. The mouth diameter is 5.3 cm., 
body diameter 7.0 cm., diameter of the small, flat base 2.5 cm., and 
total height 6 cm. Although well smoothed, the exterior is irregular 
in places. 

Burial 5. — A large, globular Mazagao Plain jar with a vertical neck 
was found intact 25 cm. west of the tail jar of Burial 4, with its base 50 
cm. below the surface. The diameter of the smaU, flat pedestal base 
is 13 cm., of the body 43 cm., and of the mouth 31.5 cm. The total 
height is 34 cm., and the neck height 10 cm. Two human faces, 
modeled on opposite sides of the neck, are set off by paired vertical 
bars in the same style as the jar from Site A-4, Burial 3 (fig. 13, h). 
The eyes and mouth are represented by small, relief buttons 5 mm. 
above the surface, with depressions in their centers. Flanking the 
face are two parallel, applique bars 8.0 cm. long, 1.0 cm. high, and 1.2 
cm. wide with horizontally incised, parallel lines. The four sets of 
these vertical bars seem to mark the limits of each face. The surface 
of the vessel is badly eroded but originally was well smoothed in spite 
of being slightly irregular. 



58 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 1«7 



Miscellaneous sherds, sandy loam, fragmentary scraps of long 
bones, and "bone dust" were intermixed inside the jar. Unfor- 
tunately, the poor preservation of the bones did not permit any phy- 
sical anthropological observations. Several fragments of an Anauera- 
pucti Incised lid, with incisions on the exterior and interior, were 
found upon the rim and shoulder of the jar. This pedestal-basin lid, 
identical in shape and design to one from Site A-4 (figs. 13, a; 14, b), 
has a high, cylindrical, pedestal base supporting a shallow, wide basin, 
which was inverted over the jar mouth. 

SITE A-4 — -VALENTIM 

Valentim cemetery is on a high hilltop about 2 km. from the north- 
east bank of the Rio Pigaca and 4 km. above its confluence with the 
Rio Vilanova. The top of the hill levels off for an area of about 25 
meters in diameter, but the vessel fragments were all on the north edge 
of the hilltop, spread over a 5- by 4-meter area on the surface or buried 
in light orange clay. Our guide asserted that he could remember 
when the vessels had been intact and said that they had been broken 
by the children from a house formerly nearby. At our arrival only a 
few scattered sherds were visible on the surface and the area was cov- 
ered with a dense, secondary forest growth and underbrush. Many 
trees were gro\ving through the vessels, making the problem of exca- 
vation extremely difficult (pi. 3, b). Most of the vessels were so 
badly broken, decomposed, or disturbed by root action that all meas- 
urements had to be obtained in situ with a re-check made in the 
laboratory from a partial reconstruction of the fragments. 

In addition to the excavation of several burial groups where some of 
the vessels were stiU partially intact, a concentration of badly broken 
fragments of numerous vessels was recovered from the center of the site 
in association with 42 European glass trade beads. The beads in- 
cluded the following varieties: 

Table B. — Glass Beads from A-4 — Valentim 



Color 



Description 



Count 



Opaque, sky blue. 



Porcelain white with red 

stripes. 
Porcelain white with red 

spots, bordered with blue. 
Porcelain white with red lines. 



Porcelain white. 



Opaque blue 

Porcelain white- 



Long, tubular, square in cross section (6 mm.), 7.5 cm. long with 
rounded edges, hole 1.5 mm. in diameter. Middle layer of 
white surrounded on both sides by opaque, sky blue. Desig- 
nated by one bead authority as a variety of "Bugle Bead" (fig. 
11, a). 

Round, to egg shaped, 6.5 mm. diameter with 3 vertical, red 
stripes, 1 mm. wide (flg. 11, b). 

Spherical, 8 mm. in diameter, with three red spots, 2-4 mm. diam- 
eter, bordered with blue; hole diameter 1.5 mm. (fig. 11, c). 

Small, tubular, round cross section, 3 mm. diameter, 6 mm. long 
with thin (0.5 mm.), straight red lines running lengthwise. 
Designated by one bead authority as a "short Bugle Bead" 
(fig. 11, d). 

Spherical to egg shaped; diameter 5-8 mm., length 6-8 mm. 

(fig.ll,«). 
Round, diameter 5 mm. Broken fragments only 

Small, "Seed Beads" round, discoldal or barrel shaped with diam- 
eter from 2.5-4.5 mm, length 2.0-3.0 ram. (fig. U, /). 



Total. 



42 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 

EVANS] 



59 



■: ' . ' .\ '!\ 





d 

— > 



J I 








o 



O 



f 

— ^ 




I CM 



RED 



/^•:/: BLUE 



□ WH 



ITE 



Figure U. — Glass trade beads from A-4 — Valentim, Mazagao Phase. 



60 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

A total of 434 sherds, mostly rims, bases or diagnostic body sherds 
was taken into the laboratory for analysis and found to represent at 
least 34 separate vessels. Twenty-two percent of the sherds are 
Mazagao Plain, 39.0 percent Vilanova Plain, 27.0 percent Camaipi 
Plain, 5.1 percent Anauerapucii Incised, 6.4 percent Pigacd Incised, and 
0.4 percent Unclassified (pi. 16, c). 

Burial 1. — A Vilanova Plain jar was at the western edge of the site, 
with its base 28 cm. below the present surface. It had a round base, 
a globular body 42 cm. in diameter, curving up to join a gently sloping 
narrow neck 8 cm. tall and 24 cm. in diameter at the rim, with the total 
height of the jar 28 cm. The jar lip was rounded and slightly thick- 
ened on the exterior. The surface, now badly eroded, was originally 
smoothed and fairly even. A deep (13 cm.), thin-walled basin was 
inverted over the mouth, with its rim resting on the shoulder of the 
jar. This lid, also Vilanova Plain, had a rim diameter of 32 cm., 
with the side walls slanting inward to a convex base, 24 cm. in diameter. 
The rim of the jar was slightly thickened on the exterior. 

Burial Group 2. — A badly broken, anthropomorphic, cylindrical 
burial urn, seated on a clay bench, jar A (fig. 12), was located 5 meters 
southeast of Burial 1. During the excavation of jar A, jar C, a vessel 
like the one from Burial 1 and with a similar lid, was found adjacent to 
the back of the anthropomorphic figure at a depth of 35 cm. Twenty- 
two centimeters to the southeast of jar A, a second large globular 
vessel, jar B, with a body diameter of 58 cm., was buried 30 cm. be- 
neath the surface. Both vessels B and C were badly damaged by 
root action from a small tree growing out of the center of jar C (pi. 
3, 6). The earth inside the two vessels contained more humus and 
was slightly grayer and darker than the surrounding light-brown, 
natural clay, probably indicating decomposition of bone from the 
secondary burials. Near the right side of the anthropomorphic figure, 
at the same depth as the bench, was a small jar with a strap handle, 
designated as jar D. 

Jar A, a Vilanova Plain male anthropomorphic figure, contained 
bone-flecked dirt considerably grayer than the surrounding soil. 
It was broken off just above the knees where it protruded from the 
ground, and the pieces were scattered in the surrounding area. The 
situation of this jar corresponds to that of the anthropomorphic jar 
excavated at Site A-3, in that both were incompletely buried so that 
from the waist up they projected above the ground and both were 
associated with a group of nonanthropomorphic funerary jars buried 
at substantially greater depth. 

In execution, jar A represents a pronounced divergence from the 
Maracd style, although the generalized features show close affihation 
(cf. fig. 12 and pi. 18). The cylindrical body is 19 cm. in diameter 



^^?^^^^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 61 




Figure 12. — Reconstruction of jar A, Burial Group 2, A-4 — Valentim, Mazagao 

Phase. 



and 48 cm. in height from the top of the bench to the rim. Six cm, 
below the rim, the walls curve sharply inward for 3 cm. to join a 
short (3 cm.) vertical collar ending in a direct rim with a mouth 
diameter of 13 cm. The head, which forms the lid, fits over the 
short neck to rest on the shoulders. The long arms (upper arm 
length 18 cm., lower arm length 12 cm., hand length 4.5 cm.) made 
from solid clay rolls 3 cm. in diameter, are attached forward from 
each side just below the shoulder. Elbows jut out toward the sides 



62 BUREAU OF AMERICAN ETHNOLOGY [bull, wt 

and hands with concave palms rest on the knees. A small clay cyl- 
inder, 3 cm. in diameter and 3 cm. long, braced the left arm against 
the vessel wall. The legs rise against the body to the knee; the lower 
leg hangs free with the five-toed feet unsupported. Anatomic details 
include elbow, wrist, and ankle bones, swollen calf, vertebral column, 
male genitalia, and six fingers on each hand. An applique band 5 
cm. high and 1 cm. thick fits around the base of the body at its 
junction with the bench. 

The bench, 32 cm. long, 15.5 cm. wide, and 7 cm. high, was supported 
by two legs running lengthwise with the center of each cut out. The 
bench was unadorned except for a low rib along the upper edge (cf. 
bench of Barama River Carib, Gillin, 1936, pi. 18, a). 

The lid, which forms the head, is approximately dome-shaped, 
with the sides curving inward to give a basal diameter slightly less 
than the maximum diameter of 21 cm. The reconstructed height is 
11 cm. The interior edge of the front has a shelflike projection, 
widest at the center and fading into the rounded rim at each side in 
front of the ear, possibly made to assure a firm, rigid mounting of the 
lid upon the neck of the body. The mouth, 7 cm. long and 2 cm. wide, 
was made by a continuous applique 5 mm. thick around a depressed 
center. The eyes, depressions cut evenly 2 mm. into the surface 
and smoothed, measure 4.0 cm. long and 1.5 cm. wide, with the 
corners slightly rounded. Eyebrows, represented by long, slender 
appliques 3 mm. high, curve downward to join the top of the ear, 
which is a curved appendage 5.5 cm. long, projecting 1.4 to 2.0 cm., 
with a small hole punched in the lobe. 

The surfaces of both lid and jar are light tan, even and well smoothed 
leaving faint polishing tracks. Wall thickness of the lid is 5 mm. and 
of the body 7 mm. 

Burial 3. — Three meters north of the anthropomorphic m-n was a 
large, globular Mazagao Plain jar covered with an Anauerapucu Incised 
pedestal-basin lid. Both the burial jar and the lid were badly broken, 
further complications being added by a tree growing through the 
center of the jar. Apparently, the burial jar had originally been 
only partially interred, for the base was only 23 cm. below the surface. 
The waist and neck were broken off and sloughed 30 cm, to the side 
with part of the lid still in position. The dirt inside the vessel had 
been too badly disturbed by root action to leave any traces of bone 
fragments. 

The burial jar was similar in size and features to the large Mazagao 
Plain jar of Burial 5, Site A-3, although the relief modeling was more 
prominent (fig. 13, b). The globular body was 50 cm. in diameter, 
with a flat base 15 cm. in diameter, and a mouth diameter of 38 cm. 
The vertical neck was 10 cm. high; the total jar height was 43 cm. 



MBOGBBB AND ARCHEOLOGY AT MOUTH OF AMAZON 

KVANS] 



63 





6 CM 



Figure 13.— Pottery vessels from Burial 3, A-4— Valentim, Mazagao Phase, 
a, Pedestal basin lid (Anauerapucu Incised), b, Burial jar (Mazagao Plain) 
upon which the lid had been placed. 



64 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

The face was constructed of appliques. The nose and mouth were 
fillets of equal size and prominence, 3.5 cm. long and 7 mm. wide. 
Each eye was a fiat nubbin 1.8 cm. in diameter with a small depression 
in the center 5 mm. in diameter and 3 mm. deep. The nearest vertical 
bar on each side of the face was 4 cm. from the eye. These two parallel 
bars with two deep notches across them were 7.5 cm. long, 1.3 cm. 
high, 1.2 cm. wide, and 2.5 to 3.0 cm. apart. The faces, on opposite 
sides of the neck, were separated by pairs of vertical bars. The 
flat rim of the vessel was rectangular in cross section with exterior 
thickening, giving a thickness of 1.5 cm. as compared to the 1 cm. 
at the body wall. 

If the lid had not been found in situ on the jar mouth and neck, 
one might reconstruct quite differently the position and usage of this 
inverted basinlike lid (fig. 13, a, h). It appears as if these elaborately 
incised basins on a high pedestal might have been made for some other 
use than as lids because the applique and incising along the inner 
edge of the bowl were completely hidden by inversion over a burial 
jar. The large, open basin measured 40 cm. in diameter and 1 1 cm. 
in depth (fig. 14, h). This basin was raised on a cylindrical pedestal 
10.5 cm. tall, the sides of which flared outward from a diameter of 18.5 
cm. at the point of attachment to the basin to 22 cm. at the base. 
A peculiar feature of the basin bowl was a series of 4 equally spaced 
holes, 8 mm. in diameter, pierced through the waU just above the 
juncture with the pedestal. They were aU made when the clay was 
wet. The holes do not seem to have been functional in holding the 
lid in place, as no corresponding holes are found in the jar. The 
position and angle of the holes would have permitted suspension by 
means of thongs or fiber rope with the basin upright and the pedestal 
down, which, as already stated, would seem to be more in accordance 
with the positions of the decorations. However, sherd fragments of 
several of these pedestal basins were recovered from this cemetery as 
well as cemetery A-3, indicating their repeated usage as Hds for burial 
jars. 

The decorated motifs were the angular spirals and straight, parallel 
lines typical of Anauerapucti Incised. All the incisions were originally 
filled with white chalk, traces of which remain in over 50 percent of 
the lines. The pedestal had a 4.5 cm. -wide band of incised decoration 
beginning just above the externaUy thickened edge of the base. The 
rim of the bowl originally had four adorno lobes: two faces 12 to 14 
cm. long and extending 1.0 to 1.2 cm. beyond the rim edge opposite 
each other, and two smaller, plain lobes 4.5 to 5.0 cm. long extending 
1.0 to 1.2 cm. beyond the rim, opposite each other and equally spaced 
between the face lobes (fig. 14, a). The two faces were made with 
applique, ranging from 3 to 5 mm. high. The mouth was made more 



HEOGERS AND 
BVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



65 



realistic by the addition of a long, horizontal incision crossed by two 
short, vertical ones. Each face covered the entire lobe, thus breaking 
the incised border in two places. The incision was on the flat, slightly 
raised border formed by the beveled surface of the interior rim 
thickening. The bowl exterior and the pedestal interior were un- 
decorated. 





Figure 14. — Anauerapucii Incised lid from Burial 3, A-4 — Valentim, Mazagao 
Phase, a, Detail of the modeling and incision of the inner lip of the pedestal 
basin lid shown in figure 13, a. b, Exterior and cross section view of the same 
lid. (The rim is inaccurately drawn; for correct profile, see fig. 16-1.) 



SITE A-5 CAFEZAL 



This site and its excavations have already been described under the 
Arua Phase (see pp. 37-38) . It was also occupied by the peoples of the 
Mazagao Phase, their ceramics accounting for 609 or 72.9 percent of 
the 839 sherds recovered from the site. Of these 50.4 percent were 
Mazagao Plain, 30.5 percent Vilanova Plain, 15.9 percent Camaipi 
Plain, 0.4 percent Anauerapucii Incised, 2.6 percent Pigacd Incised, 
and 0.2 percent Uxy Incised. 



66 



BUREAU OF AMERICAN ETHNOLOGY 



[bull. 167 



SITE A-6 ILHA DAS IGAgABAS, IGARAPlfi DO LAGO 

Halfway between the mouth of the Rio Vilanova and the entrance 
of its tributary the Rio Pigacd, a second tributary, the Igarap^ do 
Lago, branches off on the north side (fig. 1). Although the mouth is 
shallow and somewhat hidden by foliage, this stream is the largest and 
longest that flows into the central part of the Rio Vilanova. In the 
wintertime the meadows along the stream flood, forming a large 
lago (lake) from which the name derives. The region is well known 
for sites, having been surveyed in 1896 by Lima Guedes for the 
Museu Goeldi, and in 1913-16 by Farabee for the University Museum, 
Philadelphia, Pa. Owing to the illness of the oldest and best guide 
in the area, we were able to locate only one site in our limited time. 
Fortunately, this was an occupation site rather than a cemetery, and 
added considerably to our knowledge of the Mazagao Phase. 

About 10 km. northwest of the Fazenda Santa Maria (which is 8 
km. upstream from the mouth of the Igarap6 do Lago) and 5 km. 
due west of the igarape, the land rises 25 to 30 meters above the low, 
flooded meadowland into rolling hills with scattered patches of forest 
and gi-assland. The heavy, red soil of the region is highly mineralized, 
containing large quantities of small iron concretions. A large 
grove of forest mth thick underbrush covers the summit of one of 
the highest hiUs m the region (pi. 3, a), 30 meters above the river level 
and 10 meters above the bottom of the nearest ravine. The habitation 
site was located near the north end of this forest, beginning 20 meters 
in from the edge, with the sherds extending over an area 75 by 83 
meters (fig. 15). All the nearby ravines, 0.5 to 1.0 km. away, con- 
tained small springs which flowed the entire year; therefore, the 



(l 




' CAMPO 




1 ■ ■ 


. . 1 


1 





100 


200 M 

I 






/ / /° • 


— 


"^ 


^ 


Wk 



Figure 15, 



-Ground plan of Site A-6 — Ilha das Iga^abas, a habitation site of 
the Mazasao Phase. 



t^vA^NsT^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 67 

occupants of this site were not dependent upon the Igarape do Lago, 
5 km. away, as a source of water. The location of this occupa- 
tion site, typical of the Mazagao Phase, on top of a high hill would 
indicate a definite choice of an area easily defended, dry in the flood 
season, and yet conveniently near the igarapes and river for trans- 
portation and food. Ilha das Igagabas was undoubtedly one of the 
habitation sites associated with the many cemeteries in the region 
represented by the specimens in the Museu Goeldi in Bel6m and in 
the University Museum, Philadelphia. 

Although it was named "Ilha das Igagabas" (Island of the Jars) 
nothing but sherds had ever been found there by the local caboclos 
who had dug, looking for jars filled with treasure. The sherds were 
small and sparse, with only a few visible on the surface. The area 
was widely tested to determine the extent and depth of the deposit. 
The sherds extended from the surface to a depth of only 10 cm. in a 
black, clayey, loose soil containing abundant iron-concretion gravel. 
Below this refuse there was a 5-cm. layer of brown humus and then 
a sterile orange-red, heavy clay. Because of the shallowness of the 
refuse deposit and the sparsity of the sherds, a 2 by 2 meter test 
excavation was made to obtain the largest possible sherd sample. 
Of the 782 sherds collected 35.8 percent were Mazagao Plain, 27.9 
percent Vilanova Plain, 32.4 percent Camaipi Plain, 3.8 percent 
Uxy Incised, and 0.1 percent Unclassified (pi. 16, i). 

Nonceramic specimens from the habitation site included: 1 small, 
rectangular piece of red ochre with one surface polished and slightly 
convex from use and with the other surfaces irregular (4.0 X 2.5 X 1.9 
cm.) ; 1 lump of white chalk native to the area, with one surface 
fairly flat and covered with fine use scratches (3.0 X 2.2 X 1.3 cm.); 
1 piece of yellow ochi-e with use scratches and one polished surface, 
with the rest of the surface area irregular (3.0 X 4.0 X 1.6 cm.); 
1 granite hammerstone fragment with the edges roughly rounded 
by gross percussion chipping, original shape indeterminate but present 
fragment flattened with a slightly tapered, blunt end showing extensive 
battering (fragment length 6.5 cm., width 7 cm., thickness 2 to 3 
cm.); 1 bm'nt-clay fragment, and 14 fire-burnt fragments of quartz 
and iron concretions. 

DATA FROM OTHER INVESTIGATIONS 

The greater amount of material from the southern part of the 
Territory of Amapd recorded in the literature, as well as found in 
museum collections, permits a more detailed comparative study of 
the Mazagao Phase than of the other Phases. Most of these collections 
were made in the late 1890's and early 1900's by explorers, travelers, 
and ethnologists, with the result that much of the information valuable 



68 BUREAU OF AlVIERICAN ETHNOLOGY [boll. 167 

for detailed archeological analyses is lacking. It was possible to 
supply ceramic details, however, by our extensive examination of the 
collections of the Museu Goeldi in Belem. 

RIO lEATAPURtJ SITES 

Evidence from the Rio Jari drainage is provided by Nimuendajii 
in his published account (Nimuendajii, 1927, pp. 356-358 and map) 
and in the collection of 248 sherds that he deposited at the Museu 
Goeldi. He recorded 5 sites. Novo Anno, Alto Alegre, Bom Destino, 
Sao Joao de Iratapuru, and one unnamed, along the Rio Iratapurii 
(fig. 1), the first large tributary entering the Rio Jari from the north. 
The sites begin about 25 km. from the mouth and extend 10 km. 
farther upstream in a region of high (230 meters) uplands. Two 
additional sites, Uxy and Campoeira do Mestre Aprigio, are located 
on the Igarape Amazonas (fig. 1), which branches off the Rio Iratapurii 
just north of the site of Alto Alegre, and whose headwaters join the 
Rio Maraca drainage. Of the sites themselves, we know nothing 
beyond Nimuendajii's statement that they consisted of areas of black 
soil. His brief published description of the ceramics is misleading 
and would, in the absence of analysis of the sherds, lead to the affi- 
liation of the incised style with Anauerapucii Incised rather than 
with Uxy Incised.^ 

Analysis of the samples from Alto Alegre, Bom Destino, Sao Joao, 
and Uxy revealed sherds of Uxy Incised, Jari Scraped, and Mazagao 
Plain (Appendix, table 1). The frequency of decorated sherds is 
45.5 percent or higher, which is far in excess of the proportions derived 
from our excavations in Mazagao Phase sites. A conscious selection 
of decorated sherds undoubtedly accounts for this unusually high 
percentage of decorated sherds and low frequency of plain sherds. 

Among the unclassified decorated are two sherds, 1 from Uxy 
(pi. 16, /) and 1 from Bom Destino, representing parts of small faces 
with applique nubbins and fillets forming the eyes, nose, mouth, and 
eyebrows. A small human foot (pi. 16, d), 7.5 cm. long and 3.6 cm. 
wide just behind the toes, broken off where it joined the leg, is from 
Sao Joao. One flat sherd (Sao Joao) 1 cm. thick is punctured with 
numerous holes 3 mm. in diameter, arranged 4 to 6 mm. apart in 
rows (pi. 16, e). A rim sherd (Sao Joao) is ornamented by thumb- 
made depressions along the rim edge, applique and a band of incision 

' This statement, based on a study of the actual sherd material, is in no way intended to condemn Nimu- 
endajti's work or to belittle his aid to an archeological understanding of the Amazon. It is presented merely 
as a correction of published data. More than once Nimuendajii stated his lack of training in archeological 
methods and his desire to leave excavation of sites to specialists and for the future when ai'cheological tech- 
niques would be improved and perfected. Actually, he is to be commended for his collection of sherds 
from various parts of the Amazon together with the accurate recording of their provenience — an invaluable 
aid to the archeological interpretation of this vast unknown area and a service in which too few present-day 
ethnologists are willing to cooperate. 



^.^.^.?I'^^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 69 

on the exterior (pi. 16, a). A solid, cylindrical object (Bom Destino), 
6.5 cm. long and 3.5 cm. in diameter, broken at one end and decorated 
with crudely incised lines and two small asymmetrically placed 
bmnps, may be a fragment of a figiu-ine. A disk (Bom Destino), 
unevenly concave on one face and convex on the other, 6.3 cm. in 
diameter, with a perforation through the center, possibly represents 
a spindle whorl (pi. 16, g). 

RIO VILANOVA SITES 

The Kio Vilanova (then called the Rio Anauera-pucu) area was 
the scene of an expedition from July to September 1896 under the 
direction of Emilio Goeldi and the field leadership of Aureliano Lima 
Guedes. The specimens collected, now in the Museu Goeldi in 
Belem, were carefully studied on the basis of our analysis and classi- 
fication of the Territory of Amapa ceramic types. Although no 
field notes exist other than what is contained in Lima Guedes' brief 
published account with his accompanying map (Guedes, 1897, 
pp. 55-59), the collection remains one of the best documented from 
the region. However, either because this map is not as accurate as 
the ones used today based on aerial photography, or because the 
local names in the interior frequently change, or because the guides 
who knew of them had died, many of the sites could not be relocated 
in 1949. On February 1, 1916, Farabee visited the same region and 
collected several specimens for the University Museum, Philadelphia. 
Although his field notes are quite sketchy, the University Museum 
catalog identifies several specimens as coming from the same large 
cemetery, Ilha da Canoa, on the Igarape do Lago do Rio Vilanova, 
that was excavated by Lima Guedes in 1896. Farabee's (1916 a) 
description of the site is extremely limited, for his field journal at 
this point contains more general description of the country than 
archeology, but a few passages are pertinent : 

Work: Igarap6 do Lago, Feb. 1, 1916: The place was out in campo in midst of 
ant hills and some of pots were buried in these hills which are about 4 feet high 
and 8 feet across. Many pots have been buried apparently even with the ground. 
Now their tops extend 3" to 6" out of ground and all broken. One was in sight 
in edge of ant hill. This one had evidently been set in the ground halfway and 
the ant hill built over it . . . No evidence that hole had been dug into side of 
ant hill. At another place near there were several bottoms of pots set in ground — 
tops nearly all gone . . . but no burial in it. 

These and others . . . appear to have been set in ground just enough to keep 
upright. Some evidently had plates over them. . . . The same are reported from 
Region of Igarap^ do Lago and north on east bank of Rio Vilanova for 100 miles. 
That is in the campo region between Vilanova and Matapy rios. 

A study of Farabee's collection from this site supplements our detailed 
classification made on Lima Guedes' material. 

391329—57 -7 



70 BUREAU OF AMERICAN ETHNOLOGY [bull, m 

Along the Igarap6 do Lago, Lima Guedes found 5 sites, one of which 
he excavated in search of complete vessels. This site, Ilha da Canoa, 
was located in an island of forest in the savanna. Although there 
were many large sherds and broken urns visible on the surface, he 
excavated three days without obtaining any vessels in an unbroken 
condition because of the hardness of the ground. Most of the jar 
fragments still in situ had traces of bone inside and one contained a 
piece of odoriferous resin, about the size of a hen's egg and commonly 
called "cuanuaru," with the bones (Guedes, 1897, pp. 55-56). 

Lima Guedes mentions visiting additional sites on this igarape, 
called "Ilha das Igagabas" (om- Site A-6), "Ilha das Pombas," and 
"Tabeleiro do Gentio," but he comments that the material was very 
decomposed or broken by trees and roots and not worthy of bringing 
in (op. cit., pp. 48-49). Throughout his report, when he makes this 
kind of observation, he is undoubtedly dealing with habitation sites 
(although he calls all sites ^'necretorios" or cemeteries), which produce 
only scattered, badly broken sherd fragments in comparison to the 
cemeteries with large numbers of complete or partially complete 
vessels. 

After spending 15 days on the Igarape do Lago, Lima Guedes 
explored the upper branches of the Rio Vilanova, especially two sites 
called, "Vila Nova da Rainha" and "Campos da Rainha." He 
describes the second site as a "cemetery whose disposition of urns is 
exactly the same as Marajo" and mentions excavating some of the 
vessels with great difficulty and carrying them over land back to the 
river in a sling across a pole (op. cit., p. 58). 

In the Museu Goeldi there are 1 1 complete vessels and miscellaneous 
fragments labeled as coming from the "Igarape do Lago do Rio Vila 
Nova" and from the "Rio Anauera-pucii" (Vilanova). Unfortunately 
there is no catalog or other means of identifying which vessels come 
from which of the two excavated sites, Ilha da Canoa of the Igarape 
do Lago, or Campos da Rainha of the Rio Vilanova. Of the speci- 
mens preserved in the Museu Goeldi, 4 are Mazagao Plain, 1 Vila- 
nova Plain, 5 Camaipi Plain, and 1 Anauerapucu Incised. A brief 
description is offered to elaborate the characteristics of the types. 

Mazagao Plain Vessels: The surface color is dull, orange-red to orange-brown, 
smoothed but gritty due to the quartz particles and white mica temper; oxidized 
firing. One globular-bodied vessel (pi. 9, b), 32 cm. in diameter, with a slight, 
flat pedestal base, and a 14 cm. high neck with an applique of a human face and ver- 
tical, notched bars is identical in decorative motif to the large, globular vessel of 
Burial 5, from Site A-3. Another fragment from a vessel measuring 36 cm. in 
mouth diameter, body diameter estimated around 50 cm., is a variant of this 
style, having a face surrounded by an arched applique ornamented with fine 
punctates. The other two vessels are high-waisted, globular jars with flat bases. 
Small birds are modeled upon a wide, bulging collar around the neck. On one 
a single bird with a modeled head is represented (pi. 9, c) ; while the other vessel 



BV^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 71 

has two birds with outspread wings produced by low appliques on opposite sides 
of the neck (pi. 9, a). 

Vilanova Plain: The only specimen of this ware is a jar with a flat base 12 cm, 
in diameter, globular body 25 cm. in diameter, with the walls curving inward to 
a short lieck 11 cm. high and a mouth 19 cm. in diameter with a flattened rim 
thickened externally with a wide coil (pi. 9, d). The total height of the vessel is 
31 cm. It is typical of Vilanova Plain with light-tan surface color and cariapi 
temper. 

Camaipi Plain: The surfaces are M^ell smoothed, with the color ranging from 
light orange to brownish orange. Temper is a mixture of white quartz particles 
and cariapi. Three vessels have depressed-globular bodies, flat bases, and long, 
inward-slanting, straight-sided necks ending in fiat-topped, externally thickened 
rims (pi. 8, a, b). The body diameters are 29, 32, and 43 cm. with the total 
heights 27, 32, and 44 cm. One tall, badly broken, anthropomorphic vessel with 
the body oval in cross section is crudely modeled in the form of a seated man 
(pi. 8, c) . The face is on the short, straight-sided neck, which ends in a thickened 
rim with punctates along the exterior surface just below the lip. Several frag- 
ments of a large, globular jar with a body diameter of 63 cm., a mouth diameter 
of 45 cm., and a neck height of 20 cm. indicate that two faces were modeled on 
opposite sides of the neck. One face is much larger than the other but they are 
alike in style (pi. 8, d). The eyes are formed by applique rings 6 cm. in diameter; 
the mouth, a narrow, oval applique, is 9 cm. long and 3.2 cm. wide with teeth 
indicated by vertical incisions; eyebrows, nose, and ears are long, applique fillets, 
and holes are punched in the ear lobes. The face measures 20 cm. wide and 13 
cm. high. 

Anauerapucd Incised: There is only one fragment from a large bowl with typical 
incisions of parallel lines, frets and rectilinear spirals with traces of chalk in the 
incisions; the typical brick-red color, quartz temper, and sandj'^ paste is repre- 
sentative of the type. 

ILHA DO PARA SITE 

When traveling from the Rio Maraca to the Rio Anauerapucii 
(Vilanova), Lima Guedes stopped off on the Ilha do Para, a large 
island just off the coast between the mouths of the two aforementioned 
rivers (fig. 1). His description locates the site in the forest on the 
south-central part of the island where the sherds were scattered over 
an area 300 meters in diameter, all badly broken and disturbed from 
the excavations of "treasure" seekers. Although none of the vessels 
had originally been buried, the mass of sherds was partially covered 
by debris and trash. He found only one fragment of a ceramic human 
leg with ligatures and swollen calf, similar to the anthropomorphic 
urns from the Rio Maraca. All the other vessels and fragments were 
from the large burial urns shaped hke a jaboty or land turtle. Frag- 
ments of bones were inside some of the unbroken vessels. 

From the Lima Guedes expedition there are three complete zoomor- 
phic urns and several zoomorphic head and feet fragments in the Museu 
Goeldi today. LTnfortunately, a few have been mislabeled since they 
were deposited in 1896 and studied a few years later by Goeldi, so 
that today two urns carry no information other than "Ilha dos Por- 
cos." This is a recent cataloging error because the finds are clearly 



72 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

identified in Lima Guedes' report (1897, p. 54) as coming from Ilha 
do Para, as well as being illustrated with this provenience on the litho- 
graphed plates from an unpublished manuscript of Goeldi (MS., 
Estampa 7). Also, Lima Guedes did not undertake archeology on 
any island in this area except the Ilha do Para. A comment by 
Farabee confirms that the correct provenience of these zoomorphic 
burial urns is the Ilha do Para. He visited very briefly both the Ilha 
do Para and the Ilha dos Puercos on February 14, 1916, and made 
the following comment: 

On the Ilha dos Puercos, Island of the Pigs, there are numerous village sites; 
but apparently, the people removed their dead to a small island nearby called 
Ilha do Para. On this small island we were unable to find evidence of occupation 
or village sites . . . [Farabee, 1921, p. 154.] 

Farabee's field journal gives a vivid description of the condition of 
the cemetery: 

The place is 2 or 3 acres in extent and possibly 2 feet higher ^ than the general 
level of the island and does not flood in rainy season. Now even we waded half 
knee deep much of the way — only a few hundred yards from the river. Where 
the pots were found there were [was] no evidence of a village site. I dug in many 
places but found nothing, not even black earth. The pots had been placed on the 
top of the ground — now they are sunken to the bodies of the pots. Many not 
requiring more than to be lifted out without digging — others needed a little 
digging to free the legs. They had been set two or more side by side — one place 
I found six in a group, several places two — others where they had been so dis- 
turbed it was impossible to tell how many there had been but more than six and 
less than 12. All of the pots were in the form of some animal with large short 
legs with from 3-5 toes — head at one end and tail at the other. Many heads 
appear to have been meant for men's heads but even these have short tails — 
many may have been tigers. All had fragments of bones inside except when too 
badly broken to hold them. None had ashes. None had anything else inside. 
All have had covers over the mouth — These were not plates which had been used 
for other purposes but apparently were made for pot covers. All pots were near 
same size, all plain and unpainted. 

They are all too small to admit a body entire even if small and demaciated. If 
bodies had been cut up to enter the pots, the animals and insects would have 
destroyed the pots to get the remains. The people who would put their dead 
in urns would not sufifer their bodies to be destroyed so I think the bones — dry 
only — were placed in these pots . . . [Farabee, 1916 a.] 

Unfortunately no sherd collections or ceramic observations are 
available from the Ilha dos Puercos — none were made by Farabee 
and none exist in the Museu Goeldi. Therefore, it is impossible at 
this time to verify Farabee's conclusion that the Ilha dos Puercos 
was the habitation site for the cemetery on Ilha do Para. The Ilha 
do Para zoomorphic vessels have a boxlike, rounded body on four 
short, stout legs, with the head projected forward on a thick, cylin- 
drical neck (pi. 17). Some also have a short, curved tail. In the 
middle of the back there is an elliptical opening, 16 by 24 cm., to the 

'In the published report, however, he says, "to about 3 feet" (Farabee, 1921, p. 154.) 



EVANsT^ ^^" ARCHEOLOGY AT MOUTH OF AMAZON 73 

hollow interior, over which a flat, oval, disk lid fits. The average 
body measures 45 cm. in length by 30 cm. in width and 11 to 15 cm. 
in height. The legs are hollow, 8 to 12 cm. high and 10 to 15 cm, in 
diameter, and have 4 applique toes along the front of each foot. 
The face diameter is slightly greater than the hollow neck, 7 to 8 cm, 
in diameter and 8 cm. long, on which it is supported. 

Although the bodies are similar in shape and proportions, the faces 
vary considerably in expression. The face on the end of the neck is 
flattened or slightly convex, and has eyes, nose, mouth, and eyebrows 
made from apphque nubbins and fillets. Along the top or the head, 
a row of applique knobs form a kind of headdress. Two faces have 
appendages on the chin similar to the beards found on some of the 
Rio Maraca anthropomorphic urns. The ceramic type is Vilanova 
Plain with cariape temper and a light gray-brown to tan surface color. 
The surfaces are smoothed but uneven and slightly irregular. 

These vessels collected by Guedes and Farabee provide the explana- 
tion for the large, stumpy, hollow foot found at Site A-3 (fig. 6). 
Although it is Mazagao Plain and shows slight deviation in the manner 
of construction of the toes, there can be no doubt that it belonged to 
one of these zoomorphic urns. Although the emphasis of modeling 
in this cemetery appears to be centered on the zoomorphic form of the 
turtle rather than anthropomorphic figures, a few fragments of 
anthropomorphic, tubular burial urns of the Maracd style were found 
with these zoomorphic vessels on the Ilha do Para (Guedes, 1897, 
p, 54). 

KIO MAZAGAO SITES 

Our only information on the drainage of the Rio Mazagao, the short 
river that flows south to empty into the Amazon slightly north of the 
halfway point between the Rios Vilanova and ^Maraca, is also fur- 
nished by Lima Guedes and Farabee. In the headwaters of one of its 
northern tributaries, the Igarape Frechal, in a region of higher land, 
Lima Guedes encountered a large cemetery with the jars buried in the 
ground. He worked there 2 days in order to obtain a few specimens 
not completely damaged by the roots of the large trees growing on the 
site. Lima Guedes reports finding a few fragments of tubular anthro- 
pomorphic m-ns and zoomorphic specimens of the Maraca type, but 
states that the majority were of forms similar to those on Marajo 
although extremely poor in ornamentation (op. cit., p. 55). Since he 
makes no reference to painted decoration, it is likely that what he 
looked upon as Marajo similarities are in reality similarities to vessels 
of the Mazagao Phase, which are plain or crudely ornamented and 
often vaguely similar in shape to Marajoara Phase examples. The 
relative scarcity of anthi'opomorphic and zoomorphic jars gives the 
impression that this site, like A-3 and A-4, is basically of the Mazagao 



74 BUREAU OF AMERICAN ETHNOLOGY [bdll. lez 

Phase with some Maraca influence. It is unfortunate that the inac- 
cessibility of the site prevented the salvaging of sufficient ceramic 
material to permit a more definite statement. 

In the Rio Mazagao drainage, Farabee visited a different site from 
the one explored by Lima Guedes. Our efforts to locate any specimens 
from this area in the Universit}^ Museum of Philadelphia failed; 
therefore, quoting directly from Farabee's field journal (MS., 1916 a) 
gives the full extent of our information: 

Sat. 19th Feb, 1916 — Punto das Panellas. Got up at 2:30 for tide and started 
by canoe with 3 men to Punto das Panellas in Lago do Rio Ajudante an eastern 
branch of the Rio Mazagao, an hour below the city. Arrived at 8 and went to 
work. 

This P. das P. is a peninsula reaching out a long way into the lake, which is a 
lake only in the wet season when it is very large, but even then it is so full of grass, 
rushes and piri that it is difficult to get through with a canoe — must pole with fork. 
There are islands of high land and other points jutting out into the lake. The 
burial place is not more than an acre in extent and a foot or two only above high 
water. 

The place has been known for a long time and the neighbors, rubber gatherers, 
have been going there to dig up water jars and flower pots. The larger pots were 
originally about level with surface and not difficult to find because if not in sight 
as many were, they all were covered with other inverted pots and if not broken 
before they would break with the weight of man walking over and leave hole into 
the pot. 

They in a rude way would attempt to dig them up and if they broke them they 
left them on surface so now dozens are to be seen on the ground covered with 
moss among the trees . . . [Here, Farabee diverts into a discussion of "treasure" 
and a snake guarding a pot full of "treasure" as told by the caboclos.] 

Instead of money there were a lot of bones and shell teeth beads. Pots 1 
and 2.' I found an interesting looking fragment half buried with top broken and 
gone. It had been a man seated on a stool. Inside was smallest pot and a lot 
of glass beads — some plain round blue, some oblong blue and white and red and 
white. Pots 3 and 4.^ By the side of this seated man was Pot 5 with four legs, 
tail and human face — like ones found on Ilha do Para. This had bones diseased ?; 
pieces saved. These (Pots) 3-5 belonged together no doubt. We spent 3 days 
digging with four men and found a number of small pots and fragments. One in 
form of jaboty [tortoise] — head and tail — a number of faces and heads of men and 
animals. These had no important relation the one to the other because of 
former digging. , . . 

In only one of many we dug up were glass beads and these in the most important 
burial of all. These rubber men were looking for gold so took everything out of 
scores of pots but they never found glass beads — they might easily have missed 
seeing bone and shell and teeth beads but glass ones are so noticeable they must 
have seen them if there were any. 

Farabee's description of the ceramics leads one to the conclusion 
that the Punto das Panellas site is related to the pottery from the 
Ilha do Pard and probably represents the same cultural fusion of the 
Mazagao Phase with the Maracd tradition. 

• No map or sketch accompanies the text or notes. 



m^NsT^ ^^^ ARCHEOLOGY AT MOUTH OP AMAZON 75 

IGARAP£ DO URXTB^ SITES 

On the Igarap6 do Urubti, which flows southeast to empty into the 
same mouth as the Rio Maracd, Lima Guedes (1897, pp. 54-55) 
heard of a site. Being unable to go himself, he sent four of his men 
to investigate it. They reported that they made various tests in the 
ground but found only sherds of vessels without any decoration. 
Since no anthropomorphic or zoomorphic urns were reported, in all 
likelihood this was a habitation site, 

RIO MARACA SITES 

The area best known from an archeological standpoint in the Terri- 
tory of Amapd prior to 1949 was the Rio Maracd. It was visited and 
excavated by Ferreira Penna in 1871, Lima Guedes in 1896, Farabee 
in 1916, and Nimuendaju in 1927. Specimens occur in many museums 
throughout the world, but the largest and best documented collection, 
the one made by Lima Guedes, has remained in the Museu Goeldi. 
His field observations (op. cit. pp. 43-47; 49-53) supplemented by our 
detailed examination of the specimens show the tradition to be deviant 
from the Mazagao Phase otherwise characteristic of the southern 
part of the Territory of Amapd. 

The cemeteries excavated by Lima Guedes are on branches of the 
upper Igarape do Lago, a large tributary entering the right bank of 
the Rio Maracd some distance above its mouth. This is a region of 
many hills and rocky outcrops, providing numerous small caves and 
niches, which were used for burial. The Ilha do Cunhahy, located 
on an igarape of the same name 20 km. north of its junction with 
the Igarape do Lago do Maracd, is one such spot. At various heights 
in a vertical outcrop 300 meters long and about 8 meters high along 
the southwest edge of the island were four small niches, all of which 
had been used for burial. Lima Guedes found a quantity of frag- 
ments of tubular, anthropomorphic urns in the form of a human 
being seated on a bench and one zoomorphic urn in the form of a 
turtle. It was in this same cave that he found fragments of bone 
in at least one of the jars and an almost complete skull on the floor. 
Since the skull is in rather good condition and the painted jar does 
not conform to the typical pattern of any archeological pottery in the 
Territory of Amapd, these were probably placed in the cave at a more 
recent date than the anthropomorphic urn burials. 

On the left bank of the Igarape Rio Branco, which flows into the 
Igarape do Lago do Maraca from the south, at a spot called "Ilha da 
Terrapreta," Lima Guedes found three more caves, two on the south 
and one on the east side of the hill. Falling rocks and prowling 
animals had broken many of the vessels and buried others up to the 
rim. They were in the cylindrical, anthropomorphic style like those 



76 BUREAU OF AMERICAN ETHNOLOGY [bull, igt 

at the preceding site, and each cave had several of the zoomorphic 
(mostly turtle) style of urn. A significant find, indicating a post- 
Columbian date, was one tubular, anthropomorphic burial urn with 
glass beads ornamenting its arms and spine (pi. 18, b). 

A third site was on a steep-sided hdl rising some 18 meters on the 
right bank of the Igarape do Lago do Maraca about a mile above its 
juncture with the Igarape Rio Branco. This hill was called "Ilha da 
Fortaleza" because of a trench along the edge of the flat summit. 
Although the local inhabitants attributed its construction to the 
Dutch, Lima Guedes thought this out-of-the-way location and the 
fact that the adjoining cemetery was undisturbed untU recently, 
argued for Indian origin. The cemetery occupied the summit of the 
hiU and was extensive, but all the jars had been recently broken by a 
disgruntled treasure seeker. His examination showed the vessels to 
be identical to the zoomorphic and tubular anthropomorphic burial 
urns he found at the Ilha do Cunhahy. Here, he also found five stone 
axes. 

Farabee's account (MS., 1916 a) of his visit to the Igarape do Lago 
of the Rio Maraca on February 29, 1916, is not explicit, but the 
presence of several complete jars indicates that this is not one of the 
sites visited by Lima Guedes: 

This cave under a shelving sand rock was at head of a now dry stream but 
later water runs from under the rock. The cave is 3' high in front, in center 
sloping back to nearly nothing at 10' deep and 60' long. Originally there had 
been many burial jars in shape of seated men sitting together under the rock but 
no doubt animals knocked down many of them and also men looking for treasure 
had broken all the large ones. Fragments only were left scattered but we collected 
five heads, 3 figures and some pieces. Also got one skull but no long bones had 
been preserved. 

At the back of the cave there were a number of rudely outlined heads in red 
and white paint. Some were aU red, others outline red with eyes, mouth and nose 
in white. Some were small, 6" across, others double that size. AU were meant 
to be round heads but the surface was irregular so many have corners on one 
side or other — Some had red lines but white eyes inside the red circles. Nothing 
but heads can now be made out and these are so covered by the ants [termite 
tunnels] that no entire figures can be seen. 

Our analysis of the Museu Goeldi collection, numbering 29 anthropo- 
morphic lids (heads) and 17 bodies seated upon benches, produced 
detailed information on the ceramic types. The poorly mixed paste 
contains a variety of tempering materials, ranging from small black 
ash particles in a clayey paste to a moderately sandy paste with fine 
sand particles (not ground quartz) and occasional specks of black 
ash intermixed. Only one vessel shows traces of mica. Although 
Hartt (1885, p. 40) states that cariape does not appear to have been 
used, a hand-lens examination shows some fine, black ash and cariape 
present in the majority of the vessels. Since one or two vessels lack 



^f.*l^*^^^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 77 

these black ash particles, an analysis based on a limited number of 
specimens might lead to a different description of the temper. Surface 
color is an even tan to orange brown, produced in an oxidizing atmos- 
phere, with only occasional fire clouding. Firing was sufficiently 
complete to penetrate the walls in only 25 percent of the vessels; in 
the remaining 75 percent a gi*ay core remains. These features set the 
pottery apart from that of the Mazagao Phase and require its classifi- 
cation as distinct from the described pottery types of the Temtory of 
Amapa. Since the Maraca style collections were restricted to a 
limited number of burial urns without the benefit of a large sherd 
sample from habitation sites, a new pottery type has not been 
established at this time. 

In form and decoration the contrast with the Mazagao Phase is 
sharper still. The most abundant and characteristic vessel is the 
cylindrical-bodied, anthropomorphic figure seated on a bench, forming 
the jar body, with a removable head forming the lid (pi. 18). Except 
for variation in size and proportion, the bodies show a uniformity in 
execution. The cylindrical torso is fixed to the top of the low bench, 
which has two legs running from end to end. Occasionally, a zoomor- 
phic head and tail are added to opposite ends of the bench. The legs, 
attached to the torso several centimeters above its base, slant down- 
ward at an angle of 30 to 40 degrees to join the short lower leg with its 
bulbous calf. The flattened feet rest upon the ground. The arms join 
the torso a little below the upper edge or rim and are spaced the same 
distance apart as the legs. The upper arm slants downward at the 
same angle as the upper leg and the lower arm rises vertically to 
join it, making the elbow bend upward. The two arm segments are 
approximately equal in length and the hands rest upon the knees. 
The genitalia of both female and male are realistically modeled, and, 
in the case of the male, several indicate the practice of circumcision. 
Small nubbins represent the breasts and a small pit, the navel. Some 
of the figures also have nubbins at the elbows, wrists, and ankles, 
which appear to be attempts to show the prominent bones at those 
places. Most of the figures wear one or a pair of bracelets on the 
wrist and on the upper arm. Many have a curved applique fillet 
between, or just above, the arm attachment similar to the curve 
assumed by a necklace. 

The heads fall into three basic styles (pi. 19) : (1) A straight-sided, 
truncated cone with a flat disk top; (2) a domelike shape with a 
rounded top and curving sides; and (3) a small rounded head with 
a flat top, constricted neck, and shoulders which widen out to the 
diameter of the jar mouth. The first group is the largest and most 
conventionalized; the third, the rarest and the most naturalistic. 
In aU three, the features are formed by appliques, sometimes supple- 



78 BUREAU OF AMERICAN ETHNOLOGY [bdli-. i6t 

mented with incising on the eyes and mouth, A long jfillet runs 
across the forehead down both sides and ends in a short curve or 
lateral extension. The long vertical nose is often joined to this 
at its root. About half of the heads with flattened tops (forms 1 and 3) 
have this area covered with short, conical nubbins arranged haphaz- 
ardly or in rows. The tops of two heads are decorated with 
incised lines. A typical feature of form 2 is a bunlike projection from 
the upper part of the back of the head, although this is sometimes 
also found on form 1. Form 3 is set apart not onlj'^ by its natm'alism 
but by the fact that each of the five examples in the Museu Goeldi 
collection and the one specimen in the University Museum collection 
(Farabee, 1921, fig. 44) have a slight protrusion running along the 
chin that suggests a beard. The fact that glass beads have been found 
associated with the Marac^ urns makes it seem probable that the 
beards were fashioned after European models and provides a lead for 
a possible time distinction in the three head styles. 

About half of the jars are painted a solid black, yellow, or red 
over which parallel-line and spiral designs in white or black were 
sometimes added. 

Many of the vessels were so badly broken that the contents had 
been lost; however, several of the specimens in the Museu Goeldi were 
still filled with cremated bones. Ferreira Penna (1879 b, pp. 50-51) 
states that the urns contained complete skeletons, disarticulated, with 
the pelvis at the bottom, the other bones along the sides, and the 
skull on top. Although the evidence is in part contradictory, it 
appears that both cremation and secondary burial were associated 
with tubular, anthropomorphic urns. 

Although our assessment of this Marac&. tradition cannot be 
complete until something is known of the habitation sites belonging 
to it, the ceramic type and the burial pattern are distinct from those 
of the Mazagao Phase. It appears to be concentrated geographically 
in the Igarape do Lago region of the Rio Maraca with a limited 
spill-over on the Ilha do Par^ and to have been contemporary with 
the latter part of the Mazagao Phase, upon which the Maraca tradition 
exerted a minor influence as evidenced from certain burial-urn styles 
found at Sites A-3 and A-4, on the Rio Pigac^. 

ANALYSIS OF MATERIALS OP THE MAZAGAO PHASE 

Pottery Type Descriptions 

All the sherds and vessels obtained in our excavations (5,126 sherds 
and 16 vessels), plus those specimens examined in various museum 
collections, were classified into pottery types using the currently 
accepted, binomial system of nomenclature, in which the first word 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



79 



refers to a local geographical name and the second is descriptive. The 
detailed pottery type descriptions, arranged in alphabetical order, 
follow: 

ANAUERAPUC1& INCISED 

Paste and surfaces: This decorated type is on Mazagao Plain material; see 

that type description for details of temper, surface, firing, and color. 
Forms: 

Common vessel shapes: 

1. Pedestal basin used as a lid on burial urns (fig. 16-1). 

Rims: Rim of the basin is generally thickened on the interior with 
a wide flange, 2.5-3.0 cm., upon which the incisions are placed. 
The rims of many have small adornos or lobes varying in length 
from 4-12 cm. and extending 1.0-1.2 cm. beyond the rim edge, 
which is rounded. Rim of the pedestal base is flat topped or 
rounded with a slight external thickening, measuring 1.0-1.3 cm. 
Body wall thickness: Basin, 8 mm.; pedestal, 9 mm. 
Body dimensions: Basin mouth diameter, 36-40 cm.; with the bowl 

depth 10-12 cm. Total height of the vessel 20-22 cm. 
Base: Tall, cylindrical pedestal with a slight outflare, open at the 
bottom. Base diameter 20-22 cm. and 18 cm. at the point of 
attachment with a height of 12 cm. 
Decoration: Basin — Panels of incised horizontal lines with diagonal 
and squared spirals along the inner lip of the rim, separated on 
some by human faces modeled in applique. Many basins have 






1 2 3 CM 
Rim Seal* 



I I ' ■ I I ' 

4 8 12 CM 

Vastel Seal* 



Figure 16. — Rim profiles and vessel shapes of Anauerapucu Incised, Mazagao 
Phase (Appendix, table 2). 



80 BUREAU OF AMERICAN ETHNOLOGY [bull, le? 

no decoration of the exterior, but a few have parallel, incised lines 
without diagonals, rectilinear spirals, etc. 

Pedestal — Exterior is incised in a paneled band, 4-6 cm. wide, 
beginning just below the thickened lower rim, with the typical 
motifs. 

2. Bowls with everted, exteriorly thickened or direct rims (fig. 16-2). 

Rims: Everted v.dth a wide, flat surface and a round lip; externally 
thickened, ranging from 1.0-1.5 cm. in thickness; sometimes 
direct. 

Body wall thickness: 8-10 mm. 

Body dimensions: Height 6-15 cm.; mouth diameter 20-26 cm. 

Base: Rounded. 

Decoration: Horizontal, parallel lines limited to the flat, upper 
surface of the flanged rim; an occasional incised line on the ex- 
terior below the rim. 

3. Small, open, sometimes carinated, bowls with direct rims (fig. 16-3). 

Rims: Direct rim with either rounded or flattened lip, mouth 

diameters ranging from 14-26 cm. 
Body wall thickness: 5 mm. 
Body dimensions: Height 10-15 cm. 
Base: Rounded. 

Decoration: Incisions limited to the upper portion of the exterior 
walls of bowl, just below the lip in a panel averaging 2 cm. wide. 
Less common shape: 

1. Small carinated bowl with inslanted rim and a strap handle (pi. 7, d). 
Rim: Inslanted, rounded lip; mouth diameter 9 cm. 
Body wall thickness: 5-6 mm. 

Body dimensions: Diameter 13.5 cm.; height 5.6 cm. 
Base: Slightly flattened. 

Appendage: A single, strap handle with a slight groove down the 
center, 1.5 cm. wide, and attached from the lip edge to the 
shoulder. 
Decoration: Rectilinear, squared spirals between pairs of straight 
lines on the insloping wall of the bowl. 
Decoration (pi. 11): 

Technique: Fine-line, sharp, moderately deep incisions on inner lips of bowls 
or on the exterior walls of the pedestal bases or on small bowls ranging in 
technique from carefully executed incisions to a few crude specimens, but 
always done with greater care than Uxy Incised. Many lines are so 
straight and accurately drawn that they appear as if drawn with a straight 
edge. Seventy-five percent of the precise, well-executed incisions are 
filled with white chalk (a type found in natural outcrops in the region), 
with faint traces often suggesting a more widespread usage than the badly 
eroded sherd material indicates. The depth of the incisions before filling 
with chalk ranges from 1-2 mm. with the width from 0.5-1.0 mm. 
Motif: Typically horizontal, parallel and diagonal lines, evenly spaced in 
combination with frets, rectilinear and squared spirals. Curved lines are 
rare. The designs are all well spaced and regular and commonly arranged 
in panels (fig. 14, b, and pi. 11). A few forms combine a crude applique 
modeling of human faces on rim lobes with panels of elaborate incision in 
betv/een (fig. 14, a). 
Tempobal differences within the type: The rim and vessel form analysis 
(Appendix, table 2) suggests a decline in popularity of the pedestal basins 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 81 

EVANS] 

(form 1) in the cemeteries in favor of an increase of smaller open bowls without 
a pedestal (form 2). The pedestal basin form 1 does not appear until the middle 
of the Mazagao Phase sequence. 
Chronological position of the type: Appears only in the upper (late) part 
of the Mazagao Phase sequence. 

CAMAIPf PLAIN 

Paste: 

Method of manufacture: Coiling. 

Temper: Cariapi (ash of siliceous bark) and finely ground quartz particles. 
An explanation of cariapi temper is best given by a direct quotation: 
"Usually it is said that the bark is taken from the cariapi tree. This name 
seems to me to be a common designation for various plants, such as, for 
example, Bignoniacea, the genera Moquilea and Licania utilis, Turiuva, etc. 
Unfortunately, our sources are not explicit concerning the species of the tree 
in question. . . ." (Linn6, 1931, pp. 206-207). "The bark is burnt, 
whereupon it is ground and mixed with the clay. The burning is done for 
the purpose of removing the organic components which otherwise would 
lessen the durability of the vessels in the firing" (Linne, 1925, p. 38). 
The quartz is less than 5 percent of the mixture, but is definitely a conscious 
mixture and not merely due to a naturally sandy clay. All the quartz 
particles are quite granular and sharp, indicating the deliberate crushing 
and intermixing, rather than waterworn sand. Cariapi and quartz make 
the paste gritty. It is also very porous and has numerous black spots from 
the burnt ash; the sihceous particles from cariap4 are white, columnar, and 
cellular as viewed under the microscope. 
Texture: Fine, sandy, gritty. Halfway between Vilanova and Mazagao Plain 

in paste and texture. 
Color: Majority of all sherds have a gray core flecked with black and white 
particles of cariap6 and are bordered on the exterior and interior with a thin 
band of gray tan or orange tan. Rest have a solid gray-orange color. 
Firing: Oxidizing fire with 75 percent of the sherds incompletely fired; fire 
clouds rare. 
Surfaces: 

Color: Exterior and interior — Generally, both surfaces are the same, ranging 
from a dull gray brown to a light gray to a tile orange. The majority 
are gray orange or tile orange. Less than 1 percent are gray brown on the 
interior with a bright-orange exterior. 
Treatment: 

Exterior— Well-smoothed, even, with smoothing striations visible in 
most cases. Thirty-five percent with a very slick, even surface but 
not highly poUshed with a luster. All coils erased. 
Interior— Typically, the same treatment as the exterior; however, 25 
percent are slightly rougher and less well smoothed. One percent 
have a slick, gray-brown interior. 
Hardness: 2-2.5. 
Form: 

Common vessel shapes: 

1. Globular jars with short neck and vertical or everted rim (fig. 17-1; 
pi. 8, a, 6, d). 
Rim: Vertical or everted, unthickened or externally thickened 

usually with a flat or sloping flange, and rounded lip. 
Body wall thickness: 6-11 mm.; majority 8 mm. 



82 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 





Rim Seal* 



m 



'''III' 

4 8 12 CM 



V««««l Scol* 




)J) 



Figure 17. — Rim profiles and vessel shapes of Camaipf Plain, Mazagao Phase 

(Appendix, table 3). 

Body dimensions: Diameters 32-40 cm.; majority 32-36 cm. 

Base: Slightly flattened; a few rounded. 

Appendages: Occasionally, a small, irregular loop handle from the 
lip downward to the neck of the jar; cross section is oval to round, 
measuring 8-10 mm. Diameter of loop handle ranges from 2-4 
cm. 



KV^^Nsf ^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 83 

2. Cylindrical jars (fig. 17-2; pi. 8, c). 

Rim: Direct rim; lip is rounded or squared with rounded corners. 
Body wall thickness: 6-10 mm. 
Body diameters: Usually 18-25 cm. 

Base: Flattened; usually a slight pedestal 0.5-1.0 cm. tall, diameter 
8-14 cm. 

3. Open bowls with outcurving or slightly incurving sides (fig. 17-3). 

Rim: Rounded or square with rounded edges; occasionally exteriorly 
thickened. Mouth diameters 20-40 cm.; majority 24-32 cm. 

Body wall thickness: 4^10 mm.; majority 6-8 mm. 

Base: Rounded or flat. 
Temporal differences within the type: In habitation sites the globular jar 
with a short to medium neck (form 1) tends to decrease in popularity while 
open bowls with outcurved sidewalls (form 3) increase (see Appendix, table 3). 
At the point of time in the sequence where Camaipi Plain appears in highest 
percentage (i. e., Site A-6) form 1 is as high as 71.5 percent of all vessel forms. 
The tall, cylindrical jars (form 2) do not appear until the nliddle-upper part 
of the Mazagao Phase sequence. 
Chronological position of the type: The type was introduced in the early 
part of the Mazagao Phase when the differentiation of paste between sand 
and canape-tempered pottery was not too clear and the firing was not too well- 
controlled; then as the pottery tends to crystalize into Mazagao Plain and 
Vilanova Plain, Camaipi Plain decreases in popularity. 

JARf scraped 

Paste: 

Method of manufacture: Coiling. 

Temper: Crushed quartz particles, generally finely ground. 

Texture: Coarse, gritty, friable, irregular fracture. Very uneven mixture of 

temper throughout the paste. 
Color: Dull orange to rusty brown. 
Firing: Oxidizing fire; generally complete; no fire clouds. 

Surfaces: 

Color: Dull, gray tan to dusty brown. 

Treatment: Exterior and interior — All coiling lines erased, smoothed, but 
slightly irregular and uneven. A few show the smoothing striations on 
the surfaces. 
Hardness: Sand temper gives an abnormal hardness, for the sand particles 
protrude; actually the paste is fairly soft (2.5). 
Forms: 

Common vessel shapes: 

1. Small bowls (fig. 18-1). 

Rims: Rounded lip with slightly everted rim or rounded direct 

rim, both with incurved sidewalls. 
Body wall thickness: 5 mm. 
Base: No sherds found; probably same as Mazagfto Plain form 

4, i. e., round or flat. 
Body dimensions: Maximum bowl diameter is 2-4 cm. greater than 

the mouth, which is 12-15 cm. 
Decoration: Scraped in a series of parallel lines, diagonal units or 
curvilinear elements on the exterior. 



84 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 187 





4 8 12 CM 
Vessel Scole 





I 2 3 CM 
Rim Scale 



Figure 18. — Rim profiles and vessel shapes of Jari Scraped, Mazagao Phase 

(Appendix, table 4). 

2. Small jar with everted rim (fig. 18-2). 

Riin: Slightly thickened on the interior, everted; thin, tapered and 

rounded lip. Mouth diameter 8 cm. 
Body wall thickness: 6 mm. 
Body diameter: 12-14 cm. 
Base: No sherds found. 

Decoration: Exterior scraped in a series of parallel lines, diagonal 
units or curvilinear elements, sometimes with applique nubbins 
on jar shoulder. 
Decoeation (pi. 15, h-g): 

Technique: Exterior is lightly scraped with a narrow (3 mm.), flat, blunt 
tool, with the scrapings spaced 3-5 mm. apart, each line separately applied, 
but sometimes overlapping. 
Motif: A series of parallel lines, diagonal units or curvilinear elements. 
Temporal differences within the type: None observable in the small sample 

(Appendix, table 4). 
Chronological position of the type: An early pottery type, limited to the 
lower part of the Mazagao Phase sequence. 



BvTNsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 85 

mazagao plain 
Paste: 

Method of manufacture: Coiling. 

Temper: White mica particles (muscovite) and ground quartz, ranging from 
fine particles to large angular chunks, sometimes 1.5 cm. long (pi. 10). 
Texture: The naturally gritty clay, coupled with the quartz, mica temper 
makes a very coarse, friable and irregularly fractured paste. All the 
sherds have a hollow, high-pitched ring, like bricks. The poor mixture 
of quartz hunks often makes weak zones which fracture easily. 
Color: Ranges from a light, tile orange to a light, brick red; 25 percent of 
the sherds have a thin gray core bordered with bands 2-3 mm. wide of 
orange or light brown. 
Firing: Oxidized fire, generally complete; few fire clouds. 
Surfaces: 
Color: 

Exterior — Ranges from brick red to an orange red, to a brown or a 

light orange tan to a rusty brown; the majority are a dull, brick red. 

Interior — Identical in range to the exterior, except that 1 percent of the 

sherds are a gray black and slick as if rubbed with jutahy resin before 

firing. 

Treatment: 

Exterior — Well-smoothed, even and fairly regular surfaces with all 
the coiling lines erased. Fifty percent of the sherds show distinct 
smoothing striations. A few have a filmy surface finish developed 
from smoothing by hand when the clay is very wet. A small percent 
of the sherds have a slick exterior as if polished when leather hard. 
Interior — Identical treatment to that of the exterior is typical. Occa- 
sionally the interior is slightly more irregular and rough compared to 
the exterior. One percent of the sherds are well polished and slick. 
Hardness: Sand and quartz particles make it seem harder; actually 2.5. 
Form: 

Common vessel shapes: 

1. Round-bodied jar with thickened and everted rim and a short inslop- 
ing to outsloping neck (fig. 19-1; pi. 9, a-c). 

Rim: Exterior rim thickenmg in the form of an added strip or coil 
varies the profile from a rounded to a flat top with squared edges. 
Mouth diameter, 15-36 cm. 
Body wall thickness: 4—11 mm. with 25 percent of all sherds a heavy, 

coarse variety ranging from 1.2-2.2 cm. Majority 7-8 mm. 
Body diameters: 25-50 cm.; neck height 3-12 cm.; total height 20- 

45 cm. 
Base: Slightly thickened; round or flattened. Flat the most 

typical. 14^16 cm. in diameter. 
Appendages: Handles not typical but sometimes a short, strap 

handle extends from the lip to the jar neck. 
Occasional decoration: Applique fillets, nubbins, vertical and hori- 
zontal bars with occasional light incision on the appliques form 
human faces on the lids or necks of jars; sometimes bird or 
animal head adornos on jar collars; 10 percent of the large jars 
have such ornamentation. 



391329—57- 



86 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 




rf({ 



I a sen 



iy/// 




4 a 12 CM 
VttttI 8c9l« 



Figure 19. — Rim profiles and vessel shapes of Mazagao Plain, Mazagao Phase 

(Appendix, table 5). 

2. Tall, cylindrical jars with short pedestal bases and direct rims 
(fig. 19-2). 

Rim: Direct, flat- topped and square with slightly rounded edges; 

mouth diameter 16-22 cm. 
Body wall thickness: 4r-ll mm.; majority 7-8 mm. 
Body dimensions: 18-26 cm.; height 35-40 cm. 
Base: Flat, short pedestal; 0.5-1.5 cm. high, 10-15 cm. in diameter. 

3. Round-bodied jar with unthicliened rim and vertical or outsloping 

neck (fig. 19-3). 

Rim: Direct; lip rounded or square with rounded edges; mouth 

diameter 16-30 cm.; neck 3-12 cm. 
Body wall thickness: 4—11 mm.; majority 7-8 mm. 
Body diameters: 25-50 cm. 
Base: Slightly thickened, rounded or flattened; flat base usually 

14r-16 cm. in diameter. 
Appendages: Rarely, a short strap handle extends from the lip to a 

few centimeters on the jar neck. 

4. Small, open bowl with gently curved sidewalls (fig. 19-4; pi. 7, c). 

Rim: Rounded, sometimes squared with rounded edges, or tapered 
or with a slight thickening on the interior; mouth diameters 18- 
36 cm. 

Body wall thickness: 4-11 mm.; majority 7-8 mm. 

Base: Rounded or flat. 



k^vTnsT^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 87 

Rare vessel shapes: 

1. Anthropomorphic figure seated on a clay bench (after the Maracd 

urn style). 

2, Flat- topped lid with a short annular neck. Found only at sites 

along the Igarap6 Muriaca on the Rio Iratapurii. See Uxy 
Incised — form 2 for details. 

Temporal differences within the ttpe: A steady decline in popularity of the 
large globular to round-bodied jars with short to medium necks and thickened 
rims (form 1) from 80 percent to 20 percent in the habitation sites (see Appendix, 
table 5) with a slight increase in small to medium round-bodied jars with ver- 
tical to outsloping necks and un thickened rims (form 3) from the lower (earliest) 
to upper (late) part of the Mazagao Phase sequence. The early varieties of 
Mazagao Plain tend to be a grayish brown (cf. Site A-6) while the late varieties 
of Mazagao Plain (cf. Sites A-2 and A-1) are a bright red to brown red. An- 
thropomorphic and applique features on the neck are more common in the later 
part of the Phase. The tall, cylindrical jars (form 2) do not appear until the 
middle to upper part of the Mazagao Phase sequence. 

Chronological position of the type: Decreases in popularity during the 
MazagSo Phase, but is present throughout the entire time span of the Phase. 

PigAcX INCISED 

Paste and surface: This incised varietj' occurs typically on Vilanova Plain 
paste, occasionally a few decorated sherds are on Camaipi Plain; see those 
pottery type descriptions for details of paste, temper, color, etc. 

Forms: 

Common vessel shapes: 

1. Open bowls with direct rims and outcurving sides, sometimes carinated 

(fig. 20-1). 

Rims: Direct, shghtly rounded or tapered; rarely thickened. 

Body wall thickness: 6-10 mm.; majority 8 mm. 

Body dimensions: Mouth diameters 18-34 cm.; height 12-20 cm. 

Base: Usually rounded. 

Decoration: Incision is usually limited to a band, 1-2 cm. wide, of 
horizontal, parallel lines extending from 1 cm. below the rim and 
found only on the exterior. Motifs limited to wide panels of 
diagonal lines or rectilinear meander and parallel bands bordered 
by a row of punctates. A combination of punctates bordering 
or interspersed with parallel lines often appears on the inner lip, 
especially with the punctates around the lip edge. 

2. Pedestal-basin hds (fig. 20-2), 

Rims: Beveled with an outflaring, roundsd Hp on the basins. The 

pedestal base is rounded with the exterior thickened with a 

smoothed-over coil. 
Body wall thickness: Basin 8 mm; pedestal 10 mm. 
Body dimensions: Basin mouth diameter 32-38 cm.; basin height 

10-12 cm.; base diameter 25 cm. with a height of 9-10 cm. 
Base: Pedestal type, which is a tall, cylindrical collar attached to 

the bottom of the bowl. 
Decoration: Basin — Interior and exterior of the beveled rim is 

incised with horizontal, parallel lines as on the smaller bowls. 

Pedestal base — Usually the pedestal is incised with 8 parallel 

lines in a band 2 cm. wide bordered with deep, sharp, circular 



88 



BUREAU OF AMERICAN ETHNOLOGY 



[BDLL. 167 




I 2 3 CM 
Rim Seals 




I I ' I ' ' I 

4 8 12 CM 



Vessel Scale 






Figure 20. — Rim profiles and vessel shapes of Piijacd Incised, Mazagao Phase 

(Appendix, table 6). 

punctations (2 mm. in diameter) in a slightly wavy line around 
the upper edge of the incisions. 
3. Open, carinated bowls (fig. 20-3). 

Rims: Exteriorly thickened, outcurviug, rounded lip; mouth 
diameter 36 cm. 



"vA^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 89 

Body wall thickness: 8 mm. 

Body dimensions: Body diameter 30 cm.; estimated height 14 cm. 
Base: Rounded. 

Decoration: Parallel lines and rectangular units in a band on the 
carinated exterior sidewall. 
4. Small jars with short neck and slightly outcurved rim (fig. 20-4). 

Rims: Outcurved, slightly thickened on the exterior, lip rounded or 

squared with rounded edges; mouth diameter 12-18 cm. 
Body wall thickness: 6-10 mm.; majority 7 mm. 
Body dimensions: Body diameter 16-26 cm.; height 14-20 cm. 
Base: Rounded. 

Decoration: Parallel lines around the neck, sometimes with punctates 
on the lip. 
Decoration (pi. 12) : 

Technique: Sharp incised lines, very straight and well executed, evenly spaced 
and 0.5-1.0 mm. wide with an average depth of 1 mm. Chalk-filled in- 
cisions like those of Anauerapucii Incised are very rare. 
Motif: Parallel lines, rectilinear meander and horizontal lines with an occa- 
sional use of diagonal lines. No use of the squared spiral so typical of 
Anauerapucu Incised. A limited use of light circular punctates bordering 
the incised units. Three to five parallel lines are the most common 
combination. 
Temporal difference within the type: The pedestal-basin lid (form 2) appears 
in the middle-late part of the Mazagao Phase sequence; no other form trends 
are evident (see Appendix, table 6). 
Chronological position of the type: It is contemporaneous with Anauerapucii 
Incised, that is, the upper (late) part of the Mazagao Phase sequence, increasing 
slightly in popularity throughout time. 

UXY incised 

Paste: 

Method of manufacture: Coiling. 

Temper: Quartz and mica the most typical with about 25 percent of the 
sherds with only quartz particles; temper fairly well mixed in the paste. 
Texture: Sandy, gritty, friable with an irregular fracture. 
Color: Dull gray orange to rusty brown with a few sherds showing a thin 

gray core bordered by Ught orange. 
Firing: Incompletely oxidized. 
Surfaces: 

Color: Exterior and interior — Dull gray tan to dusty brown or rusty orange. 
Treatment: Smoothed but irregular and uneven, very porous and rough and 

gritty. 
Hardness: 3. 
Form: 

Common vessel shapes: 

1. Carinated or slightly carinated bowl (fig. 21-1). 

Rims: The most common form is slightly everted, thickened on the 
interior and gently curving outward with a rounded or tapered 
lip. This rim is 1-3 cm. long and forms a rather pronounced 
angle at the shoulder where it joins the bowl. Less commonly 
the rim is either rounded, unthickened and outcurving or everted 
and thickened externally with a flat top (flange) measuring 1.0- 
1.8 cm. wide. 



90 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 




T-r'f 



— D 





I I ' I ' ' I 

4 8 12 CM 

Vessel Scale 



LjU I 1 

I 2 3 CM. 

Rim Seal* 



Figure 21.— Rim profiles and vessel shapes of Uxy Incised, Mazagao Phase 

(Appendix, table 7). 

Body wall thickness: 6-9 mm. 

Body dimensions: Mouth diameters range from 18-38 cm.; majority 

22-26 cm.; bowl depths 5-15 cm. 



BVANsf^ ^^'^ ARCHEOLOGY AT MOUTH OF AMAZON 91 

Base: Rounded. 

Decoration: A few specimens have a series of wavy lines on the 
inner lip; however, the typical design is limited to the exterior 
surface from the shoulder to the rim in a band from 2-5 cm. wide. 

2. Shallow bowl with flat-topped, everted rim (fig. 21-2). 

Rims: Flat top (flange) 1.0-1.5 cm. wide, everted rim with rounded 
lip, sometimes slightly tapered. Mouth diameter 22-26 cm. 

Body wall thickness: 6-9 mm. 

Base: Rounded. 

Decoration: Same as for form 1 with incisions sometimes on the flat 
top (flange) of the rim. 

3. Flat lid with short, annular base (fig. 21-3). 

Rims: Rim of the lid is rounded with no thickening; edge of base is 

rounded and thickened. 
Body wall thickness: 8 mm. 
Body dimensions: Lid top diameter 12-18 cm.; base diameter 12-14 

cm.; height of annular base 3-6 cm., with upper edge inset 1 cm. 

from edge of top; overall height 4-7 cm. 
Decoration: Typical motifs of the type, limited to the flat top of lid. 

4. Short, vertical-necked jar with rounded body (fig. 21-4). 

Rim: Externally thickened as if by the addition of a coil. Vertical 
or slightly outslanting. Lip squared with rounded corners. 
Mouth diameter 18-26 cm. 
Body wall thickness: 7-10 mm. 

Body dimensions: Body diameters 20-30 cm.; height 18-30 cm. 
Base: Rounded. 

Decoration: Design limited to the neck in a band 2-5 cm. wide. 
Rare vessel shapes: 

1. Shallow, open bowl. 

Rim: Interior thickened, producing a broad band 1.5-3.0 cm. wide; 

rounded lip. Mouth diameter 22 cm. 
Body wall thickness: 7 mm. 
Base: Rounded. 

Decoration: Incised lines around the exterior in a band in the form 
of incised concentric circles, with lines in between which form 
crosses. 
Decoration (pis. 13, 14, 15, a): 

Technique: Deep, sharp incisions applied with a sharp instrument when 
the clay is extremely wet. A characteristic trait of the incisions is the 
very irregular, jagged, uneven lines, giving the impression of their having 
been done in haste. Depth of incisions varies from 1-3 mm. with the 
width typically 1.0-1.5 mm.; a few are 2.5 mm. wide. Occasionally, 
applique adornos or rim nubbins are found. 
Motif: Semirectilinear and/or curvilinear motifs, ranging from simple 
curved or wavy lines to interlocking frets, spirals, diagonals, parallel lines 
and triangular units. 
Temporal differences within the type: The various forms show little trend 
of change through time (Appendix, table 7). All the forms carry on in later 
pottery types except form 3, the flat lid, which is limited to Uxy Incised and the 
lower (early) part of the Mazagao Phase sequence. 
Chronological position of the type: Uxy Incised is the principal decorated 
type in the early part of the Mazagao Phase, declining through time as it is 
replaced by Anauerapucd and Picacd Incised. 



92 BUREAU OF AMERICAN ETHNOLOGY [bdll. le? 

vilanova plain 
Paste: 

Method of manufacture: Coiling. 

Temper: Black and white flecks of burnt cariape (see Camaipi Plain for 
detailed discussion of this temper) which varies from 10-25 percent of the 
mixture with the largest particles 5 mm., the average speck only 1 mm. 
and the white cellular siliceous particles only visible under magnification. 
Texture: Fine, siliceous cariap4 temper gives a pumicelike feel. Light 
weight of the sherds is due to the temper. Broken edges rub off easily 
into a white, chalky powder. Tensile strength weak, but not friable. 
Dull, flat ring like a poorly cast bell. Fine air pockets are visible in 
cross section. 
Color: Ninety percent of all sherds have a gray core flecked with black 
cariap& and bordered on the exterior and interior with a thin band of 
gray tan or orange tan. Ten percent are a solid gray orange. 
Firing: Oxidizing fire with 75 percent of the sherds incompletely fired; fire 
clouds relatively rare. 
Surfaces: 

Color: Exterior and interior — Generally, both surfaces are the same, ranging 
from a dull gray brown to a light gray to a tile orange. Most are gray 
orange or tile orange. Less than 1 percent are gray brown on the interior 
with the exterior a bright orange. 
Treatm,ent: 

Exterior — Well-smoothed, even with smoothing striations visible in most 
cases. Thirty-five percent -with a very slick, even surface but not 
highly polished with a luster. All coils erased. 
Interior — The same treatment as the exterior is typical; however, 25 
percent are slightly rougher and less well smoothed. One percent 
have a slick, gray brown interior. 
Hardness: Very soft, 2. 
Forms: 

Common vessel shapes: 

1. Globular jars with short neck and vertical or everted rim (fig. 22-1; 
pi. 9, d). 

Rim: Vertical or everted, unthickened or externally thickened usu- 
ally with a flat or sloping flange and a rounded lip. 

Body wall thickness: 6-11 mm.; majority 8 mm. 

Body diameters: 32-40 cm.; majority around 32-36 cm. 

Base: Slightly flattened, a few rounded. 

Appendages: Occasionally, a small, irregular loop handle from the 
lip downward 2-4 cm. on the jar neck with an oval or round 
cross section, 0.9-1.8 cm. in diameter. 

Occasional decoration: Not typical, but on a few globular jars an- 
thropomorphic faces and body features are modeled with applique 
and light incisions on the neck and collars and jar bodies. One 
vessel was modeled in the form of an anthropomorphic figure 
seated on a clay bench. 

2. Cylindrical jars (fig. 22-2). 

Rims: Direct rim; rounded lip or squared with rounded corners. 
Body wall thickness: 6-10 mm. 

Body diameters: 18-25 cm.; mouth diameter 12-20 cm.; height 
30-40 cm. 



^J.^l^ff^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 93 

EVANS J 




Rim Scala 





I ' I I I I I 

4 8 12 CM 

Vessel Scale 




Figure 22. — Rim profiles and vessel shapes of Vilanova Plain, Mazagao Phase 

(Appendix, table 8). 



Base: Flattened, usually -Rith a slight pedestal 5-10 ram. high; 
base diameter 8-14 cm. 
3. Open bowls with outcurving or slightly incurving sides (fig. 22-3). 

Rims: Rounded or square with rounded edges, occasionally exte- 
riorly thickened. 

Body wall thickness: 4-10 mm.; majority 6-8 mm. 

Body dimensions: 20-40 cm.; majority 24-32 cm. 

Base: Rounded or flat. 



94 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

Less common vessel shapes: 

1. Modeled anthropomorphic figure seated on a clay bench in the 

Maracd urn style (see fig. 12 and pi. 3, b). 

2. Zoomorphic urns in the form of a jaboty (land turtle) with a flat lid 

(pl. 17). 

Temporal differences within the type: Large, round or globular jars with 
short to medium necks and thickened rims (form 1) increase in popularity 
throughout time while open bowls (form 3) decrease. Note that this is just 
opposite the rim and vessel form trend in Mazagao Plain, suggesting that as 
Mazagao Plain loses in popularity to Vilanova Plain there is a comparable shift 
in vessel and rim forms. The tall, cylindrical jars (form 2) do not appear until 
the middle-upper part of the Mazagao Phase sequence (Appendix, table 8) . 

Chronological position of the type: Vilanova Plain appears in the lower- 
middle part of the Mazagao Phase, at the same time as Camaipf Plain is in- 
troduced, and continues to increase in popularity until it is the dominant 
plain pottery type in the upper (late) part of the sequence. 

unclassified decorated 

A few sherds from Mazagao Phase sites show untypical types of decoration, 
which were too rare to warrant the establishment of a separate pottery type. 
The techniques represented are scraping, red painting, and punctate. 

Unclassified scraped: 

1. Small, short-necked jars with an everted, externally thickened rim and 

rounded lip. Mouth diameter 12 cm.; strap handle from lip to upper 
shoulder. Vertical or slightly diagonal scrapings on shoulder and neck; 
evenly spaced, 7-8 mm. apart, 1 mm. wide and 0.5-1.0 mm. deep. 
One sherd from Site A-2 — Lauro and 1 sherd from Site A-5 — Cafezal. 

2. Body sherds with markings similar to those described above but less 

regular and less evenly spaced. Two sherds from A-3 — Pigacd 
Cemetery. 
Unclassified painted: 

1. Shallow bowl with a flat base, outcurved sides and a double, scalloped rim 

(2 rows of scallops) formed by gently curved lobes 1 cm. wide and 3-4 
cm. long (pl. 16, h). Base diameter 15 cm.; lip diameter 20 cm.; height 
3.5 cm.; flat base slightly thickened to 9 mm. from a body wall thickness 
of 5-7 mm. Red ochre rubbed on the inner lip of the scallops. From 
Site A-2 — Lauro. 

2. One sherd from a globular jar with a medium-length, outcurved to vertical 

neck and a thickened to everted rim. Mouth diameter 18 cm.; body 
diameter 27 cm.; neck height 4 cm. Bright red ochre rubbed on the 
neck and upper shoulder in a band 6 cm. wide. Below this are low 
applique ribs. From Site A-4 — Valentim. 
Unclassified punctate: 

1. One sherd from a shallow bowl with small rim lobes and a single row of 
small punctates along the interior edge (pl. 16, c). From Site A-4 — 
Valentim. 

Pottery Artifacts 

No pottery artifacts were found in the 1949 excavations. However, 
the collections in the Museu Goeldi made by Nimuendajii from sites 
on the Rio Iratapurij, which seriate in the early part of the Mazagao 



^vA^Nsf^ ^'■''* ARCHEOLOGY AT MOUTH OF AMAZON 96 

Phase, produced a possible spindle whorl (pi. 16, g) and fragments that 
may have belonged to figurines (e. g. pi. 16, 6). A small modeled foot 
that may have had a similar function came from Site A-6 — Ilha das 
Igagabas (pi. 16, d). 

NoNCERAMic Artifacts 

Unfortunately, the number of nonceramic artifacts is so few that no 
significant classification is possible. Although available in this part 
of the Amazon, stone was not a primary source of materials for tools, 
and the wealth of objects that must have existed of wood and other 
plant fibers did not survive for the archeologist in a tropical climate. 
Tabulating the nonceramic materials from the Mazagao Phase sites 
presents the following : 3 pieces of yellow ochre with smoothed surfaces 
and scratches suggesting use as rubbing stones; 1 red ochre fragment 
Avith the same features and use; 2 pieces oijutahy resin, use unknown; 
1 piece of white chalk, probably the material used to fill the incisions of 
Anauerapucii Incised; 16 burnt-clay fragments, probably from clay 
used in a hearth on the floor of pile dwellings or upon a pedestal stand 
as is the custom today among the Indians and caboclos of the Amazon ; 
105 natural rock fragments of which 47 were fire burnt; 4 hand-ax 
fragments roughly shaped from natural, waterworn rocks which also 
might have been used as hammerstones ; 1 well-polished, ungrooved ax 
fragment; 2 pebble pottery smoothers; 1 grooved, sandstone "shaft- 
smoother"; and 2 unworked percussion flakes which could have been 
used as scrapers. The fairly rich pottery traits of the Mazagao 
Phase make the sparsity of the other artifacts all the more noticeable. 

Glass trade beads were listed in detail at each site from which they 
were found; hence no repetition is required here. The reader is re- 
ferred to table A (p. 51) and table B (p. 58) for details. 

CERAMIC HISTORY 

The region bounded on the north by the Rio Araguari-Amapari 
and on the south by the Rio Jari (omitting the Rio Marac^, tradition 
for the moment) produced archeological material belonging to a single 
complex, designated as the Mazagao Phase. The refuse deposits of 
the vUlage sites were too shallow for stratigraphy except at Site 
A-2 — Lauro, where the refuse 45 cm. thick provided a partial basis 
for seriation of the ceramic types from the other sites. This shallow- 
ness would suggest that a short period of time is involved; never- 
theless, the changes in popularity of various wares are evident (fig. 
23; Appendix, table 1). Since there was apparently a conscious selec- 
tion of certain decorated pottery types for burial purposes, such as 
Anauerapucii Incised, it would have created false impressions and 
trends to interdigitate the cemeteries directly into the habitation 



96 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 



:5y 



-<!=! 



Q - 






> bi O 
Ul o 



< a 
UJ oot 

2 ? 

< 




■ I 



D B 





C 




o 




03 


a-z 


a; 


<< 


CG 


2_j 




<Q. 


(U 


O 


c3 




^ 




Ah 



< a. 



-T^^ 



j2 
u in 

> — 



<3 — 



i< 1 


n 


o 


- 


o 


J 


o 



CM CJ ji :l 



S3IU31 
-3W30 



SNOIlVXiaVH 



"vA^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 97 

sites. However, an examination of the general popularity trends of 
the various pottery types, as well as a comparison of the decorative 
style and technique of certain wares, correlates cemetery Site A-3 — 
Pigaca Cemetery with the upper level of both cut 1 and cut 2 of 
occupation Site A-2 — Lauro and cemetery A-4 — ^Valentim with habi- 
tation Site A-1 — Pigaca. 

European glass trade beads, found in cemeteries A-3 and A-4, give 
a post-Columbian terminal date to the seriated time sequence of the 
Mazagao Phase. Unfortunately, in spite of the fact that the beads 
include distinctive types, no more precise date can be attributed to 
them. Extended consultations with bead experts and exhaustive 
efforts to use these beads in a more specific way has produced no 
evidence to indicate what type of beads were traded first and by 
which Europeans in South America, information that would make 
it possible to assign an absolute date to the terminal sites of the 
Mazagao Phase sequence. Although in North America bead ex- 
perts have assigned specific dates to certain of the trade beads 
which also appear in the Amazon, all these dates are too late and do 
not shed further light on facts already known from historical records 
(see The Historical Aftermath, pp. 556 ff.). Europeans were trading in 
the area from A. D. 1500 onward, almost 150 to 200 years earlier than 
the dates assigned to the same types of trade beads in the North 
American area. 

The ceramic sequence in the Mazagao Phase is characterized by a 
general, consistent decrease in popularity of the sandy, crushed- 
quartz-tempered Mazagao Plain and a concomitant increase in abund- 
ance of the smooth, soft, canape-tempered Vilanova Plain, and by 
a shift in the decorated wares from the unstylized, and sometimes 
carelessly executed, Uxy Incised to the consistent and precise Anauera- 
pucii and Pigaca Incised styles (fig. 23). 

In the earliest site excavated in the Rio Vilanova region (Site A-6), 
Vilanova and Camaipi Plain are both present in addition to Mazagao 
Plam. The collections made by Nimuendaju from the Rio Iratapuru 
of the Rio Jari drainage, however, completely lack both Vilanova 
and Camaipi Plain but contain the best and most varied examples 
of Uxy Incised. Although these undoubtedly represent selected 
samples due to a conscious choice of decorated rather than plain 
wares, the fact that the collection numbers 248 sherds, nearly half 
of which are plain ware, makes it seem unlikely that Vilanova and 
Camaipi Plain would be totally missing in the collection had they 
been present at the sites. If we may seriate these sites at the bottom 
of the sequence (the only possible position when the decorated types 
of Uxy Incised and Anauerapucii Incised are considered) , the position 
of Site A-6 — Ilha das Igagabas becomes easier to explain. The 



98 BUREAU OF AMERICAN ETHNOLOGY [bull, le? 

pottery from this site differs from that of later ones both in its high 
percentage of Camaipi Plain and in the relative lack of differentiation 
in surface color between the three plain wares. Although more 
cream-colored in Vilanova and Camaipi Plain, the grayish surfaces 
approach the Mazagao Plain represented at the Iratapuru sites more 
closely than the brick-red and light-cream wares of the other Rio 
Vilanova and Rio Pigaca drainage sites. The high percentage of 
Camaipi Plain (32.4 percent) from Site A-6 may be related to the 
introduction of the new sandless plain ware and its, as yet, incomplete 
differentiation, Camaipi Plain being in reality a hybrid or transi- 
tional form with the temper of Mazagao Plain and the texture and 
color of Vilanova Plain. 

If m.ore sites in the southern part of the Territory of Amapa had 
been available for the seriation, the abrupt changes in several places 
in the Mazagao Phase sequence would be erased. A gap now exists 
in the lower part of the sequence but there is no evidence as to its 
length. Although the seriation of Site A-6 near the bottom of the 
sequence instead of near the top may appear questionable on the basis 
of the plain-ware percentages, the correctness of this position is con- 
firmed by the analysis of vessel shape (fig. 24). 

By the time of the lowest levels at Site A-2 — Lauro, the distinction 
between Vilanova and Mazagao Plain has become pronounced and the 
transitional Camaipi Plain has been reduced to an insignificant 3 
percent of the total ceramics. Mazagao Plain has become an orange- 
red, oxidized-fired ware in contrast to the light-tan surfaced Vilanova 
Plain. Their subsequent history lies in the reduction of Mazagao 
Plain from 76 percent in the lowest level of cut 2 or 79 percent in the 
lowest level of cut 1 at Site A-2, to 32 percent at Site A-1, correlated 
with the increase in percentage of Vilanova Plain from 17 percent or 
12 percent to 50 percent in the same sites and levels. Seriation of 
our habitation sites on the basis of plain wares gives the following 
order of antiquity, beginning with the earliest: Site A-6 — Ilha das 
Igagabas, Site A-2 — Lauro, Site A-5 — Cafezal, Site A-1 — Pigaca 
(fig. 23). 

Of the decorated wares, Uxy Incised has the longest history with 
the greatest change in popularity. From 3.8 percent at the earliest 
site, A-6 (omitting at this time a consideration of the Rio Iratapurii 
sites because of their undoubted selectivity of decorated sherds), it 
drops to 0.3 percent at Site A-1. Part of this decHne has its explana- 
tion in the sudden appearance in the top level at Site A-2 of Anauera- 
pucu and Pigaca Incised, both of which are characteristic of the latter 
part of the Mazagao Phase. Their occurrence in the habitation sites 
runs less than 3.0 percent with the exception of the 7.5 percent 
occurrence of Pigaca Incised at Site A-1. No distinction can be 



Ev^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 99 

drawn between the percentages of Pigaca Incised in the cemeteries as 
opposed to the habitation sites, but the generally higher percentage of 
Anauerapucii Incised in the cemeteries indicates it to be primarily a 
burial ware. 

In the Mazagao Phase incision was not only the primary, but almost 
the exclusive decorative technique. Two major types of incised 
designs are distinguishable, with the crudely applied, curvUinear 
designs of uneven and irregular lines (Uxy Incised) common in the 
early part of the Phase in sharp contrast to the late, carefully exe- 
cuted, rectilinear motifs of both Pigaca and Anauerapucii Incised. 
Not only does this latter style show greater care in the workmanship 
but it also demonstrates a more advanced ceramic design technique 
in the filling of the incised lines with white chalk. There is little 
correlation on styUstic and technical grounds to suggest that Anauera- 
pucii or Pigaca Incised evolved out of Uxy Incised. Rather, the 
abruptness of their appearance in fully developed form indicates that 
they were intrusive. 

The other decorative techniques of Mazagao Phase pottery are 
minor in importance. Scraping occurs on only a few sherds from the 
Rio Iratapurii sites. Only one sherd (a bifurcated-scalloped rim of 
Mazagao Plain paste) from A-2 and one from A-4 have any traces 
of paint ; this appears to be the result of rubbing the surface with red 
ochre. Modeled or applique ornament was employed principally 
in connection with burial jars, where it was used to produce faces 
and parts of either zoomorphic or anthropomorphic body anatomy, 
such as genitalia, backbone, nipples, navel, toes, and fingers. The 
most common antliropomorphic style is that typical of Mazagao 
Plain and VUanova Plain (figs. 13, b; 19-1), where the face was 
made on the vessel neck with applique and light incision or slight 
depressions, with the vessel body modified only slightly to suggest 
male or female sex by the addition of a thin, appliqued strip as a back- 
bone, nubbins as small breasts or genitals, and a slight depression for 
the navel. Examples of this technique include a vessel found by 
Lima Guedes at Igarape do Lago (pi. 9, b) ; jars B and C (fig. 8) of 
Burial Group 1, Site A-3; the large urn of Burial 5, Site A-3; and the 
burial urn (fig. 13, b) of Burial 3, Site A-4. The only Mazagao 
Phase examples of true anthropomorphic modeling are vessel Z, 
Burial Group 1 from Site A-3, and vessel A, Burial Group 2, from 
Site A-4 (fig. 12). Presumably these are copies of the Rio Marac^ 
style and were not an indigenous development of the Mazagao Phase. 
Zoomorphic figures are not common except on the Ilha do Para., where 
burial urns in the form of the turtle predominate (pi. 17). With the 
exception of a large foot of a turtle (fig. 6) and the head and tail of a 
cutia from the pottery bench of the anthropomorphic vessel found at 



100 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

Site A-3 (pi. 7, a), no other animals were represented on the ceramics 
of the Mazagao Phase. A few bird heads were affixed to the necks 
of some of the vessels found by Lima Guedes at Igarape do Lago sites 
(pi. 9, a, c). 

In addition to the seriation and study of the change in the popularity 
of various pottery types through time, the rim and vessel shapes of 
each pottery type were analyzed in the temporal framework provided 
by the pottery type seriation. The most distinctive trends in vessel 
and rim shapes occur in the three major plain wares — Camaipi, Maza- 
gao, and Vilanova Plain. Form 1 of Mazagao Plain, a round-bodied 
jar with thickened, everted rim and short, insloping to outsloping neck 
(fig. 19-1), decreases steadily from a high of 80 percent at the earliest 
sites to 20 percent in the latter part of the sequence (Appendix, table 5) . 
While form 1 is declining, Mazagao Plain form 3, a round-bodied jar 
with an unthickened rim and short vertical neck, is increasing slightly. 
The most distinctive time marker withiu Mazagao Plain is the late 
appearance of form 2, tall, cylindrical jars with slight pedestal bases. 
Concomitant with these developments in Mazagao Plain, the opposite 
trend is taking place in Vilanova Plain. Form 1, the round-bodied 
jar with thickened, everted rim and short, vertical to outsloping neck, 
increases in popularity (Appendix, table 8) while form 3, a large, open 
bowl, decreases. This bowl shape was also found in Mazagao Plain, 
but had an erratic history, generally ranging from 14 to 20 percent, but 
reaching 28 to 30 percent at a few sites where the small sample might 
account for the larger percentage. Camaipi Plain form 1, the round- 
bodied jar with thickened, everted rim and short vertical to outsloping 
neck showed the same decrease as that demonstrated for a sunUar form 
in Mazagao Plain (Appendix, table 3) . These rim and vessel-shape 
trends reflect the history of the plain pottery types of the Mazagao 
Phase: while Mazagao Plain decreases, Vilanova Plain increases. The 
decrease in popularity of Mazagao Plain form 1 while a similar shape 
in Vilanova Plain is increasing suggests a retention of the popular plain 
ware shape on whatever plain pottery type was most common through- 
out the history of the Phase. 

The shape trends within each decorated pottery type are not as 
pronounced as in the plain wares; however, some shapes restricted to 
certain pottery types show a distinctive distribution through time. 
Uxy Incised form 3, flat lids (fig. 21-3), is absent in all the other deco- 
rated types except in Pigaca Incised where it constitutes rare form 2. 
Pedestal-basin lids, a popular shape of Anauerapucu Incised (form 1) 
and Pigacd Incised (form 2) are not found on any other decorated 
pottery type. This point is highly significant for it further demon- 
strates that the development of Anauerapucu and Pigac^ Incised is 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 101 

EVANSJ 

the result of some outside influence and is not the direct outgrowth of 
the earHer Uxy Incised. 

Since the vessel and rim shapes were designated by form numbers 
for each pottery type, arranged generally in the order of their popu- 
larity within the type, considering the jars first and the bowls second, 
a common ground of comparison was needed to study the shape trends 
thi'oughout the Mazagao Phase as a whole. Consequently, all the 
shapes of each pottery type were analyzed and lumped into seven 
common forms. These were given an alphabetical designation and 
descriptive term (e. g., form A — carinated bowl; form B — jar, thick- 
ened rim, usually vertical neck; form C — tall, cylindrical jar, etc.; see 
fig. 24 for drawings). The various shapes of each pottery type were 
plotted with their respective form numbers to give a common unit of 
reference. The results are shown in Appendix, tables 9 and 1 0, and on a 
chart (fig. 24) where, for example, form A — carinated bowls contains 
form 2 of Anauerapucti Incised, form 3 of Pigacd Incised, and forms 
1 and 2 of Uxy Incised. Using these common shapes, the count from 
the individual tabulations of vessel and rim shape of each pottery type 
was plotted on a graph in which the vertical factor consisted of the 
various sites arranged according to their seriated sequence based on 
trends of pottery types, and the horizontal factor was the common 
vessels shapes ranging from form A through form G. 

The completed chart (fig. 24) presents a clear and graphic picture of 
the shifts in popularity of the various shapes during the Mazagao 
Phase, In fact, the shape trends are so distinctive and limited in 
their distribution to a specific point in the time sequence that the 
position of certain sherds in the Mazagao Phase sequence could be 
ascertained by shape alone. It is important to note then that the 
sequence established on the seriation of pottery types from the various 
sites is confirmed by this independent shape analysis. Site A-6 did 
not fit neatly into the lower part of the sequence because of its high 
percentage of Camaipi Plain, but its location was defended mainly on 
the absence of the late decorated types, Anauerapucti and Pigacd In- 
cised, and the presence of the early type, Uxy Incised, as weU as the 
consideration of certain color distinctions and variations in the plain 
wares. The plotting of the common vessel shapes for the Mazagao 
Phase into the order of the seriated sequence based on pottery types 
verifies, without any question, the position of the sites in the sequence, 
especially Site A-6 — Ilha das Igagabas. To be specific: form G, flat 
lids, is found only in the lowest part of the sequence where it ranges in 
popularity from 4 to 20 percent; form D, pedestal-basin lids, is not 
found in the lowest part of the sequence, but is restricted to the middle 
and upper sectors; form C, tall, cylindrical jars, is found solely in the 

391329—57 9 



102 BUREAU OF AMERICAN ETHNOLOGY [bull, m 

upper (late) sections, where it has a 6.3 to 9.0 percent popularity. 
On percentage of plain wares alone Site A-6 might have been con- 
sidered for seriation near the top of the sequence instead of near the 
bottom; however, this position would be impossible according to the 
vessel and rim form analysis. The placing of Site A-6 at the top would 
produce an isolated late appearance of vessel form G and an absence 
of vessel form C, which otherwise occurs consistently in the latter part 
of the Phase. 

The trends of the other vessel and rim shapes are also marked. For 
example, form A, carinated bowls, declines from as high as 57.0 percent 
to 12.4 percent, while form E, open bowls, increases from a low of 
between 4.0 percent and 7.2 percent at the bottom to 20.3 percent and 
25.0 percent at the top of the sequence. With the exception of form C, 
tall cyhndrical jars, whose distinctive history has already been dis- 
cussed, the trend of the jar shapes is not as clear cut as that of the 
other vessel forms. Disregarding the sites with small samples, which 
unfortunately warp the percentages, the trend of form F, round-bodied 
jar with an outcurved, direct rim, is irregular, but form B, the round- 
bodied jar with a thickened, vertical or outslanting rim, increases to a 
peak at the middle of the sequence and then begins to decline again 
(fig. 24). 

As the discussion in the preceding section on nonceramic artifacts 
indicated, nonpottery artifacts are too rare and undifferentiated to 
permit any statement about change in technique of manufacture or 
style during the Mazagao Phase that would supplement the sequence 
based on pottery. 

DIAGNOSTIC FEATURES OF THE MAZAGAO PHASE 

Sites of the Mazagao Phase are found throughout the region between 
the Rio Araguari-Amapari and the Rio Jari, with some indication of a 
late concentration in the Rio Vilanova at the time that the Rio Maraca 
was developing its own local tradition. The habitation sites are all 
located on naturally high land near a constant water supply, above 
the flooded lowlands, offering advantages from a defense standpoint. 
The refuse varies in thickness from the surface only to 45 cm. at the 
deepest site and is scattered over areas ranging from a small site 
10 meters in diameter to the largest, 75 by 83 meters. The nature 
of the sites and their associated cemeteries suggests that at no time 
was the total population large. 

The burial pattern is consistent: secondary burial with offerings in 
small bowls and occasionally with glass beads or a stone ax placed 
inside the urns. The cemetery occupies a high spot and the vessels 
were partially buried and were typically covered with a lid. The 
arrangement of the urns in the cemetery appears to have been hap- 



VES! 




A-2 



A- 
CUT 
'l= 

2-ol 

2-.ii 

I -.I 
I'. 3 



m 



SA 
BOf 
UX' 
ALT 




lllMlilllll 



FORM 

F 



FORM 
6 



NUMBER 
OF RIMS 



16 
24 
II 
16 

7 
9 

50 

22 
25 
5 
14 



ible 10). 



VESSEL SCALE 

I I I 

8 16 CM 






NUMBER 
OF RIMS 



A-2- 



A- I 

A- 5 
CUT LEVEL 
'l= 0-.I5M 

2 0-. 15 

2 .15-. 30 

I-.I5-.30 

1^1 -.30-. 45 



t 



lll l l l lll l Illl l ll I 



llllllllllll 



rrrm 



nrm 



mm 



rrm 



lllll l llllll 



16 

24 

II 

16 
7 
9 



A-6 



SO 



SAO JOAO 

BOM DESTINO 

UXY 

ALTO ALEGRE 






1 




1 1 1 


FORM 
A 




10 20% 



FORM 
B 







mmmm 


■ 


^i" 


22 






m 


HHI^BM 


■ 


25 








1 


1 


S 
14 


FORM 



FORM 
D 


FORM 

E 


FORM 

F 


FORM 
6 





Figure 24. — Trends in the popularity of common vessel shape.s of the Mazagao Phase (Appendix, table 10). 



Ev^Nsr ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 103 

hazard, often with no consideration of the location of a previous 
biu-ial urn. 

Nonpottery artifacts are rare, Hmited to a few axes, hammerstones, 
pebble smoothers, a shaft straightener, ochre and chalk. 

Seriation of the ceramic styles shows a distinct shift in emphasis 
through time. The earliest excavated sites are characterized by the 
equal popularity of Mazagao Plain, a sand}^ quartz-and-mica- 
tempered ware, Vilanova Plain, a carm^^e-tempered, smooth, soft pot- 
tery, and Camaipi Plain, a ware possessing both cariape and sand tem- 
per. Mazagao Plain shows an immediate sharp rise in popularity, but 
then begins gi-adually to give way to Vilanova Plain. ^'^ The earlier 
periods emphasize curvilinear, deep, crudely incised designs (Uxy In- 
cised) which tend to decrease in frequency as the well-developed, recti- 
linear incised varieties, Anauerapucii and Pigaca Incised, are intro- 
duced and increase in popularity. The decorated wares have a much 
higher percentage of occurrence in the cemeteries than in the occupa- 
tion sites, giving some indication of a conscious manufacture or use of 
certain wares for burial and others for domestic use. 

The effect of a strong outside influence is manifested in the sudden 
appearance of complex vessel shapes and in the introduction of the 
precisely executed design motifs of Pigaca and Anauerapucii Incised, 
which are carried to an acme of perfection in the latter type. Some 
contact or borrowing without amalgamation or assimilation of either 
culture occurred between the Mazagao Phase and the tradition occupy- 
ing the Rio Maraca, as is evidenced by the copies of anthropomorphic 
urns in the cemeteries of the Mazagao Phase. 

Although the actual duration of the Mazagao Phase cannot be de- 
termined from the time sequence derived from the seriated ceramic 
styles and stratigraphy, evidence suggests that a long period of occupa- 
tion of this region is not probable. The terminal date of the Mazagao 
Phase cannot be given absolutely, although glass trade beads are 
found in limited numbers in some of the cemeteries, indicating a post- 
European contact date for at least a part of the inhabitants of the 
Mazagao Phase and the local tradition on the Rio Maraca. 

THE ARISTE PHASE 
DESCRIPTION OF SITES AND EXCAVATIONS 

With the exception of one cemetery, Site A-14, located in the present 
city of Macapa, and a smaU camp site, Site A-13, in the headwaters of 
the Rio Matapi, the known archeological sites of the Ai'iste Phase are 
restricted to the region north of the Rio Araguari-Amapari. These 

w This transition from a sandy, gritty ware to a soft, smoother ware tempered with another material is 
repeated in the Arlstfi Phase. 



104 BUREAU OF AMERICAN ETHNOLOGY [bull, le? 

two exceptions present certain peculiarities (which will be explained 
later) and are not important enough to interfere with the general con- 
clusion that this river must have served as a boundary or frontier 
between two distinct and generally contemporaneous cultures. Not 
only the ceramic traditions but also the burial patterns indicate that 
we are dealing with separate groups. Data on the Ai'iste Phase comes 
from 7 habitation and 7 cemetery sites in addition to the famous 
CunanI burials dug by Goeldi in 1896, and sites described by Coudreau 
and Nimuendajii. 

SITE A-7 — AMApX city 

The present town of Amapa covers an Indian village and cemetery 
site. This superposition is more than coincidental, for the only con- 
tinuously dry land in the region, regardless of season, is a long narrow 
jBnger rising 4 meters above high-water level along the south side of 
the Rio Amapa Pequeno. At the time of his visit in 1895, Goeldi 
noticed the presence of Indian remains, but dismissed them as un- 
important because he saw no decorated ware (Goeldi, 1900, pp. 7-8). 
The ridge of high land ranges in width from 100 to 250 meters and a 
fragment of a stone tool was picked up as far as a kilometer south of 
the riverbank. Throughout the high area residents have uncovered 
pottery and stone fragments while cultivating their gardens or con- 
structing new houses. The continuous row of buildings along both 
sides of the main street, Rua Senador Lemos, which now occupies 
the center of the ridge, made extended excavations out of the ques- 
tion. Tests in various parts of the street produced sherds (Evans, 
1951, fig. 3) that had been ground almost to powder by the impact of 
feet over several centuries and showed the black refuse layer to be 
10 to 15 cm. deep, resting on sterile, orange to brown clay. Some 300 
meters from the riverbank, the ridge is cut by a ravine, now crossed by 
a concrete bridge and culvert. At the time this was built, some 5 
years previous to our visit, the north bank was cut off 0.75 to 1.00 
meter, removing the tops of several large jars and leaving a broken 
edge to show where the bases were still buried in the street. One of 
these, excavated with great difficulty because of the hard-packed 
clay in and around it, was 50 cm. in maximum existing diameter, 15 
cm. in existing height, and 13 cm. in diameter at its flat base. It was 
identified as Serra Plain. Associated ^vith it were a few small sherds 
of a well-polished, red variety of Serra Painted. Subsequent to our 
visit a globular jar of Serra Plain with a 2-cm.-wide red band on the 
neck and additional sherds were collected in the main street some 15 
meters north of our excavation. 

The 421 sherds collected from the site of Amapa are Serra Plain 
(96.4 percent) except for 14 sherds of Serra Painted (3.3 percent), 



MBGOBRS AND 
BVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



105 



and 1 sherd of Ariste Plain (0.3 percent). Nonceramic artifacts 
from Site A-7 include an ax, hammerstones, and a grinding stone: 

Stone ax. — A large, flat ax was made from a fine-grained, gray-black diabase by 
pecking and abrading. The sides of the poll are rounded with a tapered butt, 
pecked on all surfaces, and with a convex bit polished on both surfaces for a dis- 
tance of 1 cm. back from the blade edge; length 21.7 cm., bit width 6.8 cm., poll 
width 7.6 cm., poll thickness 3.5 cm., width of butt end 4.5 cm. 

Sto7ie tool. — The fragment of a stone tool of fine-grained granite made from a 
naturally shaped, waterworn rock which was rounded off to form a blunt end but 
with very little pecking or polishing to give it shape. This worked end shows 
slight usage as a hammerstone with several flakes removed; the other end is 
broken off. By its shape, the artifact could be a fragment of an ax or a hammer- 
stone. Present fragment measures 5.0 cm. long, 5.0-5.6 cm. wide, 3.5 cm. 
thick with a half-round cross section. 

Small hammerstone. — Of fine-grained, grayish-white diabase, shaped by abrasion 
with no pecking marks visible and with the larger end showing signs of use as a 
hammerstone. Although the corner of one edge is chipped off, the object measures 
9.3 cm. long, 4.0 cm. wide at base, 2.4 cm. wide at center of poll, and uniformly 
2.1 cm. thick (fig. 25, a). 

Grinding stone (jnano) . — This fragment is so badly eroded that most of the surface 
is pitted giving it an unnatural roughness due to the differential weathering of 







Figure 25. — Stone artifacts from A-7 — Amapd, Ariste Phase, a, Small hammer- 
stone. h, Grinding stone fragment, also used as a hammerstone. 



106 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 



the large, quartz particles in the coarse granite. One surface is well polished 
and smoothed from usage as some sort of grinding stone. Only half of the 
specimen exists with the rounded end slightly battered from use as a hammerstone; 
perhaps this latter usage occurred only after it had broken and become useless 
as a grinding stone. The present fragment measures 7.8 cm. long, 8.5 cm. wide, 
and 2.4 cm. thick, tapering slightly to 1 cm. thick at the end (fig. 25,6). 

SITE A-8 AURORA 

Although the site was constructed by peoples of the Arua Phase 
and probably used as a place of ceremonial significance in their 
transitory occupation of the Territory of Amapa before going to the 
islands of Mexiana, Caviana, and Maraj6, the later peoples of the 
Ariste Phase apparently placed a vessel or two around the standing 
stones (pi. 2) . There is no need to redescribe the site for the details 
have been given in the Arua Phase (pp. 38-40). Twelve of the fifty 
sherds found at Aurora belong to Ariste Phase pottery types. These 
fragments were scattered just beneath the surface and represent 
only a few vessels. They classify as 2 sherds of Serra Plain and 10 





Vt<'. 



ELEVATION 



I L—L. 



100 200 M N 



Figure 26. — Ground plan of A-9 — Rel6gio, a habitation site of the Arista Phase. 



BvTNsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 107 

sherds of Ariste Plain. Since there is no evidence of extensive use 
of this site or any of the other stone aHnements by the Ariste Phase, 
it deserves no further comment in this section. 

SITE A-9 — REl6g10 

The region immediately south of the town of Amapa is dotted 
with hundreds of lakes, large and small, with rolling meadows or 
thick forest along their shores. The habitation site, A-9, begins 
about 10 meters back from the edge of the Rego do Cajii, a channel 
joining the east end of Lago PracuAba with the west end of Lago 
Socaiozabinho (fig. 26; pi. 4). It occupies an area about 100 meters 
in diameter, now indistinguishable in the dense secondary forest 
growth and underbrush from the surrounding region. The land re- 
mains 1.50 meters above the water at its highest level. Although 
the caboclo who was our guide said he had found two small jars near 
a house he at one time built on the site, we found only sherds. A 
2- by 2-meter test excavation in the north end of the site, where the 
sherds seemed most abundant, showed the deposit to extend from the 
surface to a depth of 10 cm. Beneath the sherd-bearing black loam, 
the sterile soil was light brown. This test produced 439 sherds, of 
which 387 or 88.2 percent were Serra Plain, 49 or 11.2 percent were 
Ariste Plain and 3 or 0.6 percent were Davi Incised. In addition, 
there were 27 bm'nt-clay fragments and 15 granite and quartz natural 
rock fragments, only a few of which showed traces of having been 
subjected to fire. 

SITE A-10 — MONTANHA DA PLUMA 

The Igarape da Serra drains out of rocky, rugged hills with numerous 
granite outcrops into the northwest side of the Rio Flexal. Halfway 
upstream from the mouth of the Igarape da Serra and 2 km. inland 
from the north bank is a granite outcrop (pi. 5) covering an area 
about a kilometer in length and half a kilometer in width and rising, 
at its highest point, 50 meters above the surrounding tree-dotted 
savanna. Among the boulders scattered on the south flank of this 
hill, 8 meters above the level plain below, is a granite block 5 meters 
high and 4 meters wide. Beneath it runs a tunnellike cave 4 meters 
long. The south mouth, 2.45 meters wide by 70 cm. high, opens 
upon a natural shelf, 4 meters wide and 8 meters long, while the north 
mouth, 2 meters wide and 55 cm. high, was blocked on the east and 
north by two large boulders, leaving only a small entrance way 
from the west. 

Large sherd fragments from burial urns were scattered for a distance 
of 1 meter outward from the cave at the north mouth and on the south 
side covered an area 3 meters out from the cave and 4 meters along 
the base of the rock (pi. 5, 6). The interior of the cave was littered 



108 BUREAU OF AMERICAN ETHNOLOGY [bull, lei 

with broken burial urns, and many fragments were covered with a 
thin (5 cm.) layer of bat dung and dry, fine, powdery dust. The 
cave floor was irregular, with a narrow rock shelf on each side slightly 
higher than the center aisle, which was filled with sterile talus wash. 
The guide stated that he had seen the cave at a time when the burial 
urns were intact, but that years ago caboclo children, vying to see who 
could break the most, had reduced them to their present condition. 
As a result of this destruction, the only possible excavation procedure 
was to make a collection of material. Since there appeared to have 
been no gross disturbance of the position of the vessels during the 
breakage, the sherds from within the cave were kept in separate bags 
and given different catalog numbers from those on the outside on the 
assumption that the latter vessels represented later burials after the 
cave had been filled. During excavation, 12 small, European glass 
trade beads were found at the south mouth intermixed with sherds 
and dirt. All the beads were the small, "seed" variety of porcelain 
white color, discoidal in shape, 2 to 3 mm. in diameter and 2.0 to 2,5 
mm. long with the perforation 0.5 mm. or less in diameter. Fine bone 
scraps were scattered throughout; some showed evidence of cremation; 
all were too fragmentary to classify. All the sherds from inside the 
cave had a black patina, an unnatural hardness and a high, metallic 
ring resulting from the continuous percolating of mineral-laden 
waters into the porous ceramic until the pottery became mineralized. 
Since it was clear that the cave had been used solely as a depository 
for burial urns and since it was not possible to preserve or transport 
all the sherds from Montanha da Pluma, a selection was made of all 
rims, bases, and diagnostic body sherds. An analysis of these 
showed a total of 24 different burial vessels from outside the cave, of 
which 11 or 46 percent were Serra Plain, 3 or 12.4 percent Serra Painted, 
5 or 20.8 percent Ariste Plain, and 5 or 20.8 percent Arista Painted, 
and 61 vessels from inside the cave, representing 16 or 26.2 percent 
Serra Plain, 3 or 4.9 percent SeiTa Painted; 30 or 49.2 percent Ariste 
Plain, 9 or 14.8 percent Ariste Painted, 2 or 3.3 percent Flexal Scraped, 
and 1 or 1.6 percent Unclassified. 

Careful examination of the area both inside and outside the cave 
revealed no artifacts besides pottery vessels and the aforementioned 
glass beads. 

SITE A-U MONTANHA DE ARIST]6 

On the Igarape da Serra, about 8 km. upstream from Site A-10 
and some 2 km. inland from the north bank, is the Montanha de 
Ariste (fig. 1). The entire region is broken into steep-sided hills whose 
granitic substructure is revealed in sheer rock walls and innumerable 
shelters and small caves. Tall virgin forests with little or no under- 



M£60ERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



109 



growth cover the rocky slopes and summits, which rise 75 to 100 
meters above the riverbed. 

Cave 1, perhaps more correctly called a slight rock shelter, was on 
the east side of the mountain at the base of a large, sheer granite 
outcrop 100 meters long, with its flat face rising almost vertically for 
15 meters. Burial urns had been placed along the base of this outcrop 
for a distance of 15 meters, and spalling rocks had broken all of them 
badly (pi. 6, a). Sherds of the broken burial vessels, scattered in the 
rock chips and talus dirt, were most highly concentrated toward the 
center, thinning out toward both limits of the cemetery area. Of the 
31 vessels represented by the sherds collected, 3 or 9.6 percent were 
Serra Plain, 23 or 74.4 percent Arist6 Plain, 4 or 12.8 percent Ariste 
Painted, and 1 or 3.2 percent Flexal Scraped. A small Serra Plain 
sherd, 4.0 cm. long, 1.9 cm. wide and 8 mm. thick, had the edges 
curved and well-rounded from extended use, probably as a pottery 
scraper (fig. 27). 

Cave 2, some 10 meters above and southwest of Cave 1 on the 
opposite side of the mountain, was formed by the undercut base of an 
enormous granite boulderhke outcrop. The ceiling sloped from a 
height of 81 cm. at the front to 40 cm. at the back toward the middle 
of the cave and then dropped sharply to only 15 cm. at the rear in the 
west end, forming a completely protected shelter 5 meters from side 
to side and 1.80 to 3.00 meters deep. The ground in front of Cave 2 
sloped gently for about 4 meters and then dropped vertically to the 
Igarape da Serra 35 meters below. A number of the vessels that stood 
upon the dirt floor had been broken by large rocks falling from the 
ceiling (Evans, 1951, fig. 5) and by the spreading roots of a large 
sumamera tree growing against the east side (fig. 28). Others were 
damaged and almost buried in the dirt thrown back by the burrowing 




Figure 27. — Worked sherd scraper from the Arista Phase cemetery of A-11- 
Montanha de Arista, Cave 1. 



110 



BUREAU OF AMERICAN ETHNOLOGY 



[bull. 167 




> 

o 



t3 
O 

5 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



111 



of an animal. In spite of these disturbances about a dozen remained 
intact or nearly complete. 

Cremation was the common practice and the vessels with 
undisturbed contents were filled to within a few centimeters of the 
rim wdth a mixture of light-tan, sandy loam, and burned bone frag- 
ments. Two uncremated secondary burials, both adults, were 
found; one of them was in a jar that also contained remains from a 
cremated body. None of the jars had lids in place and, except for 
one stopperlike fragment near vessel B, no lids or covers were 
identifiable. 

Of the total of 56 vessels represented by the sherd material and the 
complete specimens, 16 or 28.6 percent were Ariste Plain, 21 or 37.5 
percent were Ariste Painted, 6 or 10.7 percent were Serra Plain, 3 or 
5.4 percent v/ere Serra Painted, 9 or 16.1 percent were Flexal Scraped, 
and 1 or 1.7 percent Davi Incised. 

A pottery figurine and a fragment of a stone chisel, possibly burial 
offerings, came from the eastern part of Cave 2 : 

Stone chisel. — The stone chisel (fig. 29, a, h) consists of about one-half of a small, 
highly polished, slate chisel with a groove 1-3 mm. deep along one edge. The 
function of the groove is unknown. The chisel is well made, very sharp, 2 cm. 
wide at the bit, enlarging to 2.7 cm. wide and 1.8 cm. thick at the center, with 
the existing fragment 4.7 cm. long. 




I CM 




Figure 29. — Small stone chisel from A-11 — Montanha de Ariste, Cave 2, Arista 

Phase, 



112 



BUREAU OF AMERICAN ETHNOLOGY 



[bull. 167 







I CM 




b 



FiGUEE 30. — Pottery figurine (Ariste Plain paste) from A-11 — Montanha de 
Arista, Cave 2, Arist6 Phase. 

Figurine. — The crude Arist6 Plain human figurine (fig. 30, a, b) is a rusty- 
brown except the diadem hairdo and face, which are painted black. A band 
was left unpainted on the face in the area corresponding to the forehead. No 
anatomical features are indicated except the head with the eyes, nose, and mouth 
gouged out on a very grossly modeled face. The workmanship is extremely 
crude and the result asymmetrical. The lower part flares outward slightly to a 
flat base with an oval cross section measuring 3.4 cm. from front to back and 
5.0 from side to side. The narrowest portion of the body is 2.5 by 4.0 cm. with 
the full figurine 8.0 cm. long. 

No small jars or bowls were found associated with the burial 
vessels, nor did any of the sherd fragments represent miniature 
vessels. No European glass trade beads were found either inside any 
of the vessels or scattered in the cave. 

The location of the whole or partially complete vessels in the fol- 
lowing descriptions is shown on the ground plan (fig. 28) : 

Vessel A, a small, globular body of a flat-bottomed Serra Plain jar with the rim 
broken oflf, was found lying on its side with the mouth toward the rear of the east 
end of the cave. The body of the vessel with smoothed, light-tan surfaces is 
20 cm. in diameter and 17 cm. high. The base diameter is 10 cm. and the neck 
diameter 9 cm. No bones were found inside, but many cremated bone scraps 
were scattered in the vicinity. 

Vessel B is a large Arista Plain jar with a flat base, an egg-shaped body (broken 
out on one side) and a short, vertical neck ending in an irregular, collared rim, 
averaging 4.7 cm. wide (pi. 20, b). One side of the body had been broken by 
falling rocks. The dimensions of the jar are: body diameter, 30 cm.; base dia- 
meter, 18 cm.; rim diameter, 21 cm.; body height, 21 cm.; neck height, 10.5 cm. 
The surface is well smoothed, with temper particles of quartz sand protruding. 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



113 



A heavy, flat Arista Plain disk, with a short, nubbin handle in the center of one 
face, lay near the mouth of vessel B, but its diameter of only 16 cm. makes it too 
small to have served as a lid to this particular vessel. 

Vessel C, a small Arista Plain jar, has a slightly concave base 9 cm. in diameter, 
an asymmetrical, globular body of 20 cm. in diameter, a slightly constricted neck 
with a diameter of 11 cm., and a cambered collar 14 cm. in diameter narrowing 
to a diameter of 12 cm. at the mouth. The wall thickness is 4 mm. Most of the 
rim was broken oflf, probably before burial. The vessel was found lying on its 
side, but remained partly filled with large gravel mixed with traces of cremated 
bones. 

Vessel D is a round Arista Plain bowl with a flattened bottom 12 cm. in diameter 
and walls curving out to a maximum body diameter of 28 cm., and inward again 
to a constricted mouth diameter of 17 cm. (pi. 20, d). The total height is 18 cm. 
The rim is of a form typically found on Ariste Phase bowls, a kind of collar pro- 
duced by the exterior beveling of the rim edge. The exterior surface is well- 
smoothed. The interior was filled with fine, dry, sandy-loam containing 
cremated bones. 

Vessel E, a Flexal Scraped bowl, is 15 cm. tall and 24 cm. in maximum diameter 
(fig. 31). Above the maximum diameter the walls slant inward slightly forming 
a beveled rim with a slightly constricted mouth 20 cm. in diameter, similar to that 
of vessel D. Below the waisc, the walls extend inward at a much greater angle to 
join the base. Seen from above, the mouth is not circular and the jar is generally 
asymmetrical. In addition to the scraped decoration above the shoulder and 
along the rim exterior, paired applique buttons are placed at three equally-spaced 
intervals along the rim. The contents were missing because the bowl had been 
turned over. 

Vessel F had been knocked over, broken and partially buried by spalled rock. 
It is a small asymmetrical, globular-bodied, Serra Plain jar with a flattened bottom 
15 em. in diameter and a short neck measuring 10 cm. in diameter. The rim 
had been broken off. The existing height is 22 cm. and the maximum body 




Figure 31. 



-Flexal Scraped bowl (vessel E) from A-11- 
Cave 2, Arist6 Phase. 



-Montanha de Arista, 



114 



BUREAU OF AMERICAN ETHNOLOGY 



[bull. 1G7 




Figure 32. — Flexal Scraped jar (vessel G) from A-H — Montanha de Arista, 

Cave 2, Ariste Phase. 



diameter 20 cm. The wall thickness varies from 3-5 mm. A few scraps of 
cremated bones remained inside. 

Vessel G, an excellent example of Flexal Scraped (fig. 32), rested on one side 
against the back of the cave. It has a slightly depressed, globular body 25 cm. 
in diameter, a flat base 10 cm. in diameter, a short (2 cm.) neck 14 cm. in diam- 
eter, and a cambered rim measuring 18 cm. in diameter. The overall height is 
21 cm. Two strap handles, one of which remains in place, joined the rim with 
the shoulder. The upper body wall and the rim exterior are decorated with spirals 
and parallel lines in triangular fields made with a blunt, brushlike tool 5 mm. 
wide, leaving fine, parallel marks. The interior was filled with fine, tan soil, 
and small scraps of cremated bones. 

Vessel H is a small and badly smashed Arist6 Painted (red) jar. Fragments 
of cremated bones were mixed with the dirt surrounding it. 

Vessel I, Arista Painted, is similar in shape to vessel G, but lacks the strap 
handles (pi. 21, a). Base diameter is 8 cm., maximum body diameter 24 cm., 
neck diameter 11 cm., and rim diameter 13.5 cm. with the body wall thickness 
5 mm. and the total height 22 cm. The exterior surface is badly eroded except 
on one side of the neck, just below the rim, where traces of red paint are visible. 
The vessel was partly filled with cremated bones and dirt. 

Vessel J, part of a large, flaring rim, carinated Arista Painted bowl lay partially 
buried beside a large rock in the dirt excavated by an animal in the east end 
of the cave. The surfaces are poorly preserved but traces of a black-on-white 
design remain on the inner side of the flaring rim. The reconstructed body 
diameter at the rim attachment is 26 cm., the rim diameter 36 cm., and the total 
height 14 cm. The small, slightly-concave base is 9 cm. in diameter. 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 115 

EVANS J 

Vessel K, Serra Painted, was the largest jar in the cave. It had remained 
upright but was buried up to the rim in dirt thrown out from the animal burrow. 
The long, straight-sided, insloping neck and flaring rim was joined to rounded 
shoulders (pi. 24, b). The total height was 40 cm., neck height 12 cm., mouth 
diameter 23 cm., diameter at the base of the neck 24 cm., body diameter 40 cm. 
and the flat base 20 cm. in diameter. The entire exterior is painted red. 

Vessel L, a small Arista Painted bowl was slightly incurved on the sides, ter- 
minating in a beveled rim (p. 21, b), which is painted red. It was the smallest 
vessel removed from Cave 2. The bowl measured only 9 cm. high, 15 cm. in body 
diameter, 10.5 cm. in mouth diameter, and 7 cm. in diameter at the slightly 
concave base. Fine white ash, particles of burnt bone, and one or two small 
burnt fragments of the cranium of a child were upon the bowl bottom with a 
mixture of black loam and gravel on top of them. 

Vessel M, a large, tall Arista Plain jar, was broken into two large pieces and 
almost completely buried against the rear wall. It is similar in shape to vessel B 
except for a short, outflaring rim. The total height is 35 cm., body diameter 
28 cm., neck diameter 18 cm., and mouth diameter 23 cm. The surfaces were 
eroded, revealing abundant quartz-sand temper. 

Vessel N was partly covered by the talus wash that had come around the edge 
of the cliff into the west end of the cave. It is an Ariste Plain bowl 19 cm. tall, 
with a slight, flat pedestal base 5 mm. high and 11 cm. in diameter. The sides 
curve outward to the maximum body diameter of 28 cm. and then rise almost 
vertically to a slightly flaring rim 30 cm. in diameter (pi. 20, a). The bowl had 
been filled with cremated bones, fine gravel, and light-tan, sandy soil. 

Vessel 0, a carinated bowl, represents Flexal Scraped. It has a simple scraped 
design around the rim and is similar in shape to vessel E except that the rim 
is cambered instead of beveled. A much larger bowl than any so far described, 
it had been broken into two large fragments and many smaller ones by spalled- 
off rocks. The reconstruction measured 56 cm. in diameter on the body, 15 cm. 
at the base, and 48 cm. at the mouth, with the total height 36 cm. 

Vessel P was broken and completely buried by dirt from the animal's burrow. 
The globular Serra Painted jar, measuring 45 cm. in diameter and 50 cm. in 
body height, did not have the neck or rim intact and no fragments were found. 
Around the shoulder is a band 8 cm. wide occupied by a curvilinear, meandering 
scroll (fig. 33) painted in red on the natural, light-tan surface. The limits of 
this design area are marked by broad (3-4 mm.), shallow, incised lines. The 
exterior surface is smooth and even, with smoothing tracks visible except in 
the region of the painting. 

Cave 3, located 300 meters north of Cave 1 and a little northeast 
of Cave 2, was formed by a large boulder supported on two granitic 
outcrops. The mouth, which opened to the south, measured 2 meters 
wide and 3 meters high, with the cave itself 6 meters wide and 4 
meters deep. Animals had used it as a lair and the floor was thickly 
covered with bat dung. The sherds of 7 vessels were found, all 
located in a cluster 1 meter inside the cave mouth. The jars and 
bowls were so badly broken that partial reconstruction could be 
made of only two. Four were Ariste Plain and 3 Arista Painted. 
One of the large Arista Painted jars with red paint had an elaborate, 
insloping, cambered rim 5 cm. wide with two human faces modeled 
on it (figs. 34, 35). Mouth diameter was 12.5 cm., combined neck 



116 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 




Figure 33. — Serra Painted design on the slioulder of vessel P from A-11 — 
Montanha de Ariste, Cave 2, Ariste Ptiase. Stippled area denotes red paint 
upon the natural tan surface of the vessel. 

and rim height was 7.0 cm. and the estimated body diameter, 36.0 
cm. Scattered flecks of cremated bones were noted on the cave floor 
near the broken vessels. 

SITE A-12 CRUZEIRO 

The habitation site of Cruzeiro is almost 10 km. due north of Site 
A-10, on a bank 5 meters high along the west side of the narrow and 
deep Igarape da Rasa (fig. 1; pi. 6, 6). On the opposite side of the 
igarape the Campos do Cruzeiro, an open rolling plain with scattered 
groves of trees, stretch to the north, east, and south. The Amapa 
Air Base is about 15 km. to the northeast. When this area was cut 
and bm-ned for a garden of the Fomento de Agricola, sherds were 
observed on the surface. They were sparsely scattered to a depth of 
5 cm. in an area roughly 100 meters square in a portion of the forest 
where the trees had been much smaller than the surrounding virgin 
growth. Many of the sherds were refired in the intense heat generated 
during the burning of the slash, a factor which had to be taken into 
consideration in making the ceramic classification. Of the 529 sherds, 
339 or 64.0 percent were Serra Plain, 176 or 33.1 percent Ariste Plain, 



^^?^^i^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 



117 




Figure 34. — Reconstruction of Ariste Painted vessel with anthropomorphic face 
from A-11 — Montanha de Ariste, Cave 3, Ariste Phase. The stippled area 
denotes red paint. 




Figure 35. — Detail of the anthropomorphic face on the Ariste Painted vessel 

shown in figure 34. 

391329—57 10 



118 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

11 or 2.3 percent Davi Incised, 2 or 0.4 percent Flexal Scraped, and 
1 or 0.2 percent Unclassified. In addition to the pottery, the follow- 
ing objects came from Site A-12: 2 natural iron concretions, a frag- 
mentary hammerstone, and 2 granite fragments, probably also from 
hammerstones. The broken hammerstone fragment (4.3 X 3.0 cm.) 
is rounded, of granite and suggests deliberate shaping by pecking. 
The larger fragmentary pieces of granite without definite shape have 
slightly battered edges suggesting use as hammerstones. 

SITE A-13 MATApf 

One of the two appearances of the Ariste Phase south of the Rio 
Araguari-Amapari is this campsite in the headwaters of the Igarape 
Ingles, a branch of the upper Rio Matapi (fig. 1). The site was found 
in the garden of the Minas de Ferro, which is located in a rocky, hilly, 
heavily forested area surrounded by upland savanna about 15 km. 
south of Porto Grande. In spite of extensive excavations for the 
preparation of gardens, sherds were found in only one spot approx- 
imately 2 by 2 meters on a slope 5 km. above the igarape. These 81 
sherds belong to 3 vessels, 2 Ariste Plain, and 1 Davi Incised. The 
absence of soil discoloration together with the sparsity of sherds and 
their concentration in one spot suggests that this was a temporary 
campsite rather than a village. 

SITE A-14 — MACApX 

During the 1945 excavations for water mains in the Praga Barao 
do Rio Branco of the city of Macap§;, a row of burial urns had been 
found with their rims just a few centimeters below the surface. One 
complete specimen and the sherds were deposited in the Museu Terri- 
torial in Macapa. During similar excavations in December 1948, on 
the east side of the same Praga, three similar burial urns were en- 
countered about 100 meters east of the original find. The bases of 
two of these were some 50 cm. below the surface, that of the third was 
1.05 meters down. Fragments of unburned bone were said to have 
been found in the jars mixed with dirt. The deepest vessel contained 
a necklace of 10 drilled shells. Each individual shell bead is 4.8 to 6.0 
cm. long and tapers toward each end from a diameter in the middle 
between 1.1 and 1.5 cm. The smallest end of each is biconically 
drilled 1 cm. from the tip with the hole narrowing from 4 mm. at the 
surface to 2 mm. in the center. 



EVANS r ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 119 

The fragments of the jars, as well as the complete one, were studied 
and analyzed in the light of the ceramic types of the Territory of 
Amapa. Eight vessels were Serra Plain and 3 Serra Painted with 
bands of red 9 to 10 cm. wide from the rim to the shoulders. The com- 
plete jar has high shoulders, a short neck, an outflaring, thickened, 
folded-over rim, 2.0 cm. wide and 1.4 cm. thick. The jar is 35 cm. 
high with a waist diameter of 31 cm., a mouth diameter of 30 cm. and 
a slightly rounded base 19 cm. in diameter. The sherds contain 
fragments from two flat-based, open bowls, measuring 22 and 25 cm. 
in mouth diameter with an estimated height of 12 cm. 

In spite of these extensive excavations, which have involved the 
digging of deep trenches along nearly every street in the entire city 
of Macapa, the finds just described are the only ones that have been 
made. No loose sherd material or refuse trash has ever appeared. 

SITE A-15 VILA VELHA " 

An Indian cemetery has been known for some time to exist on a 
slight rise of land within the present limits of Vila Velha on the north 
margin of the Rio Cassipore (fig. 1). Sr. Eurico Fernandes examined 
the site before it was destroyed by the expansion of the village. He 
excavated one complete burial urn, talcing photographs and notes. A 
dark humus layer extended from the surface to a depth of 12 cm. with 
brown clay beneath, indicating a cemetery rather than habitation 
refuse. 

The mouth of the jar was covered with an inverted, open, plain-ware 
bowl. Inside were cremated bones mixed with sand, 373 European 
glass trade beads, a small (6.2 X 3.5 X 0.8 cm.), notched, semi- 
polished ax of diabase with a well-ground bit and a 1-mm. notch 1.2 
cm. from the butt end, and 7 murakitdos or pendants of nephrite. Of 
these pendants 5 were cylindrical, perforated beads and 2 were sty- 
lized pendants, one in the form of an insect. Unfortunately, these 
objects have been scattered among various individual collectors; 
therefore a more detailed description was not available. 

Some of the glass trade beads were fused into a mass as if they had 
been subjected to the same cremation as the bones; these were not 



11 The following seven sites were not excavated by us; but their relatively good documentation, plus the 
fact that we were able to study firsthand the ceramic material from each, warrants their inclusion here to give 
a more complete picture of the Arist§ Phase. 



120 



BUREAU OF AMERICAN ETHNOLOGY 



[bull. 167 



included in the analysis or the tabulation of 373 beads representing 
the following varieties (pi. 25) : 

Table C. — Glass beads from A~15 — Vila Velha 



Color 


Description 


Count 


Dark blue, almost black 

Clear glass or dark blue 

Dark, serpentine green 

Milky, opalescent, bluish 
white. 

Clear glass, or dark blue, or 
amber. 

Clear glass with white 
stripes. 

Dark blue 


Oblong to spherical, diameter 8-10 mm., length 12-15 mm., 

hole 2 mm. 
Large, hand-molded, 8-faceted beads ranging in length from 

11-20 mm. with approximately the same diameter. Hole 3-5 

mm. 
Small, hand-molded, and 8-faceted, but only 8-11 mm. in 

length and diameter. 
Elongated to spherical but irregular. Called "wire-wound" 

by bead experts. Hole 3^ mm.; bead diameter 3-15 mm., 

length 10-15 mm. 
"Bunch of grapes" or "raspberry" according to bead experts. 

Very irregular in both length and diameter, ranging from 7- 

10 mm. Hole 2-4 mm. Most of this sample clear glass color. 
"Gooseberry" according to bead experts. Spherical of irregular 

lengths with widely to closely spaced, milky-white stripes 

inside. The ends are usually square but some are irregular. 

Length ranges from 4-13 mm. and diameter 5-10 mm. 
STTiall, fiphfirinal; diamptpv R mm 


8 
45 

8 
5 

26 

274 

2 


Clear glass . .- 


Spherical to oval, small; diameter 3-4 mm .. 


5 








Total -- 




373 









The jar, an excellent specimen of Serra Painted (red and black on 
white) both in form and ornamentation, has a symmetrical, double- 
recurved body profile, a flat base and a short, vertical neck (pi. 25). 
It is 43.5 cm. high with a mouth diameter of 20 cm., a base diameter of 
10 cm. and a maximum body diameter of 42 cm. with the recurved 
collar measurmg 28 cm. in diameter at the largest bulge. The exterior 
surface is smoothed and, in the area where painting occurs, white 
slipped. The designs are executed in red and black and consist of 
square and curved spirals composed of paired or triple, parallel lines, 
typical motifs of Serra Painted. The specimen and most of the beads 
are deposited with the Comissao BrasUeira Demarcadora de Limites in 
Bel6m. 

SITE A-16 ILHAS DO CAMPO 

The Rio Uaga flows into the Atlantic Ocean a few kilometers east of 
the mouth of the Rio Oiapoque, forming a narrow peninsula (fig. 1). 
This finger of land is covered with grass except for the small, scattered 
groves of forest that mark slight increases in elevation. A habitation 
site, measuring about 30 meters in diameter, is reported by Fernandes 
(personal communication) to be located in one of these patches of 
forest near the eastern tip of the peninsula, only a few kilometers from 
the Rio Oiapoque. A piece of fire-burnt quartz, 104 sherds, and a 
small, stone, hand ax were collected from the surface; no excavations 
were undertaken. The ax was made of a waterworn, basalt pebble, 
well polished, with a sharp, beveled bit and a butt end that had been 
used as a hammerstone; measurements: 8.0 cm. long, 2.2 cm. thick, 
bit 4.0 cm. wide, poll 5.0 cm. wide, butt 1.6 cm. wide. The sherds, 



Ev^A^Nsf^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 121 

now in the private collections of Srs. Eurico Fernandes and Frederico 
Barata in Belem, represent the following types: 56 sherds or 54.0 
percent Serra Plain, 2 or 1.9 percent Serra Painted, 29 or 27.9 percent 
Ariste Plain, 1 or 0.9 percent Ariste Painted, 1 or 0.9 percent Davi 
Incised, 13 or 12.6 percent Uagd Incised, and 1 or 0.9 percent Un- 
classified. A few fragments of Serra Plain are from unusually deeply 
grooved graters. 

Three face adornos or figurines are included, 2 Serra Painted and 1 
Serra Plain. The latter is modeled in the form of a head, probably 
monkey rather than human, with the eyes, mouth, and nose formed by 
low relief, incisions, and punctates; the head measures 4.7 cm. wide, 1.6 
cm. through the thickest part, and extends 2.2 cm. from the slightly 
constricted neck. Although it is impossible to determine with certainty 
whether the face is a fragment of a figurine or a rim adorno, in this case 
the latter possibility is suggested by the curvature of the lower edge 
(pi. 26, c). The other two examples appear to represent human faces. 
The largest (pi. 26, h) has an elongated body topped by a realistically 
modeled face with a complicated hairdo ; the features are made by low 
applique, punctate, and light incision. Total height is 9 cm., maximum 
head width 5.5 cm., thickness 2.6 cm. The entire surface is smooth, 
showing traces of white slip upon which red paint had been applied. 
The lower edge has a smooth break resulting from application when 
too dry, thus making a poor bond. Nevertheless, it was attached to 
something else; whether a rim or the basal part of a figurine cannot be 
determined. The features of the third face (pi. 26 a) are more gross 
than the other two. Again the eyes, nose, mouth and hair are shown 
by a combination of low modeling, incision, and punctates. The front 
of the head is white slipped with fine crackle lines, while the back is 
unslipped. Width of the face at the ears is 8.5 cm., thickness at the 
nose 2.5 cm. Although the tips of the base are broken off, the lower 
edge is smoothed over and finished. This clue suggests that some of 
these faces may be small figurines of the style from Site A-1 1 (fig. 30, 
a, b) rather than rim adornos. 

SITE A-18 — MAICA 

On the north side of the Rio Cunani, about 8 km. upstream from the 
present city of Cunani, rise a series of low, rolling, forested hills sur- 
rounded by savanna. On one of these, called Maica, a site was located. 
This hill is 200 meters in a north-south length and 90 meters in an 
east-west width and rises 30 meters above the plain, with its flanks 
forming the bank of the Rio Cunani. After a cahoclo had cleared the 
summit for a garden, sherds were noticed scattered sparsely over an 
area conforming to the general contour of the top of the hiU, 75 meters 
long and 10 meters wide, with the deepest sherd 5 cm. below the sur- 



122 BUREAU OF AMERICAN ETHNOLOGY [bull, is? 

face. Both topography and ceramic features closely duplicate those 
at Site A-12 — Cruzeiro. Of the 222 sherds collected in March 1949 
by Sr. Newton Cardoso of the Museu Territorial, Macapa, and later 
analyzed by us, 124 or 56.0 percent are Serra Plain, 72 or 32.4 percent 
Ariste Plain, 2 or 0.8 percent Davi Incised, 22 or 10.0 percent Uaga 
Incised, and 2 or 0.8 percent Unclassified. Nonceramic objects 
included 6 quartz and 4 miscellaneous natural rock fragments and 1 
fragment of a roughly shaped granite hammerstone (6.5 X 5.0 X 3.0 
cm.). 

SITE A-19 RENOVADO 

A few kilometers below the village of Cunani, Sr. Cardoso visited 
another site on a hilltop along the Igarape Holandia near its junction 
with the Rio Cunani. At the time of this survey work he was not 
aware of the fact that he was working in the same site excavated by 
Goeldi in 1895, for, today, the hill is called Renovado instead of Monte 
Curu.^^ He found one shaft grave empty, while another had 56 
sherds representing 4 vessels (2 Serra Painted and 2 Serra Plain) 
scattered among cremated bones, suggesting previous excavation. 
The sherds, including a flat base sherd mth several pierced holes, 
conform to the style reported by Goeldi. The tabulation of the 
number of vessels in each potterj^ type from Site A-19 (Appendix, 
table 11) includes Goeldi's specimens and Cardoso's, both of which 
were analyzed by us. 

SITE A-20 VILA CUNANf 

A few artifacts were recovered from a small excavation in the 
center of the city of Cunani on the Rio Cunani. The sample collected 
by Sr. Cardoso represented a mixture of colonial clay brick, porcelain 
ware, hunks of coal, modern caboclo pottery and Ariste Phase ceramics. 
There were 15 sherds of Serra Plain, of which 2 were from graters, and 

1 sherd of Serra Painted. The ceramics resemble the material from 
Maica, Site A-18. Owing to the present location of buildings upon 
the aboriginal habitation site, extensive excavations were not feasible. 

SITE A-21 PKACUtrBA 

Habitation Site A-21, on the north shore of Lago Pracu(iba, about 

2 km. northwest of Site A-9 — Relogio, was excavated by Sr. Cardoso. 
Although he attempted the stratigraphic method, the discovery of an 
airplane latch in the second level confirmed the suspicion that the site 
was too much distm'bed from modern manioc gardening to give reliable 
stratigraphic information. The refuse layer was a black loam in con- 
trast to the light-brown, sterile clay. Of the 241 sherds recovered 
from the 1.5- by 1.5-meter test, 208 or 86.5 percent were Serra Plain, 

'2 The descriptive details of the site, burials, etc. are in Goeldi, 1900, pp. 22-24, and are discussed in the 
comparative section of this Phase, pp. 136-128. 



evTns?^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 123 

5 or 2.0 percent Serra Painted, 27 or 11.1 percent Ariste Plain, and 1 
or 0.4 percent Davi Incised. Six of the Serra Plain sherds are from 
graters. A partially restorable Serra Painted bowl is similar in shape 
and design motif to the carinated bowls from Cunani found by Goeldi 
(1900, est. Ill 2a, 4, 8). 

SITE A-22 — CONCEigio 

Sr. Newton Cardoso visited this site on Fazenda Santa Maria da 
Prainha on the north bank of the lower Rio Amapa Pequeno during 
the rainy season. The slight, forested elevation surrounded by 
savanna on which the site was located is known by the name of Ilha 
das Igagabas, and is reputed to contain whole vessels. However, it 
is one of the few areas to remain above water during the rainy season, 
and the weight and tramping of thousands of cattle that seek refuge 
on it have reduced the ceramics, which are 5 cm. or less below the 
surface, to gravel. Only 91 sherds could be collected in spite of exten- 
sive digging. These resemble the ones from Site A-7 — Amapa. 
Seventy-nine or 87.0 percent were classified as Serra Plain, 3 or 3.3 
percent Serra Painted, and 9 or 9.7 percent Ariste Plain. Non- 
ceramic objects included 1 small fragment of a hammerstone and 3 
hunks of burnt clay. 

DATA FROM OTHER INVESTIGATIONS 

Less previous work has been done in the region between the Rio 
Oiapoque and the Rio Araguari-Amapari than in the southern part 
of the Territory of Amapa. Beyond the information from Nimuen- 
dajli's work and Goeldi's excavations and collections of Cunani 
materials, there are only occasional references by early explorers and 
travelers to an "Indian site" or a "pot believed to be of Indian 
origin" without details that would permit their use in a specific com- 
parative study. The following information is arranged geogi^aphicaUy 
beginning \vith the Rio Oiapoque and moving south to the Rio 
Araguarl-Amapari. 

RIO OIAPOQUE SITE 

Hamy describes an aboriginal burial urn found by Mgr. Emonet 
during one of his trips on the Rio Oiapoque (Hamy, 1897, pi. 57). 
The jar is plain ware with a reddish color, which could be Serra Plain, 
and has a lid "in the form of a hat" with a reddish, well-smoothed 
surface (fig. 36). The jar measures 40.0 cm. in height and 36.6 cm. 
in diameter, with a flattened base and a high bulbous waist. The 
lid, inverted over the jar mouth, is 17 cm. deep with a mouth diameter 
of 35 cm. Both the shape and the comments about the sm'face of the 
lid suggest that it represents the red variety of Ariste Painted. 
Experience in dealing with these early accounts suggests that there is 



124 



BUREAU OF AMERICAN ETHNOLOGY 



[BOLL. 167 




^^V^ 


w 






''■',k 








¥ 


L 


1 1 


1 1 1 




5 


CM 



Mii^^i^'^M^^^^^^ 



Figure 36. — Burial urn and lid of the Ariste Phase found by Hamy (1897) on 
the Rio Oiapoque. The urn is probably Serra Plain and the lid Arista Painted. 

little reason to take much stock in the author's caption, "Urne 
funeraire des Oyampis de I'Oyapok" (ibid.)- The tendency is to 
attribute all such finds as belonging to the Indians living in the area 
at the time the jar was found and there is no evidence to indicate that 
this is an ethnographical specimen. All of its characteristics associate 
it with archeological examples of the Ariste Phase. 



JfvA*'M«f^ *^° ARCHEOLOGY AT MOUTH OF AMAZON 125 



EVANS] 



RIO vaqL sites 



In his discussion of the archeology of the Lower Amazon, based on 
Nimuendajti's notes and specimens in the Goteborg Museum, Linne 
reproduces a map showing the location of 4 cemetery and 2 habitation 
sites on the middle to upper drainage of the Rio Uaga (1928 a, p. 584). 
His information is unfortunately very sketchy, but Ryden's study of 
Nimuendaju's notes and collection (Ryden, MS.) adds a few more 
details. 

At the burial site of Courbaril, Nimuendaju found a large, flat 
platter with a white-slipped interior painted with black lines; this 
vessel is undoubtedly Serra Painted. Other fragments comprise 
graters with deep grooves on the interior, comparable to the ones 
from Sites A-12 and A-16. 

Another burial site, Coumarouman, had most of the vessels broken 
by roots. From Ryden's (MS.) description of traces of white slip with 
black painting or red-brown to black surfaces, of elongated flanges 
opposite each other on the rim, and of carinated vessel shape with 
flat bases, there seems little doubt that this pottery is typical of 
the Ariste Phase. 

RIO ARACAUX SITES 

Kaupi. — Nimuendaju (1926, pp. 85-86; Ryden, MS.) mentions that 
the modern Palicur Indians when digging a grave often encounter an 
old burial site of their ancestors; they show no reverence for the site, 
dig up the urns, break them, and take any offerings such as beads. 
While exploring the Rio Aracaud, Nimuendaju went to a burial site 
known as Kaupi. Here a local Palicur magician dug up about 12 
burial urns, cleaned them out, washed them, and when told that they 
should have been left in the ground, he broke them and threw the 
fragments away. Nimuendaju rescued some of the fragments which 
Ryden illustrates and describes (Ryden, MS., figs. 30 A-E, 28 A-B). 
One vessel of this group was illustrated in an annual report of the 
Goteborg Museum (1927, p. 77). From these data it is evident that 
the pottery is representative of the Ariste Phase. The flat perforated 
base is typical of the vessels and sherds from the Cunani burial urns 
(cf. p. 128); the polychrome or bichrome painting in black, brown, 
red brown or orange brown on a white slip featuring parallel lines 
and interlocking units is typical of Serra Painted; the sherds grooved 
on the interiors are identical to the graters found from Sites A-12 
and A-16. Nimuendajii (Ryden, MS.) undertakes an extensive 
discussion on the subject of these "grin ding-bowls," as he calls them, 
and indicates that old Palicur Indians in the Kaupi region mention the 
fact that their ancestors used pottery "grinding bowls" for manioc 
whereas today they use wooden boards with inserted fragments of iron. 



126 BUREAU OF AMERICAN ETHNOLOGY [bull, le? 

Ulakte-Uni. — Near Ulakt6-Uni on Mont Ukupi along the Aracaud 
River is a small cave which had fragments of burial urns scattered 
on the floor. The pottery was white slipped and painted in red 
curvilinear designs, with one sherd modeled in the style of the anthro- 
pomorphic faces on the necks of Cunani vessels. Comparison of the 
CunanI vessels with illustrations (Nordenskiold, 1930, pi. 23; Linne, 
1928 a, fig. 2, p. 585) and descriptions (Ryden, MS.) of these sherds 
show them to be identical to the Serra Painted pottery found by 
Goeldi at Cunani and by us at Sites A-11 and A-21. Nimuendajii 
made special effort to look for trade materials and the previous 
contents of the vessels or bone scraps on the cave floor, but none 
could be found. 

Maimr-Mini. — At this burial site Nimuendajii (Ryd6n, MS.) found 
a vessel that was related in shape to those he had found at Monte 
Maye. No traces of paint or slip remained on the exterior, but it was 
so badly weathered that it is impossible to determine if the vessel 
originally had been plain or painted. Since its shape is similar to 
Monte Maye vessels, this burial site undoubtedly belongs to the 
Ariste Phase. 

CUNANf SITES 

The best known collection from the Territory of Amapd is the one 
made by Goeldi in 1895 from the Rio Cunani (1900, pp. 1-43). It 
includes a large number of jars and bowls recovered from the bottoms 
of two shaft graves sunk vertically into the ground with enlarged 
alcoves at the base (fig. 37) . All the urns contained traces of human 
bones mixed with dirt, and although he mentions some being cre- 
mated, Goeldi is not always explicit on this point. The existence of 
secondary burials, as well as cremated remains, would place this 
cemetery in the Ariste Phase pattern, such a combination having 
occurred at Site A-11, Cave 2. No glass trade beads were associated 
with any of the vessels. However, the fact that the faces of three of 
the five anthropomorphic jars have weU-modeled beards along the chins 
(cf. Rio Maracd anthropomorphic urns, p. 78) might be evidence 
of their post-Columbian manufacture in attempting to copy the 
full beards of the European conquerors onto ceramics. 

Our pottery analysis of these vessels collected by Goeldi, now 
deposited in the Museu Goeldi in Belem, and the sherds taken by 
Sr. Cardoso from the same spot. Site A-19, showed 17.4 percent to 
be Serra Plain and 82.6 percent Serra Painted. The flat-bottomed, 
angular-sided, cambered-necked jars and wide-mouthed, carina ted 
bowls are typical Ariste Phase forms. The interlocking, curvilinear, 
meandering scroll motifs of vessels 1, 3, 5, and 8 in Goeldi's report of 
the site (1900, est. I, II, III) are duplicated in style, color, and method 
of execution on several sherds from Sites A-11 and A-21 (fig. 33). 



MEGGERS AND 
EVAN'S] 



ARCHEOLOGY AT MOUTH OF AMAZON 



127 




Figure 37. — Profile and top view of shaft-burial at the Cunani Site, Ariste 
Phase (After Goeldi, 1900.) 



128 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

The perforations through the base of the majority of the bowls and 
jars occur also at Kaupi on the Rio Aracaud. Since all these vessels 
were used for burial, the only plausible explanation of the holes would 
be to allow drainage, since no lids covered the vessels in the shaft 
graves. 

The unusually constructed shaft grave with a widened-out recess 
and alcove 2.10 meters in diameter, reached by a vertical shaft, 1.20 
meters in diameter and 2.50 meters deep, and covered with a large 
rock-slab lid (fig. 37) does not recur at any other known site in the 
Territory of Amapa except one visited by Nimuendajii (Ryden, MS.) 
at Rio Novo (pp. 41-42). With the exception of two small cemeteries 
in which the jars were buried, Sites A-14 and A-15, burial urns were 
always set on the floor of small rock shelters, caves, or along the faces 
of large rock outcrops. 

Although the habitation Site A-18, Maica, is upstream on the same 
river, the difference in ceramic styles and percentage occurrence of 
pottery types (Appendix, table 11; fig. 46) between it and the Cunanl 
shaft cemetery. Site A-19, argues against a close relationship. A 
better correlation on ceramic type can be made with Site A-20, situated 
underneath the modern village of Cunanf ; its proximity also makes this 
the more likely associated site. 

Coudreau (1887) describes his excavation of funerary urns from an 
Indian cemetery on the upper Rio Cunani about 15 km. from our Site 
A-18 (fig. 1) and the existence of other sites along the river with esti- 
mates of their age: 

P" I found seven funerary urns in a burial shaft in the village, and I compared them 
with the burial urns I had visited at the mountain of Counani. These urns, added 
to those of the large encampment located on an island in the river three days 
distant by canoe, the remnants of which I have been able to study, permit me to 
infer a general history of the Indian civilization on the river. . . . 

Fleeing the European, the Indians moved successively upriver until the site of 
the present town of Counani. Even today, the remnants of four or five ancient 
encampments, together with the cemeteries . . . can be distinguished between 
the town and the mouth . . . 

When the Jesuits, in the 18th Century, founded a mission on the river, they had 
a reason for establishing it not far from the location of the modern town, and this 
reason was probably that of the contemporary existence of the principal village on 
this site. The urns I found in the burial shafts are doubtless of that epoch, to 
judge by the remarkable perfection of the designs. . . . The three-quarter cre- 
mated bones, rotted by humidity and filled with dirt, that I found in the urns, 
did not seem to me to be of any use for the chemical determination of their antiq- 
uitJ^ [Coudreau, 1887, pp. xx-xxi.] 

MONTE MATf SITE 

The site of Monte Maye, on a small mountain 100 meters high on 
the coast along the south side of the mouth of the Rio Cunani, has 
been known for a long time by explorers of the region (Goeldi, 1900, 



"vANsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 129 

footnote, p. 17). Both Goeldi and Nimuendajii (Ryd6n, MS.) develop 
in some detail the historical data about the Indians of the area gath- 
ered by the missionaries and explorers in the 17th and 18th centuries. 
However, the most specific information from an archeological standpoint 
comes from the explorations of Nimuendajti in 1923, part of which 
have been published (Linne, 1928 a, pp. 587-588) and the rest com- 
piled and annotated in manuscript form by Ryden (MS.). 

There were two surface burial sites on the ridge of Monte Maye, 
both of which had been so badly plundered that sherds, bones, and 
miscellaneous objects were partially buried and mixed with earth and 
leaves. Local inhabitants had carried away pendants and beads and 
even bm-ial urns, which they used as water or cooking jars. 

At the first burial site (called Urn Site A) Nimuendajti found frag- 
ments of 12 jars scattered haphazardly over the surface with two 
complete vessels in front of a small rock shelter. From the surface 
near several broken jars came bone fragments, human teeth, a 
tooth necklace, beads, rusty iron nails, a mirror, murakitdos (pend- 
ants) of greenstone, and pierced thimbles. Cahoclos stated that they 
had also found lockets with pictures of Catholic saints in them. From 
Ryden's descriptions of the vessels and sherds Nimuendajti sent to the 
Goteborg Museum, plus our examination of some of the photographs 
and drawings, there is no doubt that the first burial site of Monte 
May6 is representative of the late Ariste Phase material as typified 
by the polychrome or bichrome painting and vessel shapes so 
characteristic of Serra Painted. 

The second burial urn site (Nimuendajti's Urn Site B) is located on 
the same ridge, 100 meters away, in a place without natural stone out- 
crops. The fragments of 7 different urns were on the surface along 
with scattered bones, a broken iron knife, glass beads, a mirror, brass 
bells, and a greenstone murakitdo, 

Nimuendajii mentions several upturned lids at the foot of the hill, 
which he designated as Urn Site C, but he concludes that since these 
were the only fragments, they were probably left there by a traveler 
who started to carry things away from Monte Maye but changed his 
mind (Ryden, MS.). 

From Ryden's descriptions of the vessels and sherds (ibid.), Linne's 
comments (1928 a), Nordenskiold's illustrations (1930, fig. 2a), plus 
our examination of some of the photographs and drawings, there is no 
doubt that these burial urn sites of Monte Maye contain late Ariste 
Phase material.*^ From the presence of glass beads, mirrors, and 
metal objects it is obvious that the cemeteries date from post-Colum- 
bian times. The brass shells and thimble were sent to a specialist in 

" For full descriptive and Illustrative details see Ryden's compilations and annotated study of Nim- 
uendajfi's archeological investigations in the Territory of Amap& (Ryd6n, MS.). 



130 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

European cultural history, Prof. Nils Lithborg of the Nordic Museum 
in Stockholm, who "without the least knowledge of where or when they 
had been found, placed the period of manufacture between 1450 and 
1530 AD" (Linne, 1928 a, pp. 587-588). Nimuendajii (Ryden, MS.) 
gives several pages of discussions about the age of these sites based 
on the glass beads as suggesting 17th century, but Ryden feels they 
could be as late as 18th century. 

Some of the burial urns were perforated along the upper rim with 
corresponding holes along the edge of the lid. Fragments of small 
animals modeled on top of the lids came from this site. Several schol- 
ars have noted that these zoomorphic figures on perforated lids show 
similarity to those from the Atures on the Rio Orinoco (Linn^, 1928 a, 
p. 589, fig. 4; Nordenskiold, 1930, fig. 2, a-b, p. 18). Although nothing 
like them came from other Ariste Phase cemeteries, small modeled 
animals are found on the outer rim of certain of the large bowls from 
Goeldi's excavations at Cunani (1900, est. I 7a, b; and III la, b, c; 
2a; 22). 

ILHA DO CARAO SITE 

Nimuendajti reports briefly on his excavations of Ilha do Carao in a 
swamp at the mouth of the Igarape Mayacar^, south of the Rio Calgoene. 
Unfortunately, the descriptive data are confusing, contradictory, and 
inadequate (Linn^, 1928 b, p. 75-76; Meggers, 1948, p. 162; Ryden, 
MS.). Nimuendaju speaks of a mound 2.20 meters high with three 
stratified layers, but says these laj^ers do not appear to correspond with 
those of distinct cultures. Unfortunately, the catalog of the specimens 
does not list them by level, but Ryden felt that he was able to separate 
some of the materials. However, this is not as fruitful as it might 
seem, for the illustrations and descriptions of the incised designs and 
traces of red and white paint on the pottery suggest that all the sherds 
represent the Ariste Phase pottery types of Ua5a and Davi Incised 
and Serra and/or Arista Painted. Since there is such a high percent- 
age of incised pottery, this site is probably related to that part of the 
Arista Phase represented by Site A-16 — Ilhas do Campo. Even 
though Linne publishes Nimuendajii's map of the site and profile of 
the mound with the numbered artifacts in place (1928 b, fig. 3), the 
details of stratigraphy and the meaning of the numbered artifacts 
are not given. Although Nimuendaju mentions a dozen stone markers 
scattered irregularly on the mound (op. cit., pp. 75-76; Ryd^n, MS.), 
the profile shows stones throughout the mound arranged in such a 
way that the site suggests a burial place in which the urns were in- 
terred and then covered with dirt and stones. From the scanty data 
the stones at this site do not suggest alinements similar to those of 



BVANsP ^^° ARCHEOLOGY AT MOUTH OF AMAZON 131 

the Arua sites even if the stones were as large as some from the Rio 
Novo or Jos6 Antonio sites described by Nimuendaju. Although much 
is wanting in the way of more concrete information, without any 
doubt this site belongs to that part of the Ariste Phase characterized 
by a high percentage of incised pottery types. The stone axes are not 
distinctive enough to be assigned to a cultural horizon. 

AgAHYZAL SITE 

This stone alinement on the Rio Flexal (Frechal) has been de- 
scribed in detail in the Arua section (p. 43). Nimuendajti mentioned 
that "contrary to all other stone alinements a great number of vessel 
fragments were found. . . . The majority of these were coarse and 
undecorated; no painting; and some had coarsely incised decora- 
tions . . ." (Ryd6n, MS.), An examination of Ryden's illustrations 
of these sherds (op. cit., fig. 26) establishes the incised ones as typical 
examples of Uagd Incised and Davi Incised. Since no complete vessels 
were found, the description of Agahyzal site suggests an old habitation 
site of the Arista Phase, a portion of which had previously been used 
by the Arua for the construction of stone alinements. 

IGARAP^ TARTARUGALZINHO SITES 

Coudreau, in the description of his voyages to the Guianas (1887, 
pp. 49-50), mentions an Indian cemetery on the right bank of the 
Igarap^ Tartarugalzinho : 

The burials were situated in a line, running exactly east-west and perpendicular 
to the direction of the river, which was about 20 meters away. They were about 
a meter apart and level with the ground in the middle of an area slightly wooded, 
indicating a relatively recent exploitation. 

These urns are of crude workmanship, without ornament, of poor clay. . . . 
Each had a little cover provided with holes for attachment. When discovered 
(about a decade ago, at the time when Magalhens excavated), they contained 
bones, which have since been dispersed. Above the small cover was a large cover 
of coarse pottery; this cover was hidden under a miniature mound. The eastern 
urn, probably the most beautiful, was the one taken by the apostolic prefect 
[to Cayenne, French Guiana]. 

In the urn at the extreme west, a small urn more elegant than the others and prob- 
ably that of an infant, I found some blue and white beads, the size of grains of 
wheat, which must have come from the necklace of the little Indian. . . . The 
second urn, going toward the east, contained large red and blue beads, a necklace 
of a man perhaps. The third contained only dirt . . . 

Although his comments are not specific enough to classify the 
pottery with certainty, it is very likely that the vessels represent 
either Arista Plain or Serra Plain. Sites of the Ariste Phase are 
common in the vicinity and the burial pattern suggests that of Sites 
A-10, A-11, and A-19. 



132 BUREAU OF AMERICAN ETHNOLOGY [boll, xei 

ANALYSIS OF MATERIALS OF THE ARISTE PHASE 

Pottery Type Descriptions 

The ceramic study of the Ariste Phase is based upon an examination 
and classification of 2,156 sherds from habitation sites and 215 com- 
plete or restorable vessels from cemeteries. Utilizing the binomial 
classificatory system the following pottery types, arranged in alpha- 
betical order, were established for the Ariste Phase. 

arist£ plain 

Paste: 

Method of manufacture: Coiling. 

Temper: Sand and crushed quartz particles ranging from 1-9 mm. 

Texture: Clayey paste, very gritty with angular cleavage due to the large 

sand particles and hunks of crushed quartz. Paste mixture poor, weak, 

and friable. Sherd rings like a hollow tile or brick. 
Color: Ranges from light, rusty orange to reddish orange; about 10 percent 

of all sherds have a gray paste with orange bands on either side of the core. 
Firing: Oxidized, generally complete; fire clouds rare. 

Surfaces: 

Color: Ijight orange to dark rusty brown to a light tan on both surfaces; 
no constant correlation between exterior and interior color. White 
quartz particles of temper on the surfaces often give the sherds a speckled 
appearance. 
T?-eatment: 

Exterior — Well smoothed, fairly even and regular but not polished. 
Smoothing striations visible on only a few sherds. Large temper 
particles on the surface often give it a rough feel with crackle lines 
formed around these exposed particles. Coiling lines completely 
erased. 
Interior — Not as well smoothed as surfaces just mentioned, but scraping 
tracks visible on 10 percent of the sherds. 
Hardness: 3. 

Form: 

Common vessel shapes: 

1. Small to medium low-waisted or round-bodied jar with narrow, 
short neck and cambered collar (fig. 38-1; pi. 20, b-c). 

Rim: Cambered collar with a short nm vertical or extending 
outward 1-3 cm. from the neck. The Up is tapered, rounded or 
square with rounded edges. Mouth diameter 18-32 cm.; con- 
stricted neck diameter 10-26 cm. 

Body wall thickness: 7-12 mm.; majority 7-8 mm. 

Body dimensions: Neck and rim height 5-15 cm.; maximum body 
diameters 26-38 cm.; overall height 22-47 cm.; majority 30-35 cm. 

Base: Thickened, flattened, usually slightly concave with a promi- 
nent angular junction with the side walls. Depth of the con- 
cavity is 2-5 mm. with the base diameter 8-18 cm. 

Appendages: Rarely, a small strap handle from the collar to the 
jar neck or applique nubbins or ribs. 



il^.*l?ef® ^^" ARCHEOLOGY AT MOUTH OF AMAZON 133 

By AN S J 






Vessel Seal* 




Figure 38. — Rim profiles and vessel shapes of Ariste Plain, Arista Phase (Ap- 
pendix, table 12). 



39132&— 57 11 



134 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

2. Short to medium-necked round or low-waisted jar with exteriorly 

thickened everted or outcurving rim (fig. 38-2). 

Rim: Everted, thickened exteriorly with a coil or strip, lip rounded 
or squared with rounded edges; rarely a direct lip. Mouth 
diameters 26-38 cm. 

Body wall thickness: 7-11 mm.; majority 7-8 mm. 

Body dimensions: Neck diameter 20-32 cm., neck height 3-10 cm., 
overall height 25-48 cm.; body diameters 36-40 cm. 

Base: Thickened, flattened, usually slightly concave with a promi- 
nent angular junction with the side walls. Depth of the con- 
cavity is 2-5 mm. with the base diameter 8-18 cm. 

3. Open bowls, shallow or deep, with gently outcurving sides (fig. 38-3). 

Rims: Rounded lip, sometimes slightly thicker than the body 
wall; mouth diameters 14-25 cm. 

Body wall thickness: 7-10 mm. 

Body dimensions: Body diameters 12-27 cm.; depth 5-15 cm. 

Base: Slightly thickened, flattened and usually concave with promi- 
nent angular junction with the sidewalls. Thickness 1.5 cm.; 
diameter 4-12 cm. 

4. Bowl with incurving sides and constricted opening (fig. 38-4; pi. 

20, a, d). 
Rims: Incurving rim with squared lip with rounded edges, often 

slightly thickened on the exterior; mouth diameter 8-14 cm. 
Body wall thickness: 7-10 mm. 
Body dimensions: Maximum body diameter 12-24 em.; depth 

8-15 cm. 
Base: Slightly thickened, flattened and usually concave with 

prominent angular junction with the side walls. Thickness 

1.5 cm.; diameters 5-12 cm. 
Less common vessel shapes: 

1. "Graters" — Open, carinated bowl with outcurving rim with parallel 

grooves on bowl interior (pi. 23, d) . 

Rims: Outcurved, unthickened, with rounded Up; mouth diameters 
38-40 cm. 

Body wall thickness: 12-14 mm. 

Base: No complete vessel found, probably flattened. 

Grater groovings: Parallel grooves, 2-4 mm. wide, 3 mm. deep 
and 4-7 mm. apart, cut into the interior of the bowls but not 
upon the short, vertical or slightly outslanting side wall. In 
cross section the grooves are V-shaped with a rounded base to the 
V, Since no base sherds were found with these incisions on 
Arist6 Plain, it is assumed that the parallel grooves are limited to 
the lower part of the curved bowl interior. 

2. Flat lid with a central nubbin handle. Crude and irregular, 16 cm. 

in diameter, 1.6-2.4 cm. thick; nubbin handle 3 cm. high, 1.5 cm. 
in diameter at the top and 5 cm. at the base. 
Temporal difference within the type: Cambered collared jar (form 1) 

increases in popularity through the sequence (See Appendix, table 12). 
Chronological position of the type: Most popular in the lower part of the 
Arist6 Phase sequence in both the cemetery and habitation sites, fading out in 
the middle part in the cemetery sites with only a trace in the habitation sites of 
the upper (late) part of the sequence. 



^.f?^^^^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 135 

arist£ painted 

Paste and unpainted surface: The painted pottery is on Arista Plain; see that 
type for details of paste, temper, firing, and treatment of the unpainted surface. 
Form: 

Common vessel shapes: 

1. Small to medium low-waisted or round-bodied jar with narrow, short 

neck and cambered collar (fig. 39-1; pi. 21, a). 

Rim: Cambered collar with a short rim wall, usually vertical, ex- 
tending 1-3 cm. out from neck. Lip tapered, rounded or square 
with rounded edges. Mouth diameter 18-32 cm.; constricted 
neck diameter 10-26 cm. 

Body wall thickness: Range 7-12 mm.; majority 7-8 mm. 

Body dimensions: Neck and rim height 5-15 cm., maximum body 
diameter 26-38 cm., overall height 22-47 cm. with the majority 
30-35 cm. 

Base: Thickened, 1-2 cm., flattened and usually slightly concave 
(depth of concavity 2-5 mm.), prominent angular junction with 
side walls. Base diameter 8-18 cm. 

Appendages: Rarely a small strap handle from collar to jar neck or 
applique ribs or nubbins on collar or upper part of jar body. 

Decoration: Commonly a red band on neck or collar or on both and 
extending to the upper shoulders of the jar body. Polychrome 
design on neck and upper part of body on a few sherds. 

2. Low-waisted jar with a long, outcurving neck (fig. 39-2). 

Rim: Usually curves gracefully outward with a direct or slightly 

exteriorly thickened rim; lip rounded or square with rounded 

edges. Mouth diameter 26-38 cm. 
Body wall thickness: 7-11 mm.; majority 7-8 mm. 
Body dimensions: Neck diameter 20-32 cm., neck height 3-10 cm., 

overall height 25-48 cm., body dimensions 36-40 cm. 
Base: Same as form 1. 
Decoration: Same as form 1, plus the common occurrence of red on 

the inner lip, rim exterior, and upper part of the jar. 

3. Shallow, open bowl with outcurving sides, sometimes slightly cam- 

bered, and usually an everted rim (fig. 39-3). 

Rim: Everted, unthickened with lip tapered, rounded or square 
with rounded edges. Mouth diameters 14-26 cm. 

Body wall thickness: 7-10 mm. 

Body dimensions: Depth 5-12 cm. 

Base: Slightly thickened, flattened and usually concave with promi- 
nent angular junction with the side walls. Thickness 10-15 mm. 
Diameter 4-12 cm. 

Decoration: Red paint on inner lip, exterior of rim, and in a band 
extending for a few centimeters down on the side wall. 

4. Bowl with incurving sides and constricted mouth (fig. 39-4; pi. 

21, b). 
Rim: Incurving, sometimes slightly thickened on the exterior, 

beveled and tapered to rounded lip. Mouth diameter 10-28 cm. 
Body wall thickness: 4-10 mm. 
Body dimensions: 14-32 cm.; depth 8-16 cm. 
Base: Same as form 3. 
Decoration: Red band around exterior of rim, especially on the 

upper suface of the beveled type. 



136 



BUREAU OF AMERICAN ETHNOLOGY 



[BUIiIi. 167 




FiGUBE 39. — Rim profiles and vessel shapes of Arista Painted, Ariste Phase 

(Appendix, table 13). 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 137 

BVANSJ 

5. Large, open, carinated bowl with wide, flaring, flange rim (fig. 39-5). 

Rim: Strongly everted with a wide (6-10 cm.), outcurving flange, 
sometimes slightly thickened on the interior. Lip either rounded 
or squared with rounded edges. In about half the sherds the rim 
is lobed. These lobes or protrusions with rounded edges range 
from 1.5-4.0 cm. in width and 8-10 cm, in length. Mouth 
diameters 28-38 cm. 

Body wall thickness: 5-9 mm; majority 7 mm. 

Body dimensions: Body diameter at carination 20-32 cm.; height 
12-16 cm. 

Base: Same as form 3. 

Decoration: Usually polychrome (red, yellow, black on white slip 
or tan surface) or bichrome (red on tan) complicated designs of 
meanders, frets, interlocking spirals, lines and dots on bowl 
interiors, especially on the wide flange (fig. 40, a-b). Sometimes 
the interior is white slipped with a painted design on the interior 
of the flange surface with pattern outlined by a black line. 
Rarely, this shape is painted only red on the unslipped surface. 

Rare vessel shape: 

1. Jar with anthropomorphic face modeled on the neck and cambered 
collar. Shape usually the same as common form 1, with the face 
modeled in low relief with applique eyes, nose, and ears. The 
entire face is painted red (fig. 35) . 

Decobation: 

Technique: Most commonly a dull-red ochre thickly applied, sometimes 
almost thick enough to be a slip. A few are bichrome or polychrome with 
red or red and black fine to medium lines upon a thick white sUp. Painting 
is usually on the interiors of bowls, exterior of jar necks and bodies, bowl 
or jar Ups or in a band around the cambered rim of jars. 
Motif: Ninety percent of all sherds are a red painted monochrome in parallel, 
wide bands or a thick red band around the neck, exterior of the rim or 
from the rim to the base of the neck. In a few cases the band extends 1-2 
cm. onto the shoulder of the jar. Ten percent of the painted forms are 
polychrome or bichrome in complicated designs of meanders, frets, inter- 
locking spirals and lines with an occasional use of dots (fig. 40, a, b) . 
Temporal difference within the type: Collared, cambered jars (form 1) 
increase as the jar with thickened rim and outcurved neck (form 2) decreases. 
The cambered bowl with the flangelike rim is limited to this type and is most 
popular in the early part of the Phase sequence (Appendix, table 13). 
Chronological position op the type: Arista Painted seems to be principally a 
cemetery ware and is not found later than the middle of the Arista Phase 
sequence, after which time it is replaced by Serra Painted. 

DAVf incised 

Paste: 

Method of manufacture: Coiling. 

Temper: Quartz sand particles. 

Texture: Gritty, sandy paste, friable and fairly weak with irregular cleavage 

planes. 
Colar: Light orange to reddish orange; only a few have a gray core. 
Firing: O.xidized, generally complete. 



138 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 





Figure 40. — Arista Painted sherds from A-11 — Montanha de Arista, cave 2, 
Ariste Phase: a, Black (solid lines) and yellow (hachured lines) on a thick 
white slip, b, Black (wide lines) and red (narrow lines) on white slip. 

Surfaces: 

Color: Exterior and interior, light orange to a dull, tile orange. 
Treatment: Smoothed, but gritty and rough due to sandy paste. 
Hardness: 3. 
Form: 

Common vessel shapes: 

1. Low-waisted jar with a short neck and Vv-ide, cambered collar (fig. 
41-1). 
Rim: Slightly incurved, cambered collar (4—5 cm. wide) with squared 

lip. Mouth diameter 16 cm. 
Body wall thickness: 5-11 mm. 

Body dimensions: Body diameter 30 cm.; height 24-26 cm. 
Base: Flat with slightly concave center. 

Decoration: Vertical and slightly diagonal incisions on the collar 
exterior; diagonal lines arranged in crude triangular zones on the 
upper body wall. 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



139 



2. Round-bodied jar with flat base and short neck with thickened, 

everted rim (fig. 41-2). 
Rim: Exteriorly thickened with a wide coil, slightly everted rim 

with rounded lip. Mouth diameter 30 cm. 
Body wall thickness: 11-13 cm. 
Base: Flattened, slightly thickened. 
Decoration: Vertical and diagonal incised lines on rim exterior 

haphazardly spaced, ranging from 8-15 mm. apart. 

3. Carinated bowl (fig. 41-3). 

Rim: Carinated with outcurving flange ending in a thinly tapered, 

rounded lip. Mouth diameter 24 cm. 
Body wall thickness: 9 mm. 

Body dimensions: Sherds too fragmentary to reconstruct. 
Base: Flat, slightly thickened. 




Figure 41. — Rim profiles and vessel shapes of Davi Incised, Ariste Phase (Ap- 
pendix, table 14.) 



140 BUREAU OF AMERICAN ETHNOLOGY [bull. 16T 



Decoration: Diagonal lines crudely arranged in triangular zones 
on the exterior. 
4. Open bowl with outslanting walls (fig. 41-4). 

Rim: Slightly everted or direct; rounded lip. Mouth diameter 

12-26 cm. 
Body wall thickness: 11 mm. 

Body dimensions: Sherds too fragmentary to reconstruct. 
Base: Flat. 

Decoration: Incised lines vertical or diagonal on the upper part 
of the side walls. 
Decoration (pi. 21, c-f): 

Technique: Deep, sharp, uneven lines crudely and irregularly incised when 
the clay is extremely wet leaving very jagged and rough incisions 1-2 mm. 
wide and 2-4 mm. deep. Some lines are V-shaped in cross section. 
Motif: A series of irregular, short lines arranged diagonally or vertically on 
the rim and upper surface of the body walls. Irregularity suggests hasty 
application. Roughly arranged in triangular blocks; a few sherds suggest 
curvilinear meanders but these are executed by means of connecting short 
lines. 
Temporal difference within the type: None evident from the limited sample 

(Appendix, table 14). 
Chronological position of the type: Limited to the middle to lower (early) 
part of the Arista Phase sequence. 

flexal scraped 

Paste and surfaces: On Arista Plain, see that type description for details of 

temper, firing, color, and surface treatment. 
Form: 

Common vessel shapes: 

1. Low-waisted jar ■wdth short neck and wide, cambered collar (fig. 32, 
42-1). 

Rim: Wide (4.5-5.0 cm.), cambered collar, slightly incurved or 
straight, with a rounded or squared lip with rounded edges. 
Mouth diameter 16-26 cm. 

Body wall thickness: 6-8 mm. 

Body dimensions: Body diameters 28-32 cm; height 22-26 cm. 

Base: Thickened in center to 1.2-1.5 cm., flat, 12 cm. in diameter 

Appendages: Sometimes two strap handles on opposite sides con- 
necting the collar with upper shoulder of body; oval cross section 
1 cm. in diameter. Small nubbin appliques (1 cm. in diameter 
and height) sometimes on the shoulder. 

Decoration: Scraped decoration of triangles and spirals from neck 
to the jar waist and on the wide, cambered rim. 

2. Round-bodied jar with long, everted, direct rim (fig. 42-2; pi. 22, c). 

Rim: Strongly everted, long (3.0-4.5 cm.) direct rim with rounded 

lip. Mouth diameter, 24-34 cm. 
Body wall thickness: 6 mm. 

Body dimensions: Body diameter 26-36 cm,; height 16-18 cm. 
Base: Same as form 1. 
Appendages: Small nubbin and fillet appliques, 5-10 mm. high, 

8-10 mm. wide, and 1-6 cm. long. 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 

EVANS] 



141 




I I I I I I 1 

4 8 12 CM 
Vessel Scale 



Lj_1 I 1 

I 2 3 CM 



Rim Scale 







Figure 42.— Rim profiles and vessel shapes of Flexal Scraped, Ariste Phase 

(Appendix, table 15). 



142 BUREAU OF AMERICAN ETHNOLOGY [bull, lei 

Decoration: A complex mixture of wavy lines and meanders 
separated by almost parallel straight lines and enclosed in a 
rectangular zone formed by scraped lines. 

3. Deep bowl with constricted orifice and pronounced angular waist 
(fig. 42-3). 

Rim: Exteriorly thickened, squared lip, uneven and wavy. 

Body wall thickness: Uneven, 5-7 mm. 

Body dimensions: Maximum body diameter 28 cm.; height 17 cm. 

Base: Flattened, juncture with side walls rounded, interior thickened, 
diameter 8 cm. 

Appendages: Three pairs of small, round, "button" appliques (1.2- 
1.5 cm. in diameter) on the exterior rim band; pairs equally 
spaced around the rim. 

Decoration: Squared meander on the rim between the paired "but- 
tons"; vertical and slanting lines haphazardly arranged and over- 
lapping to cover the area from the waist to the rim. 

4. Carinated bowl with outcurved side walls and direct rim (fig. 42-4; 

pi. 22, a). 
Rim: Direct, slightly everted, joining the outcurved body walls 

with a prominent angle; lip squared. Mouth diameter 38 cm. 
Body wall thickness: 5-6 mm. 

Body dimensions: Diameter at carination 34 cm.; height 16-18 cm. 
Base: Flattened. 
Decoration: Body wall covered with lines diagonally crossed to 

form diamond and triangular-shaped areas which are filled with 

squared spirals or parallel lines. Exterior of rim covered with a 

series of intertwined meanders. 

5. Large, open, slightly carinated bowl with vertical side walls and 

thickened, everted lim (fig. 42-5; pi. 22, e). 
Rim: Externally thickened, strongly everted with a rounded lip, 

mouth diameter 40 cm. 
Body wall thickness: 10 mm. 

Body dimensions: Maximum body diameter 37 cm. 
Base: No fragments found; probably flattened. 
Decoration: Diagonally scraped lines on the 4 cm. wide vertical 
wall. Lower edge of lines bordered by an incised line. 
Decoration (pi. 22): 

Technique: Surfaces scraped with a blunt, flat tool. Scraped lines range 
from 3-5 mm. in width and up to 1 mm. in depth. Most of the scrapings 
are flat troughed, but in some cases the tool was held at an angle cutting 
the line deeper at one side than the other. Designs are made with sepa- 
rate, individual stokes, not with a comb. 
Motif: A few of the designs are carefully executed with each line distinct 
from the other in a series of roughly parallel scrapings on the vessel exterior. 
Most, however, are haphazardly done with the lines overlapping each 
other in a series of diagonal, horizontal, and vertical lines without any 
particular design motif except to cover the entire surface with scrapings. 
Motifs range from a double spiral or "S" separated from the next unit 
by a series of parallel, vertical or slightly diagonal lines to a series of 
triangles arranged vertically around the shoulder of carinated bowls with 
the spaces filled with individual scrapings. A few cambered, collared jars 
have a series of rectilinear meanders or vertical scrapings with "S" spirals 
on the waists of the jars. 



Ev^Nsf^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 143 

Temporal difference within the type: Sample too limited to note any 

changes. (Appendix, table 15). 
Chronological position of the type: Restricted to tlie lower (early) part of 
the Arist6 Phase sequence. 

sebra plain 
Paste: 

Method of manufacture: Coiling. 

Temper: Medium-coarse, ground sherd intermixed in the paste in moderate 
amounts; a few sherds have some rounded, quartz particles suggesting 
accidental mixture. 
Texture: Coarse, granular, poor mixture of clay and temper makes a very 
irregular fracture plane. Sherds are easy to break but not soft or friable 
and all have a dull thud. 
Color: The majority have a fine, thin gray core banded on either side with 

a light tan to orange; few have a full core of light orange to tan color. 
Firing: Oxidized, generally incomplete; very few fire clouds. 
Surfaces: 

Color: Interior and exterior of majority are a light, orange tan to a light, 
grayish brown; a few have a grayish tan to orange-tan exterior and a 
grayish black to gray-brown interior. 
Treatment: Most well smoothed on exterior and interior with a very even 
surface and only a few smoothing tracks visible. Fine pores from water 
bubbles are quite prominent on the surfaces, suggesting smoothing when 
the clay was extremely wet. About one-quarter of the sherds have a 
well-smoothed exterior with a less regular and uneven interior. A few 
sherds have both surfaces smoothed to a high polish. 
Hardness: 2.5-3. 
Forms: 

Common vessel shapes: 

1. Round, high- to low-waisted jar with short to medium neck, externally 

thickened, everted rim (fig. 43-1; pi. 23, b, c, e, /). 

Rim: Ranges from a very pronounced to a moderately everted rim, 
thickened with a coil added to the exterior surface. The short to 
medium length neck ranges from an insloping neck with a pro- 
nounced, everted rim to an almost vertical neck with a slightly 
everted rim. Lip is most commonly rounded but sometimes 
squared with rounded corners. Mouth diameters 18-38 cm. 

Body wall thickness: 8-14 mm., most 10-11 mm. 

Body dimensions: Body diameters 28-50 cm.., most 36-40 cm.; re- 
constructed height 36-50 cm. 

Base: Flattened, slightly thickened (1.2-2.0 cm.) sometimes with a 
slight pedestal 5-8 mm. high. Diameter 10-18 cm., majority 
16 cm. 

Appendages: Rarely nubbins or small riblike appliques are on the 
upper part of the jar body. 

2. Large, low-waisted jar with long vertical or slightly incurving neck 

and an unthickened rim (fig. 43-2; pi. 23, a). 

Rim: Gently incurved to a vertical, direct rim with a rounded lip. 

Mouth diameters 36-52 cm.; majority 44-50 cm. 
Body wall thickness: 7-9 mm. 
Body dimensions: Diameters 36-60 cm.; reconstructed height 30-50 

cm. 
Base: Same as form 1, 



144 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 





4 e 12 CM 



I 2 3 CM 
Rim ScoK 





Figure 43. — Rim profiles and vessel shapes of Serra Plain, Arist6 Phase (Ap- 
pendix, table 16). 

3. Slightly carinated open bowl with everted lip (fig. 43-3) . 

Ri7ns: Everted, usually exteriorly thickened but sometimes direct 

with a rounded, tapered or squared lip with rounded edges. 

Mouth diameter 18-32 cm.; most 22-26 cm. 
Body wall thickness: 7-10 mm.; most 7-8 mm. 
Body dimensions: Diameters 14-28 cm.; depth 6-16 cm. 
Base: Same as form 1. 
Appendages: Occasionally an adorno or lobe on the lip of the 

Serra Plain sherds from the Ilhas do Campo Site, 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 145 

KVANSJ 

4. Small to'inedium open bo\vl with outcurving sides (fig. 43-4). 

Rim: Outcurving and upcurving with rounded lip; mouth diameters 

16-28 cm. 
Body wall thickness: 5-10 mm. 
Body dimensions: Depth 10-18 cm. 
Base: Same as form 1, but rarely a rounded base occurs. 

5. Bowl with incurved sides and constricted mouth (fig. 43-5). 

Rim: Incurved with lip which is usually rounded but sometimes 

squared with rounded edges. Mouth diameters 24-28 cm. 
Body wall thickness: 8 mm. 

Body dimensions: Diameters 26-30 cm.; depth 12-18 cm. 
Base: Flattened, 8-13 cm. in diameter. 
Less common vessel shapes: 

1. "Graters" — Open, carinated bowl with outcurving rim with parallel 
grooves on the interior. 
Rim: Outcurved, unthickened, carinated, with a rounded lip. 

Mouth diameter, 38-40 cm. 
Body wall thickness: 12-14 mm. 
Base: No complete vessel found but fragments suggest probably 

flattened. 
Groovings on the bowl interior: Parallel grooves 2-4 mm. wide, 3 
mm. deep and 4-7 mm. apart, incised on the upper interior of the 
bowls but not upon the short, vertical or slightly outcurving side- 
wall (pi. 23, d). "V" with a rounded base in cross section. Since 
no base sherds of Serra Plain were found with groovings on them, 
it can probably be assumed that the groovings are limited to the 
lower bowl interiors and do not extend to the base. 
Temporal difference within the type: Jars with short to medium neck, thick- 
ened, everted rim (form 1) decline in popularity while bowl forms 3 and 4 
increase. Bowls with constricted mouths (form 5) are limited to the tniddle to 
early part of the sequence (Appendix, table 16). 
Chronological position of the type: Increases in popularity throughout the 
Arist6 Phase sequence. 

SERRA painted 

Paste: 

Same as Serra Plain; see that type description for details. 
Surfaces: 

Unpointed surfaces: Same as Serra Plain. 

Slipped surfaces: In about }i of the painted sherds, the painting is applied on 
a thickly white-slipped surface. This occurs on the exterior of jars and 
bowls and on the interior of a few carinated bowls. 
Form: 

Common vessel shapes: 

1. Round or low to high-waisted jar with a short to medium length 
neck, externally thickened lip and constricted mouth (fig. 44-1; 
pi. 24, a-b) 
Rim: Exteriorly thickened with a coil, usually everted, short to 
medium neck either slightly insloping or vertical; lip rounded or 
squared with round edges. Mouth diameter 22-24 cm. 
Body wall thickness: 8-10 mm. 

Body dimensions: Maximum body diameter 28-50 cm.; estimated 
height 30-50 cm. 



146 



BUREAU OF AMERICAN ETHNOLOGY 



[bull. 167 




I 2 3 CM 
Rim Scole 






Figure 44. — Rim profiles and vessel shapes of Serra Painted, Arist6 Phase 

(Appendix, table 17). 



Base: Flattened, slightly thickened; diameter 10-18 cm.; sometimes 
a slight pedestal (5-8 mm. high). Some jars have holes, usually 
5, punched in the base while the clay was wet, tapering from 7-8 
mm. in diameter on the interior to 3-4 mm. on the exterior. 

Decoration: Red slip or paint on the lip and exterior of the thickened 
rim, and/or a band on the neck, but rarely extending onto the 
body walls. 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 147 

EVANS] 

2. Tall jar with vertical to outcurving cambered rim and carinated or 

double-carinated body (fig. 44-2; pi. 24, c-d; pi. 25, b). 

Rim: Outcurved, cambered collar developed out of a short to long 
neck. Either direct or externally thickened with a coil. Lip 
rounded or squared with rounded edges. Mouth diameters 20-36 
cm. ; majority 28-32 cm. Occasionally, especially on the Cunani 
materials, an anthropomorphic face is modeled on the rim and 
neck with low appliques to form the eyes, ears, nose, mouth and 
eyebrows (see pi. 24, c, and Goeldi, 1900). 

Body wall thickness: 8-13 mm.; majority 11 mm. 

Body dimensions: Diameter 28-40 cm.; total height 30-50 cm.; 
neck height 10-20 cm. 

Base.- Flattened, thickened on interior, diameter 10-16 cm. 

Appendages: Applique modeling of anthropomorphic faces on 
neck and rim, and arms, breast and navel on the jar body; some- 
times an adorno of an anthropomorphic face or an animal is ap- 
plied to the shoulder of the body. 

Decoration: Predominantly a red paint on white slip, or less com- 
monly on the natural, tan surface, in combinations of inter- 
locking curvilinear spirals, meanders, frets and steps. The 
curvilinear interlocking spirals are the most common motif. 
Rarely the polychrome technique is found on this form. 

3. Carinated bowl (fig. 44-3; pi. 24, e). 

Rim: Carination ranges from a vertical to a strongly outcurving 
side wall with an exteriorly thickened, everted, direct or tapered 
rim and a lip which is rounded or squared with rounded edges. 
Mouth diameters 22-44 cm.; majority 32-36 cm. 

Body wall thickness: 6-13 mm.; majority 8-10 mm. 

Body dimensions: Body diameters 26-34 cm.; height 10-24 cm. 

Base: Flattened, internally thickened; diameters 8-14 cm. Some 
have holes in the base; see form 1, "base" for details. 

Decoration: Most commonly, red curvilinear interlocking spirals 
on natural tan-orange surfaces or on white-slipped surface on the 
exterior of the vertical or outcurving side wall. Sometimes simpler 
curved lines or more complicated frets, meanders, and simple 
spirals are on same area. Rarely the designs are in true poly- 
chrome of red and black, sometimes yellow, on a white slip or 
natural surface. 

4. Open bowl with outcurving to nearly vertical side walls (fig. 44-4). 

Rim: Direct, outcurving to nearly vertical rim with rounded lip. 
Mouth diameter 20-26 cm. 

Body wall thickness: 7-12 mm.; majority 8 mm. 

Body dimensions: Height 10-14 cm. 

Base: Flattened, slightly thickened, 8-12 cm. in diameter. 

Decoration: A combination of curvilinear spirals, frets, mterlocking 
meanders, etc. on the interior surface. Most commonly in poly- 
chrome of red, black and yellow on white slipped surface but 
sometimes only red on a white slip or red on the natural colored 
surface. 

5. Bowl with incurved sides and constricted mouth (fig. 44^5). 

Rim: Direct, incurved, with the lip either rounded or square with 

rounded edges. Mouth diameter 24^28 cm. 
Body wall thickness: 8 mm. 



148 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

Body dimensions: Diameters 26-30 cm.; height 12-18 cm. 
Base: Flattened, 8-12 cm. in diameter. 

Decoration: Band on exterior 6-10 cm. wide with the typical com- 
bination of curvilinear spirals, frets, etc. usually in red on white 
slip or red on the natural buff -tan surface. Occasionally the 
designs are in polychrome. 
Decoration (pi. 24, 25): 
Technique: 

1. Monochrome: Red paint or a thick red slip in bands around the 

collar, rim or upper part of the body; polishing striations visible 
on the red-slipped surface. Red paint in fine (1-2 mm.) to broad 
(10-20 mm.) lines on the natural tan to buff surfaces. 

2. Bichrome: Red designs of fine to broad lines on white-slipped sur- 

faces. 

3. Polychrome: Red, black and/or yellow designs on natural buff sur- 

faces or on a white slip. 
Regardless of the color technique only about one-half of the designs 
are well executed, with the others appearing as if they were done in 
great haste with the lines and units very irregular and uneven. 
Motif: Simple bands or complicated designs of a mixture of curvilinear 
and rectilinear interlocking units, spirals, lines, rectangles, meanders, 
undulating lines and waves (fig. 33). Sometimes these are combined 
with anthropomorphic motifs, i. e. faces, arms, breasts, and navel shown 
by applique with the painted design curving around and accentuating these 
features (pi. 24, c). 
Temporal differences within the type: Tall jars with vertical to outcurving 
rims, cambered and carinated or double carinated, low-waisted bodies (form 2) 
appear only in the late part of the sequence (Appendix, table 17). 
Chronological position of the type: Increase in popularity throughout the 
Ariste Phase sequence; especially common in the cemeteries. 

vxgk incised 

Paste and surfaces: On Arista Plain, see that type description for details of 

paste, color, temper and surfaces. 
Form: 

Common vessel shapes: 

1. Bowl with incurved rim, slightly constricted mouth (fig. 45-1). 

Rim: Incurved, either slightly thickened on interior or tapered, 
lip rounded. Mouth diameters, 26-40 cm.; majority 36 cm. 

Body wall thickness: 7-10 mm.; majority 8 mm. 

Body dimensions: Maximum diameters 28-46 cm; reconstructed 
height 12-20 cm. 

Base: Flattened, slightly thickened on the interior and slightly 
concave on the exterior; diameters 8-14 cm. 

Decoration: Usually a deep (1-2 mm.) groove around the rim and 
1-3 cm. below the lip. This marks the upper margin of a band of 
decoration limited to the upper shoulder of the bowl which 
consists of diamonds, triangles, squares and parallel lines. These 
motifs are usually in lighter, finer lines than the grooved incisions 
which form the border. 

2. Large, slightly carinated, open, basinlike bowl (fig. 45-2). 

Rim: Slightly carinated, vertical to slightly outslanted side walls, 
exteriorly thickened rim with a rounded lip; mouth diameters 
36-38 cm. 



Ev^Nsf ^ ^ND ARCHEOLOGY AT MOUTH OF AMAZON 



149 




I I I I I I J 

4 8 12 CM 
Vessel Scale 






l...i..,l 1 I 

1 2 3 CM 

Rim Scale 



Figure 45. — Rim profiles and vessel shapes of UaQd Incised, Arist6 Phase 

(Appendix, table 18). 



391329—57 12 



150 BUREAU OF AMERICAN ETHNOLOGY [bull, ibt 

Body wall thickness: 10-13 mm. 
Body dimensions: Diameter 34-36 cm., vertical side wall height 

6-8 cm., total height 12-14 cm. 
Base: Flat, same as form 1. 

Decoration: Band of incisions (3-6 cm. wide band) around the 
exterior of vertical to slightly outslanted side walls, just below 
the rim. Band bordered by deep (1-2 mm.) and wide grooved 
incisions. Motif in the form of triangles and diamonds or curvi- 
linear spiral and waves in this band. 
3. Short-necked jar with cambered collar and everted rim (fig. 45-3; 
pl. 26,/-^). 

Rim: Vertical, cambered collar, slightly thickened, 4 cm. wdde; 
unthickened, everted rim with a squared lip with rounded edges. 
Mouth diameter 24 cm. 
Body dimensions: Neck and collar height 7 cm.; globular body 

reconstructed to about 32 cm. in diameter. 
Base: No fragments found; probably flattened. 
Decoration: Series of almost parallel horizontal lines on cambered 
collar and neck, 1 mm. wide and 1 mm. deep averaging 8 mm. 
apart, bordering a curvilinear pattern. 
Decoration (pi. 26, d-g): 

Technique: The incisions, varying from 0.5-1.5 mm. in depth, are all U-shaped 
and made with a blunt tool, ranging from 1-3 mm. in width. All are 
carefully executed with the incised lines and grooves very regular. 
Motif: Most typically a series of short, parallel lines, intertwining wavy 
lines, short serpentine undulations, adjoining diamonds and triangles. 
There is only one example of curvilinear spirals. The designs are limited 
to the shoulders of bowls and neck, collar and shoulders of jars. Occasion- 
ally incisions occur on the exterior face of lips. Most of the light incisions 
are bordered by a deeper, broader incised groove. 

Temporal differences within the type: None observable due to the small 
sample (Appendix, table 18). 

Chronological position of the type: Limited to the early (lower) part of the 
Arist6 Phase sequence. 

unclassified decorated 

Brushed sherds: On a few body sherds, the exterior surface is lightly brushed in 
vertical to diagonal lines. Some overlap of the lines suggests application with 
a bunch of pointed sticks, which is further attested by the irregularity of the 
spacing between the lines. 

Pottery Artifacts 

The only pottery artifacts from the Ariste Phase are small figurines 
(fig. 30; pi. 26, a-c). 

Nonceramic Artifacts 

The number of nonpottery artifacts from the Arist6 Phase is so 
limited that characteristic types cannot be defined, A tabulation of 
the total materials from all the sites of this Phase gives: 2 well- 
polished celts; 1 small, notched, -well-polished ax; 1 small stone chisel; 
6 roughly dressed hammerstone fragments, a few of which may have 



fjf^^^^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 151 

been originally the poll or butt end of hand axes; 1 grinding stone 
fragment also used as a hammerstone ; 7 nephrite objects, of which 5 
are cylindrical beads and 2 stylized pendants; 27 burnt clay lumps; 
25 natiu"al rock fragments, of which only 5 are fire burnt; and 10 drilled 
shell pendants. In the Ariste Phase the role of stone artifacts was 
obviously secondary to that of pottery and probably also to perishable 
objects made of wood and plant fibers, which have all disappeared 
because of climatic conditions. 

Glass trade beads were described in detail with the site discussions, 
hence no repetition is required here. These were found at Sites A-10 
and A-15 (pi. 25). 

CERAMIC HISTORY 

In spite of the wide region, extending from the Rio Oiapoque south- 
ward to the Rio Araguari-Amapari, over which the principal ceramic 
types of the Arista Phase are found, the pottery exhibits only minor 
local differences. This consistency suggests a stable and well- 
integrated culture, undisturbed after the initial period of adjustment. 
The shallowness of the habitation sites, never m.ore than 10 cm. 
deep, made a stratigraphic approach impossible and required that 
other factors be made the basis for seriation. The presence of 
European glass trade beads in several of the cemeteries provides a 
rough terminal point. The absence of several decorative styles (e. g. 
Flexal Scraped, Davl Incised, and Uagd Incised) at the sites with the 
trade material suggests that sites producing these pottery types are 
earlier than the others and belong at the beginning of the time sequence 
for the Phase. A further clue comes from the percentage analysis of 
the ceramic types from Site A-10 — Montanha da Pluma. Since the 
vessels stacked around the two entrances to the cave may be taken to 
represent the overflow after the interior was filled, the percentage 
difference between the inside and outside should be representative of 
the trend in pottery change. This interpretation is affirmed by the 
fact that glass trade beads were found with vessels at the entrance to 
the cave and none came from inside. As in the Mazagao Phase the 
cemetery wares show selection for the decorated types, making it 
impossible to interdigitate them in the sequence of occupation sites. 
The latter were seriated separately on the basis of trends revealed in 
the cemeteries, from which they mainly differ only in the relative 
amount of decorated ware (fig. 46). 

The major trend in the plainware is similar to that in the Mazagao 
Phase. Gritty, quartz and sand tempered Ariste Plain begins as the 
dominant ware and gives way steadily, with a few minor fluctuations, 
to the increasingly popular, smooth, sherd-tempered Serra Plain. 
Ariste Plain decreases from 74.4 percent to 20.8 percent in the ceme- 



152 BUREAU OF AMERICAN ETHNOLOGY [boll, m 

teries and from 98.7 percent to 0.3 percent in the habitation sites, 
with the concomitant increase of Serra Plain from 9.6 percent to 72.5 
percent in the cemeteries and from percent to 93.8 percent in the 
habitation sites (Appendix, table 11). The relative position of the 
sites in the Ariste Phase sequence is shown on the accompanying 
chart (fig. 46). 

As was the case in the Mazagao Phase, the Ariste Phase can be di- 
vided into an early and late period on the basis of changes in the 
decorated ware. Sherds decorated in the incised or scraped tradition 
(Flexal Scraped, Uaga Incised, Davi Incised) are diagnostic of the 
earlier sites in the sequence. They are associated with painted styles 
in the cemeteries, but, except for the 1 percent occurrence of Serra 
Painted at Site A-16, these plastic traditions are the exclusive deco- 
rative style at the four earliest habitation sites. 

Flexal Scraped, the most popular of the three nonpainted decorated 
pottery types, may be likened to Jari Scraped of the Mazagao Phase. 
The technique of execution with a flat, blunt tool is identical in both 
Phases but the motif is slightly more elaborated in the Ariste Phase 
material. Whereas Jari Scraped of the Mazagao Phase was usually 
a series of parallel lines or rectilinear units, sometimes with a limited 
number of curvilinear lines, Flexal Scraped of the Ariste Phase has 
at least 25 percent of the examples executed in a carefully laid out 
design of double spirals and S motifs in units separated by parallel 
or vertical lines, triangles, and rectilinear meanders (fig. 32; pi. 22). 
The rest of the examples of the type, however, consist of separate, 
individual strokes or scrapings applied rather haphazardly over the 
whole vessel or sherd surface. 

Distinctive Uaga Incised ( pi. 26, d — g) seems to be limited to the 
extreme northern part of the region occupied by the Ariste Phase. 
The combination of decorative elements common to this type — short, 
parallel lines, intertwining wavy lines, short serpentine undulations, 
adjoining diamonds and triangles, and an occasional spiral — have no 
counterpart outside the Ariste Phase. Although Uaga Incised has a 
shorthistory,hmi ted to the early part of the sequence, the decorative 
motifs appear to have been transferred in part to the painted pottery, 
with many of the combinations of decorative elements showing up 
in the elaborate designs of Serra Painted. 

Davl Incised, the crudest of the three decorated pottery types 
utilizing plastic traditions, was rarely used as a burial ware, but shows 
great persistence in the habitation sites. It is found in small percent- 
ages at all but the latest occupation sites, A-20, A-7, and A-22 (where 
the unusually small sherd sample might account for the absence). 
The simple decorative motifs of crudely appUed, irregular hues ar- 
ranged diagonally or vertically on the rim, neck or upper body wall 



(O 
UJ 

q: 
ui 

»- 

UJ 
UJ 

o 



0) 

z 

O 

< 
I- 

03 
< 



ICAL 
PED 

table 11). 



DAVI 
INCISED 



UAQA 
INCISED 



UNCLASSi 
FIED 



UJ 

a. 

Ill 

UJ 

LU 
O 



A -15 
A- 19 
A - 14 

A- 10= OUTSIDE 
A-IO INSIDE 
A-ll CAVE 2 
A - 11 C AV E 3 
A-ll ■■ CAVE I 



□ 



□ 



]Z3 




D 



D 

[=1 



CZI 





A 


-20 






A - 7 




in 


A -22 




z 






o 


A-9 




1- 






< 


A-21 




1- 






_ m 


A-12 




- < 






i 

3 


A- 16 
A-18 
A- 13 








1 1 
20 


1 

40% 



1 

ARISTE 
PLAIN 


1 

SERRA 
PLAIN 


1 

ARISTE 
PAINTED 


1 

SERRA 
PAINTED 


1 

FLEXAL 
SCRAPED 


1 
1 

DAvT 
INCISED 


1 

UA^A 
INCISED 


1 

UNCLASSI 
FIED 



Figure 46. — Seriatjon of Arist6 Phase sites on the basis of pottery type frequency (Appendix, table 11). 



"TnsT^ ^^^ ARCHEOLOGY AT MOTJTH OF AMAZON 153 

surfaces (pi. 21, c~f) do not appear to be directly related to any of 
the other decorated potteiy types of the Arista Phase. The trace 
occurrence of the type throughout the entire Phase offers little in 
the way of diagnostic features that might be useful as time-markers. 

The late part of the Ariste Phase is characterized by the flowering 
of painting as the mode of decoration. This is most clearly demon- 
strated in the habitation sites, where the percentage of Serra Painted 
increases from 1 to 6 percent as the incised and scraped techniques 
die out. In the cemeteries belonging to the late part of the Phase, 
decoration in the plastic medium is absent and Ariste Painted, which 
had dominated the early period, is succeeded by Serra Painted as the 
major decorated ware. Since no conscious preference seems to have 
been shown in the plain ware types chosen for decoration, the increase 
in frequency of Serra Painted from percent at Cave 1 and Cave 3 
of Site A-11 to 82.6 percent in Goeldi's Cunani material (the 100 
percent occurrence shown at Site A-15 on the chart, fig. 46, is based 
on a single vessel) is apparently a reflection of the growing predomi- 
nance of sherd-tempered pottery in its undecorated form. Ariste 
Painted shows a less clearcut, but stiU definite decrease in the ceme- 
tery sites from a high of 42.8 percent at the bottom of the sequence 
to a low of 20.8 percent at the top of the sequence. Of the painted 
types only Serra Painted occurs at the habitation sites. Ornamen- 
tation is executed in red or black paint alone, or in combination, 
either directly upon the natural surface or upon a white slip. It 
is applied in bands on the neck or body of the vessel or in complex 
patterns including spirals, dots, waves, meanders, curvilinear mean- 
ders, squares, and paired and tripled parallel lines in curvUiuear and 
rectilinear motifs (fig. 40, a, b; pis. 24, 25). 

The study of these motifs according to the seriated position of the 
sites on the time-sequence chart reveals a shift from simple, painted 
bands to highly complex and intricate designs. Of the 28 Ariste 
Painted vessels represented at Site A-11, Caves 1,2, and 3 (the earliest 
site in the sequence) , only 3 have a complicated design ; the other 25 
have a plain, single-colored (usually red) band around the neck or rim. 
Of the 3 Serra Painted vessels from the same site, 1 has a complicated 
design and 2 have plain red bands. At site A-10, of the 14 Ariste 
Painted vessels, 13 have plain red bands and only 1 the complicated 
design; of the 6 Serra Painted vessels 3 are of the plain-red-band 
type and 3 of a complicated motif. At Site A-11 all but one of the 
complexly painted vessels were found outside the cave and are of 
more recent deposition than the material inside the cave. This 
transition from simple band motifs to complicated designs reaches a 
culmination in Site A-19, which in decorated pottery types includes 
only elaborately executed Serra Painted. 



154 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

In addition to the seriation study of the popularity change of various 
pottery types throughout time, a detailed analysis was made of the 
rim and vessel shapes. Several distinctive shapes occur in the two 
plain wares of this Phase. The collared, cambered, round-bodied, 
or low-waisted jar, form 1 of the Ariste Plain (fig. 38-1) does not 
occur in Serra Plain; while Serra Plain, form 2, a large jar with direct 
rim and long neck (fig. 43-2) does not appear in Ariste Plain. Al- 
though minor popularity trends occur in the various shapes for 
each plain ware, the time span is too short and the rim sample 
too small to produce any significant trends from a quantitative analy- 
sis ; for this reason the data was left in tabular form (Appendix, tables 
12 and 16) and not converted into a graphic presentation. 

A study of the rim and vessel shapes of the painted pottery types 
offers similar data — within the type the shapes are helpful in a desig- 
nation of its ceramic features, but the rim sample is not large enough 
to extract meaningful percentage trends in the various shapes al- 
though some slight trends are suggested (Appendix, tables 13 and 17). 
Again it is significant to note that Ariste Painted, form 1 (fig. 39-1), 
which comprises from 16.7 to 37.5 percent of the vessels of this pottery 
type, does not occur in Serra Painted. Likewise, Serra Painted, form 
2 (fig. 44-2), one of the most popular shapes of this type, does not 
appear in Ariste Painted. In other words, except for the more stand- 
ard common vessel shapes, such as open bowls, bowls with constricted 
mouths, and round-bodied jars with short neck and thickened rims, 
the major shapes of one painted or plain pottery type do not occur on 
another painted or plain pottery type. If the related plain and 
painted potter}^ types are analyzed together, other distinctions 
emerge. Specifically, since Ariste Painted appears to be a decorated 
type placed on Ai'iste Plain paste, and Serra Pamted one on Serra 
Plain paste, it might be expected that the same vessel shapes would 
be present. However, a comparison of Ai-iste Plain forms with 
Ariste Painted forms reveals that two shapes occurring in Ariste 
Painted never occur in Ariste Plain. This suggests that decoration 
of certain pottery forms with painting did not involve merely the de- 
cision to paint some of the standard Ai'iste Plain shapes, but rather, 
two distinct vessel forms — Ariste Painted, form 3, shghtly carinated 
bowl with everted lip, and Ariste Painted, form 5, carinated bowl 
with flangelike rim — were consciously manufactured for painted 
pottery only. Since Ai-iste Painted pottery is almost exclusively 
limited to the cemeteries and these two shapes are not found in the 
related plain ware, there is no question that there was a conscious 
manufacture and exclusive use of certain vessel shapes for burial pur- 
poses in the early part of the Ariste Phase. 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 155 

EVANSJ 

The rim and vessel shape study of Serra Plain and Serra Painted 
offers similar correlations to those just described for Ariste Plain and 
Ariste Painted. Serra Painted is applied to Serra Plain paste with 
both types sharing certain common vessel forms and shapes. How- 
ever, the most common vessel shape (74 percent at one cemetery site) 
of Serra Painted (form 2, tall jars with vertical to outcurving cam- 
bered rims and carinated or double-carinated low-waisted bodies) is 
exclusive to Serra Painted and is not found in any other pottery types 
of the Ariste Phase. Although a few fragments of Serra Painted 
vessels are found in the habitation sites, the pottery type was ap- 
parently manufactured primarily for burial purposes. 

For the three pottery types utilizing plastic techniques of decora- 
tion, Davi Incised, Flexal Scraped, and Uaga Incised, the rim and 
vessel shapes are represented by such small samples that no shape 
trend is observable within each type, although certain forms are re- 
stricted to each of the pottery types (figs. 41, 42, 45). 

In an attempt to see if certain common rim and vessel shapes 
showed distinct trends throughout the Ai-iste Phase, similar forms 
in the various pottery types were combined into nine common shapes, 
each of which was given an alphabetical designation for form and a 
short descriptive name. Into this scheme (Appendix, tables 19 and 
20) the forms of each pottery type were tabulated and the percentage 
occurrence calculated in order to establish the common rim and vessel 
shapes for the Ariste Plain. 

Two observations are outstanding: (1) Certain shapes are restricted 
to the cemetery sites, and (2) there are distinct trends of certain of 
the common shapes throughout the Ariste Phase sequence. Although 
the time span for the Ariste Phase is apparently not very long, and 
hence the amount of change in some of the forms is not outstanding, 
certain gross trends in the rim and vessel shapes are apparent: (1) 
Common form A, a collared, cambered jar with a round or low- 
waisted body, decreases from a high of 46.1 percent at the lower part 
of the cemetery sequence to 11.7 percent at the top; (2) common 
form E, a bowl with a constricted mouth, decreases from 20.0 percent 
to 11.2 percent in the habitation site sequence and from 20.0 percent 
to 11.7 percent in the cemetery site sequence; (3) common form F, 
carinated bowl with flangelike rim, decreases from 15.3 percent and 
15.6 percent to 5.2 percent and 5.9 percent in the upper part of the 
cemetery site sequence; (4) common form D, a slightly carinated bowl 
with an everted lip, increases from between 7.8 percent and 3.1 per- 
cent to 17.3 percent in the cemetery sequence. In the habitation 
sites the trend of common form D is obscured by certain small samples 
which warp the percentages, however, if these are ignored the larger 



156 BUREAU OF AMERICAN ETHNOLOGY [bull, is? 

samples show an increase in popularity from 21.2 percent to 43.8 
percent throughout the habitation site sequence of the Ariste Phase. 

With the sequence established from the larger samples from our 
own sites and the first-hand study of documented museum collections, 
it is possible to interpret some of the sketchy data resulting from other 
investigations (pp. 123-131). Although much is to be desired from 
the comparative materials, at least there appears to be no gi'oss 
conflict of any sort with the ceramic trends established for the Arista 
Phase. Since the sites with a high percentage of incised pottery 
(i. e. Uaga and Davi Incised) are in the earliest part of the sequence, 
Ilha do Carao and Agahyzal probably seriate in this position. The 
absence of trade materials from both of these sites adds confirmation 
to this conclusion. Those sites with a large number of graters (like 
our Sites A-12 and A-16) and a high percentage of the elaborately 
painted, bichrome and polychrome Serra Painted (as typified by 
A-19 — Cunani Site), belong to the middle and upper part of the 
Arista Phase time sequence. As a result, sites with pottery resembling 
the Cunani pottery shapes, surface painting, anthropomorphic and 
zoomorphic modeling, and peculiar holes punched in the vessel bases 
must be considered as approximately the same time period, which is 
just before European contact. Such sites are the Rio Aracaua sites 
of Kaupl and Ulakte-Uni, the Igarape Tartarugalzinho site, the 
Rio Oiapoque site, and possibly Com'baril and Coumarouman on the 
Rio UaQd. The presence of so much European trade material with 
the Monte Maye finds of Serra Painted seriates this site with the post- 
European site of A-15— Vila Velha. 

As the discussion in the preceding section on nonceramic artifacts 
indicated, stone artifacts are too rare and undifferentiated to reveal 
any trends or changes in style to supplement the time sequence 
based on pottery. 

DIAGNOSTIC FEATURES OF THE ARISTE PHASE 

The Ariste Phase is distributed thi-oughout the northern half of the 
Territory of Amapa, bounded b}^ the Rio Oiapoque on the north and 
the Rio Araguarl-Amapari on the south. Variations of nonceramic 
artifact material through the entire region are minor. A seriation of 
the ceramics indicates a shift from the popularity of incised (Uaga 
Incised) and scraped (Flexal Scraped) decorations and very gritty, 
sand-tempered plain ware (Ariste Plain) to a preference for painting 
(Serra Painted) and a smooth, sherd-tempered plain ware (Serra 
Plain). Painted wares occm* in all periods in the cemeteries, which 



EVANsT^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 157 

consistently show a much higher frequency of decorated wares than 
do the village sites. 

Burial urns are characteristically deposited in high rock shelters or 
caves, but when such natural recesses are not available, either direct 
interment of the jars or specially constructed subterranean shafts 
with alcoves take their place. No plan could be ascertained in the 
arrangement of the burial urns in the caves. The burial urns, often 
modeled or painted in anthropomorphic figures or faces, are generally 
not provided with lids. They contain a mixture of dirt and bones of 
cither secondary or cremated bmial, cremation being the more 
common and also the more recent practice. In rare instances a small 
stone ax, a small figurine, nephrite pendants or glass trade beads 
were placed with the bones of the deceased, but miniature jars or 
bowls were never among the offerings. 

Occupation sites, averaging 100 meters in diameter, are not found in 
close proximity to the burial sites, and in all cases consisted of extreme- 
ly shallow refuse, usually no more than 5 cm. in depth. They are 
typically located on natural high land free from flooding and with 
good drainage, near a constant water supply provided by either a 
lake or an igarape. No architectm-al features could be ascertained, 
but the fact that the refuse deposits consist of heavy concentrations 
of sherds upon sherds with little intermixture of dirt and no evidence 
of a dirt floor suggests the use of houses on piles with raised floors 
similar to the structures used in the region today. The ceramics of 
these habitation sites are generally nondescript with the exception of 
sherds with deep parallel grooves on the interior that may have come 
from graters. 

The presence of a few Ariste Plain sherds at the stone alinements 
of Aurora (A-8) and Agahyzal is indicative of the occasional use by 
the Ariste Phase of former Arua Phase sites. 

Besides pottery, only stone artifacts have been preserved. A few 
ungrooved axes and hammerstones, all well-polished gi'anite or 
diorite with a slightly curved, bifaced bit, indicate a well-developed 
stone-polishing technique or trade with a group practicing such an 
art. 

No specific information is available on the length of occupation in 
the region, and the application of absolute-dating techniques does not, 
at the moment, seem feasible. We could be dealing with a small 
population over a long period of time, or a large population over a 
short period of time. However, all factors considered, it appears 
that the culture of the Ariste Phase was not present in the region for 
any great length of time before the arrival of the first Europeans in 
the early years of the 16th century. 



158 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

CONCLUSIONS AND INTERPRETATIONS 

The historical position of each Phase, its role in the cultural develop- 
ment of the Territory of Amapa and its relationships to the total 
picture of the prehistory of the Lower Amazon and northern South 
America can now be outlined. 

No preceramic sites have been found in the Territory of Amapa, 
although shell middens are reported from the Middle Amazon and 
from British Guiana (Gillin, 1948, p. 821; Osgood, 1946, pp. 23-37; 
Evans and Meggers, MS.). The possibility of discovering pre- 
ceramic hunting, fishing, nonshellfish-gathering cultures is virtually 
nonexistent owing to the limited use of stone artifacts and the perish- 
able nature of other materials employed instead. 

The Arua Phase, the first pottery-producing culture, entered the 
Territory of Amapa with an established ceramic tradition. Although 
the early form of Piratuba Plain is often crude, it by no means suggests 
a group just learning the ceramic art or having just received the idea 
by diffusion. By tracing similarities in pottery decoration, stone 
artifacts and stone alinements, it is possible to make a good case for 
a northern derivation of this culture (see pp. 548 ff. for details). This 
makes the Arua the only archeological Phase to have come to the 
mouth of the Amazon from this direction. Their occupation of the 
Territory of Amap§, appears to have been brief, and shortly prior to 
European contact they abandoned the mainland for the Islands of 
Mexiana, Caviana, and Maraj6, where they survived into the pages 
of recorded history. The absence of a time gap between the departure 
of the Arua and the arrival of the next cultures suggests that the former 
were forced either by gradual population pressure or by more overt 
coercion to vacate the mainland coast. 

Who were the groups that displaced the Arua? The archeological 
record shows two Phases, the Mazagao and the Ariste, developing con- 
temporaneously in adjacent regions of the Territory of Amapa until the 
time of their disruption by European contact. Although these Phases 
are distinct throughout their local history, there is evidence that they 
may have differentiated from a common base about the time of their 
entry into the Territory. Jari Scraped sherds from Sao Joao and 
Bom Destino, both early Mazagao Phase sites, are almost identical 
in design, motif, method of execution, and general ceramic features 
(sandy temper, rough surface, tan color) to Flexal Scraped from 
Sites A-16, A-12, A-11, and A-10 — Inside Cave, all early sites of the 
Ariste Phase. Associated with the scraped wares in both regions is a 
type of incised decoration not found in the later sites, or found in 
greatly diminished percentage: Uxy Incised in the south, and Davi and 
Uaga Incised in the north. This combination of scraped and incised 
traditions with similar motifs argues for an early aflSliation or a first- 



^vTnsT^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 159 

cousin relationship between the two Phases. The initial use of a 
light- orange to grayish-tan surfaced, sandy, quartz or sand-tempered 
pottery in both Phases lends f mother weight to the theory of a common 
origin, while the parallel transition from this pottery to a smoother, 
sherd or canape-tempered ware may be attributable to independent 
evolution. In other words, while it is not peculiar to see the Ariste 
Phase emphasizing curvilinear painted decoration in its later stages 
of development, or to see the Mazagao Phase developing rectilinear 
incision and apphque, it seems more than coincidence that both of 
them began with plastic decoration of a similar type and with a 
dominant sandy-textured plain ware. 

The immediate origin of the ancestral Mazagao-Ariste Phase can 
only be vaguely postulated at this time because of the paucity of 
scientific archeological information from a major part of northern 
South America. The known materials from coastal Venezuela, the 
Lower Orinoco, Trinidad, the Antilles, and British Guiana bear no 
resemblance to the early level materials of the Mazagao and Ariste 
Phases. The limited collections from Dutch Guiana (Goethals, 
MS.) show affiliations with the Ariste Phase of the northern part of 
the Territory of Amapa, but these appear to represent influence from 
the later rather than the ancestral form. The various cultures 
delineated by stratigraphic excavations on the islands in the mouth 
of the Amazon (see fig. 205) are not even remotely related to the 
Mazagao or Ariste Phases in the Territory of Amapa. This evidence, 
although admittedly incomplete, seems to indicate that the ancestral 
Mazagao-Ai'iste Phase was not derived from the north by the coastal 
route, and that some other area and route within northern South 
America must be sought. However, comparative material is even 
sparser and more poorly documented in the Amazon area than in 
Venezuela and the Guianas. Examination of sherds from Itacoatiara 
on the Amazon River in the collection of Sr. Frederico Barata revealed 
a few with complicated, interlocking, rectilinear, incised patterns 
(pi. 88, e-f) similar to the comphcated incised designs of Uaga Incised 
from the Ariste Phase (pi. 26, d-g) ; however, the paste characteristics 
are not similar. 

There is no doubt that such connections are tenuous, but in our 
present stage of knowledge of the archeology of the Middle and Upper 
Amazon, every scrap of evidence must be used. The negative evidence 
of movements from the north plus this scanty, but related, material 
from the Amazon suggests that the Mazagao-Ariste Phases of the 
Territory of Amapa must have entered, or at least been influenced, 
from the Amazon or one of its many upriver tributaries rather than 
from the Guianas, coastal Venezuela, or the Antilles. Although the 
northeastern part of Brazil is poorly known from an archeological 



160 BUREAU OF AMERICAN ETHNOLOGY [bdll. m 

standpoint, the few bits of data from this region do not suggest any 
influence of importance on the cultures of the Amazon until historical 
times. 

After the postulated ancestral Mazagao-Ariste culture became 
estabhshed over the whole of the Territory, the initial unitj^^ gave way 
to diversity resulting both from evolution in the absence of further 
contact with each other and from influences independently received 
from different sources. WhUe there is nothing to indicate that the 
Mazagao and Ariste were hostile to one another, neither is there any 
evidence of intercommunication. Not a single trade sherd from the 
Arist6 Phase was found in any Mazagao Phase site, or vice versa. 
With two exceptions, the sites of the two Phases are separated by the 
wide, swift-running, forest-bordered, low-banked Rio Araguari- 
Amapari. The exceptions, two Ariste Phase sites south of this river, 
have already been discussed (p. 118), but in review it is pertinent 
to mention that Site A-13 — Matapl was nothing more than a camp- 
site, probably not used more than once or twice, and Site A-14 — 
Macapd represents such a variation from the typical Ariste Phase 
materials that it is undoubtedly explained as a late transplantation 
of the Indians by the Europeans when they began to fortify and 
settle the area (p. 564). A week spent in survey of the Rio Araguari- 
Amapari brought forth not a single site, adding to the distributional 
evidence which leads to the conclusion that it was a frontier and a 
sort of aboriginal "no-man's land," between the Mazagao Phase to 
the south and the Arista Phase to the north. 

The first half of the Mazagao Phase represents an undisturbed 
evolution of the various ceramic styles. Jari Scraped dies out, the 
quality of Uxy Incised declines, the plain wai-e continues to be a 
light gi'ay to light tan in surface color. By the time of Site A-6, 
a distinction is beginning to emerge between a sandy, quartz and 
mica-tempered ware and a smoother, canape-tempered ware. At 
the succeeding site, A-2, several innovations suddenly appear, arguing 
for a strong outside influence. These are: (1) Precisely executed, 
chalk-filled, rectilinear, incised designs (Anauerapucu and Pigaca In- 
cised); and (2) completely oxidized firing of Mazagao Plain, producing 
a brick-red surface color. 

The late incised styles of the Mazagao Phase are distinctive and 
their origin should be easily traceable; however, the great gaps in our 
knowledge of the archeology of northern South America make the 
identification less positive than might be desired. An extensive 
search through the literatm"e and museum collections reveals some 
comparable styles. An examination of the sherds in the University 
Museum of Philadelphia from Arauquin, Apure, Venezuela, col- 
lected by Petrullo, as well as inspection of the illustrations of his 



"vTnsT^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 161 

report (1939), shows a great similarity in incised designs to the 
Mazagao Phase materials. Further examination of the Late Araii- 
quin and Late Ronquin Aspects on the Orinoco, as defined by Howard 
(1943), adds more illustrative information. The incised rectilinear 
spirals, parallel lines, nicked rim edges associated with lines, diagonal 
miits aud meanders of the Middle Orinoco (pi. 86) shown by Petrullo 
(1939, pi. 31, Ic, Id, If, 2f, and pi. 32, j, 1) and by Howard (1943, 
fig. 71 and pi. 6P) are motifs also executed on Pigacd and Anauera- 
pucu Incised (pi. 11, 12). 

In addition to these similarities with Middle Orinoco sherds, the 
chalk-filled, rectilinear incisions of Anauerapucu Incised compare 
rather closely in technique wdth specimen No. 24243 (pi. 87, a) in 
the Division of Archeology of the United States National Museum 
from Manizales, 60 miles southeast of Medellin, Colombia. This is 
not to argue that this Quimbaya style of incised vessel is directly 
related to Anauerapucli Incised. It is not, for the shape and paste 
characteristics are different, but the technique of incision and the filling 
of the incisions with a white substance in both areas is worth men- 
tioning. Incisions and punctates filled with lime or white paint are 
typical of another archeological zone in Colombia, the Tierradentro, 
but here the only similarity to the Territory of Amapd sherds is the 
use of a chalk, white paint, or lime in incisions or punctates (Bennett, 
1946, pi. 175). Unfortimately, the chronological sequences in Co- 
lombia are in dispute and few authorities agTee on the exact position 
of these cultures, hence no accurate check can be made to compare 
them sequentially with development of the Mazagao Phase in the 
Territory of AmapA. 

If either, or both, the Middle Orinoco of Venezuela or some part of 
Colombia can be considered a possible source of some of the incised 
traditions found on the mainland in the Lower Amazon, some explana- 
tion is needed to account for the absence in the Territory of Amapd 
of the other ceramic traditions associated with these wares in Colombia 
and in the Middle Orinoco. The modeled tradition associated with the 
incised style in the Arauquin and Late Ronquin Aspects of Venezuelan 
sites either did not diffuse or was not accepted by the Mazagao Phase. 
A third alternative explanation, that the modeled tradition reached 
the Middle Orinoco of Venezuela subsequent to the diffusion of the 
incised tradition out of the area and eventually into the Territory of 
Amapd, cannot be evaluated until more is known of the ancestry of 
the Late Ronquin and Arauquin cultures. The theoretical point of 
view, that the modeled tradition was not accepted even though the 
incised one was received in full force, might be explained by the fact 
that, while familiarity with the incised technique made the Mazagao 
Phase receptive to innovations in motif, modeling had only a rudi- 



162 BUREAU OF AMERICAN ETHNOLOGY [bull, lei 

mentary development, primarily as applique, and therefore the intro- 
duction of the elaborate modeling of the Late Ronquin and Arauquin 
style would have entailed a radical departure, which was refused. 
Granted, these theories lack adequate documentation but with such 
a vacuum in the archeological information between the upper reaches 
of the Orinoco and the Territory of Amapd in Brazil, they offer the 
only means of explaining the Mazagao Phase innovations. 

When theories of this sort are proposed, they are often evaluated 
in terms of the feasibility of the route of influence or migration that is 
implied. In this case, the route is indeterminable at present because 
the area between the Territory of Amapd and parts of the Upper 
Orinoco and Middle Orinoco is a blank, archeologically speaking. 
However, since none of the similarities just discussed are represented 
in the Ariste Phase, which occupies the northern part of the Territory 
of Amapd, the possibility of a migration from the north down and 
around the Atlantic Coast and up the Amazon does not seem to have 
archeological support even though it has been the more often suggested 
movement (Steward, 1948 b, p. 885; Willey, 1949, b, pp. 195-196 
and map 3). With this evidence, then, the inland waterways rather 
than a coastal migi-ation seem to be the answer. 

In addition to this strong influence from outside of the Territory of 
Amapa leaving its impact on the incised styles and the plain ware of 
the Mazagao, the late Mazagao Phase copied from the adjacent 
Maracd tradition. The results attained in this effort indicate a more 
tenuous relationship with the Maracd culture than with the more 
remote one in the Orinoco. Whereas the Orinoco styles are closely 
reproduced and affect certain paste characteristics as well as design 
motif, the Maracd similarities may be interpreted as reconstructions 
from memory of a form only briefly glimpsed. There is no modifica- 
tion in the basic ceramic. The two cylindrical, anthropomorphic urns 
found in the Mazagao Phase cemeteries (pi. 3, b; fig. 12), are suffi- 
ciently different in detail of construction, such as the manner of 
attachment of the legs, the attachment and angle of the arms, the 
position of the feet, the execution of the bench legs, the shape of the 
head, and the addition of modeled bracelets or necklaces and painted 
designs, to dispose of the inference of direct copying or trade vessels. 
Yet, the basic similarity in general bodj^ shape, tubular construction, 
anatomical details of wrist and ankle bones and sex identification, 
position on a bench sometimes ornamented with an animal head and 
tail, and identity in function as an urn for secondary burial, leaves no 
doubt that a Maracd influence is involved. The large jaboty (turtle) 
urns characteristic of the Maracd tradition (pi. 17) appear less fre- 
quently. However, a large foot from cemetery A-3 (fig. 6) must 
have belonged to such a jar and it shows the same kind of deviation in 



EV^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 163 

specific detail of execution that appears on the anthropomorphic urns. 

The geographical restriction and cultural uniformity of the Maraca 
tradition argues for a very short occupation of the Rio Maracd area, 
which the presence of glass beads places about the time of European 
contact. The place of origin of the tradition is not known definitely, 
but it is much less obscm-e than the source of some of the other ceramic 
styles in the Territory of Amapd. Anthropomorphic figures seated 
upon clay benches have been reported from Popaydn, Quimbaya, and 
other parts of the Cauca valley in Colombia (Bennett, 1944 a, figs. 
IIA, 17B, 17C, pi. lOF; Willey, 1949 b, pi. 39c; Imbelloni, 1950, pi. 
26 No. 24), from Manabi, Imbabura, and Carchi Provinces in Ecuador 
(Uhle, 1929, pis. 6, 10; Gonzalez Sudrez, 1910, pi. 5, fig. 1, pi. 7, fig. 1; 
Saville, 1910, pis. 86, 87, 88; Jij6n y Caamano, 1920, pi. 41) and from 
the region of Bocono and Niquitao in the State of Trujillo, Venezuela 
(Imbelloni, 1950, pi. 26 No. 25 and No. 27, figs. 14, 15; Kidder II, 
1944, pi. 17, 18-32; Kidder II, 1948, pi. 75G). Although each of these 
figures and the clay benches upon which they sit exhibit features 
peculiar to their own areas, certain similarities suggest a common 
origin. These include the small clay bench, 4-8 cm. high, usually on 
four legs; the cylindrical body of the figure, tall in proportion to its 
diameter; painting sometimes on the body; the predominance of male 
sex over female; the cylindrical arms with elbows generally akimbo 
and hands commonly resting on the knees; the hands and feet crudely 
modeled and stylized, often with 3 to 7 fingers and toes shown by 
light incisions; and swollen calves suggesting the use of ligatures. 
Similar anthropomorphic figures not, however, seated on benches 
have been found in the above areas and also in eastern Bolivia (Im- 
belloni, 1950, pis. 17, 18). 

For the purposes of this comparative study, it was felt that suf- 
ficient evidence of similarity of style was presented by restricting the 
comparison to anthropomorphic figures seated on clay benches. 
Their concentration in the Andean region of Colombia, Ecuador, 
and Venezuela suggests this area as a possible source of the Rio 
Maraca style. Whether the Lower Amazon material represents a 
migration or diffusion out of one of the above-mentioned areas with 
a local variation developing in the Rio Maraca, or whether it is one 
of many lines of diffusion and influence out of an undetermined 
central source, only future work will prove. Imbelloni has recently 
pursued this subject in greater detail, embracing all styles of anthro- 
pomorphic urns and comes to a similar point of view: 

In conclusion, the center formed by Maracd and its branches was influenced 
by a modeling tradition separated from the main Venezuelan-Colombian stem, 
to which belonged the classic, seated male figure, with or without a bench. [Im- 
belloni, 1950, p. 119.] 



164 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

The Ariste Phase in the northern part of the Territory of Amap^, 
in contrast to the Mazagao Phase, does not appear to have enjoyed 
a preHminary period of undisturbed development. The earhest 
cemetery sites excavated, the Caves of A-11, exhibit traits absent 
from the early Mazagao Phase. Most prominent of these are burial 
in caves instead of in the open, the use of painted decoration on burial 
urns, and the practice of cremation instead of secondary burial. 
Other innovations include the cambered collar and the flaring lip 
with prominent lobes on cemetery vessels and the graters in the 
habitation sites. If the argument for an early affiliation between the 
two Phases is valid, these traits just mentioned must be laid to local 
invention or to an influence that did not penetrate south of the Rio 
Araguari-Amapari. 

The early predominance of painting in bands and/or over the whole 
surface of the vessel fades away in a gradual transition to the elaborate 
linear motifs as typified in the distinctive Serra Painted style on the 
vessels collected by Goeldi from CunanI (pi. 24, c-e). This gradual 
shift of decorative styles and techniques is suggestive of local evolu- 
tion. Such a conclusion finds support in an absence of affiliation 
between the complicated painted motifs of Serra Painted and other 
known painted styles from northern South America. Although the 
later developments suggest indigenous evolution, the fact that some 
form of painting begins with the earliest occupation levels of the 
Ariste Phase rather than after the Phase was well established lends 
weight to the supposition that the initiation of a painted style is to 
be attributed to outside influence. The sherd-tempered Serra Plain 
(which closely resembles Piratuba Plain of the Arua Phase in paste 
features) begins at the same time. Both painting and sherd temper 
are Arua Phase traits, and there is a possibility that while thej^ were 
displacing the Arua, the people of the Ariste Phase were influenced by 
their ceramic styles. The fact that the Arua Phase is weakly repre- 
sented south of the Rio Araguari-Amapari would account for the 
absence of a comparable influence on the Mazagao Phase. 

Comparative material for the Ariste Phase is also restricted by 
the lack of extensive archeological work in French and Dutch Guiana. 
The Reichlens' (1947) comments on French Guiana are interesting 
but not usable from a comparative ceramic standpoint. Dutch 
Guiana is almost unknown except for the limited excavations made 
in the Paramaribo area during the summer of 1951 by Peter Goethals 
(MS.) of Yale University. Our examination of Goethals' sherds 
revealed no significant similarity to the Mazagao Phase, but a large 
amount of the pottery showed close resemblances to the sand-tem- 
pered Ariste Plain, the sherd-tempered Serra Plain, and such decorated 
types as Uag^ Incised, Davi Incised, Ariste Painted, and Serra 



wi'tfi'f® ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 165 

Painted. A comparison of some of the most distinctive and charac- 
teristic vessel shapes of the Ariste Phase with the Dutch Guiana 
materials reveals an occurrence of two shapes in both areas: the 
large, open carinated bowl with wide, flaring rim lobes (Ariste Painted, 
form 5) and the carinated bowl (Serra Painted, form 3). The flat 
base with an angular junction with the side walls, so common to 
most of the pottery types of the Ariste Phase is also the most frequent 
base shape among the vessel fragments from these coastal Dutch 
Guiana sites. The application of red paint to the lip and rim edges, 
common on vessel forms 3 and 4 of Ariste Painted and form 1 of 
Serra Painted, is repeated on the painted sherds from the Paramaribo 
area. Further similarities are evident in the Dutch Guiana scraped 
and incised sherds, which resemble in technique and design elements 
Uaga Incised, Davi Incised, and Flexal Scraped from the Territory 
of Amapa. 

Further work in the Guianas should strengthen the cultural connec- 
tions of the Arista Phase of the Territory of Amapa with the north. 
When these areas are more thoroughly surveyed from an archeological 
standpoint, it wiU probably be possible to extend the boundaries of 
the Ariste Phase into French and Dutch Guiana. The more elaborate 
painted style and the complex jar shapes of the late Ariste Phase 
may not be discovered, however, in the other Guianas. Since these 
features appear to be wholly post-A. D. 1500, they may be confined 
to the northern Territory of Amapa, where the aboriginal culture 
seems to have escaped for a longer time the disrupting effects to 
which the adjacent parts of the Guianas were subjected (see p. 565) 
and to have enjoyed as a result a longer period of indigenous de- 
velopment. 

The presence of cremation in the Territory of Amapd is more diffi- 
cult to explain. The secondary burial found in the same jar with a 
cremated body at Site A-11, Cave 2, probably represents the last sur- 
vival of this earlier method of disposal of the dead. Since cremation 
with the deposition of the ashes in a pottery vessel is absent in the 
southern part of the Territory of Amapd except in the Rio Maracd 
region and is found only in the late Marajoara Phase on the islands 
in the mouth of the Amazon, with no other rehable definition of this 
burial trait in archeological sites in the Lower or Middle Amazon, 
these areas do not seem likely to have furnished the influence that 
introduced cremation into the Arist6 Phase. It may have come from 
the north, but the sketchiness of our knowledge of the distribution of 
cremation prevents a more specific statement. Perhaps supporting 
evidence can be gleaned from the fact that it is practiced among the 
Rucuyen and Atorai of the Guianas. It is generally considered the 
rarer funeral practice, with the Rucuyen placing the ashes in a pot 

391329—57 13 



166 BUREAU OF AMERICAN ETHNOLOGY [boll. i67 

which is kept by the widow, while the Atorai bury the ashes (Gillen, 
1948, p. 851; Koth, 1924, pp. 641, 664-665). Among the Wai Wai 
Indians at the headwaters of the Essequibo River in British Guiana 
and the Mapuera River in Brazil, cremation with the placement of 
pottery vessel over the ashes was common until recent times (Evans 
and Meggers, MS.). Although Gillen (1948, p. 851) in a summary 
article on the Guianas states that archeological evidence along the 
Orinoco reveals that cremation as well as urn burial was common, the 
primary sources on this area (Kidder II, 1944; Osgood and Howard, 
1943; Osgood, 1943) mention only urn burial and secondary burial. 
The two shaft graves from Goeldi's Cunani burial site (A-19) present 
another problem. Either this practice was locally invented, as sug- 
gested by Goeldi (1900, pp. 22-23) in his explanation that it substi- 
tuted for the natural caves otherwise used but absent here, or the 
idea was received by diffusion. To date, the only reported South 
American shafts of this style come from the Rio Pichinde Complex, 
Rio Bolo Complex, and Quebrada Seca Complex of the Upper Cauca 
River, Colombia (Ford, 1944, figs. 2, 6, 7, 10, 12), and from the 
Quimbaya region of the Middle Cauca River, Colombia (Bennett, 
1944 a, p. 59; 1946, fig. 92). Although these Colombian shaft burials 
with oval antechambers are almost identical in shape and construction 
to the ones at the Cunani site, the burial pattern within the chamber 
differs. In Colombia, secondary burial or direct inhumation with 
offerings of plam ware and/or crude incised or applique-modeled 
vessels was the common practice as compared to cremation urn burial 
in painted vessels in the Territory of Amapa. If the idea of an ante- 
chambered, shaft tomb had been received by diffusion from the groups 
using the same structure in Colombia, it seems probable that other 
items of the burial complex and possibly some pottery styles would 
also have been transferred to the Ariste Phase. A third isolated occur- 
rence of the antechamber, shaft grave in the Piedmont Region of 
North Carolina (Coe, 1949) strengthens the case for independent 
invention. The formal similarity of all these burial chambers can be 
laid, in part, to the structural limitations on digging a shaft-burial 
chamber: the domed antechamber roof reduces the danger of cave in, 
and the chamber to one side of the main shaft served the function of 
facilitating removal of dirt rather than a purely aesthetic purpose. 
These structural features which limit the shape and general design, 
added to the distribution of the trait in isolated and widely separated 
areas, and the absence of associated traits indicating diffusion, 
strengthen the likelihood that the appearance of shaft burials in the 
Arista Phase in the Territory of Amapa can be considered an example 
of independent invention. 



I 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 167 

EVANSJ 

In summary, the results of the preceding analysis of cultural develop- 
ment in the Territory of Amapd and comparison of traits found in this 
region with those from other parts of northern South America can be 
digested into the following points : 

1. Ceramic-using cultures arrived in the coastal region north of the 
Amazon in late precontact times. 

2. Prior to this time, the region was probably occupied by a pre- 
agi-icultural, hunting and fishing population, or a preceramic, agri- 
cultural group. 

3. There is evidence that the first ceramic-using Phase in the Ter- 
ritory of Amapa, the Arua, is derived from the north. 

4. The Arua Phase was forced out of the Territory of Amapd and 
onto the islands in the mouth of the Amazon by the incoming ancestral 
Mazagao-Ariste Phase. 

5. The postulated ancestral Mazagao-Ariste Phase was probably de- 
rived from farther up the Amazon or one of its tributaries and did not 
come along the coast of the mainland from a northerly direction. 

6. Once established in their respective areas, the Ariste and Maza- 
gao Phases developed independently of each other, and show no 
evidence of contact in the form of trade or influence. 

7. About, or just prior to, contact times (A. D. 1500) the Mazagao 
Phase was strongly influenced by a culture with an incised style of 
decoration resembling the Late Arauquin and Ronquin of the upper 
Middle Orinoco, coming by way of an inland route rather than via the 
coast. 

8. The Ai'ist^ Phase was subjected to different influences, which 
introduced cremation and painting of pottery into the area in late 
pre-European times. 

9. The anthropomorphic, seated figures of the Maraca tradition in 
the southern part of the Territory of Amapd show the closest afl&lia- 
tions with the Andean area of Ecuador, Colombia, and Venezuela, 
and suggest a derivation from this area about the time of European 
contact. 



MARAJO ISLAND 

GEOGRAPHICAL DESCRIPTION " 

Maraj6 is the largest of a multitude of islands that divide the Ama- 
zon into many channels as it nears the Atlantic Ocean (fig. 47). Al- 
though the main flow of the river passes along its western and northern 
shores, numerous juros or narrow passages direct part of the water 
into the Rio Pard and the Baia de Maraj6, which separate the island 
from the mainland on the south. So strong is the force of the outpour- 
ing fresh water that the salt sea approaches the Cabo Maguari only 
toward the end of the dry season, when the diminished river gives 
way before the incoming tide. The equator passes 0°5' to the north of 
the northern coast, and the 49th and 50th meridians divide the island 
into almost equal thirds. The best estimate of dimensions comes 
from the World Aeronautical Charts prepared from aerial photographs, 
on which Maraj6 is shown as 265 km. (165 miles) long from east to 
west and decreasing in width from 180 km. (110 miles) near the west 
end to 150 km. (95 miles) at the mouth of the Rio Arari. 

As one goes from east to west, there are two other changes in addi- 
tion to the gradual widening of the island that are of importance: 
the decUne in elevation and the alteration of the vegetation. From 
the eastern coast, which rises 5 to 6 meters above the water at high 
tide, there is a gradual slope toward the west until the land left by the 
strainerlike mesh of streams is in constant danger of being inundated 
by the high tides that come twice each day. The surface is not per- 
fectly level, but the depressions and rises are so slight that they are 
noticeable only in the rainy season when most of the patches of forest on 
the eastern part remain free from the water that inundates the campos 
and turns them into a vast shallow lake. By the middle of the summer 
(August), most of the fioodwaters have drained off except where they 
are trapped in depressions. Evaporation and months without a 
heavy shower usually result in the drying up of these areas before the 
advent of the next winter's rains. There are a few places where the 
marsh is especially large and floored with soft oozing mud that will 
bear only the lightest weight, and here a special tenn, mondongo, is 
applied. Ferreira Penna gives a good description of one on the 
northeastern part of Marajo: 

This name is given to a marsh of vast extent, 10-12 miles inland from the north 
coast and reaching from the headwaters of the_Rio Cururu eastward almost to 

" For an excellent description giving the impressions of an early visitor ,[see Anonymous, 1807. 
168 




\'^o' 



iiO 



_i I 1 1 I 

10 20 30 40 50 KM 



• 1948-49 EXCAVATIONS 

FOREST 
i SWAMPY LOWLAND 



AFTER AAF PRELIMINARY BASE 
946 A a B. 



ogical sites. 







V- 












^-— -"^^^-^ 
















"f^^ 


^^^^ \ ^^ 


.Choves 


TtC-^*!;" 






^%^^ 














P^: 


'^M^s^ 




WP^m 


% 






^T?^'^ 




Y^TX. CABO 






"/v:-v.:v< 


;; .v^^ 


^^^ 


i . 


/ jx/^ V^rv 


y\ 


. < f y -*- \ JiL / i. -it 


^W\*r\ 


T "^ 




^^ \ ^ MAGUARI 






•? A 1 ' ' ■ 




■ ■ ' • • \ It j 


■ i\. 


V%) - 1 2^V ^,^^ 7 c"^ 




^ H^ /-^-^-V "*" ^ 


>jt • J 1^ v> ^ 


Iftiu . y 




\ 






«■"" [)/ V 


'■:;.;■.:■ -5$^ 


^^^ ^v^ 


ofe 


i^^vp^^^ 


\ 




;^-r'4< 


p^_ 


s_:r. 


y 


30' 




(^T^ 


■■ ■'■ v^ 


'^ (^ 


. V 


^'^l'^ 


v. 


-MS . ^ . 'C^ 


■*» k — : ■ 


./.P^ 


f^ 


■ik:l 30' 








^^2w 


iS^t^ 


■ r' 


./:•■• :.^ 




,T_^ r~c \..]x^ 


*\ V 


>\;A 


5, 


■' 








iT ^-^v X vLJ^ 


/y»' <»L. r^- 


I 'A J' 


•* ■ ■ '***, ^^""^ ' 




-^■y^^^^'V.- ■ /■ -f- ■ ^ ." ' -/■' 


^Y y/T^' , -Jr: A ' J/X %, 


\-/ -7 


\; 


■ J 
























J 










^^-s. 1 •'X-^^G' 


11 ^^ T^ ^\. > 


■ '^' ■ ' - 


• ''''}' -V • 




. V>Y. ■jJx.><.- / /■".•- ^ I 


\\ ' ''J-\.' -J^'^^V 


. ■ V^ ^ V\' i 


r 


V.^^ 








^ 


|Si 


ii 


ipi 


F 


7-^^|lt 


' GUAJARA V .^ 


jm 


X 


Sour* 






rNii/ ^^pcj ^ 


r^A.fc 


^ r^]^^ 


■■'••■:'>- 


3^v^-^ 


:-.-," /■'■^Ir^:^l 


>-^-32^f 


W^i- 


r^ 








l^nr^v . H 


\ ^p 1 ' ^ 


V? fJ^^ ^ ■ • ■ 


".'•.■("■ 


V Vr^ 


-i;J 


J- 15 1/. ■■■ J Y M^^. 


l/ys fV ' 


^y^Vr?" v,^- ■ 












U '^' \- 


\^y^ g- fSLf0^ ^*v 


' .* ■ /'- '- 


• ■*."".'■■ /*•*— -^- '. 




jr y^iiUJj*!:.^— • v. 


I ( Ms Y 


C ' ^ ^ - 












5 » ^ 


^^!?^^?~^ P>il 


V '. • J ■ 


■ ^"'vp-^' ■■.'.■.■■". '■.■■-^ 




V V '|- ^^"^^^-^-^-^ . /^"^^aY' 




O ''^jV^P- 






!• 






/ tst^^ 


~V~>^. L ^^ 


'%i^/' 


''"'T^s:^'^.'.-.-:-; ;'/^ 


^i 


^J^^Ur\^''^-^\rC7^ "fl- f} 


Vy>^'' 


( \^^ 




r 
































\\^^;;;~_,^ t« ^/yV^ 


""^^^^ ' ■ 


■^^?^ -^A-^ 


LV.p\. 


■ -1 • -l^ 




'■■ '■ ) ■ ■ :/'^'^?^^-^^ V '1 " ° ^^^^5>^ ' ■ ^ • " 


^^^ 










i/- \ : 1 ) < 


ii 


ill 


il 


liiii 


■/ 


,wj|» 


Vf V ■'■ ■' ^^-w' 


^ 








■n 

c 




Pre 


wMm- 






1 


;|p^Jm:pS; 


■ 1 ' \>*?vu/^ y 










o 




r^-V'; ■';•■. v. ;■•'■■•■ 




. • ^ t; . ■ -^Y"^ • ■ 




* ■'^''^Nl" * ■ "^Vl-^^^h jff^y^^ - V 


.'JL --XU/ 1 












v/\''.- -y/ r^ ^ 


^Vi'' •■j>~^ 




<."•■•.'■• 


v'" i^--:^V .■ ■ ■• 




■ v. ; '. I ,^N-~-^ v-v^'^^ 


\l ''C^ 








rao' 




Vr^Y ( -?i|^ 


m-^:^ 


f^' IZ^ M 


^■•'.'■/ 


■ . ' A 




^ ^^r^ J ^-^yf" \ V 
















^■^ 


fcl^'^ '^ 


Tc^ . ■ 


"3^ •v^'^-\ 




i"^^i^MsX<' '^""'"^'^i-^ ■'vr^^--^ 














'^ 


^reverS\,__. )) 


Pllp 


p: 


:4ii 


'>/ri 


^^".- 


ot 






1 1 1 1 1 1 


10 20 30 40 50 KM 








■N^*^ - ^S>»^- ^\ 


MF ji |, J J NiT n /*? !>-^ 


.(/ 


■ ^""X \- 'Y<rv. 1 '. ^ - I " J '\' 


)l^ 








■ 1948-49 EXCAVATIONS 








ll/o^ c^'^^^-V) 


o^^ j!^r'''"'^\v' W^"^^. 


.■\^^. ^ A 


\v\ j Yiy. /^^ 




r' 
















S.^^^^'-]?^ 


,^^^^1' __^/^ ^ 


^^"N^ 


1 ^^Tf^^jjf 












•..: FOREST 










RIO PARA 
















i. SWAMPY LOWLAND 

AFTER AAF PRELIMINARY BASE 
946 A a B 






50- 30' 

1 


5C 


• 




49" 


1 

30' 4 9* 









Figure 47. — Maraj6 Island, showing major streams, vegetation pattern, and the location of archeological sites. 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 169 

BTANSJ 

the coast. Within its limits are formidable mires, several small lakes, various 
islands and above all, an infinity of swamp vegetation, especially Atingas (Cala- 
dium arborescens) among which are concealed millions of reptiles that make the 
approaching of these solitudes a dangerous thing. . . . 

At the onset of winter, the mondongos collect a large part of the rain water, and 
rapidly filling, begin to extend themselves by means of the natural drainages. 
These outlets are the rivers Tartarugas, Ganhoao and Arapixy, which empty into 
the north coast; the Cururii, which flows westward; the Moco5es, leading south- 
west into the Anajds; and the Genipapucd and perhaps the Apehy, the former 
flowing from northeast to southwest and the latter from north to south, both end- 
ing in the Lago Ararf . [In Derby, 1898, pp. 164-165.] 

The explanation for the differing elevation lies in the geological 
structure of Maraj 6. The foundation of the eastern part of the island, 
like that of the adjacent mainland to the south, is a stratum of red 
sandstone (gres vermelho) . An outcrop of this deposit appears in the 
bed of the Kio Arari about 30 Ion. upstream from its mouth and pro- 
duces a small rapid at low tide, which gave the town of Arariuna its 
former name of "Cachoeira." Except for this and the vicinity of 
Ponta de Pedras on the south coast, the island is devoid of visible 
stone. The central, northern, and western parts are predominantly- 
alluvial deposits built up behind the original rocky obstruction by the 
silt-laden waters of the Amazon (Rich, 1942, pi. 28). 

The vegetation pattern seems at times to be correlated with the 
elevation of the land and the drainage pattern, and yet there is no 
complete consistency. It is customary to say that a line drawn from 
Afud in the northwest to Maund, opposite the mouth of the Tocantins 
on the south, marks the approximate division between the campo or 
grassland and the forest. Actually, a large part of the eastern half of 
the island is also occupied by forest, which forms a wide coastal belt 
(Rich, 1942, pi. 27) and fringes all the major rivers and many of the 
smaller ones (pi. 27, 6) . Except in the region between Lago Arari and 
the east coast (Rich, 1942, pi. 29), there are no large vistas in which 
the campo is free from scattered trees or from clumps or ilhas of forest 
(pis. 28, 29). "Island" is an appropriate designation for these clumps 
of trees for two reasons: as a figurative expression it suggests their 
dry-season appearance, irregularly scattered over the campo like 
islands in a sea, and in the rainy season it has literal accuracy, because 
the trees grow on slight rises that remain above the waters covering 
the campo. The coastal forest belt is widest along the south, where 
its elevation of some 2 meters above high tide makes it exploitable by 
slash-and-bum agriculture, which is carried on up to the present time 
(pi. 27, 6). Derby (1898, p. 168) reported that in 1898 sugarcane and 
cacao were being cultivated there profitably. 

The dense and unbroken forest covering the southwestern half of 
the island is not usable for agriculture because of the poor drainage of 
the land (pi. 27, a). As a result, in spite of the network of navigable 



170 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

rivers that provide ready access, settlements are sparse even today 
and the inhabitants are mainly rubber-gatherers. Hartt's description 
of the area around the town of Breves would be applicable anywhere 
in this part of Maraj 6 : 

Here and there along the canals of Breves are eminences of land similar to that 
where the town is located, but, in general, the borders of the river are inundated 
with each high tide and the scattered houses are built on poles to raise them above 
the water. The canals are narrow, extremely deep and filled v/ith muddy water. 
Whether the tide is in or out, they are always swollen as though it were high. 
And how rich the vegetation that borders them! Here is found the mangues 
with its beautiful green foliage, its arched main roots, its pendant areal rootlets 
with trifurcated tips and its cigar-shaped seeds; there the channel is bounded on 
both sides by walls of vegetation, the tips of the branches grazing the water at 
high tide and stopping the beautiful rafts of grass and broad-leafed mururS 
with its blue flowers; and further on we see for many kilometers the majestic 
muritis with their superb fan-shaped leaves, their dead branches yellow and 
drooping, and supporting their heavy clusters of scaly fruits. Here and there 
the graceful and slender trunks of the assal palms lift into the sunshine their deli- 
cate green fronds and seem to fringe the band of broad muriti leaves. The 
ubassii, stout and vigorous like a giant amaranth, joins with the lance-leafed 
aningas and the mimosas to fill the spaces between the trunks of the palms. This 
is the perfection of the vistas along the Amazon. The calm brown waters of the 
high tide; the floating plants; the dark green shadows in the water beneath the 
dense foliage of the bank; the bending palms; the wall of vegetation, seemingly 
as solid as a wall of stone; the glint of the sun's rays on the blue wings of the 
morpho butterfly as it flits across the river; the flock of parrots, appearing two by 
two, their wings fluttering against the dark blue sky sown with silvery clouds; the 
soar of the lovely kingfisher, poised at one moment on a high branch and suddenly 
flinging itself at a pidba, which it carries from the water sparkling in its mouth — 
all these produce a picture the traveler can never forget, and the effect of which is 
heightened by the warm still air, the perfumes and the agreeable languor of the 
tropics. [Hartt, 1898, pp. 174r-175.] 

Probably for the same reasons that make it poorly suited for human 
habitation today, the forested western part of Maraj6 seems to have 
been sparsely settled in pre-European times. At least, no sites or 
ceramics have ever been reported from the area. Of primary concern 
here, therefore, is the eastern two- thirds, in almost the exact center 
of which is Lago Arari. 

Lago Ararl is the largest of some half-dozen permanent lakes on the 
northeastern part of the island. It runs north-south, 16 km. long 
and 4 km. wide in the dry season, and is so shallow at that time of 
the year that it can be waded across (Derby, 1898, p. 165). With the 
inundation of the caiwpos, the lake becomes greatly enlarged and in- 
creases to from 5 to 9 meters in depth. Even at its shallowest, it is 
not always a placid body of water, but is often whipped by a strong 
wind into choppy waves that make crossing in a small canoe a long 
and arduous task (cf. Lange, 1914, pp. 303-305). A number of small 
streams and igarapes flow into the lake from all directions, and it in 



MEGGERS ANn ARCHEOLOGY AT MOUTH OF AMAZON 171 

EVANSJ 

turn functions as one of the major sources of the Rio Ararl. This 
river, because it provides access to the heart of the cattle region, has 
become of primary importance in the present economy of the island. 
Leaving the western shore of the lake about one-fourth the distance 
from the southern tip, and joined shortly by the Anajis-miri (Ana- 
jasinho) coming from the west, it follows a generally southeastern 
course and enters the Baia de Maraj6 opposite the modern city of 
Belem. Most of its course is through campo, usually obscured from 
view by the wooded shore. Except near the mouth, where it widens 
considerably, the Arari is 15 to 20 meters in width. The rise and fall 
of the tide are noticeable well above the town of Arariuna, where the 
only permanent obstacle to navigation, the rapid, is passable at high 
tide. Toward the end of the dry season the upper course becomes 
clogged with canarana (a coarse grass) , which maintains a feeble hold 
on the soil and is readily torn loose when the rainy season strengthens 
the current. Massed together into deceptively solid islands, these 
plants float downstream and occasionally temporarily impede ths 
passage of boats (Lange, 1914, p. 294; Derby, 1898, p. 166). 

The Rio Anajas, by far the largest river on the island, originates in 
the mondongos north of Lago Arari and in the campos not far from the 
middle Rio Arari and takes a tortuous course, gathering many tribu- 
taries along the way, until it enters the Amazon at the west end of the 
island. Except for the upper reaches, it flows through forest, and al- 
though navigable it carries little traffic today. However, as an avenue 
leading into the center of the island from the west and readily accessi- 
ble from the mainstream of the Amazon, it probably was the route by 
which some of the peoples identified with the intrusive archeological 
Phases penetrated to the northeastern half of Marajo. 

Several short but wide rivers flow into the Rio Para — including the 
Guajar4, the Sao Domingo, the Canaticii, the Pracuuba, and the 
Atua — all navigable and draining forest rich in wild rubber trees. 
Two important rivers, the Camara and the Paracuari (formerly 
Igarape Grande), flow east to the Baia de Maraj6. At the mouth of 
the latter is Soure, the largest modern town on the island, A number 
of short streams and igarapes run north into the Amazon, but few 
are passable at low tide because of sandbars and sunken tree trunks 
and none extend far beyond the coastal band of forest. The primary 
access to the interior in the north is via the Rio Curiu-u, a tributary of 
the Anaj4s that runs generally parallel to the coast for a considerable 
distance. 

At sea level, close to the equator and exposed to a constant breeze 
from the ocean, Maraj6 presents a seasonal variation marked by the 
presence or absence of rain rather than by differences in temperature, 
which has an average annual variation of only 1.5" C. (Le Cointe, 



172 BUREAU OF AMERICAN ETHNOLOGY [boll. 167 

1945, p. 82). Although there are no figures available for Maraj6, 
the annual rainfall at nearby BeI6m averages 2,551 mm. (104.3 
inches) (op. cit., p. 86). The inception of the rainy season in Decem- 
ber or January raises the level of the Amazon so that the water falling 
on Marajd is unable to drain off. The baixas fill and overflow and 
the lakes swell until the major part of the campo is transformed into 
a vast sheet of water, which the protruding tips of the grass blades 
make into a facsimile of endless rice paddies. After the water has 
stood a few months, succulent water plants and lily pads begin to 
appear where a short time before the land was parched, baked, and so 
dry that the hardiest grass withered and turned brown. The water 
also covers the low western end, creating a flooded forest known as 
igapo. Travel is possible everywhere in the interior by dugout, 
poHng in the shallow places, paddUng in the deeper ones. The short- 
ness of the season prevents the growth of a dense and impenetrable 
mat of vegetation like that sometimes formed in rivers and shallow 
lakes, and before the introduction of the horse it was probably the 
time when the inland transportation and communication were easiest. 

Although rain falls in Belem throughout the year, Maraj6, Mexiana, 
Caviana, and the Territory of Amapa have a dry season of 3 to 4 
months duration, during which almost no rain falls. Showers occur 
with diminishing frequency as May gives way to June, and the water 
begins to drain from the land, leaving soggy campos. The drying is a 
slow process, and early September often finds the baixas with consider- 
able water and the soil of the campo damp enough to keep the grass 
bright green. The pink, morning-glory-like flowers on tall stalks and 
the yellow-blooming carobeira trees scattered over the plain give this 
time of the year the beauty and freshness of a northern spring. From 
September to December or January, it is a rare cloud that darkens the 
sky, and a rarer one still that brings so much as a sprinkle. The 
campos, small streams and baixas dry quicldy, and even the pirizais 
and mondongos shi-ink considerably in extent. The soil becomes 
parched, dusty on the surface, and hard as cement. Vegetation on the 
campo becomes browner and dryer under day after day of undiminished 
sun. In December the clouds begin to build up more extensively than 
in preceding weeks, often to dissolve in the evening, but one day they 
do not and there comes the first shower, often followed by a bright 
rainbow that heralds the beginning of another j^ear. 

The marked difference between wet and dry season produces differ- 
ential conditions of food supply. The greater number of wild fruits 
mature during the rainy season, including cupuassu, bacuri, ingd, 
manga, jutahi, cacao, maracujd, assai, bacaba, and mamdo. Not only 
does man find wild food more plentiful at this time, but the predomi- 
nantly vegetarian animals are well fed and plump. The modern 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 173 

EVANSJ 

caboclos say that the rainy season is the best time for hunting, and even 
today when thousands of head of cattle roam the campo and men 
search the forests armed with guns, game is to be found in most areas. 
Among the most important animals are the porco do matto (Dicotyles 
labiatus; peccary), paca {Coelogenys paca), anta (Tapirus americanus; 
tapir); cutia (Dasyprocta aguti; agouti), capivara {Hydrochoerus 
capivara), coati de bando (Nasus socialis), onga pintada (Felis onga; 
jaguar), preguiga (Bradipus tridactylus; sloth), tatu (Prionodontes 
gigas; armsidilio) , tamandud-bandeira (Myrmecophagajubata; anteater), 
and numerous species of monkeys, of which the guariba (Mycetes sp.) 
or howler is one of the largest. 

From the water come the bdto or porpoise, the jacare or cayman 
(several species), and a great variety of fish of all sizes, flavors, and 
consistencies. Since these disperse over the campo in the wet season, 
fishing is most profitable when the lakes and rivers are smaller and the 
fish are more confined. Today weirs are used in the igarapis, where 
the fish are trapped with the falling tide, nets are employed along the 
coast and in the lakes, and pirarucu {Sudis gigas) are generally speared. 
The latter attain a length of more than 2 meters, a weight of 80 kilos, 
and are equally delicious fresh or dried. An early account speaks of 
fishing by poisoning the streams (Anonymous, 1907, p. 295). The 
water also yields turtles of several varieties, and their eggs, as well as 
those of the jacare and camaledo {Iguana tuberculata) , are considered 
delicacies by the people today. 

Last, but by no means least, are the birds, present in variety and 
profusion. Of all the fauna, they are the most in evidence. No trip 
up an igarape is without its glimpse of the blue and brown kingfisher 
perched on a high branch, the strange, brown-plumed ciganas roosting 
in the foliage along the shore, or a pah' of gaudy, screaming araras 
(macaws) flying high overhead. The white-plumed garga (heron), the 
rosy colhereira (spoon-bill), the immense tuyuyu, ih&jacu, httle brown 
marrecas (ducks), and great numbers of other edible bhds, including 
parrots, are to be caught in the forest or on the campo. 

In regard to useful plants of a nonedible natm-e, it is perhaps suffi- 
cient to mention that here as elsewhere in the Amazon there are the 
palms, fibers, timbers, and reeds yielding the raw materials for every- 
thing from baskets to houses, from dyes to canoes. 

The resources of Marajo might be summarized from the point of 
view of human exploitation as rich in wild foods, but poor for agri- 
culture except tree crops. A description of the potentialities about 
the year 1675, when cattle raising was still on a minor scale, probably 
gives a good approximation of the aboriginal condition: 

The island of Joannes has clear air, good water and good lands and pastures . . . 
There are . . . several fields in one part that are full of mango trees that give 



174 BUREAU OF AMERICAN ETHNOLOGY [bull. 16T 

very fine mangos, and others with turtles, and forests where wild pigs can be hunted, 
as well as deer and various other edible animals. Also, its lakes and rivers are so 
abundant in fish that the national fishery located on its shores takes in enough 
tainhas to provide the city of Pard [Bel6m] with fresh and salted fish. The earth 
of this island is good for cane and in some parts for tobacco, and also for the 
planting of cacao trees, which in places grow wild, to the great advantage of the 
inhabitants [Betendorff, 1910, pp. 25-26]. 

Possessed of the Indian's knowledge of fish poisons, of the manufacture 
of traps and weapons, of the habits of the game, and of patience, 
stealth, and skill, no one woidd be threatened with starvation. Kather, 
the island could support a relatively large population, principally 
because of the excellent fishing conditions found in the combination of 
large and small streams, lakes and the Amazon itself. In view of this, 
and considering the slight amount of agriculture on the island today 
(almost totally concentrated in the southeastern border) , it is probable 
that the cultures that came and went in prehistoric times lived pri- 
marily by hunting, fishing, and gathering, and only secondarily by 
agriculture. 

TROPICAL FOREST PHASES 

Four of the five archeological Phases discovered on the Island of 
Marajo exhibit characteristics in settlement pattern and in ceramics 
that show them to have been fundamentally like living cultures of 
the Tropical Forest Pattern. They have been designated as the 
Tropical Forest Phases to emphasize this similarity and to underhne 
the great contrast between their simple culture and the vastly more 
complex one possessed by the Marajoara Phase. The four Tropical 
Forest Phases, to be described in chronological order, are (1) the 
Ananatuba Phase, (2) the Mangueiras Phase, (3) the Formiga Phase, 
and (4) the Arua Phase. 

THE ANANATUBA PHASE 

Description of Sites and Excavations 

Four habitation sites identified as belonging to the Ananatuba 
Phase are located on the Fazenda Santa Catarina, which occupies the 
section of the north coast of Maraj6 lying opposite the western half 
of Mexiana Island (fig. 48). 

site j-7 — sip6 

On the right bank of the Igarape Tapera, about 10 km. inland 
from the north coast of Marajo, is a village site 70 meters in length 
and averaging 25 meters in vvadth. It is composed of two nearly 
circular mounds about 4 meters apart, joined by an area of less ele- 
vation but still higher than the surrounding terrain so as to produce 
an hour-glass-shaped mound with a northwest-southeast axis (fig. 49). 




Figure 48.-North coast of Marajd Island, showing vegetation pattern and the location of sites J-6 through J 13. 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



175 




Figure 49. — Plan of J-7 — Sipo, a habitation site of the Ananatuba Phase. 

The existing height of 75 cm. represents the accumulation of refuse 
during occupation. Finely broken sherds were heavily concentrated 
on the northwest rise and scattered over the rest of the surface. The 
forest growth, which extends along both banks of the igarape, covers 
the site, but is sparser there than in the surrounding area. The 
largest tree was some 60 cm. in diameter, and the undergrowth in- 
cluded spine bushes and cane (pi. 30, a). An impression left by an 
excavation made 18 years prior to our visit was still distinct. The 
edge of the campo lay about 15 meters away, on the opposite side 
of the igarape. 

Two stratigraphic tests were made : cut 1 in the center of the south- 
eastern end and cut 2 near the center of the northwestern rise. Both 
were 2 by 2 meters square and carried down by 15-cm. levels into 
sterile soil. Cut 1 produced black loamy soil in the first two levels, 
becoming gray brown at 30 cm, and light gray below 45 cm. Sterile, 
water-deposited sand was encountered at 52 cm. and continued to 80 
cm., the greatest depth tested. Hard conglomerate lumps of clay, 
often cemented to each other or to the sherds, appeared in the north- 
east corner of the cut below 30 cm. and continued to the bottom. 
The following cultural remains were recovered : 

Level .00-. 15 m.: 1,198 sherds, 8 clay lumps (6 burnt red). 
Level .15-.30 m.: 1,016 sherds, 22 clay lumps (2 burnt; the remainder 
conglomerate of charcoal, bone, and clay). 



176 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 



Level .30-.45 m.: 609 sherds and 18 clay lumps (10 burnt, 8 conglomerate). 
Level .45-. 60 m.: 144 sherds, 11 clay lumps (4 burnt, 7 conglomerate). 

In cut 2, the black loamy soil continued to a depth of 56-58 cm., 
where it changed suddenly to light tan because of sand mixture. ^^ 
Sherds were thickly concentrated in the sandy soil at this point of 
transition. The sandy mixture, containing greenish concretions and 
bone scraps in addition to sherds, continued to 80 cm., where it gave 
way to sterile, water-deposited sand containing many conglomerate 
lumps. The cut produced: 

Level .00-.15 m.: 608 sherds, 3 clay lumps (1 burnt). 
Level .15-.30 m.: 649 sherds, 1 concretion. 
Level .30-.45 m.: 986 sherds, 1 clay lump. 

Level .45-. 60 m.: 2,604 sherds, one reworked sherd 4.5 cm. diameter in- 
completely perforated slightly off center (fig. 50; pi. 
42, d), 6 clay lumps (3 burnt, 1 with parallel depres- 
sions as if pressed against branches). 
Level .60-. 75 m.: 1,730 sherds and 10 clay lumps (2 burnt). 
Level .75-.90 m.: 226 sherds, 8 clay lumps. 




I CM 

Figure 50. — Partially drilled sherd from J-7 — Sip6, Ananatuba Phase. 



» This stratigraphic division between sandy soil and black loam corresponds to a cultural division, the 
intrusion of Mangueiras Phase ceramics. 



m^A^a}^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 177 

SITE J-8 — MAGUAEf 

This small habitation site is 3.5 km. southwest of J-7, in a clump 
of trees about 100 meters in diameter surrounded by open campo 
(fig. 48). A barely perceptible elevation above the surrounding area 
prevents this spot from being inundated during the wet season and 
permits the growth of trees. The underbrush is thin, as at J-7, and 
the trees are with a few exceptions 35 cm. or less in diameter. The 
soil is hard, gray clay. 

In a cut 1.5 by 1.5 meters, begun in the east half of the site, the 
sherds were found to be sparse, small, and in a poor state of preser- 
vation. Since the deposit was less than 15 cm. deep, the cut was 
widened into a test trench to increase the sherd sample. A few small 
sherds were scattered on the surface. Only 127 sherds and 15 burnt 
clay lumps were collected. 

SITE J-9 ANANATUBA 

An extensive stretch of forest, part of the coastal fringe, begins 
1 km. northwest of J-8 (fig. 48). Just inside its limits is a habitation 
site, which sherds on the surface reveal to be roughly circular and about 
20 meters in diameter. The tree growth on this spot is sparse in 
contrast to the heavy vegetation in the adjacent area, and the soil is 
black instead of light gi'ay. The surface of the site has an elevation of 
about 50 cm. above the surrounding terrain, resulting from the 
deposition of refuse. 

A surface collection was made, and a cut 2 by 2 meters was exca- 
vated in 15-cm. levels near the center of the site. The soil was black, 
sandy loam containing hard, irregularly shaped concretions, sometimes 
cemented to sherds, which were extremely abundant and relatively 
large. Sterile gray-brown sand was reached at 50 cm. and white sand, 
also sterile, at 60 cm. This cut produced: 

Level .00-.15 m.: 1,465 sherds, 60 clay lumps (a few burnt, but most hard, 

black clay concretions with orange flecks) , and 1 broken 

conical pottery object (A) . 
Level .15-.30 m.: 1,419 sherds, 30 clay lumps (Like the preceding level), 

and 1 small cylindrical pottery object (B). 
Level .30-.45 m.: 1,360 sherds, 25 clay lumps and a cylindrical pottery 

object (C). 
Level .45-.60 m.: 352 sherds and 10 clay lumps. 

Cylindrical clay objects. — Three small conical to cyHndrical objects 
made of pottery were recovered from successive levels of cut 1. All 
are Ananatuba Plain pottery, solid, circular in cross section and show 
no wear. 

Object A (fig. 51, a) has a flat end and sides that constrict slightly toward 
the opposite end, which is broken off. Maximum diameter 1.7 cm., 
existing length 2.4 cm. 



178 



BUREAU OF AMERICAN ETHNOLOGY 



[bull. 167 






Figure 51. — Cylindrical pottery objects from J-9 — Ananatuba, Ananatuba 

Phase. 



Object B (fig. 51, b) is cjdindrical, rounding to flattened ends. Diameter 

2.4 cm., length 3.3 cm. 
Object C (fig. 51, c) resembles Object B but has a nubbinlike rounded tip 

at one end. Diameter 2.7 cm., length 4.3 cm. 

SITE J-IO SOROROCO 

Another village site is 500 meters north of J-9, inside the edge of 
the same stretch of forest. It is a low mound, measm-ing 50 meters 
north-south and 10 meters east-west, with a small elevation in the 
center 5 meters in diameter and 75 cm. above the rest of the mound. 
The surface is thicldy covered with sororoco, a spiny palm. 

A cut 1.5 by 1.5 meters, controlled m 15-cm. levels, was excavated 
just south of the central elevation. Scattered sherds were encountered 
in the first 5 cm., corresponding to the hummus layer. Beneath this were 
25 cm. of sterile sand. At 30 cm., there was a sudden transition to 
black loam containing sherds and conglomerate concretions, the two 
sometimes cemented together. This condition continued to the 
bottom of the cut, the concretions becoming more abundant until at 
80 cm. they formed a cementlike deposit that had to be broken with 
a pick, but which had sherds embedded in it. A small, complete 
Ananatuba Plain jar (pi. 42, a), 14 cm. tall and 11 cm. in body diam- 
eter, with a direct rim 6 cm. in diameter, curved walls and a pointed 
bottom was found at 55 cm. Gray sand was encountered at 1.00 meter 
and white sand streaked with orange at 1.05 meters. Both sands were 
sterile, lacking concretions as well as sherds. 

A test was made in the central elevation and it was found to be 
caused by a thicker deposition of sand just below the humus layer. 
Since conditions below that level corresponded to those in the cut, 
the sherds from the two excavations were combined, giving the follow- 
ing totals: 



S'Jilfi'f^ ^^° ARCHEOLOGY AT MOUTEZ OF AMAZON 179 

Level .0O-.15 m.: 68 sherds, 2 clay lumps. 

Level .15-. 30 m.: sterile. 

Level .30-.45 m.: 530 sherds, 16 clay lumps (3 burnt). 

Level .45-.60 m.: 498 sherds, 15 clay lumps (4 burnt). 

Level .60-.75 m.: 807 sherds, 26 clay lumps (concretions). 

Level .75-.90 m.: 450 sherds. 

Level .90-1.05 m.: 247 sherds, 1 burnt-clay lump. 

The deposit of sterile sand overlaying the habitation remains 
indicates the possibihty that this site was abandoned as the result of a 
severe flood. Inquiries made as to the closest watercourse brought the 
information that the only igarape in the vicinity is now about 1.5 km. 
to the south. In view of the evidence of waterborne sand on the site 
and of the known impermanent nature of igarape beds, it can be safely 
assumed that this stream has changed its course since Ananatuba 
Phase times. 

Data Fkom Other Investigations 

The only additional information on the Ananatuba Phase comes 
from surface collections made by Peter Hilbert at two sites in central 
Marajo, included in the site numbering as J-19 and J-20 (fig. 47). 
Although only one sample is large enough to be used for seriation, 
these data are important because they extend the known distribution 
of the Phase inland from the north coast toward the center of Mara j 6 
Island. 

SITE J-19 

An Ananatuba Phase village site, J-19, is located 2 km. inland from 
the left bank of the Rio Anajas-miri, 4 km. above its junction with 
the Rio Arari. A surface collection sent to us for classification 
contained 75 sherds, of which 54.6 percent were Ananatuba Plain, 
41.4 percent Sororoco Plain, and 4 percent Sipo Incised. Seriation 
places this site contemporary with J-10 (fig. 56). 

SITE J-20 

On the upper Rio Camutins, just above the last of the Marajoara 
Phase mounds on that stream is an Ananatuba Phase village site 
(J-20). The surface collection contains only 23 sherds, of which 
11 are Sip6 Incised, 7 Ananatuba Plain, and 5 Sororoco Plain. 

Analysis of Materials op the Ananatuba Phase 

Pottery Type Descriptions 

The pottery type classifications are based on the analysis of 13,843 
sherds of the Ananatuba Phase. The detailed descriptions are ar- 
ranged in alphabetical order. 



180 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

ANANATUBA PAINTED 

Pabtk: On Ananatuba Plain; see that pottery type description for details of 

temper, firing, color, etc. 
Sukface: The surfaces are treated like those of Ananatuba Plain with the excep- 
tion that the painted sherds are always well smoothed and usually given a white 
slip on the surface to be painted. 
Form: 

Rim: Generally direct with a square or rounded lip; occasionally the rim 

expands to a slightly rounded lip. Rim is rarely exteriorly thickened. 
Body wall thickness: 5-12 mm.; majority 8 mm. 
Bases: None found with paint; probably rounded as is typical of Ananatuba 

Plain. 
Common vessel shapes reconstructed from sherds: 

1. Shallow to rounded bowls with direct rim and flat or rounded lip. 

Mouth diameter 14-36 cm. Painted in any of the variations listed 
under "decoration" (fig. 52-1). 

2. Bowls with rounded body. Wall thickness gradually increases 1-2 

cm. below the rim giving it a thickness 2-5 mm. greater than that 
of the lower body wall. Mouth diameter 24-32 cm. Paint on 
the rim top, carried over to the upper exterior waU on one (fig. 52-2) . 

l_i_l 1 I 

I 2 3 CM 

Rim Scale 



''I'll' 

4 8 12 CM 

Vessel Scale 





Figure 52. — Rim profiles and vessel shapes of Ananatuba Painted, Ananatuba 
Phase (Appendix, table 24). 

Rare vessel shapes: 

1. Bowls with rounded carination and slightly everted rim with a rounded 

lip. Rim diameter 24 cm. Paint is a horizontal stripe on the 
interior of the body. 

2. Jars with globular body, constricted mouth and exteriorly thickened 

rim. Mouth diameter 30 cm. Paint on exterior of rim. 
Decoration: 

Technique: Bright red coloring is applied in a paper-thick coat to the slipped 

or well-smoothed surface. 
Motif: The paint was applied in the following variations: 
Solid areas: 

Rim top only — 6 sherds. 
Rim top and upper exterior — 1 sherd. 
Rim top and interior — 4 sherds. 
Interior — 5 sherds from body of vessels. 



"v^ANsf^ AND ARCHEOLOGY AT MOUTH OF AMAZON 181 

Patterns: 

Diagonal stripes of varying width — 2 sherds, 1 on the exterior, and 

1 on the interior. 
Horizontal stripe 1-2 cm. wide on the interior wall — 10 sherds. 
TEMPOBAii DIFFERENCES WITHIN THE ttpe: None (Appendix, table 24). 
Chronological position of the type: Present throughout the Ananatuba 
Phase, but never exceeding a fraction of a percent in frequency. 

ananatuba plain 
Paste: 

Method of manufacture: Coiling, coils 1.5-2.5 cm. wide. 

Temper: Ground sherd with most of the particles small, under 2 mm. 

Texture: Good mixture of clay and temper particles, well-kneaded into a 
compact mass. Pinhole air pockets frequent from air bubbles in the wet, 
moist clay during manufacture. Good tensile strength, broken edge very 
rough and angular. Good clear ring when knocked together. A sharp 
line often distinguishable between the floated surface and the core. 

Color: A gray core with thin (0.5-1.0 mm.) tan to white-tan bands on both 
surfaces in 85 percent of the sherds. Light-tan core in the remainder. 
Lighter particles of ground sherd temper often speckle the gray core. 
Black specks are sometimes present. 

Firing: Incomplete, oxidized firing; 25 percent of sherds have fire clouds. 
Surface: 

Color: Exterior and interior — On 75 percent of the sherds the exterior and 
interior range from light, dull tan to a tannish white to a cream to a 
grayish white. On the remainder, the interior is light gray and the exterior 
one of the above-mentioned shades. Erosion of surfaces and exposure 
of the light-gray core gives many sherds a false gray surface color. 

Treatment: There is no absolute correlation between the exterior and interior 
but sherds well polished or better smoothed on one surface are often 
better smoothed than usual on the other surface also. Exceptions are 
probably from large bowls or jars where one surface was emphasized. The 
surface treatment ranges from irregular, uneven surfaces mth only hand 
smoothing to scraped or smoothed surfaces with smoothing tracks to well- 
floated, semipolished surfaces with a velvety feel. Floating blends into 
slipping, which is unmistakable on some sherds as an added layer in cross 
section. Condition of preservation of sherd surfaces closely correlates 
with the degree of smoothing; otherwise the whole surface is badly pitted 
and often gone. 

Hardness: Soft and easily scratched with the fingernail; 2.5. 
Form: 

Rim: Direct, exteriorly thickened, or everted with rounded or flat lip (pi. 35). 

Body wall thicknsss: 5-13 mm. ; majority 8 mm. Well-smoothed sherds range 
from 5-9 mm. 

Body diameter: 20-50 cm. except for a few miniature vessels. 

Base: Rounded and unthickened on bowls; slightly thickened and more 
pointed on jars. 

Common vessel shapes reconstructed from sherds: 

1. Globular-bodied jars with constricted mouth, direct rim, rounded or 
flattened lip. Mouth diameter range 8-34 cm.; majority 12-26 cm. 
Occasional exterior thickening just above the maximum diameter 
produces an angular shoulder (fig. 53-1). 
391329—57 14 



182 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 




FiGtTRE 53. — Rim profiles and vessel shapes of Ananatuba Plain and Sororoco 
Plain, Ananatuba Phase (Appendix, tables 23 and 26). 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 183 

EVANS] 

2. Jars with globular body, short vertical or slightly everted neck, 

direct rim with rounded or flattened lip. Rim diameter range 
8-40 cm.; majority 10-22 cm (fig. 53-2). 

3. Jars with rounded body, upper walls insloping to direct rim with 

flattened lip. Rim diameter 14-36 cm. Three to four unsmoothed 
coils sometimes ornament the exterior just below the rim (fig. 53-3). 

4. Globular-bodied jars with constricted mouth and exteriorly thickened 

rim with a rounded lip. Mouth diameter 12-26. Rim cross 
section is 1.5 cm. thick or less (fig. 53-4). 

5. Globular-bodied jars with constricted mouth and heavy, exteriorly 

thickened rim and rounded lip. The cross section of the rim is 
more than 1.5 cm. in thickness. Mouth diameter 12-20 cm. 
(fig. 53-5). . 

6. Wide-mouthed jars with rounded body, exteriorly thickened rim and 

rounded lip. Maximum body diameter not more than 4 cm. greater 
than the exterior rim diameter. Rim diameter 8-36 cm. (fig. 53-6) . 

7. Globular-bodied jars with a collarlike, everted rim and rounded lip. 

Rim diameter 16-30 cm. (fig. 53-7). 

8. Bowls with rounded bottom, walls curving outward or upward to a 

direct rim with rounded, flattened or pointed lip. Rim diameter 
20-40 cm.; depth 6-15 cm. (fig. 53-8). 

9. Bowls with rounded bottom, sides curving outward then upward to 

an everted rim with rounded or flattened lip. Rim diameter 16-36 
cm.; depth 5-9 cm. (fig. 53-9). 
Rare vessel shapes: 

1. Miniature vessels: 1 percent of the sherds are from miniature jars 
(pi. 42, a) and bowls usually 4-5 cm. in diameter and 5-10 cm. high 
with the typical vessel shapes and rim forms of the type. 
Appendages: Loop-shaped handles with an oval cross section (1.3 by 1.8 cm. 
ranging to 2.0 by 3.5 cm.) are made as a separate unit with a round extension 
at each end and attached vertically to the vessel wall by punching two 
holes for the plug insert (fig. 53; pi. 35). Edges then smoothed over, but 
the method of construction is clearly visible in cross section. The loop 
handles range from 5-10 cm. in length and project 3-6 cm. from the body 
wall. 
Temporal differences within the type: 

Specks of black ash are abundant in the paste of about 50 percent of the 
sherds from J-7, cut 2, levels 30-45 cm., 45-60 cm., and 60-75 cm., and 
in one sherd from J-10, cut 1, level 90-105 cm. (lowest level). Microscopic 
examination suggests that this is a natural characteristic of the local clay, 
which would explain its occurrence in the pottery from only one site. The 
sherd from J-10 may be from a vessel taken from the old village of J-7 to 
the new one. 
A few temporal differences can be discerned in the frequency of vessel shapes. 
Jars of shape 7 are restricted to the earliest site. Shapes 1, 4, and 9 show 
a slight decline in frequency, while shape 8 is on the increase (Appendix, 
table 23). 
Chronological position of the type: Present throughout the Ananatuba 
Phase as the most abundant type, but shows a decrease in frequency from the 
early to late part of the sequence. 



184 



BUREAU OF AMERICAN ETHNOLOGY 



[bull. 167 



CABMO BRUSHED 

Paste and surface: On Ananatuba Plain with a small minority on Sororoco 
Plain; see those pottery type descriptions for details of temper, firing, color, etc. 
Form: 

Rim: Majority thickened on the exterior with a rounded lip; a few direct, 

with rounded or angular lip. 
Body wall thickness: 7-9 mm., majority 8 mm. 
Base: Rounded with slight thickening at the center on the interior. 
Common vessel shapes reconstructed from sherds: 

1. Jars with large, globular body, 32-36 cm. in diameter, with a con- 
stricted mouth and exteriorly thickened rim. Mouth diameter is 
23-34 cm. (fig. 54-1). 






4 8 12 CM 
Vessel Scale 



I 2 3 CM 
Rim Scole 



Figure 54. — Rim profiles and vessel shapes of Carmo Brushed, Ananatuba 
Phase (Appendix, table 24). 



2. Jars with globular body, short vertical or slightly insloping neck, 4 

cm. high, and a direct rim with rounded or flattened lip. Body 
diameter ranges 32-36 cm.; rim diameter 13-22 cm. (fig. 54^2). 

3. Shallow bowls with rounded bottom, outsloping sides, direct rim and 

rounded or flattened lip. Rim diameter 14-38 cm., majority 
26-34 cm. (fig. 64^3). 
Rare vessel shapes: 

1, Deep bowls with vertical or slightly insloping sides and exteriorly 
thickened rim. Rim diameter 16-18 cm. 
Decoration (pi, 36): 

Technique: Lines not evenly spaced or regularly parallel, indicating that the 
tool was an irregular bunch of twigs. Spacing 0.5-4.0 mm. apart, depth 
0.5-2.0 mm. 
Motif: Parallel brushings on the exterior, executed horizontally, diagonally 
or vertically to the rim. The horizontal variety is typical of bowls while 
the vertical or diagonal type is most often found on jars. Brushing ex- 
tends over the entire body, including the base. A minority of the sherds 
were brushed twice in different directions, giving a criss-cross pattern. 



BV°A°Nsf^ *^° ARCHEOLOGY AT MOUTH OF AMAZON 185 

Temporal differences within the type: Criss-cross brushing occurs at J-9 
and J- 10 but is absent at 3-7. Vessel shape 1 dechnes and shape 3 increases 
in popularity throughout the Ananatuba Phase sequence (Appendix, table 24) . 

Chronological position of the type: Carmo Brushed is present throughout 
the Ananatuba Phase, expanding in popularity toward the middle of the 
sequence. 

8Ip6 incised 

Paste and surface: On Ananatuba Plain except that the surfaces of this incised 
type are somewhat better finished and always smooth. See the type descrip- 
tion of Ananatuba Plain for details of temper, firing, color, etc. 
Form: 

Rim: Majority direct with rounded or angular lip, some with exterior 

thickening and rounded lip. 
Body wall thickness: 6-10 mm., majority 8 mm. 

Base: Body wall curvature on the incised sherds and Ananatuba Plain 
sherds indicate the base was rounded, but no base sherds are included in 
Sip6 Incised since the decoration was limited to the upper body. 
Common vessel shapes reconstructed from sherds: 

1. Jars with globular body, walls incurving to a direct rim with a rounded 

or flat lip. Mouth diameters range from 22-28 cm. at Site J-7, 
from 16-25 cm. at Site J-10. Body diameter is 26-29 cm, (fig. 
55-1). 

2. Jars with globular body, walls incurving to an exteriorly thickened 

rim with a rounded lip. Rim diameter 14-28 cm, (fig. 55-2). 

3. Jars with short, vertical or insloping necks and direct rim with rounded 

lip. Rim diameter ranges 18-34 cm., majority 18-28 cm. (fig. 
55-3). 

4. Deep bowls with outcurving sides, direct rim and rounded or flattened 

lip. Rim diameter 12-28 cm. (fig. 55-4). 
Rare vessel shapes: 

1. Bowl with sloping side walls joining rounded base at a pronounced 

angle. Mouth diameter 34 cm., body diameter at angle 25 cm., 
vertical height above angle 6,5 cm,, estimated total height 10 cm. 

2. Bowls with outcurving sides and everted, flat-topped rim with 

rounded or pointed lip. Diameter 12-16 cm, 

3. Jars of common shape 2 but with a very heavy, exteriorly thickened 

rim. Rim diameter 14-34 cm. 
Decoration: 

Technique: Band of incision applied on the upper exterior surface between 
the maximum diameter of the vessel and the rim edge. Lines are typically 
cleanly made, U-shaped cuts, done with a blunt, round-ended tool when 
the surface of the clay was leather hard. Majority of incisions are 1.5 to 
3.0 mm. wide and 0.5 to 1,0 mm, deep. Some of the motifs combine this 
type of line with fine crosshatching made with a sharp-pointed tool leaving 
a mark about the width of a fine pencil line. Although the total effect of 
the designs is pleasing and suggests regularity, close examination reveals 
the lines to be somewhat unevenly applied, with overlapping strokes and 
unequal spacing. 
Motifs: The designs can be classified under 7 major types: 

1. Inverted scallops just below the lip (pi. 37). The width of the 
scallop is generally uniform on a single vessel, but has a range of 



186 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 








III I I 

I 2 3 CM 
Rim Scale 




r\ 



'It'll' 

4 8 12 CM 

Vessel Scale 



Figure 55. — Rim profiles and vessel shapes of Sip6 Incised, Ananatuba Phase 

(Appendix, table 25). 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 187 

EVANS] 

variation within the type of 1-3 cm. The typical arrangement is a 
single row placed 0.5-1.5 cm. below the rim edge. The area im- 
mediately below is occasionally covered with parallel, diagonally 
incised lines. 

2. Zoned, fine Crosshatch (pi. 38). Irregularly shaped areas of fine 

Crosshatch with straight, stepped or scalloped boundaries are 
defined by broad incised lines. These are alternately left blank and 
filled with fine crosshatching, done with short strokes so that the 
lines overlap. The design occupies a band around the shoulder of 
the vessel and is set off at the upper and lower edge by a hori- 
zontally incised line, the upper one being about 1 cm. below the 
rim edge, 

3. Zoned, large Crosshatch (pi. 39, a-b). Similar to type 2 but with 

the crosshatching composed of broad incised lines like those used 
to outline the zones. 

4. Diagonal Crosshatch, unzoned (pi. 39, c-e). A band of lines incised 

diagonally in one direction around the vessel circumference and 
crossed by a similar number of lines running diagonally in the 
opposite direction. The area they occupy may or may not be 
bounded above and below by a horizontally incised line. 

5. Zoned parallel lines (pi. 40). Zones of incised parallel Hues, usually 

stepped, alternating with unincised zones. The band they occupy 
on the upper part of the vessel is demarcated by horizontal incised 
lines at the upper and lower limits. 

6. Unusually broad, parallel lines(pl.41,a-rf). Parallel lines assuming 

the appearance more of shallow scraping than true clearcut incision. 
The motifs are dominantly rectilinear, often triangular. 

7. Row of circles along the rim (pi. 41, e). Small, somewhat uneven 

and irregularly spaced circles in a row along the rim edge. This 
variety is rare, represented on only 2 sherds. 
Temporal differences within the type: All of the design types are present 
only at J-7, where the execution is also the best. Types 4 and 7 are absent 
at J-10; types 1 and 7 are absent at J-9. Only types 5 and 6 are represented at 
J-8, possiblj'^ because of the small sherd sample from that site. No trends are 
evident in vessel shape (Appendix, table 25). 
Chronological position of the type: Present through the Ananatuba Phase 
with a slight increase in frequency. 

SOROROCO plain 

Paste: 

Method of manufacture: Coiling; coils range from 1.5-2.5 cm. wide. 
Temper: Ground sherd with most of the particles small to moderate in size; 

no hunks. 
Texture: Good mixture of clay and temper particles giving a well-kneaded 

compact mass. Sherds hard to break and leave an irregular, angular edge. 

All sherds have a clear ring when knocked together. 
Color: Orange-tan to pinkish-orange core in 25 percent of the sherds. Others 

have a thin (0.5-1.0 mm.), pinkish-orange band on both surfaces with 

a uniform, light-gray core. Lighter-colored particles of sherd temper 

often speckle the gray core. Some sherds are speckled with black. 
Firing: Oxidized firing, more complete than in Ananatuba Plain; only a 

few fire clouds. 



188 BUREAU OF AMERICAN ETHNOLOGY [bdll. 167 

SUEFACES: 

Color: 

Exterior — A light pinkish orange to dull, deep red, the latter color 

occurring on only 5 percent of the sherds. 
Interior — Usually a light to dark gray. Only 5 percent of the total 
sherds are pinkish orange or red on both surfaces. This pottery 
type is easy to distinguish from Anjos Plain by the lightness and 
pinkishness of the orange. 
Treatment: Exterior and interior — Majority are smoothed on the exterior 
and interior with the surfaces slightly irregular and uneven; only a few 
sherds (less than 1 percent) have the surfaces floated. All the surfaces 
tend to be badly pitted and easily eroded due to the poor surface finish. 
Hardness: Easy to scratch with the fingernail; 2.5. 
Form: 

Rim: Direct, exteriorly thickened or everted with a rounded or flat lip. 

Body wall thickness: 5-13 mm., majority 8 mm. 

Body diameters: Range from 20-50 cm. 

Base: Typically rounded and unthickened; a few with slightly thickened, 

blunt, pointed bases. 
Vessel shapes: The same range of shapes as Ananatuba Plain (fig. 53), but 
with different relative frequency. See type description of Ananatuba Plain 
(pp. 181-183) for details and the "Ceramic History of the Ananatuba Phase" 
(p. 191) for the discussion of forms. 
Temporal differences within the type: 

Black ash particles are present in about three-fifths of the sherds from J-7, 
cut 2, levels 30-45 cm., 45-60 cm., and 60-75 cm.; and in about one-half 
of the sherds from cut 1, levels 15-30 cm., 30-45 cm., and 45-60 cm. 
None were noted in the samples from J-9 and J- 10. Examination under 
a microscope indicates that this is a natural characteristic of the clay 
source used during the occupancy of J-7, rather than a conscious addition. 
Its seemingly greater abundance in Sororoco Plain as compared to Anana- 
tuba Plain probably results from a difference of firing of the two types. 
A few trends are discernible in vessel shape. Shape 7 is found exclusively 
in the earliest site, and shape 1 declines in frequency from early to late. 
Shapes 2 and 5 exhibit a sharp increase toward the end of the Ananatuba 
Phase sequence (Appendix, table 26). 
Chronological position of the type: Present at all sites and showing an 
increase in frequency from the early to the late part of the Ananatuba Phase 
sequence. 

unclassified decorated 

The majority of the sherds in this category are either too small or too badly 
eroded for classification. Those that are well preserved are either unique or too 
rare to warrant the creation of a separate decorated type. They represent three 
techniques: incising or brushing, relief, and punctation. 
Incised sherds: 

From Site J-7 — Sip6: 3 sherds with light scratches; 2 with deep grooves; 

6 with indistinct designs. 
From Site J-9 — Ananatuba: 2 sherds with faint scratches; 1 with deep 
grooves made when clay was very wet; 6 with faint and irregular incised 
lines (possibly a crude variety of Sip6 Incised); 1 with nicks along the 
exterior rim edge and faint incisions on the exterior. 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 189 

BVANS J 

From Site J-10 — Sororoco: 3 from the same jar are ornamented with a row 
of broad grooves 1.5-2.0 cm. long, placed diagonally around the neck; 13 
with faint or badly eroded incised designs. 

From Site J-8 — Maguari: 3 with incised designs (probably Sip6 Incised). 
Modeled sherds: 

From Site J-7 — Sip6: 3 sherds with a raised ridge on the exterior that looks 
like an unsmoothed coil. 

From Site J-9 — Ananatuba: 1 sherd with a modeled knob. 

From Site J-8 — Maguari: 1 sherd with pinched surface superficially resem- 
bling corrugation (pi. 42, e). 
Punctate Sherds (pi. 42, b-c): 

From Site J-7 — Sip6: 1 sherd with a relief rib, 2.0 cm. wide and 1.6 cm. high, 
is covered with punctates, 2 mm. in diameter, spaced irregularly 1-4 mm. 
apart. 

From Site J-9 — Ananatuba: 2 sherds from a deep bowl with the exterior 
covered with horizontal rows of generally triangular punctates; 1 with 
6 rows of shallow, generally oval punctates occupying a broad interior rim 
thickening, and with the rim and interior painted red; 1 with rows of fine, 
evenly spaced punctates that may have been made with a dentate tool. 

Pottery Artifacts 

Objects of pottery other than vessels are exceedingly rare in the 
Ananatuba Phase. Three small, cylindrical objects of fired clay (fig. 
51) came from three successive levels at Site J-9. No wear is visible 
and their use is unknown. A worked sherd (pi. 42, d; fig. 50) came 
from Site J-7 and was probably a spindle whorl. 

Nonceramic Artifacts 

No chips, fragments,, or objects of stone, bone, or shell were found 
at any Ananatuba Phase site, except for an occasional small iron con- 
cretion that appears to be a natural inclusion in the local soil. 

Ceramic History 

The four strata cuts and the surface collections produced 13,483 
sherds of Ananatuba Phase pottery types, which were analyzed by 
levels and seriated to give the sequence shown on the adjacent chart 
(fig. 56). Prior to the Mangueiras Phase intrusion in the upper levels 
of J-7 — Sip6, the seriation gives a relatively smooth picture of ceramic 
change, in which the whitish-surfaced, gray-cored Ananatuba Plain 
declines from 93 percent of the total sherds in the lowest level at J-9 — 
Ananatuba to 48 percent in the upper level at J-10 — Sororoco. In 
this same period, Sororoco Plain, a pink- to orange-surfaced ware, 
increases from 6 percent to 49 percent (Appendix, tables 21 and 22). 

The attempt was made to subdivide Ananatuba Plain into a 
polished or slipped and an unpolished variety. The resulting per- 
centages gave the polished type a frequency of 5 percent at J-9, 



190 BUREAU OF AMERICAN ETHNOLOGY [bull, ist 

4 percent at J-10, and 21 percent at J-7. This seems likely to be a 
reflection of differences in soil rather than differences in manufacture, 
however, when it is considered that the soil at J-9 and J-10 contained 
a high percentage of clay, which eroded the surfaces of the sherds at 
these two sites badly, whereas the sandy soil at J-7 left the surfaces 
well preserved. Added to this difficulty is the fact that the gradation 
between unpolished and polished is so gradual that it was often 
impossible to decide into which category a sherd should be put. 
Until evidence from a larger number of sites is available, it seems best 
to consider this variation as a more careful finishing applied to the 
surface of a small percentage of Ananatuba Plain vessels. 

The decorated type diagnostic of this Phase is Sip6 Incised, in which 
boldly drawn, broad, incised fines are applied to the exteriors of bowls 
and small jars in a variety of patterns ranging from a simple, scalloped 
line to a complex, zoned band. The popularity of this type grows 
from 0.5 percent at J-9 to between 1 percent and 3 percent at J-10 
and the first occupation of J-7 — Sipo. It reaches its peak during the 
Mangueiras Phase occupation of J-7, during which time it increases to 
from 8 percent to 25 percent of the total Ananatuba Phase wares. 
This is also the period of the greatest variety and best execution of 
the incised designs. 

The other important decorated ware in the Ananatuba Phase is 
Carmo Brushed, which differs only in paste from many sherds of 
Croari and Bacuri Brushed associated with the Mangueiras Phase. 
From an occurrence of 1 percent or less at J-9, it increases to from 
4 to 6 percent in the lower levels of J-7 and continues thereafter to 
fluctuate between 1 percent and 4 percent until it disappears just 
before the end of the Ananatuba Phase. 

Sherds of Ananatuba Painted occur scattered throughout the Phase 
in the amount of a fraction of a percent in the levels where they 
appear. It is possible that painting was originally present in greater 
frequency, but has not survived the erosion to which the sm-faces of 
the sherds were subjected in the soil. On the existing samples, the 
red paint was applied either on the rim top or edge, or in parallel, 
diagonal, or horizontal bands on the body. Ked-painted sherds occur 
with greatest f requeue}^ at J-7, and as in the case of polished surfaces 
on Ananatuba Plain, this may be a reflection of less hostile soil condi- 
tions, rather than of an actual increase in the popularity of the pamted 
technique. 

An analysis of the vessel and rim shapes characteristic of Ananatuba 
Plain and Sororoco Plain and calculation of their percentage frequen- 
cies reveals remarkably little change in popularity of various forms 
(Appendix, tables 23 and 26) in spite of the relatively long period of 
time that the sequence appears to represent. It is also interesting to 



J -7 S 



r I o-.if 

20-.!" 
2M5-^ 
2=.30 
L 1= .15- 



AS 



.30- 
.45- 
J -19— SURF 
I -lOi .60- 
.75- 
L .90-1 
J-8 — SURF 

r i=.3oi 



J-7 



s; J -9 

o 



2 = .4g 

2 .6( 

2- .7^ 

L \- .a5 
o-.i 

.15- 
.30- 
L .45- 



les 21 and 22). 



ANJOS 
PLAIN 



ESPERANQA BACURI 
RED BRUSHED 



L_i I I I 

20 40 % 



s J 




\^ ^ ^ / X 



\/ ^ / / ^ / \ 

\ / / / / /-r-i 



MIIIIIIIIIIIIIMIIIII 

II i] 1 1[ 1 1 1 1 1 1 1 1 II nrn 

iiiiiiiimiii 

iiiiiiiiiiiiiiiiiiiiiiii 

ANANATUBA 
PLAIN 



\ / / / /-^ 



\/ / / \ 

1/ / /I 



Hill 


1 


mil 


1 


m 


1 


m 


1 


SOROROCO 
PLAIN 


SIPO 
INCISED 



CARMO ANANATUBA 
BRUSHED PAINTED 




MANGUEIRAS 
PLAIN 



ANJOS 
PLAIN 



ESPERANQA BACURI 
RED BRUSHED 



l_j_ 



20 40% 



Figure 56. — Seriation of Ananatuba Phase sites on the basis of pottery type frequency (Appendix, tables 21 and 22). 



CTANsf^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 191 

note that although there are over four times as many classifiable 
rims of Ananatuba Plain (1,172) as there are of Sororoco Plain (283), 
the same vessel forms are represented in both, and the Ananatuba 
Plain shapes show no more consistent or clearly defined trend than 
do those of Sororoco Plain. 

A comparison of the ratio of bowls to jars in the two plain wares 
shows that 46 percent of Ananatuba Plain rims are from bowls, in 
contrast to 15.5 percent of Sororoco Plain rims. Since in actual 
numbers Ananatuba Plain jars are always more abundant than those 
of Sororoco Plain, the increasing frequency of Sororoco Plain reflects 
an increasing tendency for jars to become oxidized during firing, 
whereas bowls are relatively unaffected. 

Of the nine vessel shapes, all are not equally abundant in both 
wares. Jars of shapes 2 and 5 are relatively more frequent in Sororoco 
Plain than in Ananatuba Plain and in fact represent the most com- 
mon shapes in the former ware. As mentioned above, bowls (and 
particularly shapes 8 and 9) are the dominant form in Ananatuba 
Plain. Jar shape 6 is the only form that shows a decline in Ananatuba 
Plain and a corresponding increase in Sororoco Plain; other shapes 
appear to run about the same course in both wares and to be inde- 
pendent of the shift in popularity between the two wares. The only 
shape with marked temporal significance is jar shape 7, which occurs 
only at the earliest site, J-9 — Ananatuba. 

The site of J-7 — Sip 6 presents two unusual features which raise 
special problems of interpretation: First, the upper levels cannot be 
seriated reliably into the Ananatuba Phase sequence because they 
represent a time when the culture was being subjected to strong 
disruptive pressures that had the effect ceramically of suppressing or 
selecting certain of the pottery types and thus altering the normal 
trend, and second, the size and shape of the site are somewhat differ- 
ent from the pattern exhibited by the others in the Ananatuba Phase. 
A closer analysis permits some interesting deductions about the 
history of J-7. 

Two kinds of evidence are provided by the site, one from the 
density of the sherds in the levels and the other from the position of 
the levels in the pottery seriation. The arrival of the Mangueiras 
Phase divides both of the stratigraphic cuts in the middle. Cut 1 
has 2 levels producing only Ananatuba Phase pottery and 2 with 
Mangueiras Phase mixture; cut 2 has 3 levels in each category. The 
levels in the two cuts that contain Mangueiras Phase wares produce 
an almost equal density of sherds in spite of the 15 cm. difference in 
depth of deposit. The upper two levels of cut 1 contained 2,214 
sherds as against 2,243 from the upper three levels of cut 2. How- 
ever, in the lower levels the sherd density is six times greater in cut 2 



192 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 



(4,560 sherds) than in cut 1 (753 sherds), suggestmg a considerably 
greater intensity of habitation on Mound 2 (cut 2) during the pre- 
Mangueiras Phase period at J-7. 

An examination of the position that the lower levels of the two cuts 
at J-7 occupy in the pottery type seriation (fig. 56) shows that the 3 
levels of cut 2 fit between the 2 levels of cut 1. The fact that the 
vast majority of the sherds from this part of cut 1 come from the 
upper level (30-45 cm.) suggests that the major pre-Mangueiras 
Phase occupation of Mound 1 (cut 1) is not contemporary with that 
at Mound 2, but instead subsequent to it. In other words, both the 
sherd density and the seriation evidence indicate that the original 
village at J-7 was on the site of Mound 2, and that prior to the advent 
of the Mangueiras Phase people there was a removal or expansion to 
the site of Mound 1. 

Mangueiras Phase wares replaced those of the Ananatuba Phase 
with almost the same rapidity in the two parts of the site (table D), 

Table D. — Relative frequency of Ananatuba Phase and Mangueiras Phase wares 

at J-7—Sip6 



Level (em.) 


Ananatuba Phase 
wares 


Mangueiras Phase 
wares 




Cutl 


Cut 2 


Cutl 


Cut 2 


0-15 


Percent 

5.6 

43.4 

100.0 


Percent 

11.4 

22.6 

38.5 

100.0 


Percent 
94.5 
56.6 



Percent 
88.6 


15-30 


77.4 


30-45 


61.5 


45-60- 














which leads to the conclusion that the population increase brought 
about at this time required the reoccupation of the older part of the 
village in addition to the facilities of the newer part. An equal 
intensity of habitation is also indicated by the equality of sherd 
density in the two cuts during this period. 

It remains to decide what the temporal relationship is between 
J-7 and J-10. The ceramic seriation places J-10 following the pre- 
Mangueiras Phase period at J-7 (fig. 56). However, the trends in 
vessel form of Ananatuba Plain and Sororoco Plain argue strongly 
for the position of J-10 preceding J-7 (Appendix, tables 23 and 26). 
The problem is rendered more difficult by the fact that the upper 
part of J-7 evades seriation in the Ananatuba Phase sequence because 
of the heavy Mangueiras Phase mixture and the apparent distortion 
of the normal ratio of frequency in the manufacture of Ananatuba 
Phase pottery types, particularly Sipd Incised (fig. 56) . Furthermore, 
although the ceramic seriation separates J-7 into two parts, broken by 
the period of habitation of J-10, the evidence from vessel shape 



"v^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 193 

popularity suggests that this does not represent an abandonment of 
J-7. Certain vessel shapes, best represented by Sororoco Plain shapes 
2 and 5, maintain nearly even strength throughout the existence of 
J-7 but are absent or notably rarer at J-10. If J-7 had been aban- 
doned for J-10 and the people had later returned, they would have 
carried on the vessel shape tradition that had developed at J-10 
rather than reverted suddenly to the tradition they had when they 
left J-7. The frequency of shapes 1 and 3, although fluctuating, 
also presents the smoothest curve when J-10 precedes J-7 rather 
than in the sequence indicated in the pottery type seriation. 

The alternative possibilities are: (1) The pre-Mangueiras Phase 
period of J-7 precedes J-10, and J-7 was later reoccupied; (2) J-10 
precedes J-7 ; (3) the two sites are partly contemporaneous. Against 
the first conclusion is the analysis of vessel-shape trends discussed 
above. The trends in pottery types seem equally to preclude the 
second possibility. The lower levels of J-7 have notably more 
Ananatuba Plain and less Sororoco Plain than is found at J-10 and 
fit into the seriated sequence only when placed between J-9 and J-10 
(fig. 56). These contradictions can be somewhat resolved by falHng 
back on the third possibility, namely that the two sites are partly 
contemporary, but this too raises difliculties. The absence of Man- 
gueiras Phase influence at J-10 appears to rule out the existence of 
J-10 subsequent to the pre-Mangueiras Phase period at J-7. In 
view of the proximity of these two sites and the apparently continu- 
ous contact between Mangueiras Phase sites occupying the surround- 
ing area and separated by considerably greater distances (demon- 
strated by the rapid diffusion of Pseudo-Sip6 Incised and of certain 
Ananatuba Phase vessel shapes after their adoption by the Man- 
gueiras Phase population at J-7), it is impossible to believe that J-10 
could have escaped the fate that befeU J-7 if it had been inhabited 
during this time. 

Diagnostic Featuees of the Ananatuba Phase 

Ananatuba Phase sites are typically located in the forest, close to 
the edge of the campo and well away from the coast. If there is a 
stream nearby, as in the case of J-7, it functions only as a source of 
domestic water supply and the nearest navigable igarape is about 1 
km. away (except at J-20). The ceramic refuse marking the extent 
of the former village covers an area of 300 to 770 square meters and 
is cu-cular or oval in outline. The deposit is tjrpically 0.60 to 1.00 
meter in depth, with sherds abundant and intermixed with a com- 
paratively small quantity of dirt. No cemeteries were found, nor 
was there any evidence to suggest the manner of disposal of the dead. 



194 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

Ceramically, the Phase is identified by two undecorated wares, 
Ananatuba Plain, with a whitish surface and a gray core, and Sororoco 
Plain, differentiated by the pink-to-orange color produced by more 
definitely oxidized firing. Both types are present in all levels, but 
Sororoco Plain increases Avith the passage of time, providing the basis 
for sedation of the sites. In keeping with the domestic purpose of 
the ceramics, the amount of decoration is small and its execution 
simple. Carmo Brushed, with the surface "brushed" with a bunch 
of small twigs, and Sip6 Incised, with simple but tastefully executed, 
incised designs, are the major decorated types. Both are infrequent 
at the earliest site (J-9), but soon attain a combined frequency of 
10 to 12 percent, which remains relatively constant until late Sip6 
times, when there is a marked increase in the popularity of Sip6 
Incised. 

Ananatuba Phase features of uncertain significance include the 
great abundance of clay lumps and concretions cementing the sherds 
together in the ground. Since these deposits were encountered in no 
other Phase, it seems probable that they are in some way related to 
an unidentified and exclusively Ananatuba Phase practice. A few 
lumps of clay bearing twig impressions may be indicative of wattle 
and daub in the house construction. The only ceramic artifacts 
were 3 small, solid lumps of fired clay, basically cylindrical in form, 
one of which has a small nubbin at one end ; and a sherd reworked in 
the shape of a disk and partially drilled off center. 

The duration of the Ananatuba Phase is unknown, although a clue 
is provided by the exceptional density and depth of the ceramic 
refuse (see pp. 252-253). No evidence of European contact was found, 
substantiating the conclusion drawn on the basis of seriation with 
other Phases on the Island of Maraj6 that it came to an end sometime 
before the advent of Cabral. 

THE MANGUEIRAS PHASE 

Description of Sites and Excavations 

Sites belonging to the Mangueiras Phase have been found on the 
western half of the north coast of Maraj6, on central Marajd and on 
southern Caviana (fig. 145). 

SITE 3-6 — CROARf 

The Kio Croari flows into the Amazon about 8 km. west of the 
town of Chaves, on the north coast of Marajd. It is a continuously 
winding stream, wide at the mouth but narrowing quickly to about 
10 meters, a width that it maintains as far as the site. Both banks 
are heavily forested and abound in bird life. J-5 is about 3 km. from 



MKOGBBS AND ARCHEOLOGY AT MOUTH OF AMAZON 195 

BVANS] 




Figure 57. — Plan of J-5 — Croarl, a habitation site of the Mangueiras Phase. 

the mouth of the river, on the north side of a small tributary a short 
distance above its juncture with the main stream. A large natural 
clearing, one of many small patches of campo that break the con- 
tinuity of the forest on this part of the island, begins just beyond the 
eastern limits. The sharp rise of 1 meter in elevation and the abund- 
ance of sherds on the surface set the site off distinctly from the natural 
terrain. 

Three mounds of equal height compose J-5 (fig. 57). Mound 1, 
34 meters in maximum width and 55 meters long, is on the bank of 
the igarape, the bed of which has accumulated a quantity of sherds. 
The south side of the mound follows the contour of the stream and 
incorporates a small bend, giving it a somewhat comma-shaped 
outline. Mound 2, an oval 52 meters long and 30 meters wide, 



196 BUREAU OF AMERICAN ETHNOLOGY [bcll. m 

adjoins the north edge of Mound 1. Mound 3, 25 meters in diameter, 
is separated from the west edge of Mound 1 by a 2 meter wide de- 
pression. All of the mounds, in common with the adjacent forest, 
are covered with large trees. A caboclo house was located on Mound 
2 at the time of our visit. 

A stratigraphic cut 1 meter square, was begun near the west end 
of Mound 1 and carried down by 15 cm. levels to sterile soU. The 
ground was so interlaced with large roots that digging was difficult 
and the limits of the cut had to be reduced slightly as depth increased. 
In the refuse layer, the clayey soil was black and saturated with sherds 
for the first 30 cm. At that depth the cut passed through a charcoal- 
flecked layer 1 to 2 cm, thick. Below this the sherds were a little less 
numerous, and continued to decrease in frequency with increased 
depth. By level .60-.75 meters the dimensions of the cut had been 
reduced to 50 by 50 cm. by the presence of large roots. A second 
charcoal-flecked layer, the same thickness as the first, was encountered 
at 80 cm. and sterile, yellowish-brown clay began at 85 cm. The cut 
was continued to a depth of 1 meter with no change in the condition 
of the native soil and since this depth was below the flood level, no 
further testing was undertaken. The existing height of the mound, 
therefore, represents the accumulation of refuse during occupation 
rather than an intentional construction. 

Cut 1 produced the following cultural materials: 

Level .00-. 15 m.: 393 sherds, 8 burnt clay lumps, 1 figurine head (fig. 59, a). 

Level .15-.30m.: 267 sherds, 16 burnt clay lumps, 1 figurine body (fig. 
59, b), 1 short cylindrical object (fig. 60), 1 complete 
miniature Croarf Brushed jar (pi. 44). 

Level .30-.45 m.: 187 sherds, 19 burnt clay lumps, 1 mouthpiece of a 
tubular pipe. 

Level .45-.60 m.: 199 sherds, 8 burnt clay lumps. 

Level .60-.75 m.: 253 sherds and 16 burnt clay lumps. 

Level .75-.90 m.: 80 sherds and 19 burnt clay lumps. 

To this material the surface collection added 485 sherds and 1 com- 
plete tubular pipe. 

Pipes. — The two pipes are dark, gray-brown Mangueiras Plain. 
The surface of the fragment from the strata cut is well smoothed, 
while the complete one is somewhat rough, particularly on the inside 
of the bowl (fig. 58, a). In shape and size, the two examples are 
almost identical. Both have a conical bowl tapering down to a small 
hole (6-8 mm. in diameter) thi'ough the flattened mouthpiece. The 
bowl diameter of the complete specimen is 3.0 cm. and the total 
length 6.8 cm. The mouthpiece is oval in cross section, 2.5 by 1.2 
cm. in the complete specimen and 2.6 by 0.9 cm. in the fragment. 

Figurines. — Although the two figm-ine parts were found in succes- 
sive levels, the head appears to be too small to belong with the torso. 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



197 





I CM 



I CM 



Figure 58. — Pottery tubular pipes from Mangueiras Phase sites: a, J-5 — Croarf. 

b, C-3— Porto Real. 

Both are Mangueiras Plain, with a Hght-tan, well-smoothed surface. 
The head (fig. 59, a), flat at the back and convex in front, was modeled 
by pressing several thin pieces of clay onto a small rounded lump. 
The base has a ragged break only around the edge; the central part 
shows the well-smoothed surface of the foundation ball, which was 
too diy when attached to form a strong connection with the body. 
A layer 4 to 5 mm. thick was laid over the core to form the face, and 
the eyebrows and nose are created by a small ribbon of clay that was 
pressed upward onto the greatest diameter of the head, fading into 
the siu'face below and creating a slight convex ridge at the top, ac- 
centuated where the two impressions meet. Secondary working of 
the surface produced a slight prominence in the center of the space 
on each side of the nose to mark the ej^es. A second flat bit of clay, 
2 mm. thick, was placed over the back of the head, its overlapping 
front edge forming the hairline at the top and sides. At each side, 
just above where the ears should be, there is a bun-like projection. 
Hair is realistically indicated by vertical scratches beginning at the 
hah'line and extending to the base of the head. The lower part of the 
face is missing. Existing height is 3.5 cm., width at the two buns 4.0 
cm., and thickness from front to back 2.7 cm. 




I CM 




I CM 



FiGTiRE 59. — Figurine parts from J-5 — Croari, Mangueiras Phase. 

391329—57 15 



198 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 



The torso (fig. 59, b), likewise of solid construction, is broken off at 
the neck and just below the waist. It has two outstretched arms, one 
broken off at the shoulder and the other at about the elbow. An 
applique ridge representing the spine runs up the back from the waist 
to the upper edge. There is a small depression at the navel and two 
slight rises on the chest. On the back and continuing over the top of 
the shoulders is an asymmetrical design, executed with a fine zigzag 
incised technique, that may signify a textile pattern or body painting. 
The torso is 4.5 by 4.0 cm. in diameter and 5.5 cm. in existing height. 

Lahret (?). — Of problematical use is a small, cylindrical object 
(fig. 60) with flat ends and a smooth, light-tan surface showing traces 




i CM 

Figure 60. — Labret (?; fragment from J-5 — Croarf, Mangueiras Phase. 



of red pigment (Esperanga Red). The cross section is almost perfectly 
chcular, 2.5 cm. in diameter. The total height is 1.7 cm. For a 
distance of 6 mm. from one end, the surface is slightly irregular and a 
little lower than that of the remaining 1.1 cm. The junction is a 
ra.gged edge, indicating that something modeled over one end has been 
broken off. The only clue to the function of this object is its slight 
resemblance to labrets from other Mangueiras Phase sites (cf. Site 
C-3, p. 202). 

SITE J- 7 — SIPO 

Originally an Ananatuba Phase site, J-7, was subjected to strong 
Mangueiras Phase influence during the latter part of its existence. 
Of the 6,803 sherds from cut 2, 1,649 are Mangueiras Phase wares, 
and of the 2,967 from cut 1, 1,707 are Mangueiras Phase wares. 
Details of the site and its excavation are given under the Ananatuba 
Phase (pp. 174-176). 



MBOOERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



199 



SITE J-13 — BACURf 

This habitation site is along the west side of the Igarap^ Bacurl, 
about 3 km, from where it flows into the Amazon opposite the western 
tip of Mexiana (figs. 47, 61). The refuse covers a roughly oval area, 
30 meters from north to south and 100 meters from east to west. 
From about 1 meter high at the center, it levels off gradually to the 
east in the du*ection of the iyarape, which is about 250 meters away. 
The south edge is also a gradual slope, but on the north side the eleva- 
tion is abrupt. Forest covers the site and stretches away on all sides, 
broken by occasional small, natural clearings, but there are no large 
trees on the site itself (pi. 30, 6). Small sherds are scattered over the 
surface and occm' in abundance where the soil has been uprooted. 

A stratigraphic cut 1.5 meters square and controlled in 15-cm. levels 
was excavated near the center, at the highest point. The soil was a 
loose, dark-gray, sandy loam throughout the cut, and except in the 
lowest two levels the sherds were very small. Light-gray clay with 
flecks of white was encountered at 65 cm. and although a few sherds 
were embedded in it, these were confined to the upper limit. Below 
that, the soil continued sterile for 30 cm. to the bottom of the test. 
Tabulation by levels gives the following counts: 

Level .00-15 m.: 190 sherds. 

Level .15-.30 m.: 1091 sherds and 1 clay lump. 

Level .30-.45 m.: 777 sherds. 

Level .45-.60 m.: 664 sherds, 3 burnt-clay lumps. 

Level .60-.75 m.: 309 sherds (1 worked). 

The surface collection included 388 sherds and 1 burnt-clay lump. 




eO 50 100 M 



Figure 61. — Plan of J-13 — Bacurf, a habitation site of the Mangueiras Phase. 



200 



BUREAU OF AMERICAN ETHNOLOGY 



[BULIi. 167 



SITE J-16 — CANIVETE 



J-16 is in almost the exact center of the Island of Marajo, about 250 
meters from the north bank of the Igarape Nerd, a small tributary of 
the upper Rio Anajas (fig. 47). This part of the island is almost 
equally divided between forest and campo, the forest tending to cover 
the areas where a slight elevation prevents inundation during the 
rainy season. Since the site was visited during the height of the 
rainy season, in the month of May, the reason for its location could 
readUy be perceived. Running in a north-south direction, away from 
the igarape, are two long stretches in which the land is 0.25 to 1.00 
meter higher than the maximum water level (fig. 62). Between and 
surrounding these rises the land was either soft and mucky or flooded. 
On each of the rises, on the part closer to the igarape (i. e., the southern 
part), there is an area with sherd refuse. The first is about 70 meters 
in diameter, and the second is 70 meters long and 15 meters wide. 
The former site, at the edge of the campo, produced sherds to a depth 
of 20 to 25 cm., while in the latter the refuse layer was only 5 cm. 
thick. The campo at this time of the year was covered by a few centi- 
meters to several meters of water, giving the appearance of a marshy 
lake. 



10 E9 50 H 





< 


' > ■ ->■ 


-:r •■■'■■■ 1 . ■■■ 





■ *l«0 - noOOEO OURINO KAINT SEASON 




APPROXIMATE LIMIT OF U^tlNUNDATEO LAND 



AREA OF' MAJOR 
EXCAVATION 



LIMIT OF F0RE3T 



Figure 62. — Plan of J-16 — Canivete, a habitation site of the Mangueiras Phase. 



"va^'nsT^'' ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 201 

Since the deposit was too shallow to give stratigraphic results, a 
sherd sample was secured by an excavation equivalent to a 1.5 by 
1.5 meter area in Mound 1 (the larger and deeper site) and by scat- 
tered smaller tests in Mound 2. The collection from Mound 1 num- 
bered 599 sherds and 28 clay lumps, a few of which were fired. Mound 
2 is represented by 123 sherds and 1 clay liunp. 

SITE C-3 — PORTO REAL 

One Mangueiras Phase habitation site was located on the Island of 
Caviana, 4 km. northeast of a point 5 km. above the mouth of the 
Igarape Pocoat6, which drains into the south coast (fig. 151). The 
entire area is covered with forest, which stretches unbroken by clear- 
ings for several kilometers from the site in all directions. The habita- 
tion area occupies a roughly circular space, about 25 meters in diam- 
eter, on the south edge of a slight natural rise. The lower land stretch- 
ing off to the south is covered with up to a meter of water during the 
rainy season. The nearest surface water during the dry season at the 
present time is the Igarape Pocoat6. The soil color on the site is 
distinctly darker than in the surrounding area, and small sherds were 
scattered over the surface. 

Two stratigraphic excavations were undertaken: Cut 1, 5 meters 
from the eastern edge of the site, and cut 2 north of the center. Both 
were 1 by 1 meter and controlled in 8 cm. levels. The soil in both cuts 
was black and well loosened by root action. Sterile soil, a light-gray 
clay with orange flecks, was reached at 38 cm. in cut 1 and at 34 cm. 
in cut 2. Cultural remains recovered totaled: 

Cut 1: 

Level .00-.08 m.: 170 sherds. 

Level .08-.16 m.: 277 sherds, 10 burnt-clay lumps, 1 small stone chip. 

Level .16-.24 m.: 341 sherds, 11 burnt-clay lumps, 1 ceramic labret (?) 

Level .24r-.32 m.: 153 sherds, 5 burnt-clay lumps. 

Level .32-.40 m.: 28 sherds and 1 clay lump. 

Cut 2: 

Level .00-.08 m.: 154 sherds. 

Level .08-. 16 m.: 237 sherds, 2 burnt-clay lumps, 1 broken pipestem, 1 

ceramic labret (?) fragment. 
Level .16-.24 m.: 175 sherds, 3 burnt-clay lumps. 
Level .24-. 32 m.: 147 sherds, 1 incised, biconical, pottery object. 

An additional 1,551 sherds and 35 burnt-clay lumps, and another 
possible ceramic labret were collected from the surface and miscellane- 
ous tests. 

Pipe. — The pipe fragment is part of a tubular pipe similar to those 
from J-5 — Croari, except that it has a round instead of a flattened 
mouthpiece (fig. 58, b) . The ceramic type is Mangueiras Plain, with 



202 



BUREAU OF AMERICAN ETHNOLOGY 



[BDLL. 167 



a light-tan, well-smootlied surface. The sides taper outward from a 
diameter of 8 mm. at the end toward the bowl, which the interior 
contour indicates to have been conical. The existing length is 2.5 cm. 
Labrets. — Three small objects, one from each cut and one from the 
surface collection, possibly functioned as lip or ear plugs. The com- 
plete specimen (surface collection) is collar-button-shaped, with a 
disk 2.4 cm. in diameter, slightly concave on the face, and a short shaft 
widening to a head 1 cm. in diameter, also with a slightly concave sur- 
face (fig. 63, a). Overall length is 1.4 cm. The two fragmentary speci- 
mens (fig. 63, b-c) represent the disk end, with the head broken off. 
The diameters are 2.3 cm. and 3.5 cm. All three are Mangueiras 
Plain, with light-tan surfaces and no trace of decoration. 






I CM 




Figure 63. — Labrets and biconical object of pottery from C-3 — Porto Real, 

Mangueiras Phase. 



Biconical object. — A fragment of an incised object from cut 2, level 
.24-.32 meter, is of undetermined use (fig. Q3,d). It is biconical, taper- 
ing sharply from a diameter of 2.5 cm. toward both ends, which are 
broken off. One surface is plain, the other is lightly incised with fine 
scratches in a series of crudely drawn concentric circles and quadrant 
lines. The ceramic type is Mangueiras Plain. 



Bvi^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 203 

Data from Other Investigations 

SITE J-17 FLOR DO ANAjXs 

Subsequent to our fieldwork on Maraj6, Peter Hilbert of the Museu 
Goeldi made a trip to the vicmity of Lago Arari and located a Man- 
gueiras Phase site underlying the present town of Flor do Anajds 
(fig. 47). This site, designated as J-17, is on the west bank of the Rio 
Aran, 80 meters north of its junction with the Anajas-miri (Ana- 
jasinho). The bank here is about 2.50 meters above the river level 
in the dry season, and remains free from flooding even when the 
surrounding area is inundated. The refuse deposit, mdicated by 
abundant surface sherds, extends about 150 meters along the bank 
and 30 to 50 meters inward. Sherds protrude to a depth of 10 to 20 
cm. in the eroded bank, and a 1.5 meter square excavation near the 
northwest end of the site also produced sherds to a depth of 20 cm. 

The collection, sent to us for analysis, included: 

Level .00-.15 m.: 229 sherds. 

Level .15-.30 m.: 39 sherds, 5 burnt-clay lumps, and one clay ball 2 cm. 

in diameter, possibly the foundation for a figurine head 

(cf. p. 197). 

The surface collection produced 140 additional sherds (Hilbert, 
pers. corres.). 

Analysis of Materials of the Mangueiras Phase 

Pottery Type Descriptions 

The habitation sites of the Mangueiras Phase on Mara] 6 and 
Caviana Islands produced 13,724 sherds. Analysis of these resulted 
in the classification of the following pottery types, named by the 
binomial system and listed in alphabetical order: 

anjos plain 
Paste: 

Method of manufacture: Coiling. 

Temper: Ground sherd, particles in a single specimen ranging from fine 

grains (0.5 mm.) to large hunks (4-6 mm.). Temper usually easily visible 

because it is a lighter orange or tan than the paste. 
Texture: Uneven m.ixture, leaving air pockets around larger temper particles; 

hard to break, edges irregular, granular, and crumble easily. Dull, clayey 

thud when hit together. 
Color: Ranges from light tan to bright, tile orange. A distinct orange to 

tan core is characteristic but about 10 percent have over 75 percent of the 

paste orange with a thin gray core. Light speckles from lighter-colored 

temper are often visible. 
Firing: Complete, oxidized firing; no fire clouds. 



204 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

Surfaces: 
Color: 

Exterior — Typically a tile orange with a diffuse, dusty appearance; 

small percentage are orange tan. 
Interior — Typically a dusty, dark gray with an orange hue; 25 percent 
have the same tile orange to orange tan as the exterior. 
Treatment: Exterior and interior — Usually sherds with better smoothed and 
even-surfaced exteriors are also better finished on the interior. About 
30 percent of all sherds smoothed with a scraping tool and rubbed over 
with the hand, leaving a fairly even and regular surface. Others were 
smoothed over only sufficiently to obliterate the coil lines and have a 
very irregular and uneven surface with pits and channels from dragged 
temper particles. Swipe marks from fingers and hand visible on many 
sherds. 
Hardness: Easily scratched with the fingernail; 2.5. 
Form: 

Rim: Direct or exteriorly thickened with rounded, occasionally flat or 

pointed lip. Occasionally a slight thickening on the interior. 
Body wall thickness: 5-10 mm.; majority 7-8 mm. 
Body diameters: Range 26-60 cm. with the majority of the large jars around 

50 cm. 
Base: Rounded and slightly thickened, usually 2-5 mm. thicker than the 

body wall. 
Vessel shapes reconstructed from sherds: 

1. Globular jars with walls curving inward to a direct rim with a rounded 

lip. Mouth diameter 12-30 cm.; majority 14-20 cm. Maximum 
body diameter 26-60 cm. (fig. 64-1). 

2. Globular-bodied jars with constricted mouth and prominent ex- 

teriorly thickened rim with rounded lip. Rim cross section 1.5 cm. 
thick; rim diameter 16-24 cm. (fig. 64-2). 

3. Jars with rounded body, wall slanting inward to a direct rim with 

squared or rounded lip. Mouth diameter 14-36 cm.; majority 
18-28 cm. (fig. 64-3). 

4. Jars with rounded body, walls insloping to exteriorly thickened rim 

with rounded lip. Rim diameter 14-34 cm.; majority 18-26 cm. 
(fig. 64-4). 

5. Bowls with rounded body, outcurving sides and direct rim with 

pointed or rounded lip. Rim diameter 12-34 cm. (fig. 64-5). 
Appendages: Loop handle with a round cross section ranging from 1-2 cm. in 
diameter. Attached by affixing to the vessel and kneading extra clay 
around the point of attachment. Length 5-8 cm. and extending 5 cm. 
out from the jar wall. Rare and apparently limited to large jars. 
Temporal differences within the type: Vessel shape 4 is the dominant jar 
form in the early part of the Mangueiras Phase and shows a decline in frequency 
throughout time. Shapes 1, 2, and 3 are unimportant until the latter part of 
the Phase. (Appendix, table 29). 
Chronological position of the type: Anjos Plain is absent in the earliest part of 
the Mangueiras Phase sequence but after its appearance it increases in fre- 
quency until the end of the Phase. 



Bv^^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 205 





4 8 12 CM 
VesKi Scalt 




I 2 3 CM 

Rim a Handle 
Seal* 




wvwvv 




VWVs^S 




^z// 




Figure 64. — Rim profiles and vessel shapes of Anjos Plain. Maneueiras Phase 

(Appendix, table 29). 



206 



BUREAU OF AMERICAN ETHNOLOGY 



[bull. 167 





I I I I 1 

I 2 3 CM 
Rim Scale 






L I I I I 1 I 

4 8 12 CM 

Vessel Scale 



Figure 65. — Rim profiles and vessel shapes of Bacuri Brushed, Mangueiras Phase 

(Appendix, table 30). 



Bv^Nsr ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 207 

BACURf BRUSHED 

Paste and surface: This brushed variety occurs on Anjos Plain; see that pottery- 
type description for details of paste, color, temper, etc. 
Form: 

Rim: Majority direct with rounded or angular lips; a lesser number with 

exterior thickening and rounded lip, or interior thickening. 
Body wall thickness: 5-14 mm.; majority 8-9 mm. 

Base: None had brushing on them; hence probably the same as Anjos Plain. 
Vessel shapes reconstructed from sherds: 

1. Jars with globular bodies, interiorly thickened or direct rim with 

rounded or flat lip. Rim diameter 14r-22 cm.; body diameter 
estimated as 40-60 cm. (fig. 65-1). 

2. Globular-bodied jars with constricted mouth and exteriorly thickened 

rim with rounded lip. Mouth diameter 14-20 cm. (fig. 65-2). 

3. Jars with rounded body, walls insloping to direct rim with rounded or 

flat lip. Rim diameter ranges 16-26 cm., majority 18-22 cm. 
Body diameter 32 cm. (fig. 65-3). 

4. Bowls with nearly vertical sides, direct rim with rounded lip. Mouth 

diameter ranges 20-27 cm., majority 20-22 cm. (fig. 65-4). 

Decoration: Brushing with a bunch of twigs on the exterior surface (pi. 43). 
Majority were given a single brushing but 20 percent were brushed a second 
time in a direction diagonal to the first marks, producing a crude cross hatching. 
The striations are 1-3 mm. wide, with the majority 1 mm., and spaced from 
adjacency to 4 mm. apart. Depth varies from 0.5-2.00 mm. (typically 1 mm.) 
and is probably related to the wetness of the surface at the time of apphcation. 
Direction of the strokes is predominantly vertical but about one-third of the 
jars with short necks are brushed horizontally around the neck and vertically 
or diagonally on the body. 

Temporal difference within the type: Bacuri Brushed occurs only in the latter 
part of the Mangueiras Phase. It reaches its greatest frequency in the lower 
levels of Site J-13 and its greatest refinement at Site J-16, where the brushing 
is finer and the lines are more evenly spaced and applied to produce a regular 
cross hatching. No trends are evident in vessel shape (Appendix, table 30) . 

Chronological position of the type: Late type in the Mangueiras Phase, 
possibly a reflection of Ananatuba Phase contact where brushing was one of the 
main methods of decoration. 

CROARf BRUSHED 

Paste and surface: On Mangueiras Plain; see that type for details of paste, 

temper, firing, etc. 
Form: 

Rim: Everted or exteriorly thickened with flattened top and squared, pointed 

or rounded lip. 
Body wall thickness: Range 4—11 mm.; majority 8 mm. 
Bases: Rounded, generally uneven and irregular. 
Vessel shapes reconstructed from sherds: 

1. Jars with globular body, constricted mouth and everted rim, some- 

times exteriorly thickened, with flat or rounded lip. Size ranges 
from miniatures 6 cm. in diameter and 6.5 cm. tall to large vessels 
with a rim diameter of 24 cm. (fig. 66-1) . 

2. Bowls with outsloping sides and slightly everted, flat-topped rim with 

rounded lip; rim diameter 18 cm. (fig. 66-2), 



208 



BUREAU OF AMERICAN ETHNOLOGY 



[bull. 167 



Decoration: Applied by brushing vertically, beginning just below the rim and 
sweeping downward over the upper part of the body, but not extending over 
the base (pi. 44) . Variation from fine scratches to Unes 2-3 mm. wide, according 
to the size of the sticks used. On 64 percent the first brushing was crossed over 
by another producing a crisscross effect. A few have a horizontal line 1-2 cm. 
wide along the upper limit of the marks below the rim, providing an even de- 
marcation for the beginning of the lines. 






I I I I I I I 

4 8 12 CM 

Vessel Scale 



1 2 3 CM 
Rim Scale 



Figure 66. — Rim profiles and vessel shapes of Croarf Brushed, Mangueiras Phase 

(Appendix, table 30). 

Temporal difference within the type: None (Appendix, table 30). 
Chronological position of the type: Croari Brushed occurs only at Site J-5, 
the earliest site in the Mangueiras Phase. 



E8PERAN5A RED 

Paste and surface: On Mangueiras Plain paste; see that description for details 

of temper, texture, firing, and surface treatment. 
Form: 

Rim: Interiorly thickened, direct, everted or exteriorly thickened, with 

square lip. Rarely the lip is rounded, undulating, or scalloped. 
Body wall thickness: 5-10 mm.; majority 7 mm. 
Base: Rounded and slightly thickened. 
Vessel shapes reconstructed from sherds: 

1. Bowls with broad, flat rim top produced by eversion or interior 
thickening, flattened bottom and outslopicg sides. Rim diameter 
18-44 cm.; majority 26-40 cm. Lip square and occasionally 
scalloped (fig. 67-1). 



BV^Nsf^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 



209 






I 2 3 CM 
Rim Seal* 



I I 1 I I I I 

4 8 12 CM 




FiGtTRE 67. — Rim profiles and vessel shapes of Esperanga Red, Mangueiras Phase 

(Appendix, table 32). 



2. Bowls with rounded or flattened bottom, sides curving outward and 

upward to direct rim with expanded, square lip. Diameter 10-40 
cm.; majority 24-40 cm. (fig. 67-2). 

3. Bowls with rounded bottom, vertical sides and everted rim with 

square lip. Rim diameter 9-26 cm. (fig. 67-3). 



210 BUREAU OF AMERICAN ETHNOLOGY [bdll. i67 

4. Jars with rounded body and exteriorly thickened rim with rounded 

lip. Rim diameter 20-26 cm. (fig. 67-4). 

5. Jars with rounded body and everted rim (fig. 67-5) . 
Decoration: Surfaces rubbed with a red ocher polishing stone when clay fairly 

wet, causing floating and streaking with red. Degree of polishing determines 
the thickness and color of the surface layer; bright red-surfaced sherds are the 
best smoothed. About 10 percent are deep, dull red; the remainder tannish 
red to dark brown because of differential polishing, firing, and weathering. 
The surface lacking red pigment is in most cases less well smoothed, often 
showing prominent tracks and having a regular to slightly irregular or uneven 
surface. 

Temporal difference within the type: The most typical examples of Esper- 
an^a Red are from the later levels at Sites J-5 and C-3. The small sample 
shows no trend in vessel shape (Appendix, table 32). 

Chronological position of the type: Attains a maximum frequency in the 
upper levels at J-5 and then declines and disappears just before the Mangueiras 
Phase comes to an end. 

mangueiras plain 

Paste: 

Method of manufacture: Coiling. 

Temper: Large amount of ground sherd with size range from 0.5-4.0 mm. 
Easily distinguished by angularity and flat, smoothed surfaces of some of 
the particles. 

Texture: Compact, well-mixed paste with temper evenlj' distributed; sharp, 
angular cleavage; good tensile strength, hard to break and drops with 
a good ring. 

Color: Tan to salmon on immediate surface. Dark gray or black core in 
75 percent of sherds. Remainder fired with an orange band 1-3 mm. 
wide on exterior leaving the remaining cross section black or gray. 

Firing: Incompletely oxidized with a large number of fire clouds. 
Surface: 

Color: Exterior and interior — Salmon to tan to gray to dark gray. One- 
third of the sherds have fire clouds on one surface making the complete 
color range possible on a single sherd. Light and dark gray are the most 
typical of the coarse, heavier sherds, and the highest percentage of salmon 
appears on thinner sherds. About 10 percent are dark gray on both 
surfaces and 20 percent salmon on both surfaces with the remainder ex- 
hibiting various combinations of the complete range. 

Treatment: Surfaces range from velvety smooth to rough and irregular, with 
about 30 percent in the well-smoothed category. The thinner-walled 
sherds are usually the best smoothed. The smoothing process was 
carried out when the clay was moderately wet leaving no scraping tracks. 
A few of the more rough and irregular sherds are gritty to the feel. 
Crackle lines and fine pits from water bubbles are common, even on the 
velvety-smooth sherds. 

Hardness: 2.5-3.0. 
Form: 

Rim: Direct, exteriorly thickened or everted with typically rounded lip on 
jars; direct, interiorly thickened, everted or exteriorly thickened with 
rounded or angular lip on bowls (pi. 45). 

Body wall thickness: Range from 4-20 mm.; majority 5-10 mm. 

Body dimensions: Jars 10-50 cm. in maximum body diameter. 



EVANS] ARCHEOLOGY AT MOUTH OF AMAZON 211 

Base: Rounded on bowls and plates. Jar bases are of 3 major types: 

1. Rounded, varying from almost flat to a curvature continuing that of 

the body walls and thickened on the interior. 

2. Concave, 5-10 cm. in diameter, with a depression 1-2 cm. deep in 

the center on the exterior, sometimes reflected in a slight convexity 
on the interior. In this type the junction with the body wall is 
often a marked angle of 40-50 degrees. 

3. Flattened and thickened to produce a flat or convex interior surface. 

Junction with body wall is an angle of 30-50 degrees. Diameter 
12-20 cm., thickness 1.0-1.5 cm. at the body wall and 1.5-3.0 cm. 
at the center. 
Vessel shapes reconstructed from sherds: 

1. Globular- bodied jars with flattened bottom and everted, collarlike 

rim with a rounded lip; lip rarely pointed or squared. Rim 
diameter 10-32 cm.; majority 14-22 cm. (fig. 68-1). 

2. Globular-bodied jars with flattened bottom, rounded lip, exteriorly 

thickened rim, constricted mouth. Rim diameter 12-20 cm. 
(fig. 68-2). 

3. Jars with flattened bottom, rounded body, upper walls insloping to 

a direct rim with a rounded lip. Rim diameter 10-28 cm. (fig. 
68-3). 

4. Globular-bodied jars with direct rim, rounded lip. Mouth diameter 

10-44 cm.; majority 12-24 cm. (fig. 68-4). 

5. Wide-mouthed jars with rounded body, slightly insloping upper 

walls terminating in an exteriorly thickened rim. Rim diameter 
10-34 cm. (fig. 68-5). 

6. Bowls with flattened bottom, outsloping sides, direct rim with a 

rounded or angular lip. Rim diameter 9-34 cm. ; a few miniatures 
with diameter of 4-8 cm. and a few very large with diameters 
35-44 cm. (fig. 69-6). 

7. Bowls with flattened bottom, outsloping sides, rim interiorly thick- 

ened or everted to produce a broad level inner surface or top; 
lip rounded. Diameter 18-44 cm.; majority 26-40 cm. Occasion- 
ally the lip is undulating or scalloped (fig. 69-7) . 

8. Bowls with rounded bottom, vertical sides and everted or exteriorly 

thickened rim with rounded or angular lip. Rim diameter 8-32 
cm. (fig. 69-8). 
Appendages: 

Rim adornos: Occasionally the rims have simple adornos in the form of 
protruding lips or scalloped edges extending from 1.0-2.5 cm. beyond 
the normal rim edge (pi. 45). 
Handles: Large loops, with a round cross-section ranging from 1.3-2.0 
cm. in diameter were affixed directly to the jar wall and thickened at the 
point of attachment. Length is 5-8 cm. Handle protrudes 4-6 cm. 
from the vessel surface. 
Occasional decoration: A few jars have ornamental unsmoothed coils on the 
neck exterior (pi. 45, c). 
Temporal differences within the type: The occasional appearance of un- 
smoothed coils on the neck exterior appears to be a trait adapted from the 
Ananatuba Phase since it occurs only at Site J-7, where the initial contact 
took place, and at later sites. Vessels of shapes 1 and 7 are abundant in the 
early sites and rare or absent in the later ones. Shapes 3 and 4 are the dominant 



212 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 16T 







(«i^f 



Nsvvvvm 

wwvs 




I. i I I I 

I 2 3 CM 



Rim Scol« 



4 8 12 CM 
Vessel Scale 



Figure 68. — Rim profiles and vessel shapes of Mangueiras Plain jars, Mangueiras 
Phase (Appendix, table 31). 

jar forms of the late sites, and are absent or very rare in the early ones. Shapes 
5 and 8 show a decline in frequency from early to late sites (Appendix, table 31). 
An important trend in the appearance of Mangueiras Plain is its tendency to 
approach the Anjos Plain dusty orange surface color in the latter part of the 
Phase, in contrast with the range from salmon to black characteristic of the 
earlier sites. 



HEQOERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



213 




////i//iii 





'^rfrrr 




4 8 12 CM 
Vessel Scale 

FiGTIKE 69.- 



Rim Scole 



-Rim profiles and vessel shapes of Mangueiras Plain bowl?. 
Mangueiras Phase (Appendix, table 31). 



Chronological position of the type: Mangueiras Plain is the dominant 
plain ware at the beginning of the Mangueiras Phase and declines in frequency 
with the increasing popularity of Anjos Plain. 

pocoat6 sceaped 

Paste and surface: On Mangueiras Plain; see this type for details of paste, 

temper, firing, etc. 
Form: 

Rim: Direct with rounded or square lip; everted to produce a flat, broad 

upper edge; occasionally thickened on the exterior. 
Body wall thickness: 5-12 mm., majority 8 mm. 
Bases: Flattened on exterior and shghtly thickened on the interior to 1)-^ 

times the body wall thickness. 
Vessel shapes reconstructed from sherds: 

1. Globular jars with flattened base, everted rim and rounded or squared 
lip. Rim diameter 16-28 cm. (fig. 70-1). 
391329—57 16 



214 



BUREAU OF AMERICAN ETHNOLOGY 



[B0LL. 167 







' I 1 I I I 1 

4 8 12 CM 
Vessel Scale 



I ■ I I 1 

I 2 3 CM 

Rim Scale 



Figure 70. — Rim profiles and vessel shapes of Pocoat6 Scraped, Mangueiras Phase 

(Appendix, table 32). 



BV^Nsf^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 215 

2. Jars with globular bodies, upper walls insloping to short vertical 

neck, direct rim and square or rounded lip. Rim diameter 16-22 
cm. (fig. 70-2). 

3. Shallow bowls with direct rim and square or rounded lip. Mouth 

diameter 15-25 cm. (fig. 70-3). 

Decoration (pi. 46) : 

Technique: Pocoat6 Scraped is distinguished from Croarl Brushed by the 
broadness and angularity of the marks, which are flat channels separated 
by narrow, flat-topped ridges. The surface is left extremely irregular and 
uneven and the scraping or combing was done when the clay was moist 
enough to leave the ridges distinct. The marks are typically 1-4 mm. 
in width and about 1 mm. deep with a few examples, apparently from 
larger jars, 1.3 em. in width. The majority are around 2 mm. in width. 
Motif: The scraping marks are applied to the jar exteriors horizontally 
around the neck below the rim, or the horizontal band is omitted and the 
vertical or diagonal scrapings begin at the neck. On bowl exteriors the 
direction of scraping is around the circumference. A few examples have 
scraping lines running in two directions producing a hachured or herring- 
bone effect, but with no apparent effort at regularity so that the result is 
crude in comparison with those specimens scraped in one direction only. 
A considerable number of sherds were scraped on both interior and ex- 
terior, with the directions of the lines not always the same, since those on 
the interior are uniformly parallel to the rim. A great many are scraped 
on the interior only, which raises the possibiUty that this method of decora- 
tion may have developed from what was at first an accidental by-product 
of scraping and smoothing the interior surface. 

Temporal difference within the type; None (Appendix, table 32). 

Chronological position of the type: Limited to the sites in the early part of 
the Mangueiras Phase sequence. 

pseudo-sip6 incised — mangueiras phase variety 

Paste: All but 10 percent are on Mangueiras Plain, the remainder on Anjos 
Plain, See those pottery type descriptions for details of paste, temper, and 
firing. 
Form: 

Rim: Direct or slightly everted rim with rounded or flattened lips. Rim 

diameters from 10-24 cm. 
Body wall thickness: 5-9 mm. 
Bases: Probably rounded. 
Vessel shapes reconstructed from sherds: 

1. Globular jars with constricted mouth, direct rim and rounded lip. 

Mouth diameter 14-24 cm. (fig. 71-1). 

2. Globular-bodied jars with insloping neck, direct rim, rounded or 

flattened hp. Rim diameter 10-20 cm. (fig. 71-2). 

3. Shallow, rounded bowls with direct or slightly everted rim, rounded 

lip. Diameter 10-24 cm. (fig. 71-3). 
Decoration (pi. 47, a-g): The incised designs are copied from Sip6 Incised of the 

Ananatuba Phase (see pp. 185-187). At Site J-14 — Bacuri, design types 1, 2, 

4, 5, and 6 occur; at Site J-16 — Canivete, design types 1, 5, and 6; at Site 

J-17 — Flor do Anajds, design types 2 and 6. 
Temporal difference within the type: None (Appendix, table 32). 
Chronological position of the type: This pottery type appears as one of 

the wares of the Mangueiras Phase immediately after the conquest of the 



216 BUREAU OF AMERICAN ETHNOLOGY [bdll. i67 





4 8 I2CM I 2 3 CM 

Vtaitl Scale Rim Scale 

Figure 71. — Rim profiles and vessel shapes of Pseudo-Sip6 Incised, Mangueiras 
Phase variety (Appendix, table 32). 

Ananatuba Phase village of J- 7 — Sip6 and evidently represents a direct copy 
of the Ananatuba Phase decorative motifs. The variety and accuracy of the 
copy is best at Site J-13 — Bacuri, which is contemporary with the Mangueiras 
Phase occupation of Site J-7 — Sip6 and closer geographically to Site J-7 than 
are the other two Mangueiras Phase sites at which this type appears. 

UNCLASSIFIED DECORATED 

The largest number of sherds in this group come from the three earliest sites of 
the Mangueiras Phase, J-5, J-17 and C-3. Four techniques are represented: 
incision, punctate, excision, and corrugation 

Unclassified incised: 

1. Rims of Mangueiras Plain, shape 7, with incised designs on the broad, flat 

top. Motifs are mainly parallel, straight or zigzag lines and spirals. 
Techniques may be broad, 2-4 mm. wide, lines or narrow lines and small 
areas of fine cross-hatch. Total from J-16, 15 sherds; J-17, 1; and C-3, 
3 (pi. 47y, Z-n). 

2. Body sherds with simple geometric design of widely spaced lines, the sur- 

face covered completely or in zones with fine incisions or scratches. 
Total from J-5, 1 sherd; C-3, 2 sherds; J-16, 2 sherds. 

3. Body sherds with simple geometric designs of parallel, straight or curved 

lines. Total from J-5, 1 sherd; J-17, 4; C-3, 12; J-16, 1 (pi. 47, k). 

4. Badly eroded sherds with faint incisions. Total of 14 from Site J-13. 
Unclassified punctate: 

1. Single row of deep evenly-spaced punctates along the rim exterior of 

bowls of Mangueiras Plain, shape 8. The punctates are rectanguloid 
at Site J-5 and circtilar at C-3. Total sherds from J-5, 5; C-3, 9 (pi. 
47, h). 

2. Applique rib 1.0 cm. high and 1.5 cm. wide with a row of punctates along 

the top or one row at each side marking the junction of the rib with the 
body wall. Total of 2 sherds from Site C-3. 

3. Punctates applied in rows over vessel exterior. Rows relatively parallel, 

but punctates irregular in size and depth. Total of 2 sherds from Site 
J-13 (pi. 47, i). 
Unclassified excised: 

1. Background unevenly gouged out, leaving areas of the original surface. 
Since the sherds are all less than 2.5 cm. square, the motif is not recon- 
structable. Total of 4 sherds from C-3. 



BV^Nsf^ ^^" ARCHEOLOGY AT MOUTH OF AMAZON 217 

Unclassified corrugated: 

1. Large, deep impressions made by pinching the coil between thumb and 

fingers. Total of 3 sherds from C-3; 7 sherds from J-13. 

2. Small corrugations made by pressing downward on the coil so as to produce 

a scalloped lower edge (typical technique of Floripes Corrugated of the 
Acauan Phase). Total of 14 sherds from C-3. 

3. Blunt stick instead of finger used to jab along the coil, producing a corru- 

gated effect, but with very deep impressions between the globs. Total 
of 3 sherds from C-3. 

Pottery Artifacts 

Short, tubular pipes occur at J-5 and C-3 (fig. 58). These same 
two Mangueu-as Phase sites produced collar-button-shaped objects 
that probably were labrets (figs. 60, 63). Figurines are represented by 
a head and a torso of different figurines (fig. 59) from different levels 
at Site J-5. Both of these objects exhibit considerable detail in work- 
manship. The detailed descriptions and illustrations will be found 
with the various site descriptions. 

Nonceramic Artifacts 

No fragments or objects of stone, shell, or other nonceramic ma- 
terials were encountered, with the exception of a small, unworked 
stone chip from Site C-3, cut 1, level 8-16 cm. 

Ceramic History 

The seriation for the Mangueiras Phase is based on 13,724 sherds 
from 6 habitation sites. Of these 11,566 are plain ware: 4,088 Anjos 
Plain and 7,478 Mangueiras Plain. All but one of the sites had refuse 
accumulations greater than 15 cm. in depth and were excavated strati- 
graphically. The seriation of these levels and the changes in the 
frequencies of the ceramic types are shown on the accompanying 
graph (fig. 72). 

The ceramic sequence in the Mangueiras Phase is characterized by a 
gradual decrease in the popularity of Mangueiras Plain, a gray-cored 
ware, and the concomitant increase of Anjos Plain, an orange-cored 
ware (Appendix, table 28). The earliest site in the sequence, J-5 — 
Croari, produced only Mangueiras Plain in the lower levels. Anjos 
Plain has a frequency of 3.5 percent at the beginning of C-3 — Porto 
Real, and continues to grow in popularity until it has reached 71.2 
percent at the end of the occupation of J-13. This increasing emphasis 
on Anjos Plain is reflected in changes within the Mangueiras Plain 
ware. At J-5, C-3, and in the lower levels of J-13, Mangueiras Plain 
is most typical in surface color, which runs the gamut from cream 
through rose to black. In the upper levels of J-13, it becomes in- 
creasingly transitional in the direction of Anjos Plain, until the two are 



218 BUREAU OF AMERICAN ETHNOLOGY [bdll. le? 

indistinguishable in external appearance and differ only in the color 
of the core. Anjos Plain, on the other hand, shows a great uniformity 
in surface appearance from the earliest to the latest sites. 

Three decorative techniques are characteristic of the Mangueiras 
Phase, but only at the earhest site, J-5, do they all occur together. 
Esperanga Red, in which simple red bands are painted on the rim or 
interior of bowls, reaches its clim.ax (24.4 percent) in the upper level 
of J-5, but persists in amounts of under 5 percent almost to the end 
of the Phase. Pocoat6 Scraped, in which parallel, troughhke marks 
cover the surface, also begins in the lowest level of J-5. It reaches 
its maximum of 28.6 percent during the first part of the occupation 
of C-3, decHnes to 4.8 percent at J-17, and is absent during the re- 
mainder of the Phase. Brushed decoration has a disconnected 
history. Crude brushing on Mangueiras Plain (Croari Brushed) 
increases from 1.9 percent to 5.9 percent and then decreases to 3.8 
percent at J-5. Brushing is completely absent at C-3, but reappears 
later in the Mangueiras Phase as a technique applied only to Anjos 
Plain (Bacuri Brushed). Its sudden reappearance as the dominant 
decorated technique, and the equally sudden disappearance of Pocoat6 
Scraped, seems to have been stimulated by contact with the Ananatuba 
Phase, in which brushing was popular. A handful of sherds with 
punctate decoration occurs scattered sporadically throughout the 
Phase, but these are neither consistent enough in appearance nor 
sufficiently numerous to be interpreted as anything more than the 
results of occasional experimentation. 

Diagnostic of the early Mangueiras Phase, and lost by Bacuri (J-13) 
times, is the use of broad, shallow, incised lines in rectilinear or curvi- 
linear patterns, especially on the flat upper rim edge of shallow bowls. 
This trait appears suddenly, fully developed, at J-5 and is also promi- 
nent at J-17, but as the Phase continues it falls increasingly by the 
wayside until it finally disappears. A comparison of the technique 
and motifs with those of Carobal Incised of the Acauan Phase strongly 
suggests contact with and influence from this latter Phase. This 
conclusion is strengthened by the association of scattered excised and 
corrugated sherds, which reproduce other popular Acauan Phase 
surface embellishments, and by the resemblances between the scraped 
types present in the two Phases (Pocoato Scraped and Paciencia 
Scraped). The detailed analysis of the impHcations of this situation 
is given under the afRliations of the Acauan Phase (pp. 540-545). 

Another fascinating decorated type from the standpoint of its 
origin and history is Pseudo-Sipo Incised. This type was so named 
in order to emphasize the fact that the motifs are identical with those 
of Sipo Incised of the Ananatuba Phase. It is distinguished only by 
being applied to Mangueiras Plain or Anjos Plain, showing that it 



0-.I5 
.15- 
-I3j . so- 
ls- SURF^ 

I 

fl 1=0-. 
.45-. 

iL .60-, 
I=.I5- 
2=0- 
2=.I5 

L 2= .30 



J-7^ 



J- 



C- 





r °--'t 


'n 






L .15-. 




r 2=0- 




1=0-. 


^x 






2=.I6| 

. I=.I6^ 





;:: J- 



O-.l 
.15- J 
5^ .30- 
.45- 
.60 
L .75- 



table 28). 



D 



[31 



I 1 



BACURI 
BRUSHED 



PSEUDO-SIPO 
INCISED 





r 0-.I5M 




.15-. 30 


J-13 


.30-. 45 


1 
J-I6-SURFACE 

1 


t hO-.l5M 

1 




.45-. 60 




L .60-. 75 


J-tJ 1^.15-. 30 




2=0-. I5M 




2=.l5-.30 




. 2. 30-. 45 


r 0-.I5 M 


J-17-< 


L .15-. 30 


f 2=0-. I6M 


° '1 2=. 16-32 


L l=.l6-.32 




"" 0-.I5M 




.15-. 30 


-5- 


.30-.45 




.45-. 60 




.60-. 75 




.75-. 90 



s 



I i I I I 

20 40% 



t 



k \ \ \ \ s ^ 



EH 



V / / / / 7Z3 

1/ y / / / y-1 



II II III II 


iiiiii 


1 








llllllllll 


iiiiii 


II 








llllllllill 


IIIIII 


III 








II 1 1 1 1 1 1 1 1 1 1 




1 III 








iiiiiii mil 




im 








iiiiiiiiiiii 




INI 



HIIIMIIIIIIIIIIIII 



k\ SM 



EH] 



IZIZI 
IZI 

[D 
I 



MANGUEIRAS 
PLAIN 



ANJOS 
PLAIN 



GROARI 
BRUSHED 



EI3 

izia 



m 
m 

01 

ni 
m 

POCOATO 
SCRAPED 



S 
(3J 

I 

D 
I 

B 
I 

I I IIII I 

rrm 

m 

a 



I 

I 

ESPERANQA 

RED 



(31 



I 1 



BACURI 
BRUSHED 



PSEUDO-SIPO 
INCISED 



Figure 72. — Seriation of Mangueiras Phase sites on the basis of pottery type frequency (Appendix, table 28). 



KVA^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 219 

represents the adoption and perpetuation of this alien decorative 
technique by the people of the Mangueiras Phase. Its popularity- 
was such that it was quickly taken up b}^ other Mangueiras Phase 
villages, as witnessed by sherds from J-13, J-16, and J-17 (fig. 72). 
Because of this, Pseud o-Sip6 Incised provides a means for quick 
identification of post-Ananatuba Phase contact sites. 

An analysis of vessel shape, based on 521 rims of Mangueiras Plain 
and 203 rims of Anjos Plain, reveals that certain shapes are associated 
with one or the other type and certain others mth early or late sites 
(Appendix, tables 29 and 31). A similar ratio of bowls to jars obtains 
here as was noted between the plain wares of the Ananatuba Phase: 
58 percent of the Mangueiras Plain rims are from bowls and only 10 
percent of those of Anjos Plain. Since Anjos Plain is absent or rare 
in the earlier sites, it lacks two early shapes. These are Mangueiras 
Plain shapes 1 and 7, which are most numerous at J-5 and tend to 
fade out thereafter. Within Mangueiras Plain, there is a gradual 
shift in bowl form, with shape 8 being equally comm,on as shape 6 in 
the first half of the Phase, but declining as shape 6 becomes the 
dominant Mangueiras Plain vessel type. Jar shape 5 is also a typi- 
cally early and exclusively Mangueiras Plain form. The dominant 
Anjos Plain form, jar shape 4, occurs also to a minor degree in Man- 
gueiras Plain, and persists in both types. 

The most interesting aspect of the vessel shape analysis is the 
amount of acculturation it reveals on the part of Mangueiras Phase 
ceramics. While it was in the process of engulfing the Ananatuba 
Phase at J-7, the Alangueiras Phase was also adopting a number of 
Ananatuba Phase pottery traits. Examination of the remaining three 
jar shapes, which first appear or markedly increase in abundance in 
Mangueiras Phase wares at J-7, reveals that these are shapes of long 
standing in the Ananatuba Phase. The comparison is more pro- 
nounced when the differing frequencies of these shapes in the two 
plain wares are eliminated by adding the rim sherds of the same shape 
together and recomputing the percentage (Appendix, tables 27 and 
33). Figure 73 shows the relative pre- and post-Ananatuba Phase 
contact occurrence of Mangueiras Plain shape 4 and Anjos Plain 
shape 1 (which are the same) in contrast to the liistory of that shape 
in the Ananatuba Phase. Figure 74 treats combined Mangueiras 
Phase plain ware shape 3 in the same way. Random occurrences in 
precontact Mangueiras Phase sites may be misclassification of a small 
sherd or deviant part of a bowl rim, or they m.ay indicate that these 
shapes were present but rare until the stimulus of the Ananatuba 
Phase was felt. The shapes associated with the decorated types 
reveal an interesting dichotomy that is further evidence of the strength 
of the influence exerted by the Ananatuba Phase: Croari Brushed, 



220 BUREAU OF AMERICAN ETHNOLOGY 

ANANATUBA PHASE MANGUEIRAS PHASE 



[BULL. 107 




\ / 



POST- ANANATUBA 
PHASE CONTACT 
(J-7, J-13, J-16) 



PRE -ANANATUBA 
PHASE CONTACT 
(j-ir, J-S,'C-3) 



10 20-30% 




Figure 73. — Stratigraphic evidence for the origin by acculturation of Mangueiras 
Plain Vessel Shape 4 (Mangueiras Phase) from the Ananatuba Phase (Ap- 
pendix, tables 27 and 33). The bars show the relative frequency of the vessel 
shape at sites of both Phases in seriated order. 

Pocoat6 Scraped, and Esperanga Eed, which are characteristic early 
Mangueiras Phase decorated types, are found only on Mangueiras 
Phase vessel shapes, whereas Bacuri Brushed and Pseudo-Sipd Incised, 
representing Ananatuba Phase influence, are exclusively on Ananatuba 
Phase shapes. The significance of this correlation is enhanced by 
the fact that Esperanga Red, which continues to be made in the late 
Mangueiras Phase, is not influenced by Ananatuba Phase vessel 
shapes. 

An anachronistic feature of J-17 — Flor do Anajds is the presence of 
fragments of exceedingly thick and gross, red-slipped tangas. Tangas 
of this type have been found at Marajoara sites, but appear not to be 
frequent in the best-known ones. If they are part of the Mangueiras 
Phase culture at this site, then a drastic revision of the seriation is 
required. There are several considerations, however, that suggest 
they are intrusive: (1) they were fomid only on the surface and in 
the upper level of the cut, and the collections from both these sources 
also contained fragments of modern tile and earthenware, originating 
from the present village (the lower level produced neither modern 
ceramics nor tanga sherds) ; (2) the site seriates early in the archeo- 
logical sequence on Marajo, and Marajoara trade materials in late 
Formiga Phase and earty Arua Phase sites place the arrival of the 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



221 



Marajoara Phase much later; and (3) The caboclos are active excava- 
tors of Marajoara sites and frequently bring not only complete vessels 
but also well-preserved adornos and other curious objects back to 
their houses, which could readily account for their intrusion when 
broken into the earlier archeological refuse. Until better evidence 
for contemporaneity is discovered, therefore, it seems justifiable to 
interpret this Marajoara Phase mixture as a recent one attributable 
to the modern village occupying the spot. An examination of nearby 
Marajoara sites may show that the heavy, red-slipped tangas are 
common at one of them, which would establish their source. 

Diagnostic Features of the Mangueiras Phase 

Mangueu-as Phase sites are found over a wide area, having been 
identified so far from central and northern Marajo and southern 
Caviana. They are located in the forest where the land is not subject 
to annual flooding, but proximity to the campo seems to have been 
a less important factor in the choice of a village location than prox- 
imity to a navigable stream. Unlike the Ananatuba Phase, Man- 
gueiras Phase sites are always within 250 meters of a large igarape, 
although rarely on the immediate shore. Nor are any of these sites 
on the coast, the nearest (J-13) being 3 km. inland. Some are at the 
edge of the campo, others several kilometers distant. As indicated 
by the ceramic refuse, the villages covered an area of 2,000 to 4,000 



ANANATUBA PHASE 



MANGUEIRAS PHASE 



POST- ANANATUBA 
PHASE CONTACT 
(J-T, J-13, J-IC) 



PNC • ANANATUBA 

PHASC CONTACT 

(C-S) 



10 to 30% 




Figure 74. — Stratigraphic evidence for the origin by acculturation of Mangueiras 
Plain Vessel Shape 3 (Mangueiras Phase) from the Ananatuba Phase (Appendix, 
tables 27 and 33). The bars show the relative frequency of the vessel shape 
at sites of both Phases in seriated order. 



222 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

square meters, except at C-3 where the site is unusually small. The 
depth of the deposit varies from 0.05 meter at J-16 to 1.00 meter at 
J-5, apparently indicating great irregularity in the length of time 
various villages were occupied. No cemeteries were identified, and 
there is no evidence of practices associated with the disposal of the 
dead. 

The seriation of the sites belonging to the Mangueiras Phase is 
based on the decreasing frequency of Mangueiras Plain, a sherd- 
tempered, gray-cored ware, and the corresponding increase in Anjos 
Plain, which is more completely oxidized in firing, typically eliminating 
the gray core. Among the Tropical Forest Phases, this one is out- 
standing for its high percentage of decorated sherds, but these are 
mostly simple brushing and scraping, and never attain the artistic 
level of Sipo Incised in the Ananatuba Phase. The better quality 
of the ceramics, the presence of incised decoration, annular bases, 
and distinctive vessel shapes make for ready identification of the 
earliest sites, because none of these traits persist for any length of 
time. A late time marker is Pseudo-Sip 6 Incised, with designs copied 
from Sipo Incised the Ananatuba Phase. Typical ceramic artifacts, 
also early, are tubular pipes (J-5 and C-3), labrets (J-5 and C-3), 
and figurines (J-5). No drilled sherds or other objects identifiable 
as spindle whorls were encountered. The presence of irregular lumps 
of burnt clay in the refuse of all levels and in aU sites has no ready 
explanation. 

After contact with the Ananatuba Phase, which occurred about 
midway in the sequence represented here, the Mangueiras Phase 
ceramics underwent a strong degree of acculturation, as witnessed by 
the appearance of Ananatuba Phase vessel shapes and decorative 
techniques. Its seriated position indicates that the Mangueiras 
Phase did not survive into historic times and, in confirmation, no 
evidence of European contact was found at any of the sites. 

THE FORMIGA PHASE 

Description of Sites and Excavations 

Formiga Phase sites are located on north and central Maraj6 
(fig. 47). 

site J-4 MUCAji. 

This large habitation site is some 5 km. inland from the town of 
Chaves on the north coast of Marajo, near the upper reaches of the 
Igarap6 Atura-miri, which winds off in a northeasterly direction to 
empty into the Amazon several kilometers east of Chaves (fig. 86). 
This stream is now clogged with trees and almost drj'^ toward the upper 
end. J-4 is in a natural clearing at the edge of the dense coastal 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



223 



forest strip, and is covered with grass and a few scattered small trees. 
Lake Arapapa, a small, shallow body of water, is about 1 km. to the 
west. 

The site consists of 2 large and 4 small mounds distributed over an 
area of 150 by 150 meters (fig. 75). All are approximately a meter 
higher than the general terrain and readily distinguishable by the 
taller and greener grass growing on them, as well as by their elevation 
(pi. 31, a). That the height of the mounds is due in part to artificial 
construction is indicated by the thinness of the refuse layer and by a 
depression between the two largest that may have been left by exca- 
vation of dirt for the adjacent mounds. The two major mounds are 
parallel and run north-south. The first, on the east, measures 100 
by 20 meters, and the second, just west of it, is slightly shorter. The 
remaining 4 mounds, 2 to the west and 2 to the south, vary from 25 to 
35 meters in length and 5 to 8 meters in width. All but one have the 
longest axis oriented north-south. A circular depression at the north- 
east corner of Mound 1 may be the remnant of a well. 

No sherds could be seen on the surface, and tests made at various 
spots on three of the mounds showed the refuse layer to be 10 cm. or 
less in thickness and the sherds to be sparse and in a poor state of 
preservation in the clay matrix. Below the refuse layer, the clay 



.■.-LIMIT;. OF -rOREST. 




10 20 30 M 



Figure 75. — Plan of J-4 — Mucaj6, a habitation site of the Formiga Phase. 



224 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

became yellow in contrast to the dark gray of tlie occupation level. 
Excavation was concentrated on the north end of Mound 1, where 
the sherds were more profuse than in other places tested. 

The sample collected included 929 sherds (of which less than 100 had 
surfaces in good condition), 56 burnt-clay lumps (3 with grooves as 
though plastered against sticks) and 2 particles of u'on concretions, 
which show no use and are natural soil inclusions. 

SITE J-6 — FORMIGA 

In a wide stretch of campo dotted with clumps of forest about 4 km. 
southeast of the Ananatuba Phase Site J-7 — Sip6, are three small 
mounds (fig. 76). The grass growing on them is taller than in the 
surrounding area, and at the beginning of the dry season was markedly 
darker green (pi. 31, 6, c). The nearest forested spot is a kilometer 
away. A small igarape winding across the campo passes along the 
north edge of the site. Mound 1, the largest of the group, is oval, 
oriented slightly northwest by southeast, and measures 20 meters 
long by 8 meters wide and 1.25 meters high. Mound 2, 28 meters 
east of Mound 1, is 8 to 10 meters in diameter and 0.50 meter high. 
Mound 3, 75 meters east of Mound 2, has a diameter of 18 meters and a 
height of 0.50 meter. Few sherds are visible on the surface, and most 
of those included in the surface collection were found by the workmen 
digging on their own. 

A stratigraphic excavation was made in each mound. Cut 1 , in the 
center of Mound 1, was 2 by 2 meters square and controlled in 15-cm. 
levels. The refuse layer was black clay and of uniform character 
except at level .30-.45 m., when many lumps of unfired clay were 
encountered near the east side. Natural, sterile, yellow-brown clay 
was reached at 90 cm. The count by levels totaled: 

Level .00-.15 m.: 209 sherds and 10 burnt-clay lumps 

Level .15-.30 m.: 1,228 sherds (one worked and drilled) and 153 burnt-clay 

lumps 
Level .30-. 45 m.: 1,108 sherds and 126 burnt-clay lumps 
Level .45-. 60 m.: 554 sherds (1 worked and drilled) and 29 burnt-clay 

lumps 
Level .60-.75 m.: 150 sherds and 17 burnt-clay lumps 
Level .75-.90 m. : 116 sherds, 4 burnt-clay lumps 

Cut 2, a little northeast of the center of Mound 3, was excavated in 
the same dimensions and levels as cut 1. Soil conditions repeated 
those in cut 1, the hard, black clay becoming slightly yellower and 
dryer in level .30-45 m. Sterile soil was reached at 50 cm. At a 
depth of 45 cm. the west edge of the cut passed through a pocket of 
black clay about 15 cm. in diameter, containing fragments of burnt 
bone, which Marshall T. Newman, United States National Museum, 
was able to identify as human. There was no concentration of sherds 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



225 




226 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 



near this burial that could be interpreted as constituting gi-ave goods. 
The count from cut 2 by levels gives : 

Level .00-. 15 m. : 123 sherds and 1 burnt-clay lump 

Level .15-.30 m. : 238 sherds and 1 burnt-clay lump 

Level .30-.45 m.: 494 sherds and 19 burnt-clay lumps 

Level .45-. 60 m.: 126 sherds and 2 burnt-clay lumps 

Cut 3, 1 meter square, was put in the center of Mound 2. Here the 
sod layer extended to a depth of 10 cm., but conditions below dupli- 
cated those in the two previous cuts, with sterile grayish-brown clay 
appearing at 50 cm. This test was continued to a depth of 75 cm., 
into the sterile soil underlying the mound. Cultural remains were 
distributed : 

Level .00-. 15 m. : 218 sherds, 1 burnt-clay lump 
Level .15-. 30 m.: 739 sherds, 13 burnt-clay lumps 
Level .30-45 m.: 836 sherds and 27 burnt-clay lumps 
Level .45-. 60 m.: 166 sherds and 11 burnt-clay lumps 
To these totals, the surface collection added 146 sherds 

Drilled sherds. — The two drilled sherds from cut 1 are Formiga 
Plain (fig. 77). Both were crudely shaped into a circle, averaging 




I CM 

a b 

Figure 77. — Drilled sherds from J-6 — Formiga, Formiga Phase. 

3 cm. in diameter, and drilled through the center from both sides 
with a hole 6 to 9 mm. in diameter. Both are slightly concavo-convex 
in cross section, with a thickness of 6 mm. 

Data prom Other Excavations 

site j-18 — coroca 

The only other site that has been identified as belonging to the 
Formiga Phase is J-18 — Coroca, on the right bank of the Rio Arari, 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 227 

EVANS J 

a few kilometers south of its junction with the Rio Anajasinho (fig. 47). 
It was excavated by Peter Hilbert of the Museu Goeldi, who describes 
it as on the edge of a rise that parallels the Rio Arari 100 meters 
inland from its usual western shore. The refuse deposit overlies a 
low, artificial mound now covered with trees, which extends 25 
meters north-south and 4 to 8 meters east-west. From an elevation 
of 1.00 meter at the middle, it rises to a summit 1.50 meters high at 
the center of each half. 

A 1.5 by 1.5 meter stratigraphic excavation made in 15-cm. levels 
on the northern summit showed the refuse to extend to a depth of 
65 cm. Soil in the first level was light-brown clay, becoming lighter 
in color and increasingly sandy to a depth of 60 cm., where it changed 
to hard, whitish clay, which continued sterile to 1 meter (Hilbert, 
pers. corres.). 

The excavation produced the following materials: 

Level .00-. 15 m.: 47 sherds. 

Level .15-. 30 m.: 156 sherds. 

Level .30-.45 m.: 214 sherds and 1 fired clay lump. 

Level .45-.60 m.: 209 sherds. 

Level .60-. 75 m.: 17 sherds. 

One clay lump and 19 additional sherds made up the surface collec- 
tion. All the material was sent to us for analysis and provides im- 
portant additional information on the ceramic history of the Formiga 
Phase. 

Analysis of Materials of the Formiga Phase 

Pottery Type Descriptions 

The description of the Formiga Phase pottery types is based on 
the analysis and classification of 8,042 sherds from habitation sites. 
Using the binomial system of nomenclature, the following types were 
established, arranged in alphabetical order: 

CATARINA PLAIN 

It is possible that the sherds described as Catarina Plain are actually badly 
eroded examples of the Marajoara Phase ware, Inajd Plain. Their initial ap- 
pearance coincides with that of certain Marajoara Phase decorated types, and 
the vessel shapes resemble those of the Marajoara Phase. Since the condition of 
preservation prevented an identification and correlation based on paste and 
surface features, it was considered safer to give these sherds a separate designation, 
keeping in mind the possibility they might not constitute a new Formiga Phase 
plain ware, but rather trade material from the Marajoara Phase. 
Paste: 

Method of manufacture: Coiling. 

Temper: Crushed sherd. Sometimes visible as distinct bright-orange, cream, 
or light-gray particles with smooth surfaces which contrast with the dark- 
gray paste. Size ranges from 0.5-5.0 mm. Moderate amount visible. 



228 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

Texture: Distinctly laminated appearance in cross section with numerous air 
pockets where layers have buckled slightly. Extremely poor kneading of 
clay and temper when clay rather dry produced poor cohesion. Easy to 
break with a very crumbly, irregular edge caused by each lamination 
breaking in a different plane. Separation along old coil lines suggests 
modeling when the clay was unusually dry with poor kneading of the coils. 

Color: 80 percent of the sherds are fired orange in an irregular band along 
both surfaces and to a depth of 2 mm. leaving a medium to dark-gray 
core. The zone of contact between the two colors is irregular, with the 
orange encroaching on the gray along the lines of lamination giving the 
cross section a variegated appearance. Remaining sherds are gray cored 
with a thin brownish-tan layer on the surfaces. 

Firing: Incomplete, in oxidizing atmosphere. A few fire clouds. 
Surfaces: 

Color: Exterior and interior — Majority of sherds are leather-red-brown some- 
times shading off to a dusty-cream on both surfaces. A few have this 
color on exterior only, the interior being variegated gray-orange. The 
variegated appearance on both surfaces of 5 percent of the sherds is pro- 
duced by irregularly spaced splotches of bright orange, light tan, and dark 
blue-gray. 

Treatment: Exterior and interior — Unusually poor quality of the paste 
resulted in loss by erosion of the original surface in over 50 percent of 
the sherds. Of those in which the surface is preserved, 25 percent are 
partially smoothed with a hard-surfaced tool bringing finer particles to 
the surface and giving a dull luster, but still leaving many irregularities, 
small pits and tool-smoothing marks. The remaining 75 percent are 
superficially smoothed with the hand or fingers, erasing the coil lines but 
leaving a rough, uneven, coarse, soft surface. 

Hardness: 2. 
Form: 

Rim: Externally thickened or slightly everted with an angular inner lip edge. 

Body wall thickness: Range 5-15 mm., majority 6 mm. 

Body diameters: Range 22-38 cm. 

Base: Flattened, joining the sidewalls at an 80-degree angle, or slightly 
thickened on the interior producing a slight convexity. Diameters range 
22-24 cm. 

Vessel shapes reconstructed from sherds: 

1. Carinated bowls with flattened bottom, everted or exteriorly thick- 

ened rim. Rim diameter 26-34 cm., diameter of flat base 22-24 
cm. (fig. 78-1). 

2. Jar with flat bottom, rounded body, insloping neck, exteriorly thick- 

ened rim. Rim diameter 22 cm.; maximum body diameter 28-42 

cm. (fig. 78-2). 
Temporal difference within the type: None (Appendix, table 35). 
Chronological posiiion of the type: Catarina Plain appears suddenly at the 
termination of the Formiga Phase. 

coROCA plain 

Paste: 

Method of manufacture: Coiling indicated by breakage which occasionally 
leaves a concave edge on one sherd and a convex one on the corresponding 
edge of the part broken off. 



MB00ER8 AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



229 





I I I I I 

1 2 3 CM 
Rim Scale 



I I I I I I I 

4 8 12 CM 

Vessel Scale 

FiGtTRE 78. — Rim profiles and vessel shapes of Catarina Plain, Formiga Phase 

(Appendix, table 35). 

Temper: Ground sherd with wide variation in the size of particles in each 
specimen. Granules are not evenly distributed, probably contributing to 
the general fragility of this type. Color of temper particles is typically 
bright, orange, contrasting sharply with the gray paste. 
Texture: Cleavage very irregular and granular; many air pockets. Extremely 
friable because of poor quality of paste composition, mixture and filing. 
Knocking the sherds together produces a dull thud. 
Color: Whitish to light orange to bright orange beginning from the exterior 
surface and extending inward in a band of varying width sometimes in- 
cluding the entire cross section. Some trace of gray is present in about 
55 percent of the sherds. The oxidation is frequently complete except 
for the incerior surface, so that the gray "core" tends to be along the in- 
terior edge rather than in the middle of the cross section. 
Firing: Oxidized, incomplete to complete; no fire clouds. 
Surface: 
Color: 

Exterior — Range from light, whitish tan to light orange to pinkish 

orange. 
Interior — The same range plus a dull gray brown, which occurs on the 
majority of the incompletely oxidized sherds. 
Treatment: Superficial smoothing leaving small pits and protruding temper 
grains, and an irregular and somewhat undulating surface. Smoothing 
fines parallel to the rim are common on the interior; made by wiping the 
fingers aiound the circumference. 
391329—57 17 



230 BUREAU OF AMERICAN ETHNOLOGY [bdll. i67 

Slip appears on a majority of the sherds on the exterior, sometimes car- 
ried over to the interior on bowls. The slip is often poorly applied, with 
a marked variation in thickness in a small area from paper thinness to 
0.5-1.0 mm. The sUp was applied when the surface was too dry, making 
the bond poor and causing the slip to peel off readily. 

Hardness: Soft; 2. 
Form: 

Rim: Typically direct, or everted with a rounded lip. Occasionally with 
slight exterior or interior thickening. 

Body wall thickness: Range 4-10 mm, majority 6-9 mm. 

Base: Rounded with slight interior thickening, amounting to an increase of 
about one-third over the body wall thickness. 

Vessel shapes reconstructed from sherds: 

1. Jars with short outflaring or everted necks and curved or vertical 

sides producing a globular or ovoid body. Direct rim with rounded 
or flattened lip. Rim diameter 8-32 cm., majority 20-22 cm. 
(fig. 79-1). 

2. Globular-bodied jars with sides incurving to a direct rim with a 

rounded hp. Rim diameter 14-20 cm. (fig. 79-2). 

3. Small jars with globular bodies and rims either thickened on the ex- 

terior or slightly everted to produce the appearance of thickening. 
Rim diameter typically 7-14 cm.; sometimes 15-18 cm. (fig. 79-3). 

4. Bowls with rounded bottom, outsloping sides and direct rim with 

rounded lip. Rim diameter 18-24 cm. (fig. 79-4). 
Appendages: Three fragments of handles come from J-6 — Formiga. These 
are round in cross section and have a small pluglike protrusion on the end 
for insertion into the body wall (cf . Ananatuba Plain) . 
Temporal differences within the type: No vessel shape trends in the small 

sample of rim sherds available (Appendix, table 35) . 
Chronological position of the type: Present throughout the Formiga Phase 
with an early climax and a slow decline. 

EMBAtJBA plain 

Paste : 

Method of manufacture: Coiling. 

Temper: Ground sherd, with particles ranging from 0.5-3.0 mm. 

Texture: Poor mixture of clay with temper leaving many air pockets and 
clumps of temper particles. Cleavage plane is angular and irregular. 

Color: Often mottled in appearance because of the lighter colored particles 
of sherd temper in the grayish paste. Except for a thin band, almost 
paper thin, of orange or reddish tan adjacent to both surfaces, the core is 
gray to black. 

Firing: Incomplete oxidation; many fire clouds and splotches of bright orange. 
Surfaces: 

Color: Ranges from light tan to cream to orange-tan to dull gray on both 
exterior and interior; however, the majority are dark-tan to tannish-orange 
orange on exterior with a grayish interior. 

Treatment: Exterior and interior — Lightly smoothed, leaving a rather un- 
even and irregular surface with many pits where the temper particles 
dragged. Roughness of surfaces made them susceptible to erosion. Some 
of the thinner-walled jars and bowls are slightly more regular and better 
smoothed. 

Hardness: Soft; 2. 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



231 




I I I I I 

I 2 3 CM 
Rim a Handle 
Scale 




' I 'I I I I 

4 8 12 CM 

Vessel Scale 






Figure 79. — Rim profiles and vessel shapes of Coroca Plain, Formiga Phase 

(Appendix, table 35). 



232 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

Form: 

Rim: Direct, exteriorly thickened or everted, with rounded lip. 
Body wall thickness: Range 5-12 mm., majority 7 mm. 
Body diameters: Range 22-48 cm. 

Base: Rounded on exterior and slightly thickened on interior. A few non- 
typical bases, flat and with a slight pedestal, come from Site J-4 — Mucajd. 
Vessel shapes reconstructed from sherds: 

1. Jars with rounded body, walls sloping or curving inward to an ex- 

teriorly thickened rim with a rounded hp. Mouth diameter 10-38 
cm. (fig. 80-1). 

2. Jars with globular body, walls incurving to a direct rim with a rounded 

Hp. Mouth diameter 1&-30 cm. (fig. 80-2). 

3. Jars with globular body and collarlike, everted rim. Rim diameter 

18-24 cm. (fig. 80-3). 

4. Bowls with rounded bottom, walls curving outward, then nearly 

vertical to an exteiiorly thickened rim with a rounded lip. Rim 
diameter 12-26 cm., majoiity 18-26 cm. (fig. 80-4). 

5. Rounded bowls with direct rim and rounded lip. Mouth diameter 

10-26 cm.; majority 16-26 cm. (fig. 80-5). 

6. Bowls with rounded bottom, upcurving sides, expanding slightly 

at the direct rim with a flattened top. Mouth diameter 20-32 
cm. (fig. 80-6). 
Appendages: Rarely, loop handles with a circular cross section, 1.2-1.5 cm. in 
diameter. Ends have a small conical projection for insertion into the 
body wall but this is smaller than on Ananatuba Phase handles and 
was of little structural advantage. 
Temporal differences within the type: Vessel shape 1 appears to increase 
slightly in frequency and shape 6 to decline, but there are no well-defined trends 
(Appendix, table 36). Differences in vessel shape preference have an areal 
rather than a temporal distribution in the Formiga Phase. 
Chronological position of the type: Gradual increase in frequency from the 
beginning to the end of the sequence of the Formiga Phase. 

formica plain 
Paste: 

Method of manufacture: Coiling. 

Temper: Ground sherd, particles ranging from 0.5-3.0 mm. 

Texture: Moderately hard to break owning to tensile strength produced by 
the hard, smoothed surfaces. Cleavage is very angular and a freshly 
broken edge is soft and crumbly on the core. Admixture of the clay and 
temper is poor, with the temper often lumped together, leaving numerous 
air pockets. 

Color: Ranges from a thin band of orange to tan on the polished surface with 
a light to dark gray core, to an orange band 40 percent of the cross section 
thickness with a dark gray core. Ten percent of all the sherds are fired 
light tan to orange through the cross section. The lighter temper particles 
are often visible in the gray cores. 

Firing: Oxidized under extremely variable conditions; usually incomplete. 
Fire clouds and bright orange splotches are frequent. 
Surfaces: 

Color: Exterior and interior — Ranges from whitish cream to tan to orange 
tan to cream streaked with dull, dark red on the polished surface, to dull, 
orange red or dusty grayish orange on the unpolished surface. All combi- 



MEOGSaS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



233 




I ■ I I I 

I 2 3 CM 

Rim a Handle 
Scale 



Mi'ivivffiiS 



/// 



I I II I I I 

4 8 12 CM 



Vessel Scale 





FiGUKE 80. — Rim profiles and vessel shapes of Embaiiba Plain, Formiga Phase 

(Appendix, table 36). 

nations and ranges of colors are on both the exterior and interior surfaces, 
with the whitish cream to cream streaked with a dull, dark red being the 
most frequent. This streaked color is apparently due to a variation in the 
amount of pressure applied in polishing the slipped surfaces. Polishing 
tracks are always reddish because the pressure of the tool removed a 



234 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

slight amount of the whitish or cream slip from this area permitting the 
orange-red undersurface to show through. A well-polished, evenly smoothed 
sherd usually has a redder color than the others. 
Treatment: 

Exterior and interior — 75-85 percent of the sherds have a very smooth, 
slipped or floated surface with a high luster. The rest are unslipped 
and the polishing is more poorly done, leaving the surface uneven and 
irregular with smoothing tracks and pits, and producing only a slight 
luster. Of the highly lustrous, well-polished sherds, half were given 
a cream slip, reaching 1 mm. in thickness with poor cohesion to the 
underlying surface so that it flakes off easily. Some surfaces were 
highly irregular and uneven when the smoothing was begun and the 
final regularity came as a result of polishing, hence the streaking as 
the surfaces were worn down. On the floated sherds the surface 
was worked when wet bringing up a fine layer of clay; these are dis- 
tinguished by the thinness of the surface layer and its tan-orange 
color. Ten percent of the jar sherds appear to have been polished 
only on the rim and neck leaving the body exterior only lightly 
smoothed. A diagnostic feature of this type is the luster from 
polishing, even though the surfaces as a whole are not regular. 
Hardness: 2.5 
Form: 

Rim: Direct, exteriorly thickened or everted with rounded lip; expanding 

with a flat top. 
Body wall thickness: 5-11 mm., majority 7-8 mm. 
Bases: Typically rounded and unthickened or slightly thickened; one flat 

base joining the sides at a 55 degree angle. 
Vessel shapes reconstructed from sherds: 

1. Bowls with rounded bottom, upcurving sides and expanding rim with 

a flat top. Rim diameter 18-44 cm. (fig. 81-1). 

2. Rounded bowls with a direct rim and rounded lip. Rim diameter 

10-30 cm. (fig. 81-2). 

3. Bowls with rounded bottom, walls curving outward, then nearly 

vertically to an exteriorly thickened rim with rounded lip. Mouth 
diameter 16-30 cm. (fig. 81-3). 

4. Small jars with globular body, constricted mouth and sHghtly everted 

or exteriorly thickened rim with a rounded lip. Mouth diameter 
10-20 cm. (fig. 81-4). 

5. Jars with globular body, walls incurving to direct rim with a rounded 

lip. Mouth diameter 8-20 cm. (fig. 81-5). 
Appendages: One fragment of a handle with a circular cross section 1.5 cm. 
in diameter. 
Temporal differences within the type: None (Appendix, table 37). 
Chronological position of the type: Formiga Plain is the dominant plain 
pottery type in the early part of the Formiga Phase. It undergoes a steady 
decline in frequency and has almost disappeared by the end of the Phase. 

MUCAJA CORRUGATED 

The term "finger pinched" might have been more accurately descriptive than 
"corrugated," but the latter term was chosen because it seems highly probable 
that this technique represents an effort to copy the appearance of a corrugated 
pottery type used by another group. 



"v^N^f^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 



235 






I ■ I I I 

I 2 3 CM 
Rim Scale 




I I I I I I I 

4 8 12 CM 

Vessel Scale 



Figure 81. — Rim profiles and vessel shapes of Formiga Plain, Formiga Phase 

(Appendix, table 37). 



236 



BUREAU OF AMERICAN ETHNOLOGY 



[BULI.. 167 



Paste: Always on Embadba Plain; see that pottery type for descriptions of 

temper, firing, etc. 
Surface: 

Color: See Embatiba Plain. 

Treatment: Exterior — Coil lines left visible and embellished by pinching 
horizontally between the thumb and forefinger, creating two large and 
deep impressions separated by a narrow prominent ridge. An alternative 
was to press downward along the coil edge making a row of impressions 
but without the prominent ridge (pi. 48, a-h) . 

Interior — Smoothed suflSciently to erase coil lines, but leaving irregular- 
ities and often smoothing tracks. 
Hardness: 2. 
Form: 

Rim: Everted with slight, exteriorly thickened and rounded or pointed lip 
on the sherds from J-4. Those from J-6 are direct with a rounded lip. 
Body wall thickness: 4-20 mm.; majority 6-10 mm. 
Base: Flat, joining the wall at an angle of 40-65 degrees. 
Vessel shape reconstructed from sherds: 

1. Semicylindrical jars or bowls with flat bottom and everted, thickened 

rim. Rim diameter 16-32 cm. majority 16-18 cm. (fig. 82-1). 

2. Jars with rounded body, insloping neck and direct rim with a rounded 

lip. Mouth diameter 16-18 cm. (fig. 82-2). 




I I I 



J 



I 2 3 CM 
Rim Scale 




' ' I I I I ' 

4 e 12 CM 

Vessel Scale 



Figure 82. — Rim profiles and vessel shapes of Mucajd Corrugated, Formiga 
Phase (Appendix, table 38). 



Temporal differences within the type: Crudest and least resembling true 
corrugation at the time of its earliest occurrence in the Formiga Phase. The 
small sample shows no trend in vessel shape (Appendix, table 38). 

Chronological position of the type: Appears suddenly in the latter part of 
the Formiga Phase sequence. 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



237 



PSEUD0-SIP6 INCISED — FORMIGA PHASE VARIETY 

Paste and surface: The majority are on Formiga Plain, remainder on either 
Embauba Plain or Coroca Plain; see those type descriptions for details of paste, 
temper, color, etc. 
Form: 

Rims: Direct or exteriorly thickened with a rounded lip. 
Body wall thickness: Range 3-8 mm., majority 6 mm. 
Base: Rounded. 
Vessel shapes reconstructed from sherds: 

1. Bowls with a rounded bottom, sides curving inward to an exteriorly 
thickened rim with a rounded lip. Rim diameter 24-30 cm. (fig. 
83-1). 





Rim 



2 3 CM 
Scole 




4 8 12 CM 
Vessel Scole 



Figure 83. 



-Rim profiles and vessel shapes of Pseudo-Sipo Incised, Formiga 
Phase Variety (Appendix, table 38). 



2. Bowls with a rounded bottom, outcurving sides and direct rim with 

a rounded lip. Rim diameters average 22 cm. (fig. 83-2). 

3. Jars with rounded body, walls insloping to an exteriorly thickened 

rim with a rounded lip. Rim diameters range 6-20 cm., majority 
14-20 cm, (fig. 83-3). 
Decoration (pi. 48, i-n) : 

The incised designs are copies from Sip6 Incised of the Ananatuba Phase, 

(See Sip6 Incised, pp. 185-187, for details of technique and motif.) 
Motif: 

At Site J-6, design types 1, 4, 5, and 6. 
At Site J- 18, design type 4. 
Temporal differences within the type: None (Appendix, table 38), 
Chronological position of the type: Occurs sporadically throughout the 
Formiga Phase sequence. The fidelity of the copy of Ananatuba Phase motifs 
indicates some direct contact with either the Ananatuba Phase or the Man- 
gueiras Phase and serves to compensate somewhat for the absence of a strati- 
graphic link. 



238 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 



SAUBA BRUSHED 

Paste and surface: On Coroca Plain, see that type description for details of 

temper, color, firing, etc. 
Form: 

Rims: Slightly thickened on the exterior or interior with a rounded lip. 

Body wall thickness: Range 4-7 mm., majority 4 mm. 

Base: Probably rounded. 

Vessel shapes reconstructed from sherds: 

1. Jar with rounded body, insloping neck and exteriorly thickened rim 

with a rounded lip. Rim diameter 18-22 cm. (fig. 84-1). 

2. Jar with a globular body, walls incurving to a direct rim with a rounded 

lip. Rim diameter 14 cm, (fig. 84-2). 



I 2 3 CM 
Rim Scale 




Figure 84. — Rim profiles and vessel shapes of Sadba Brushed, Formiga Phase 

(Appendix, table 38). 

Decoration : Exterior surface covered with parallel brushings typically 0.5-1.0 mm. 
wide and from 1.0-4.0 mm. apart with the majority 2.0 mm. apart, applied when 
the clay was damp enough to leave the marks sharply defined. A small per- 
centage were brushed twice with a bunch of twigs held at different angles to 
produce diagonal or criss-cross patterns. 

Temporal differences within the type: None discernible in the small sample 
available (Appendix, table 38). 

Chronological position of the type: Present throughout the Formiga Phase. 

unclassified decorated 



A few scattered sherds with incised or punctate decoration were found at all 
Formiga Phase Sites. They are tabulated below: 
Unclassified incised: 

1. Fine-line incised, marks ranging from very fine to 1 mm. wide, usually 

running parallel, occasionally cross hatched or zigzagged. Total sherds 
from J-6, 37; from J-18, 1; from J-4, 1. 

2. Simple, rectilinear patterns with intersecting straight lines. Total sherds 

from J-18, 6. 



EvSf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 239 

3. Miscellaneous badly eroded or very small sherds with traces of incision. 
Total from J-6, 15 sherds. 
Unclassified punctate: 

1. Rows of ovoid punctates beginning below the rim and probably covering 

the exterior; shape of marks varies greatly on the same sherd, running 
from triangular to ovoid to rectangular, depending on the angle at which 
the tool was held. Total from J-6, 3 sherds; from J-18, 6 sherds. 

2. Row of punctates along the rim exterior. One sherd from J-6. 

Pottery Artifacts 

The only pottery artifacts found were two sherds that had been 
worked into crude disks and perforated through the center, possibly 
for use as spindle whorls (fig. 77). Both of these came from Site 
J-6 — Formiga (see p. 226 for details). 

Nonceramic Artifacts 

Artifacts or fragments of stone, bone or other nonpottery material 
are completely absent. 

Ceramic History 

The seriated sequence shown in figure 85 is based on the analysis of 
7,234 sherds from J-4 and J-6 on the north coast and 643 sherds from 
J-18, in central Marajo. The trends were derived from 4 stratigraphic 
cuts, 1 in each of the mounds at J-6 and 1 at J-18, the shallowest of 
which produced sherds to a depth of 60 cm. (Appendix, table 34). 

The passage of time in the Phase is marked by changes in the popu- 
larity of the three plain wares: Formiga Plain, Coroca Plain, and 
Embaiiba Plain. The earliest level produced only Formiga Plain, a 
cream to dark-red ware with a lustrous surface. From a climax of 
95 percent in the lowest levels of J-6, it declines to 2 percent by the 
end of the Phase. Taking its place are the unpolished types: 
Embatjba Plain, with a dull grayish or brownish surface, and Coroca 
Plain, with a whitish to reddish surface. The latter has an early 
climax, followed by a slow decline, while Embauba Plain increases 
gradually until the end of the Phase. The percentage of Coroca Plain 
is considerably greater in the lower levels at J-18 than in contemporary 
levels at J-6, reflecting a regional difference. J-4, a one level site, 
seriates near the end of the J-6 occupation by virtue of its high per- 
centage of Embauba Plain. 

The decorated types associated with the Formiga Phase are for the 
most part undistinguished and unclassifiable. Of the two identifiable 
types, Saiiba Brushed is present in nearly all levels at J-6, usually in 
a fraction of a percent, and is comparable in its execution to the 
brushed types of the other Tropical Forest Phases. Pseudo-Sip6 
Incised, so-called because of the similarity of the motifs and technique 
of incision to Sipo Incised of the Ananatuba Phase, is also found 



240 BUREAU OF AMERICAN ETHNOLOGY [bull, le? 

throughout the sequence, although somewhat more sporadically and 
never exceeding 1.4 percent in any level. 

In the final third of the Formiga Phase there is a sudden introduction 
of a finger-pressed surface decoration that may represent an attempt 
to imitate the appearance of corrugation. The fact that the earliest 
as well as the crudest examples of this technique occur at J-4^'' may 
indicate a stimulus from somewhere to the west. Its introduction at 
J-6 failed to take root, although the specimens from this site approach 
true corrugated ware more closely in that the coil lines remain un- 
obliterated. 

Another innovation at the end of the history of J-6, and one that 
probably brought the Phase to an end, is represented by the intrusion 
of two Marajoara Phase decorated types, Arari Excised (pi. 49, a-e) 
and Guajard Incised (pi. 49,f-j). These appear suddenly at the top 
of cuts 1 and 2, and are unquestionably of Marajoara Phase origin, 
probably acquired by trade. Although the exceedingly poor condition 
of Catarina Plain (which appears about the same time) makes positive 
identification impossible, there is a good chance that it is in reality 
badly eroded Inajd Plain, one of the Marajoara Phase plain wares. 
A few badly eroded sherds from Teso dos China, a Marajoara Phase 
site, showed similar variegation of gray and orange. 

As in the other Tropical Forest Phases, the vessel shapes of the 
Formiga Phase wares fail to show any well-marked trend of increase 
or diminution in frequency. In this Phase, the relatively small num- 
ber of rims makes the situation even more obscure. A computation 
of the ratio of bowls to jars in the three plain wares shows that 85.7 
percent of the Coroca Plain rims are from jars in contrast to 32.4 
percent in Formiga Plain. A similar predominance of jars in the 
more highly oxidized ware is also characteristic of the Ananatuba and 
Mangueiras Phases. In Embaiiba Plain the two categories are more 
equally represented, with 59.4 percent jars and 40.6 percent bowls. 

The most striking aspect of the vessel shapes of Formiga Phase 
wares is their marked spatial distribution. In Formiga Plain, shape 
4 is the dominant jar form at J-4 and J-6 but is absent at J-18 (Appen- 
dix, table 37). In Embatiba Plain, J-18 lacks shape 1, which accounts 
for 40.4 percent of the rims from J-4 and 41.2 percent of those from 
J-6, as well as shape 6, which reaches 11.2 percent at J-4 and 9.1 
percent at J-6. On the other hand, Embauba Plain, shape 3 is absent 
at J-6 and rare at J-4, but claims 58.3 percent of the rims from J-18 
(Appendix, table 36). 

A similar disparity between apparently contemporary Formiga 
Phase sites is evident in types of decoration. Only J-6 produced 

'« The trace shown farther down on the chart (fig. 85) is In the level Immediately preceding that seriating 
above J-4, and probably belongs to the end of that level. 



1= 0-.l5t 
2=0-. 15 
2M5-.3 
SURFAC 
3=0-.l5 
0-.I5 
IM5-.3 
\-.ZO-A 
.15-. 30 
2= .30-. 4 
.30-. 45 
3=.I5-.; 
.45-.6C 
.60 - .75 
3-. 30-. 
3=.45- 
2=.45- 
^ 1= .45- 



- I'. 60 

o 

S 1= .75 








!■ I 

HZ] 

I I 



1 1 



\// / // // //;n 



FORMIGA 
PLAIN 




EMBAUBA 
PLAIN 



I I 



I I 



COROCA 
PLAIN 



SAUBA 
BRUSHED 



PSEUDO-SIPO 
INCISED 



I 
I 
I 

I 

I 

a 
I 



UNCLASSIFIED 
DECORATED 



B 
B 
P7a 



MUCAJA 
CORRUGATED 



CATARINA 
PLAIN 



I I I I I 
20 40% 



ARARI 
EXCISED 



6UAJARA 
INCISED 



FiGCRE 85. — Seriation of Formiga Phase sites on the basis of pottery type frequency (Appendix, table 34). 



5f^^fv«r ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 241 

EVANSJ 

Saiiba Brushed and Pseudo-Sip6 Incised, although these occur in the 
earliest levels and so should have had ample opportunity to diffuse to 
J-18. Since innovations of this type passed rapidly between equally 
widely separated sites of the Mangueiras Phase, it seems reasonable 
to conclude that Formiga Phase villages were relatively more isolated 
and perhaps also less receptive to ceramic innovations than were those 
of the Mangueiras Phase. 

The presence of a decorated variety employing the motifs of Sip6 
Incised of the Ananatuba Phase is susceptible to an interpretation 
similar to that given for Pseudo-Sip6 Incised in the Mangueiras Phase 
(pp. 218-219). The Formiga Phase variety of Pseudo-Sip6 Incised 
appears to be another and perhaps independent example of ceramic 
acculturation. The exact origin in time or place is uncertain since no 
sites showing Formiga Phase-Ananatuba Phase contact similar to the 
Mangueiras Phase-Ananatuba Phase contact at J-7 were located, and 
the type is present in the earliest known levels of the Formiga Phase. 
However, in contrast to what happened in the Mangueiras Phase, the 
ceramic influence appears to have been purely local, since no Pseudo- 
Sip6 Incised sherds were collected from J-4 or J-18. 

Diagnostic Features of the Formiga Phase 

A typical Formiga Phase village was located in the ca7npo but 
accessible to the forest, and adjacent to a stream. At J-4 and 
J-6, this was a small igarape, large enough to provide a water supply 
but too small to be navigable except in the rainy season. The sites 
consist of 1 to 6 independent mounds, which at J-6 are produced by 
the refuse accumulation, but at J-4 and J-18 have an artificially 
constructed core. The mounds at J-4 cover a considerably larger 
area and are individually larger than those at J-6, but the refuse 
deposit has a depth of only 10 cm. in contrast to almost a meter at 
J-6. The presence of a few lumps of clay bearing twig impressions 
may indicate the use of wattle and daub in the house construction. 

All three of the habitation sites are contemporary, as shown by the 
interdigitation of the levels in the strata cuts, with J-6 covering the 
longest span of time and J-4 much the shortest. The ceramic history 
is characterized by the decreasing frequency of Formiga Plain, a 
lustrous ware with a streaked surface, an early climax and gradual 
decline in Coroca Plain, with a whitish to reddish surface, and an 
increase on the part of Embauba Plain, with a dull grayish to brownish 
surface. Decoration is typically crude and nondescript, the only 
consistent types being Sauba Brushed and Pseudo-Sip6 Incised and 
these are present from beginning to end with no notable change in 
frequency. Ceramic artifacts are limited to two worked and drilled 
sherds that may have been spindle whorls. 



242 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

The burial pattern seems to have been cremation and interment 
of the ashes beneath the house in the village refuse. One such burial 
was encountered in J-6, cut 2, at a depth of 45 cm. No grave goods 
of any description were associated. 

The inception of the Formiga Phase is uncertain, but its termination 
is apparently the result of the arrival of the Marajoara Phase, whose 
people came in contact with the village at J-6 just before it was 
abandoned. 

THE ARUA PHASE 

Description of Sites and Excavations 

In spite of the prominence of the Arua in historical accounts of 
Maraj6, only two village sites were found on the north-central coast. 

site J-2/3 — CHAVES AIRPORT 

In 1943 a small emergency landing field was constructed on the 
southeast edge of the town of Chaves, on the north coast of Marajo 
(fig. 86). Leveling operations removed 50 to 75 cm. of earth and 
uncovered 18 plain jars containing bone fragments about 25 cm. 
below the surface. The townspeople recalled that all were broken 
or so fragile that they fell into pieces when removed from the support- 
ing earth. We were unable to discover a single sherd in any one's 
possession in 1948. A small, ungrooved ax found at the same time 
was still preserved. 

Examination of the site revealed scattered sherds on the surface 
at the north and south edges of the field, which were about 60 meters 
apart. The original surface of the ground averages 25 cm. above 
the field level along the north edge and 50 to 75 cm. higher on the 
south edge. The fact that the refuse layer averages 25 cm. in thick- 
ness at both extremes, together with the information that jars were 
found in the intervening area, seems to justify the conclusion that 
we are dealing with the two parts of a single large habitation site. 
The refuse layer was dark-gray, sandy loam, with the sherds concen- 
trated between 20 and 25 cm. below the surface. Flecks of charcoal 
were noted at 15 cm. on the north side. Excavation was carried out 
by cutting back the north bank 1.5 meters inward along a section 13 
meters long, and by peeling off the south bank in a strip 30 meters 
long and 1 meter wide. One hundred and fifty-four sherds from the 
north side and 741 from the south excavation give a total of 895 
sherds from the site. 

In 1948 the north edge of the site was only 100 meters from the 
coast, where the steep, high bank is subjected to continual erosion 
by the waves. At the time the village was occupied the shoreline 
can be said with certainty to have been somewhat farther away. 
The entire area was wooded prior to its clearing for the airstrip. 



EVANsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 



243 




244 



BUREAU OF AMERICAN ETHNOLOGY 



[BDLI,. 167 





Figure 87. 



I CM 



-Stone ax from J-2/3 — Chaves Airport, a habitation site of the 
Arua Phase. 



Ax. — The only nonceramic artifact from J-2/3 was an ungrooved, 
polished ax (fig. 87) of fine-grained, dark-greenish diorite flecked with 
black. Several polishing planes toward the blade and the sides and 
a few pecking marks at the butt end make the sm-face slightly irregular. 
The ax is 9.0 cm. long and 5.6 cm, wide at the convex blade, with 
curving sides and a rounded butt. Thickness is 3.7 cm. The blade 
was chipped in the center during use. 



SITE J-11 — CARMO 

A small habitation was located on the east bank less than a kilometer 
above the mouth of the Igarape do Carmo, a small stream emptying 
into the north coast of Marajo several kilometers east of the town of 
Chaves (fig. 48) . The land in this area is comparatively high and there 
are no surface indications of village refuse. Discovery was made by 
the caboclos because of sherds once exposed in the bank of the igarape 
after a heavy rain. The coastal fringe of forest covers the region and 
the site was overgrown with cane and spiny palms, but no large trees 
were close to the excavation. A cut 1.5 meters square was begun 
about 8 meters in from the igarape, but the sherds were so sparse that 
it was enlarged on all sides in an effort to secure a more adequate 
sample. There was no soil discoloration from habitation, the color 
throughout being light gray. The sherds were 12 cm. below the 
surface and the soil above them was densely compacted with roots. 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 245 

EVANSJ 

Refuse was extremely sparse, even in comparison with other sites of 
the Arua Phase, and only 28 sherds were obtained. 

Data from Other Investigations 

Survey and excavation on the Islands of Mexiana and Caviana 
produced a large number of Arua sites, both habitations and ceme- 
teries. Additional sites were found in the Territory of Amapji. 
Details of these are given on pages 37-41 and 457-524. 

Analysis of Materials 

Since the vast majority of the cultural remains came from the 
Islands of Mexiana and Caviana, the pottery and other artifacts have 
been analyzed and described on pages 525-534. 

Ceramic History 

A seriation of the Arua habitation sites based on vessel shape places 
J-2/3 and J-U in the lower part of the sequence. For details, see 
pages 534-537. 

Diagnostic Features of the AruS Phase 

The Marajo sites are typical of Arua Phase village sites both in 
location and composition. Notable, however, is the absence of any 
report of cemeteries of the type associated with the Phase on Mexiana 
and Caviana. For a general summary of the Phase see page 538. 

COMPARISON AND INTERPRETATION OF THE TROPICAL FOREST PHASES, 
WITH A METHOD FOR COMPUTING VILLAGE DURATION 

By their lack of standardization in shape, their simplicity or absence 
of ornamentation and their exclusively utilitarian purpose, the 
ceramics of the Ananatuba, Mangueiras, Formiga, and Arua Phases 
evoke the image of a level of cultural development where technology 
was competent to remove the concern with subsistence from the 
position of immediate urgency it occupies among Marginal hunters and 
gatherers, but not yet able to assure reliable and constant production 
of a surplus that would permit the technological elaboration and the 
social and religious development attained by the advanced cultures 
of aboriginal South America. The size and composition of the 
habitation sites indicate small, scattered villages. All of these traits 
are descriptive of living tribes of the Tropical Forest culture area, 
and it is probable that the extinct Phases resembled the living tribes 
in other aspects of their culture as well: that they made and used 
baskets, mats, hammocks and canoes; that the sociopolitical unit was 
the extended family or clan, with a chief whose duties and powers 
were limited; that full-time specialists and social stratification were 

391329—57 18 



246 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

absent; that religious observances were primarily of a shamanistic 
character; that crisis rites included the couvade at bu'th and ordeals 
at puberty. 

Although there is ample justification for assigning these four 
archeological Phases to the Tropical Forest culture pattern, it does 
not follow that they were any less distinctive in particular features 
than are an equal number of living Tropical Forest tribes. Although 
settlement pattern and ceramic traits are never used as the primary 
means of differentiating living tribes, where they are overshadowed 
by more striking social and religious differences, these are the only 
aspects of the culture that remain to the archeologist and when 
examined closely they prove to be equally varied. For many of the 
historical problems the archeologist hopes ultimately to solve, a 
complete ceramic definition of the culture is sufficient to provide the 
pertinent data, and a reconstruction of the total culture is not re- 
quired. However, the description of the Ananatuba Phase in terms 
of firing, surface finish and incised designs of the pottery conveys no 
image of the sort we are accustomed to evoke when speaking of a 
living society. It provides no basis for comparison with the cultures 
on the ethnographic level. It is a name in the archeological sequence 
and nothing more. 

In the hope of adding to the Phase definitions and revealing their 
individuality more clearly, an extensive examination was made of 
the only nonceramic data available, the location and composition of 
the sites themselves. A difference in settlement pattern was im- 
mediately evident. For example, the Ananatuba and Mangueiras 
Phase sites are alike in two features: they both occur in the forest 
rather than the campo, and they are not on the immediate coast. 
The Mangueiras Phase people, however, emphasized closeness to a 
navigable stream, whereas in the Ananatuba Phase proximity to the 
campo was more important. The Formiga Phase is distinct from the 
previous two in its choice of a campo location, while the Arua Phase 
sites are typically m the forest on the shore of a good-sized igarape 
or on a smaller one not far from its junction with the coast. 

These differences in village location are likely to be associated 
with differences in less tangible aspects of the culture, especially as 
they relate to the subsistence resources. For instance, it might be 
suspected that the Arua were dominantly riparian, placing greater 
emphasis on the watercourses for food and transportation than the 
Ananatuba, with their immediate accessibility to both forest and 
campo. A difference in mobility, also suggested by the Arua em- 
phasis on coastal and inland water routes and the Ananatuba Phase 
avoidance of the same, is borne out by estimates of village perma- 
nency. 



evTns?^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 247 

No trace was found of postholes, which are usually relied on by 
archeologists to give information on house type. However, the 
assumption that pile dwellings were used is lil^ely to be correct, since 
it is derived from the fact that sites are located on slight natural rises, 
poorly drained and muddy during the rainy season, and from char- 
acteristics of the refuse accumulation. Slight historical corrobora- 
tion is found in accounts of the existence of such structures in the 
area in the 16th and early 17th centuries (Nordenskiold, 1920, p. 7). 
Another clue to house construction comes from the finding of frag- 
ments of clay \vith stick impressions in sites of the Formiga and 
Ananatuba Phases, suggesting the use of mud-plastered walls. 
Nordenskiold (op. cit., p. 3) reports this wall- type used by Indians 
around Roraima in Guiana, as well as from more remote parts of 
Colombia and Bolivia. He was uncertain as to whether it was an 
aboriginal or Em-op ean-introduced technique but our evidence sug- 
gests the former conclusion may apply to the mouth of the Amazon. 

There are several features of the sites that suggest differences in 
village composition. Unfortunately, there is no inforaiation about the 
type of refuse accumulation associated with different house types and 
village arrangements in existing Tropical Forest settlements so that 
the interpretation of the archeological situation must depend on rea- 
soning that seems logical but cannot at the moment be supported by 
ethnographic facts. 

Two basic types of village organization are found among Amazonian 
tribes today: (1) one or more large, communal dwellings, and (2) a 
cluster of separate family houses arranged in a circle or in one or 
more rows. On the one hand, the population is concentrated in one 
or more spots; on the other, it is spread out. Theoretically, these 
differences should result in differences in refuse accumulation. In a 
communal house, there is an approximately even distribution of Hving 
over a continuous floor area. Assuming that the average rate of 
breakage of ceramics over a period of time was constant for the differ- 
ent families occupying the house, and that the sherds were swept or 
dropped through the floor, they would accumulate in the protected 
area beneath the house. They would not be further broken by being 
kicked about or trampled on by the occupants or mixed with more dirt 
than filtered through the house floor or was deposited as silt during 
an exceptionally heavy rainy season. In other words, the midden 
should have the characteristics observed at the Ananatuba Phase 
sites: a relatively smaU, generally round or oval area with the sherds 
thickly concentrated and comparatively large. The use of a communal 
house in the Ananatuba Phase can be checked by comparing the site 
area to the dimensions of modern Tropical Forest communal houses. 
Those of the Tupinambd measure 75 to 90 meters long by 9 to 15 



248 BUREAU OF AMERICAN ETHNOLOGY [bdll. le? 

meters wide, and house 100 to 200 individuals. Since Ananatuba 
Phase sites are smaller (table E), the population of an average Anana- 
tuba Phase village may be estimated as not exceeding 150. 

In the second type of village pattern, the individual houses are set 
at various distances from each other. They may or may not be pro- 
vided with walls, but in either case the exit is readily accessible. If 
the assumption can be made that the path of least resistance would be 
to toss the sherds from a broken vessel out on the ground (as is done 
by caboelos today), they would become scattered in the surrounding 
area. Children playing would kick them about and adults walking 
through the village would step on them and break them into smaller 
pieces. Disintegration of old houses and construction of new ones 
would change the arrangement and allow the refuse to accumulate 
evenly over the habitation area. Exposure to these conditions, as 
well as to weathering from sun and rain would spread the sherds about 
and mix them in the dirt. If the same rate of breakage obtained as 
in the first type of village and the populations were of comparable size, 
we woidd expect the sherds to be scattered over a wider area, to be 
generally smaller, sparser, and mixed with a greater quantity of dirt. 

Mangueiras Phase villages are typically five times larger in area than 
those of the Ananatuba Phase. However, two facts suggest that this 
does not represent a difference in house type, but rather a difference in 
village size. The density of the sherd refuse is quite similar to that in 
Ananatuba Phase villages, with an average of 606 sherds per 15-cm. 
level at J-13 as against an average of 650 per 15-cm. level at J-10. 
In one Mangueiras Phase site, C-3 — Porto Real, the sherds are even 
more concentrated than in any site of the Ananatuba Phase. Further- 
more, the area occupied by C-3 is smaller than that typical of Anana- 
tuba Phase villages, and is too small to represent anything but a com- 
munal type of house. Since it is unlikely that this one village would 
have an untypical house type, it can be concluded that communal 
houses are also characteristic of the Mangueiras Phase. The larger 
area covered by the refuse at most sites suggests that typical villages 
were composed of several such houses, representing a considerably 
larger population than was characteristic of villages of the Ananatuba 
Phase. 

Turning to the Arua Phase, we find that the area of the site averages 
only 154 square meters, except in three exceptionally large sites cover- 
ing over 1,000 square meters (table G). The smaller area would not 
allow enough room for houses of the individual family type, and prob- 
ably indicates that the Arua villages were typically composed of a 
single communal house like those of the Ananatuba Phase, but only 
one-fourth to one-half as large. This interpretation of the Arua can 
be checked to some extent ethnographically. Archeology shows that 



Jf,ff.?«?^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 249 

the Arua migrated to the Islands of Mexiana, Caviana, and Maraj6 
from Brazilian Guiana. Linguistically, they have been identified by 
one authority as Arawak (Nimuendajii, 1948 b, p. 195). This being the 
case, it is probable that the Arua villages on the islands were not 
greatly different from those in the Guianas in more recent times. In 
characterizing them, Gillin (1948, p. 829) says, "Relatively small 
settlements, seldom if ever containing more than 200 individuals, 
oftener 30 to 40, are the rule. . . ." 

Extension of this analysis to the Formiga Phase runs up against 
several complicating factors. The upper levels of the 3 mounds at J-6 
were subjected to much greater damage by erosion and other destruc- 
tive agencies than the lower levels, which softened and broke the 
sherds into smaller pieces and materially increased the count per level. 
At J-4 the mounds are larger and more numerous, but the refuse layer 
is only 10 centimeters thick as compared with a maximum of 90 cm. 
at J-6. The greatest difference between the refuse deposits here and 
those of the other Tropical Forest archeological Phases is the accumu- 
lation of sherds in several independent spots with sterile areas between 
them, rather than in one continuous area. Why this should be the 
case is not clear. There is evidence at J-4 and J-18 that the moimds 
were artificially constructed and then lived upon, but at J-6 the refuse 
deposit extends to the bottom of the mound, indicating that it is purely 
a midden accumulation. These differences in the known Formiga 
Phase sites prevent a simple characterization of the village pattern. 
In regard to house type, the small area occupied by many of the 
mounds (table H) and the association of several mounds in a smaU 
village area suggests that communal houses may have been used in 
this Phase also. 

When it comes to the question of assigning dates or durations to 
these Marajo Phases, new difficulties emerge. None of the existing 
methods of absolute dating for archeological sites are applicable in the 
Amazon area, and there is little prospect of a new method being 
developed that can overcome the handicaps of rapid and complete 
disintegration of all but the ceramic remains. Prodding by the lay- 
man, who always asks, "How old is it?", as well as their own desire to 
give the reconstructed sequence of cultures some point of reference in 
time as well as in space, leads archeologists to search for some means 
of estimating the relative duration of sites and cultures. This must 
often be based on "feehng for the material" acquired by intensive 
study, because of the absence of any standard for correlating a certain 
amount of ceramic change with a given span of years. In the 
hope of achieving some more objective basis for estimating the 
relative durations of the Tropical Forest Phases on Maraj<5, the refuse 



250 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

conditions in the sites were subjected to an analysis that resulted in the 
following interpretations. 

The first aspect of an archeological site that strikes one as likely to 
be of temporal significance is the depth of the refuse deposit. If the 
refuse at Site A is 2 meters deep and that at Site B only 1 meter deep, 
Site A may be supposed to have been occupied for a longer period of 
time than Site B. While this may seem true in theory, it is an unreli- 
able yardstick in practice. Many variable and often accidental 
factors enter into the composition of a refuse deposit, and comparative 
stratigraphy has shown that a shallower deposit may actually repre- 
sent a longer period of time than a deeper one (Ford, 1951, p. 94 and 
fig. 36). A communal house is likely to give a different rate and 
density of refuse accumulation than a village of scattered, individual 
houses. Another possible source of error exists when the refuse 
deposit is a special area set aside by the village inhabitants rather than 
a gradual accumulation over the village itself. 

Prone to error as refuse accumulation appears to be as a basis for 
estimating relative duration of sites, it is the only method that can at 
present be applied to Amazon archeology. Used within a single 
geographical region, limited to cultures of comparable level of develop- 
ment, and shorn of the accidental variations just mentioned, it is 
possible to avoid some of the major pitfalls and to arrive at estimates 
that should give afc least an approximation of relative duration. 

The four cultural Phases under discussion all appear to be typical 
representatives of the Tropical Forest culture pattern. Thus, it is 
likely that a similar average rate of breakage pertained in the different 
Phases and at different times in the same Phase. If this assumption 
may be accepted as valid, then the sherd accumulations can be 
regarded as a constant among the variable features in the growth of 
the midden deposits. In estimating the relative duration of sites 
within a Phase or between Phases, the basis wiU be therefore the total 
number of sherds present in a sample of standard area rather than the 
relative depth of the total refuse deposits. 

The dimensions of the refuse area selected as the basis for computing 
the rate of sherd accumulation depend primarily on convenience. 
If all the stratigraphic excavations had been the same size, that 
figure would have been chosen. Unfortunately, however, three sizes 
are represented: 1 by 1 m., 1.5 by 1.5 m., and 2 by 2 m. The middle 
figure was arbitrarily selected as the standard and the sherd counts of 
the smaller and larger excavations were adjusted to this standard 
area in the following manner. The area encompassed by a cut 2 by 
2 m. is 4 times greater than in one that is 1 by 1 m. The area of a 
1.5 by 1.5 m. cut is 2.25 times greater than that of a 1 by 1 m. cut. 
To adjust the sherd total from a 2 by 2 m. excavation to the standard 



J?v?fvl?^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 251 

for 1.5 by 1.5 m., the total was first divided by 4 (which reduced it to 
the equivalent of a 1 by 1 m. cut) and then multiplied by 2.25 (which 
increased it to the equivalent of a 1.5 by 1.5 m. cut). For excavations 
of 1 by 1 m., multiplication of the sherd total by 2.25 was the only 
step needed. 

One exception was made to the rule that the total sherd count from 
an excavation of standard 1.5 by 1.5 m. area was used to derive the 
site duration. This was J-7 — Sipo, which began as an Ananatuba 
Phase village and later received a Mangueiras Phase influence, ap- 
parently consisting of the invasion and cooccupation of the site by 
people of the Mangueiras Phase. Since it is of interest to know how 
long the village lasted prior to this event and how long it survived 
thereafter, the levels with Ananatuba Phase pottery only were calcu- 
lated separately from those with both Ananatuba Phase and Man- 
gueiras Phase sherds. As it happens, this division occurs in the middle 
of each cut, the lower 2 levels of cut 1 and the lower 3 levels of cut 2 
representing the exclusively Ananatuba Phase occupation, and the 
upper 2 levels of cut 1 and the upper 3 levels of cut 2 the combined 
Ananatuba Phase-Mangueiras Phase period (Appendix, table 21). 

Having decided upon a standard area of excavation, the next step 
was to select a standard density of sherds. The concentration of 
pottery in the village sites of the Ananatuba and Mangueiras Phases is 
remarkable. The 2- by 2-m. square cut at the Ananatuba Phase site 
of J-9 produced 4,596 sherds, and there is no reason to beUeve that 
such a density is not typical of the site as a whole. The possibility 
that we are dealing with a special dump heap rather than a habitation 
accumulation is unlikely since such systematic refuse disposal is not 
found among living tribes of the Tropical Forest culture. Further- 
more, similar high densities occur at other habitation sites of these 
Phases. 

Before assigning a duration for the accumulation of this amount of 
refuse, an attempt was made to find out what rate of accumulation 
was recognized in other areas of the New World with similar levels of 
cultural development in pre-European times. The most obvious 
comparison is with the Southeastern United States, where run-of-the- 
mill settlements were small and scattered like those of the Tropical 
Forest. The information needed for this kind of analysis is rarely 
given in reports on Southeastern sites, but one interesting example is 
provided by WiUey. In his discussion of Mound Field, in Wakulla 
County, northwest Florida, he remarks that 'Tit I . . . had a total of 
4,789 sherds, an amazing number from an excavation three meters 
square and less than one meter deep" (1949 a, p. 60). Like the sites 
of the Ananatuba Phase, the Mound Field site is "small in extent and 
probably represents only a small population" (ibid.). The ceramic 



252 BUREAU OF AMERICAN ETHNOLOGY [bdll. 167 

analysis showed it to have been occupied from early Santa Rosa to 
late Swift Creek times, estimated as from A. D, 900 to 1200 (op. cit,, 
figs. 14 and 20). This estimate of 300 years of uninterrupted occupa- 
tion may be too high, but it seems reasonable to assume that the 
accumulation at J-9, which contained only 193 fewer sherds (i. e., 
4,596) in a cut less than one-half the area and only three-fourths the 
depth of that at Mound Field, must indicate a long period of residence. 
It is inconceivable that such a sherd accumulation could have been 
built up in less than 100 years, and if there is a gross error, it is likely 
to be on the conservative side. However, the primary goal is to pro- 
vide a basis for deriving relative rather than absolute duration, and 
for this purpose the figure of 100 years can be considered as the 
equivalent of 100 percent. A site producing half the number of sherds 
from the standard area will thus be interpreted as having lasted half 
as long, although this may represent an actual duration of either more 
or less than 50 years. 

When the sherd totals per strata cut were reduced or increased to 
agree with the results to be expected from a standard 1.5 by 1.5 m. 
excavation, the site with the largest sherd total turned out to be J-10 
with 2,600 sherds from cut 1. This is 20 more than the excavation at 
J-9, which was compared above to Willey's Mound Field, and is 
preferable to the total from J-9 because it is a round number. A 
further advantage to J-10, cut 1 is that it was excavated in the dimen- 
sions now being used as the standard size and did not have to be 
converted. Using this sherd count as the unit of measure gives the 
formula : 

2,600 sherds per 1.5X1.5 m. area=100 years. 

It should be emphasized that this formula for the rate of sherd 
accumulation is designed specifically for refuse deposits associated 
with houses of the communal type. Since there is reason to beUeve 
that the rate of accumulation may vary with house type, care should 
be taken in using it where houses of the individual family type seem 
indicated by the area and density of the refuse deposit. ^^ 

The application of this formula to the sherd totals per standard cut 
of 1.5 by 1.5 meters from sites of the Ananatuba Phase gives estimates 
of village duration ranging from 4.9 years at J-8 to 147.7 years at J-7, 
Mound 2. However, J-8 is the only village with a duration of under 
98.9 years (table E). 

" This interpretation is illustrated by sites of the MazagSo and Aristfi Phases in the Territory of AmapS. 
In both of these the habitation area is larger than in any of the sites on the islands, ranging from 4,160 to 0,600 
square meters in the Mazagfio Phase and occupying 7,854 square meters in the Arist§ Phase. The sherd 
density is very low, however, totaling on the average under 150 sherds per 15-cm. level at A-2 in contrast to 
over 650 per 15-cm. level in sites of the Ananatuba and Mangueiras Phases. This dispersed site area and 
scattered distribution of sherds fulfills the characteristics deduced as correlated with villages of individual 
family houses (p. 248), and the application of the communal house formula is invalid. For results of the 
use of this formula in the Acauan Phase, see pp. 455-456. 



UG60BRS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 

Table E. — Duration of Ananatuba Phase village sites 



253 



site number and size of original 
strata cut 



Site dimensions 



Site area 

(sq. 
meters) 



Total 
sherds 

from 
the cut 



Sherds per 
standard 

cut of 
1.5X1.5 m. 



Estimated 

duration in 

years 



J-7: 



Ananatuba Phase 



Mound 1, cut 1 (2X2 m.). 

Mound 2, cut 2 (2X2 m.). 

J-8 (1.5X1.5 m.) 

J-9 (2X2 m.) 

J-10 (1.5X1.5 m.)„ 



30 m. dlam... 
35X22 (7) m.. 
30 m. diam... 
20 m. dlam... 
10X50 m 



Mangueiras Phase mixture with 

Ananatuba Phase 
J-7: 

Mound 1, cut 1 (2X2 m.) 

Mound 2, cut 2 



30 m. dlam... 
35X22 (?) m.. 



706 

770+ 

706 

314 

500 



706 
770+ 



818 
4,560 

127 
4,596 
2,600 



2,149 
2,243 



Total duration of J-7 



Mound 1, cut 1. 
Mound 2, cut 2. 



460 
2,568 

127 
2,580 
2,600 



1,210 
1,265 



17.7 

98.0 

4.9 

99.2 

100.0 



46.0 
48.8 



17.7+46.6=64.3 
98.9+48.8=147.7 



This is an unparalleled degree of permanency for villages of the 
Tropical Forest Pattern, as attested by the ethnographical evidence 
(p. 21), but since the formula was designed to err on the conserva- 
tive side there seems to be no way to avoid attributing this duration 
to the Ananatuba Phase sites. If anything, they may have been 
occupied considerably longer. A similar degree of village permanency 
appears to have existed in the Mangueiras Phase. The formula here 
gives a range from 10.3 years at J-17 to 118 years at J-5 (table F). 
Of the 5 pure Mangueiras Phase sites, 2 lasted more than 100 years 
and 2 less than 25 years. 



Table F. — Duration of Mangueiras Phase village sites 



Site number and size of original 
strata cut 



Site dimensions 



Site area 


Total 


Sherds per 


(sq. 


sherds from 


standard 


meters) 


the cut 


cut of 
1.5X1.5m 


3,000 


1,379 


3,100 


706 


2,149 


1,210 


770+ 


2,243 


1,265 


2,250 


3,031 


3,031 


3,848 


599 


599 


7,500 


268 


268 


492 


969 


2,180 


492 


743 


1,672 



Estimated 
dvuation 
in years 



J-5 (1X1 m.) 

J-7: 

Mound 1, cut 1 (2X2 m.) 

Mound 2, cut 2 (2X2 m.) 

J-13 (1.6X1.5 m.) 

J-16 (1.5X1.6 m.)... 

J-17 (1.5X1.6 m.) 

C— S' 

Cutl (lXlm.)_ 

Cut2(lXlm.)_.- 



(3 mounds). 

30 m. dlam. 
35X22 (7) m 
30X75 m—., 
70 m. diam. 
150X50 m.... 

25 m. diam.. 

26 m. diam.. 



118.0 

46.6 
48.8 
116.0 
23.0 
10.3 

84.0 
64.5 



The high degree of village permanency exhibited in these two early 
Phases is in strong contrast to the situation in the Arua Phase. 
Here the depth of the refuse deposit was insufficient to permit strati- 
graphic excavation and the calculation of the sherd count per 1.5 by 
1.5 meter area is derived from information on the exact area covered 
in making the sherd collections and test excavations. The results 
show that the 10 Arua villages for which data are available were 



254 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 



occupied for periods ranging from 1 to 19.2 years (table G). Half of 
these lasted more than 12 years and half less than 12 years. The 
latter figure encompasses the known durations of modern villages in 
the Guiana area from which the Arua are immediately derived, and 
suggests that the estimates may have some validity. 

Table G. — Duration of Arua Phase village sites 



Site number 


Site dimensions 


Site area 
(sq. meters) 


Total 

sherds 

from the 

tests 


Sherds per 
standard 

cut of 
1.5X1.5 m. 


Estimated 
duration 
in years 


J-2/3 


100X50 ( ) m.-_ 


5,000 


895 
28 
151 
1,024 
606 
809 
607 
646 
315 
724 
525 
126 


? 

28 
? 
500 
300 
500 
455 
260 
150 
371 
392 

38 




J-11 


1 


M-7 


15 m. diam 

15 m. diam 

12 m. diam 

15X75 m 

7X20 m 


177 
177 
113 

1,125 
140 
160 

1,500 

300 

150 

19 




M-2 


19 2 


C-5 


11.5 


C-6 


19 2 


C-7 . 


17 5 


C-8 


8X20 m 


10.0 


C-10 


20X75 m 

30X10 m 

15X10 m 

5 m. diam 


5 7 


C-13 


14.2 


C-14 - . 


15 1 


C-15 


1.4 







When the formula is applied to sites of the Formiga Phase, some of 
the variable, external factors affecting its reliability are brought out. 
There is a great difference between the durations derived for the 3 
mounds composing J-6, ranging from 20.1 to 170 years (table H). A 
similar difference in duration is suggested by the seriation of the 
stratigraphic excavations in these 3 mounds (fig. 85). However, the 
mound covering the greatest span of time in the seriation is cut 1 
(Mound 1), whereas the calculation of duration gives the maximum 
span to cut 3 (Mound 2). Examination of the field notes suggests an 
explanation for this discrepancy. The condition of the sherds from 
cut 3 was poorer than from the other 2 mounds, the erosion of the 
surfaces being so advanced that the sherds were thin and fragile. The 
result was a high rate of fragmentation under the pressure of cattle, 
root action, and similar external forces after abandonment of the site. 
This produced a sherd count per 1.5 by 1.5 meter area that is very 



Table H. — Duration of Formiga Phase village sites 



Site number and size of original 
strata cut 



Site dimensions 





Total 


Sherds per 


Site area 


sherds 


standard 


(sq. meters) 


from the 


cut of 




cut 


1.5X1.5 m. 


2,000 


929 


523 


480 


3,365 


1,895 


254 


981 


553 


78 


1,959 


4,407 


150 


643 


643 



Estimated 
duration 
in years 



J-4: 

Main mound (2X2 m.).. 

J-6: 

Mound l.cut 1 (2X2 m.) 
Mound 3, cut 2 (2X2 m.) 
Mound 2, cuts (1X1 m.) 

J-18 (1.5X1.5 m.) 



100X20 m.. 

60X8 m-._ 
18 m. diam 
10 m. diam 
25X6 m.... 



20.1 

73.0 

21.3 

170.0 

24.7 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



255 



much larger than that from any of the other mounds at J-6 and cannot 
be considered as reflecting the original rate of sherd deposition. 

A comparison of the estimated durations of J-4 and J-18 brings out 
another kind of discrepancy. In both of these sites the refuse deposit 
rests on an artificially constructed earth mound, in contrast to J-6, 
where the refuse extends to the original ground surface. The dura- 
tion of 20.1 years for J-4 represents a concentration of sherds from the 
standard area of 1.5 by 1.5m. but only 10cm. in depth. The duration 
of 24.7 years for J-18 is derived from a refuse deposit of much less 
density, since only 120 more sherds were collected in a deposit of more 
than 60 cm. in depth. This situation permitted the excavation of 
J-18 in five levels and stretched out the period of time represented by 
this site on the seriation chart as compared to J-4 with only one level 
(fig. 85). In view of this lack of uniformity in the density of the sherd 
refuse in the Formiga Phase, it is safer to refrain from relying on the 
estimates of duration until more sites of this Phase have been examined 
and the variable features can be more adequately evaluated. 

Two independent checks can be made on these estimates of village 
duration. Rouse (1952, pp. 564-565) has used rate of refuse accumu- 
lation to estimate the duration of the archeological periods on Puerto 
Rico. Using the depth of the deposits belonging to the historic period 
as a basis for calculation, he concluded that 40 years were required to 
accumulate 25 cm. of refuse, giving a rate of 1 cm. per 1.6 years. 
Rouse's figures utilize the actual depth of the deposit, rather than the 
density of sherds in the refuse that forms the basis of our calculations. 
A comparison of the results given by the two methods for sites of 
the Ananatuba and Mangueiras Phases (table I) shows interesting 
concordances as well as noteworthy discrepancies. 



Table I. — Differential results of two methods of calculating rate of village refuse 

accumulation 



Phase and sites 


Our formula: 
2,600 sherds 
per 1.5X1.5 
meter cut 
equals 100 
years 


Rouse's 

formula: 

1 em. deposit 

equals 1.6 

years 


Ananatuba Phase: 

J-7, mound 1, cut 1 


64.3 

147.7 

4.9 

99.2 
100.0 

118.0 
46.6 
48.8 

116.0 
23.0 
84.0 
64.5 


96 


J-7, mound 2, cut 2 


144 


J-8 


24 


J-9 __ 


96 


J-10 


120 


Mangueiras Phase: 

J-5 


144 


J-7, moimd 1, cut 1 


48 


J-7, mound 2, cut 2 


72 


J-13. 


120 


J-16 


24 


C-3, cut 1 


64 


C-3, cut 2 . 


51 







256 BUREAU OF AMERICAN ETHNOLOGY [boll, iot 

Of the 12 duration estimates for villages of the Ananatuba and 
Mangueiras Phases, Rouse's method of calculation gives larger figures 
than ours in 8 cases and smaller ones in 4. The differences amount to 
less than 5 years for J-9, J-13, J-16, the total duration of J-7, Mound 
2 (cut 2), and the Mangueiras Phase occupation of J-7, Mound 1 
(cut 1). Considering the completely independent derivation of the 
two methods of estimating duration, this is a remarkable correspond- 
ence. However, there are 4 cases in which the discrepancy between 
the two results is 20 to 31.7 years: J-5, J-10, the total duration of 
J-7, Mound 1 (cut 1), and the Mangueiras Phase occupation of J-7, 
Mound 2 (cut 2). In each of these, Rouse's method of determining 
duration gives the larger result. It seems possible that this is caused 
by the major consideration that prompted us to substitute sherd 
count for the more usual depth of refuse accumulation, namely, the 
accidental factors that may influence the amount of dirt mixed with 
the sherds under tropical forest conditions. 

The second independent means of checking the reliability of our 
formula comes from ethnographic evidence. The writing of this 
report was interrupted by field work in British Guiana, during which 
we worked on the Upper Essequibo River among the Wai Wai Indians 
(Evans and Meggers, MS.). Since this group has not been disrupted 
by European contact and preserves its Tropical Forest Pattern of 
culture with a high degree of purity, we took advantage of the oppor- 
tunity to test the rate of refuse deposition at two recently abandoned 
Wai Wai villages, where the actual period of habitation could be 
determined from informants. Since these Indians use a communal 
house, the rate of accumulation should be comparable to that in the 
Ananatuba, Mangueiras, and Arua Phases. The only potential 
source of disagreement lies between the dirt floors of the Wai Wai and 
our interpretation that houses on piles may have been used by the 
archeological Phases. 

There were two abandoned Wai Wai villages for which information 
on length of habitation was available. E-2 was said to have been 
occupied for about 6 years and E-11 for 3 to 4 years. From each of 
these sites a sherd collection was made from a measured area and the 
sample was counted. The result was then converted into the standard 
cut dimensions and the site duration computed according to the for- 
mula. Classification showed that the sherds from this part of E-11 
belonged to an earlier (Taruma) occupation of the site, but the 
results at E-2 are almost identical to the duration given by the Wai 
Wai informant (table J). 

In summarizing this effort to derive site duration from the density 
of the sherd refuse, it may be said that the method appears to have 
definite possibilities for application to Tropical Forest archeology. 



HEOGEES AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 
Table J. — Wat Wai village duration 



257 



Site 


Excavated 
area 


Sherds 
recovered 


Sherds per 
1.5X1.5 m. 


Duration 

estimated 

from sherd 

density 


Actual 
duration 


E-2 - -- 


Meters 
2X2 


317 


177 


Years 
6.82 


Years 

6 







Before it can be accepted without considerable qualification, however, 
more tests must be made in living or recently abandoned villages where 
the conclusions derived from sherd density can be checked against 
the actual period of habitation. With civilization rapidly encroach- 
ing on the remaining unacculturated tribes in the Amazon area, it is 
to be hoped that ethnologists will cooperate in securing the necessary 
information. On the archeologists' part, it would be advisable to 
base the duration estimate on an average density of sherds derived 
from several samples at a site rather than a single excavation as we 
have used in most cases here. This would minimize errors arising 
from the accidental selection of an unusually sparse or dense part of 
a site for excavation. 

CHARACTERIZATION OF THE TROPICAL FOREST PHASES 

Based on the archeological evidence, the four Tropical Forest 
Phases on Maraj6 can be characterized as follows: 

Ananatuba Phase. — The people who left the remains identified 
ceramically as the Ananatuba Phase lived in isolated villages in an 
area whose known extent is from the north-central coast of Mara jo 
inland toward the Rio Anajds. A single communal dwelling large 
enough to house between 100 and 150 individuals constituted the 
village, which was located in a patch of forest at the edge of the campo. 
Houses were probably raised on piles and may have had mud-plas- 
tered walls. Every bit of evidence, whether derived from village 
location, ceramic quality and stability, or village duration (estimated 
as typically 100 years), points to a quiet, peaceful existence, uninter- 
rupted either by exhaustion of the food supply or by raids from 
belligerent neighboring tribes. This undisturbed type of life may be 
the reason that Ananatuba Phase ceramics are of such good quaUty 
and include the only weU-developed decorated ware (Sip6 Incised) 
present in the Tropical Forest archeological Phases. Two comments 
can be made in regard to burial practices, both negative ones: surface 
or mound urn burial was not the method of disposal of the dead, and 
abandonment of the house at the death of an occupant was not a 
custom of the culture. 

Mangueiras Phase. — Sites of the Mangueiras Phase have been 
found on central and northern Mara j 6 and on southern Caviana. A 



258 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

location was chosen in the forest accessible to a navigable stream. 
Villages appear to have been typically composed of several large 
communal houses, suggesting a village population in excess of that in 
any of the other Phases. Estimated duration of the villages varies 
from 10 to 118 years, and the number of simultaneously occupied sites 
for this Phase is indicative of a flourishing and expanding culture. 
This interpretation is further supported by the way the Mangueiras 
Phase people were able to dominate and assimilate the population of 
the Ananatuba Phase village of J-7, which they conquered. The 
promptness with which ceramic decoration of the Sipo Incised tradi- 
tion was adopted illustrates the receptivity of the Mangueiras Phase 
to new ideas, and the rapid diffusion of Pseudo-Sipo Incised to distant 
villages indicates constant intercommunication. The pottery of this 
Phase is well made and durable, predominantly incompletely oxidized- 
fired in the early period and becoming completely oxidized in late 
times. This is the only Tropical Forest Phase making pipes, figurines, 
and labrets of pottery. There is no positive evidence for disposal of 
the dead, and the same negative considerations mentioned for the 
Ananatuba Phase apply here as well. 

Formiga Phase. — This cultural complex is distinguished by a settle- 
ment pattern in which the village was situated in the campo and not 
readily accessible to navigable water (except J-18 — Coroca). Two 
sites were identified near the north coast from the 1948-49 field work 
and one more has since been found just southwest of Lago Arari. A 
low, artificially constructed, earth mound underlies the refuse accum- 
ulation where the land is low, perhaps to raise the village area suf- 
ficiently to prevent its inundation during the rainy season. Formiga 
Phase ceramics are poor quality and unresistant to erosion. A clue 
to burial customs is presented by the discovery of a cremation in- 
trusive into the refuse at J-6. Contemporary Formiga Phase settle- 
ments differ strikingly in vessel shape preference and decorative 
technique, indicating either a high degree of isolation or an unusual 
lack of interest in ceramic matters. 

Arud Phase. — The archeological evidence is supplemented with an 
occasional hint from historical (pp. 579 ff.) and ethnographical (p. 249) 
sources to produce a characterization of the Arua. These people 
lived in very small communities, probably typically a single communal 
house sheltering half a dozen or less families, located on the shore 
of a navigable stream near the coast. This proximity to a ready 
route of travel and the extremely short duration (estimated 1 to 20 
years) with which the majority of the sites were occupied give an 
impression of mobility to the Arua culture that contrasts markedly 
with the sedentariness of the other Phases. There is a possibility 
that abandonment of the village was customary at the death of an 



EVA^Nsf^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 259 

occupant, a practice typical in the Guianas today (Gillin, 1948, p. 
856). Arua dead were buried in cemeteries, the bones placed in large 
jars that were set on the surface of the ground in a forested spot. 
Practically the only ceramic decoration used by these people was 
applied to the burial urns. An occasional burial offering consisted of 
a small, crude pottery bowl, a polished stone ax or, after European 
contact, glass beads and other trade objects. 

THE MARAJOARA PHASE 
DESCRIPTION OF SITES AND EXCAVATIONS 

The existence of the Marajoara Phase has long been known, and 
prior to the 1948-49 expedition it was thought to be the only one 
occupying Marajd. It is distributed on the eastern half of the 
island, within a circle roughly 100 km. in diameter, with its center at 
Lago Arari (fig. 145). The sites composing J-14 and J-15 are near 
the southwestern limit of this area, at the headwaters of the Rio 
Anajas. 

SITE J-14 — MONTE CARMELO 

The Monte Carmelo group consists of three mounds situated on the 
main course of the upper Rio Anajas (fig. 47). Although the site 
was visited by Steere (1927) in 1879 and Holdridge (1939) in 1931, 
neither recorded an accurate description (see pp. 308-309). 

Before beginning excavation, a reconnaissance was made to de- 
termine the nature and extent of the sites. The two largest mounds 
are along the south bank of the river, separated by 100 meters of low 
land, which is flooded during the rainy season. Both are cemeteries. 
The third is about 150 meters north of Mound 1, on the opposite side 
of the river. It is considerably lower in elevation and was identified 
as a habitation site. Of the three. Mound 1 appeared to be the least 
disturbed by erosion and cultivation, and it was selected for more 
intensive examination. 

Mound 1, Guajard. — This mound measured 121 meters long by 56 
meters wide at the end of the 1949 rainy season, when the water was 
at its highest level (fig. 88). Land was inundated on all sides, making 
approach possible only by boat. The present contours suggest that 
it was constructed in two parts, leaving a "waist" a little west of the 
center produced by a depression on the top and indentations on the 
north and south sides. The east end is higher than the west, the 
former rising 6.50 meters above the water, the latter 4.75 meters. 
The sides slope upward at approximately a 30-degree angle to a flat- 
tened platform 20 by 70 meters. The north side has suffered most 
from erosion and sm-face sherds are particularly abundant there. 
Near the east end, where the slope had been cleared for a modern 



260 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 




.i.»*i» 



i ' ' ' ' A s«ii 



WOUND • • HONTG CAMCLO 




Figure 88. — Plan of J-14, Mounds 1, 2, and 3, a mound group of the 
Marajoara Phase. 



house, the circular mouths of large jars were outlined on the surface- 
The entire site is overgrown with large trees, among them cacao, 
which the present inhabitants believe to be of Indian origin (pi. 34, a). 
The surface collection was selected in a manner similar to that from 
J-15, Mound 1 (see p. 286), and produced a variety of decorated 
sherds, a miniature stone ax (fig. 132, b), and a pottery spindle whorl 
(fig. 136, b) (see pp. 372 and 380 for descriptions). 

Cut 1, 1.5 meters square and excavated by 15-cm. levels, was begun 
in the area of highest elevation, 50 meters from the east end of the 
mound. The soil for the first 20 cm. was dark gray-brown loam 
containing many sherds and roots. The broken edges of a nest of 
four vessels were encountered in the second level on the northwest 
side, with the base of the largest resting at 55 cm. (fig. 89). The 
body of this jar (A), measuring 70 cm. in maximum diameter, was 
intact below the shoulder. Large sherds from the rim were broken 
off and inverted around the neck of the second jar, Avhich had been 
placed inside. 

Jar A, Joanes Painted (fig. 90): The exterior is covered with a paper-thin, 
white slip, with lumps and irregularities where applied unevenly or dried before 
well-smoothed. Slip continues over to the interior of neck. Remnants of poly- 
chrome painted design, composed of wide (1.1-2.2 cm.) and narrow (pairs or 
threes, 1.5 mm. wide), red and black lines, covering neck and body. Two sherds 



syAN^r ^""^ ARCHEOLOGY AT MOUTH OF AMAZON 



261 




891329—07 ^10 



262 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 




Figure 90. — J-14, Mound 1, cut 1, vessels A, B, C, and D, Marajoara Phase. 

with convex bosses averaging 8 cm. in diameter, one with heart-shaped applique, 
belonged on the neck (cf. Palmatary, 1950, pi. 90, a). The thickened rim exterior 
was painted with a wide red band. As reconstructed, the jar had a globular body 
70 cm. in maximum diameter, an insloping neck and rim diameter of 50 cm. 
Wall thickness at neck and shoulder varied between 1.0 and 1.5 cm. The paste 
had a gray core. 

Four centimeters of black, ashy dirt containing fragments of bm'nt 
bone and tiny sherds remained inside the bottom of this jar. A 
small, blunt stone ax (fig. 133) lay outside near the base. 

The rim of the second vessel (jar B) was also broken off and scat- 
tered around the edge. It was similar in shape but slightly smaller 
than jar A. Inside were a small jar (D) resting on a small open bowl 
(C), and the remains of a cremation burial. The fact that these two 
jars, when intact, would project above the surface at its present level 
is an indication of the minimum amount of diminution the mound 
must have suffered by erosion since Marajoara times. 



Ev^Nsf^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 263 

Jar B, Joanes Painted (fig. 90): White slip on exterior, smooth and slightly 
undulating to the touch, occasional small lumps and scattered crackle. Slip 
carried over rim top. Interior rough because of protruding temper grains and 
insufficient smoothing. Traces of paint on neck and body indicate the design 
to have been polychrome, black predominating on the neck, with red used for 
accent, and equal or greater use of red with black on the body. Lines appear to 
have been wide (7-12 mm.). No sherds were found with relief or applique. 
Thickness of wall at neck and shoulder varied from 1.1 to 1.7 cm. Reconstructed 
height was 70 cm., maximum body diameter 65 cm., and rim diameter 39 cm. 
The paste had a gray core. 

Jar C, white slipped (fig. 90): White slip covering interior, paper-thin, evenly 
applied, with prominent coarse crackle, and showing no traces of paint. Exterior 
somewhat irregular, with finger-print smoothing marks running parallel to the 
rim, pocks and slight lumps. The profile is asymmetrical, with the side walls 
varying from curved to angular. An oval pedestal base 12.5 by 15.5 cm. in 
diameter is formed by a coil 1 cm. wide and 2 cm. high. Rim diameter, 34 cm. 

Jar D, Inajd Plain (fig. 90) : Exterior and interior well smoothed. Flattened 
bottom, 7 cm. in diameter; globular body, 24 cm. diameter; rim diameter, 16 cm.; 
height, 21 cm. Rim ornamented with applique nubbins 1.7-2.2 cm. long and 
1.2 cm. wide, with two vertical notches on each. 

These small vessels rested at a depth of 30 cm. below the surface. 
At this level, pockets of orange-brown clay began to appear. The 
base of jar E rested at 45 cm. and although badly shattered, it shel- 
tered black, ashy soil with flecks of bone and lumps of tafFylike clay 
flecked with red, yellow, and black, indicating a cremation. No 
teeth were found. 

Jar E, Joanes Painted (fig. 91): A paper-thin, white, smooth and even slip 
covers the exterior from the rim to a ridge 6.5 cm. above the base. Base of ex- 
terior and entire interior fired an even shade of orange; surface regular but not 
smoothed sufficiently to obliterate slight pits. On opposite sides of the body 
in the area of maximum diameter are two anthropomorphic faces formed by 
applique strips. The eyebrows are 2 mm. high while the nose is 7 mm. high. 
The arched eyebrows continue halfway around the side, meeting a painted red 
line that borders the chin. The remainder of the slipped surface bears traces of 
a geometric design in paired red lines 2-3 mm. wide. The jar has a flat bottom 
12 cm. in diameter, a depressed-globular body 37 cm. in maximum diameter, a 
short vertical neck 9 cm. high and 20 cm. in diameter, terminating in a direct 
rim. Total height is 31 cm. 

As the 4th level was begun, it became apparent that continuing the 
excavation in 15-cm. levels would not be practical. Sherds in the 
fill were rare except as parts of burial jars, and the position of the 
jars bore no relation to the arbitrarily divided levels. Records were 
instead kept of the position and condition of each jar as it was en- 
countered in the excavation. 

The base of jar F rested at a depth of 60 cm. in the center of the 
cut. The lid was broken but in place covering the mouth. Black 
ash inside indicated that the jar had contained a cremation. 

Jar F, Inajd Plain (fig. 92): Both exterior and interior are slipped with a 
layer of the same clay as the paste, averaging 0.75 mm. thick. This slip is fired 



264 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL, ler 




2 CM 



Figure 91. — J-14, Mound 1, cut 1, jar E, Marajoara Phase. 



orange on the exterior and steel gray on the interior, forming a distinct compact 
layer easily seen on the cross section. Brush marks cover both surfaces, making 
them uneven, and on the neck exterior the slip has been wiped in such a way as 
to produce two ridges giving somewhat the effect of unsmoothed, overlapping 
coils. The jar is 30.5 cm. tall, with a short, slightly insloping neck 5.5 cm. high, 
a globular body 38 cm. in diameter, and a short, pedestal-like base 1 cm. high and 
16 cm. in diameter. The direct rim is 18 cm. in diameter. The lid, Anajds Plain 
Incised, is likewise covered with a thin slip of the same clay as the paste, and 
bears the marks of a broad smoothing tool. It is a bowl with a rounded bottom 
and short, slightly outsloping sides. Rim diameter is 24 cm. ; height of the wall 
3.4 cm; total height, 6 cm. Decoration consists of 3 parallel, incised lines 1-2 
mm. wide running horizontally in the area between the angular junction with the 
base and the rim thickening. The core is completely oxidized. 

Toward the northeast side of the cut, with its base at a depth of 
55 cm., was the globular bottom of a small, broken, plain jar (jar G). 
The neck was missing and the body was filled with orange clay. The 
existing fragment was 21 cm. high, 27 cm. in maximum diameter, 
with a rounded bottom 8 cm. in diameter. The soil in the entire 
north half of the cut at this depth was bright orange, becoming 
browner toward the south, and contained many hard, fire-burnt 
lumps of clay. 

Jar H was encountered in the west corner with its base at a depth 



Bv^Nsf ^ ^° ARCHEOLOGY AT MOUTH OF AMAZON 



265 




2 CM 



Figure 92. — J- 14, Mound 1, cut 1, jar F, Marajoara Phase. 

of 85 cm. Except for a broken rim edge, it was intact. A bowl-like 
lid was inverted over the mouth and extended slightly beyond it. 
Inside the jar were bone ash and sherds that appeared to be frag- 
ments of a platterlike bowl. Two small bowls, superimposed, rested 
on the shoulder. The upper one, slightly the larger, contained black 
dirt and a couple of tiny sherds. At a depth of 75 cm., fragments 
of a human skeleton were found adjacent to the jar, on the outside. 

Jar H, Joanes Painted (Jig. 93): White slipped on neck interior and extending 
over exterior to just below the maximum diameter, smooth and even except 
toward the lower limit. The remainder of the surface, both interior and exterior, 
is also well smoothed. Dark-gray fire clouds are scattered on the exterior. The 
paste has a gray core. The neck is embellished with low relief bands, the inter- 
vening areas painted geometrically with red lines 2 mm. wide. Traces of paint 
on the body reveal a polychrome pattern, with narrow, paired, red lines separating 
bands of interlocking spirals and steps done in black. The jar has a slightly 
outslopiug neck, 7 cm. high; a depressed-globular body, 40 cm. maximum diameter, 
and a flattened bottom, 12 cm. diameter. Total height is 42 cm. The lid is an 
Anajds Plain Incised bowl, 28 cm. in exterior rim diameter and 6 cm. deep. The 
interior is smooth, either well-floated or slipped with the same clay as the paste, 
which has an orange core. The exterior is much rougher, with prominent smooth- 
ing marks, pocks, and some crackle. Decoration is limited to two parallel, in- 



266 



BUREAU OF AMERICAN ETHNOLOGY 



[BDLL. ler 




J!',ffv«f " *'"' ARCHEOLOGY AT MOUTH OF AMAZON 267 

EV AN o J 

oised lines around the flat upper edge of the exteriorly thickened rim. Many 
black and gray fire clouds are present on the exterior. 

Both of the small associated bowls are Inajd, Plain. The lower and smaller 
one (pi. 66, b) is 8 cm. deep, 16 cm. in diameter at the direct rim, and 7 cm. in 
diameter at the slightly rounded base. The walls are 5 mm. thick. It is regular 
and symmetrical, with the surfaces well-polished. The upper bowl (pi. 66, a) is 
8.5 cm. deep, with a flattened bottom and convex sides curving outward to a 
maximum diameter of 21 cm. and then inward to the constructed mouth with a 
diameter of 17 cm. Both surfaces are well-smoothed, with a few fire clouds on 
the exterior. 

The upper edge of jar I appeared in the 4th level, with a carinated 
bowl resting inverted inside its flaring rim as a lid. The neck and rim 
had been broken off by earth pressure and pushed down inside the 
body, which was leaning slightly to the east. The jar was filled with 
very wet earth, which had reduced the bones of a secondary burial to a 
puttylike consistency. The long bones had been laid in a pile along the 
east side. Associated was a complete red-slipped tanga. 

Jar I, Joanes Painted (fig. 94) : The entire exterior of the vessel and the flaring 
rim are covered with thin, white slip, eroded off in spots and revealing a light- 
orange undersurface. Both surface and slip are smooth. The paste has a gray 
core. The jar is 82 cm. in total height, and 22 cm. tall from the base of the neck 
to the upper rim edge. Rim diameter is 42 cm.; maximum body diameter, 58 
cm.; and the diameter of the flat base, 13 cm. Wall thickness is 1.6 cm. at the 
everted rim and 1.1-1.2 cm. on neck and body. The neck bears two anthro- 
pomorphic faces on opposite sides, with the features indicated by low relief: 
heavy-lidded eyes, U-shaped nose with prominent tip, 8-shaped ears, and pro- 
truding mouth. Two appliques 7.6 cm. long, 3 cm. wide and triangular in profile 
were apparently attached to the shoulder and may have been lugs. The entire 
exterior was painted in black and red in a geometric pattern of wide and narrow 
lines similar to that on Jar L. The lid is Ararl Plain Excised (pi. 58, i). Both 
surfaces were covered with a slip of the same clay as the paste (which had a 
gray core) and well-polished so as to produce a slick finish and a slight luster. 
The exterior is ornamented with three bands of excision, one on the thickened rim 
exterior, one on the concave side and one around the edge of the curved bottom. 
The excisions are deep and the design regularly executed. The bowl is 34 cm. in 
maximum diameter and 15.5 cm. deep. 

Resting at the same level as jar I and with a section of the rim 
broken out where it leaned against the side of the latter, was jar J. 
Although it is considerably smaller than is typical for vessels with 
secondary burials, it conforms to this earlier pattern in having the lid 
resting inside the neck. Inside were fragments of unburned bones, 
sherds from the broken lid and a sherd from the rim of a large Inajd, 
Plain jar, which had an original diameter of 54 cm. (no other sherds 
belonging to this jar were encountered). 

Jar J, Joanes Painted (fig. 95; pi. 75, a): The paste has an orange core. Both 
surfaces are covered with a white slip, with a few fire clouds on the inner mouth. 
The exterior rim edge and the pedestal base are painted red, and the remainder of 
the exterior is covered with concentric circles and spirals in black, with the inter- 



268 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 




Figure 94, — J-14, Mound 1, cut 1, jar I, Marajoara Phase. 



vening areas painted solid, giving a negative effect. The painted lines are crooked, 
wavering, and unequally spaced (cf. Palmatary, 1950, pi. 85, a). The vessel is 26 
cm. in height, 21 cm. in rim diameter, 23 cm. in maximum body diameter and 14 
cm. in base diameter. It has a slightly outflaring neck 7.5 cm. tall, a globular body 
and a flat, pedestallike base. The lid was a small, carinated bowl with Ararf Plain 
Excised decoration (pi. 57, a), 15 cm. in diameter and 4 cm. in depth. Both surfaces 
are well-smoothed and the paste has an orange core. The entire exterior is covered 
with the excised design, and the excisions were filled with white. 

In the west corner, at a depth of 1.20 meters, was a fragment of the 
upper part of a small and unusually shaped Inajd Plain vessel with 
appUque decoration. 

Inajd Plain vessel fragment: The fragment has a tall conoidal base, sloping in" 
ward toward the upper part, which is an expanded, cuplike neck terminating in an 
everted rim and flattened lip. Mouth diameter is 12.5 cm., neck height 5 cm., 
diameter at junction with base 6 cm., existing height 10 cm. Crude anthro- 
pomorphic faces ornament opposite sides of the neck. They are composed of oval, 
coffee-bean eyes, a larger similar applique for the nose-mouth, and fillets curving 
upward from the center above and around the eyes. The same ears function for 



HEGOEBS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



269 



both faces and are high relief, with an indentation just above the middle, producing 
an upper and a lower lobe. Although the top is more elaborate, the general shape 
of the base suggests that this is a variation of the tall, semicylindrical potstands 
found with relative frequency in many Marajoara Phase cemeteries. 

At the southwest side of the cut, with its base at a depth of 1.20 
meters, was jar K. The neck and rim were missing and the sherds 
were not encountered in the fill. Inside the jar, with its upper edge 28 
cm. below the broken top of jar K was a large, complete Camutins 
Plain basin, almost level and upright, and containing reddish clay. 
Directly beneath the basin were the remains of a skeleton, in relatively 
good condition as a result of the protection afforded by the basin. The 
skull had been placed at the northwest edge, 45 cm. below the existing 
top of the jar; the ribs were adjacent to the north, the pelvis was to 
the southeast, and the long bones were in the southwest half, 50 cm. 
below the existing top of the jar. Traces of red paint were visible on 
the femur. As many fragments as could be salvaged were preserved 




I I I 
2 CM 



Figure 95. — J-14, Mound 1, cut 1, jar J, Marajoara Phase. 



270 



BUREAU OF AMERICAN ETHNOLOGY 



[BDLL. 167 



and submitted to Marshall T. Newman, United States National 
Museum, for examination. He reported (pers. commun.), "fragmen- 
tary remains of apparently one individual . . . : there is no duphca- 
tion of parts, and all remains are consistent with the picture of an 
adult male. Age cannot be determined, although the teeth show wear 
approaching 4th degree (pulp cavities exposed). There are signs of 
several apical abscesses on the maxilla." 

Jar K, Inajd Plain (fig- 96): The exterior is bright orange, the interior light 
gray, both surfaces undulating and not sufficiently smoothed to remove pits 
and irregularities. The slightly concave base is 18 cm. in diameter. The walls 




4 CM 



Figure 96. — J-14, Mound 1, cut 1, jar K, Marajoara Phase. 

slope outward to the maximum body diameter of 70 cm. (at a height of 20 cm.) 
and then continue upward, curving slightly inward until the shoulder, where 
the curve becomes more pronounced. Existing height (approximately to the 
lower edge of the neck) is 62 cm. The basin found inside is Camutins Plain 
(pi. 67, c), with both surfaces smoothed, leaving the striations of the smoothing 
tool clearly visible. Several dark-gray fire clouds mar the otherwise bright, 
tile-orange surfaces. The thick, everted rim is irregular, with a diameter of 
45-46 cm. Four slight protuberances are distributed along the outer edge. 
The flattened bottom has a diameter of 14 cm. from which the walls curve out 
to the maximum diameter at a height of 7 cm., and then rise vertically to the 
everted rim. Total height is 20 cm. 



f.f?S'^^^ *'"' ARCHEOLOGY AT MOUTH OF AMAZON 271 

In the cast corner of the cut, behind and partly beneath jar I, 
was the upper edge of jar L with its base at a depth of 1.80 meters. 
The large basin that had been inverted and placed over the mouth was 
badly broken (Meggers and Evans, 1954, pi. 7), possibly during 
excavation for the burial of jar I. The jar itself was intact except 
that the widely flaring rim had been broken off, but all the fragments 
were lying around the neck. The interior was filled with dirt contain- 
ing fire-burnt lumps, small sherds, wood ash, and yellow sand, ap- 
parently taken from the fill and put inside at the time of burial. The 
remnants of a human skeleton were arranged in the bottom (depth, 
65 cm. below the top of the neck), with the long bones at the north 
side running east-west. The femurs showed traces of red paint. 
M. T. Newman, United States National Museum reports "no evidence 
of more than one individual represented. This individual has light 
gracile bones, and may represent a sub-adult or adult individual" 
(pers. commun.). A highly polished, orange tanga (pi. 82, c) lay at 
the southeast side. This was the only tanga found that was slipped 
with the same clay as the paste instead of with red, but it exceeded 
all others in perfection of smoothing and completeness of polishing. 

Jar L was resting in the broken bottom of an Inajd, Plain jar, 
the existing fragment of which was 22 cm. in height. No further parts 
of this jar were found. It contained dirt, human bone fragments, 
and a broken red tanga. On the south side of jar L, at the base of 
the neck, were found additional human bone fragments and a broken 
red tanga (pi. 82, e), representing a third burial. 

Jar L, Joanes Painted (fig. 97; pi. 76) : The entire vessel, with the exception of 
the exterior of the flat bottom and the interior below the neck, is covered with 
a white slip, 0.5 mm. thick, well-smoothed, and even. There is one small (15 
cm. dia.) fire cloud on the body, otherwise the white slip was not discolored. The 
paste has a gray core. Two anthropomorphic faces adorn opposite sides of the 
neck. The features are similar in execution to those on jar I, which stood 
above this one: bulbous, half-shut eyes, prominent bifurcated nose, protruding 
mouth, and hour-glass-shaped ears. The ears on this vessel appear to have an 
ornamental spool inserted in the lobe, with a pendant tassel shown in low relief 
(fig. 147, a). The area around the eyes and mouth is painted solid red, which 
continues below the mouth to a cylindrical, horizontally pierced lug. Between 
the ears of the two faces is a small anthropomorphic figure with a highly stylized 
face, the left arm extending upward and the right one bent downward, and the 
legs slightly buckled. The curvilinear motifs filling the background are in red. 
On the body, the design is principally in black, with red used sparingly for accent. 
The design partly carries further the anthropomorphic theme by showing 
stylized arms and hands with four fingers. Below the small figures on the neck 
are stylized faces with double, curled topknots and pronged ears. All the 
remaining surface is divided into small, irregular spaces filled with short lines 
and so expertly balanced as to give the impression of symmetry, although close 
examination shows the treatment of each area to be slightly different. Occasional 
small drops of red paint occur on the body, where they splashed during the 
painting of the neck, indicating that the body was done first. The jar has a 



272 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 




FiGXJBB 97. — J-14, Mound 1, cut 1, jar L, Marajoara Phase. 



widely flaring rim, 65 cm. in diameter, a sloping neck 22 cm. tall, prominent 
shoulders, a rounded body and a flat, slight pedestal base 16 cm. in diameter. 
Maximum body diameter is 70 cm. ; total height, 84 cm. 

Lid, Joanes Painted (fig. 97): Interior and vertical sides of exterior white 
slipped, well smoothed, with slight crackle on exterior. The paste has a gray 
core and the unslipped surfaces are tan to bright orange and smoothed with 
polishing tracks visible. About one-third of the surface is fire clouded. No 
painting is visible except on the exterior edge of the rim, which is red. The 
shape is that of a deep, small-bottom bowl with slightly carinated sides and the 
maximum diameter at the rim. Dimensions are: rim diameter, 48-51 cm.; 
body diameter at carination, 42-45 cm.; base diameter, 15 cm.; total height, 20 
cm.; height above carination, 9 cm. The exterior of the base is slightly concave. 



"vTwIf ^ *~° ARCHEOLOGY AT MOUTH OF AMAZON 273 

Jar L was flanked by two Inaj^ Plain jars, placed at the same 
depth and with their body walls touching those of jar L, so that it is 
evident that all three jars were interred as a unit. The elaborateness 
of the central jar makes it probable that it contained an important 
personage whose comfort in the next world needed to be assured. 
Jar M, on the north side of jar L, was covered with an inverted bowl 
2 cm. smaller in rim diameter than the jar. The interior was filled 
with whitish sand in which were embedded two bowls and the bones 
of two individuals. When found, the bowls were resting at an angle 
with their mouths on the same slope, the smaller one to the north and 
46 cm. deep, the larger to the south and 48 cm. deep (fig. 98). Both 
were filled with dirt, that from the larger containing fragments of 
charcoal. Beneath the bowls were the disarticulated bones of one 
individual, miscellaneous sherds, and, in the center of the jar, a 
complete red-slipped tanga. About 15 cm. below this first set of 
bones was part of a second skeleton. The jar also contained a number 
of miscellaneous nonhuman bones. Field identification gave an 
inventory of 3 skull fragments from small rodents, 4 parts of bird 
skulls including one from a large species like the tuyuyu, and numerous 
postcranial fragments. Both human and animal bones had been 
painted red, traces of the pigment being discernible even on the 
smallest scraps. 

The human bones were submitted to Marshall T. Newman, U. S. 
National Museum, for examination and he was able to identify both 
age and sex: "These bones represent the very fragmentary remains 
of a rather rugged male over 26 years of age, and a gracile female 
between about 18 and 25 years (distal end of clavicle unfused)." 
The male, which was the upper burial, shows one outstanding feature: 
"The glabellar fragment is particularly interesting since it shows al- 
most positive frontal deformation of the sort that levels out the brow 
ridges and glabella, reduces the nasion depression to almost no depres- 
sion at aU, and makes for an almost straight profile from hair line to 
nasal bones" (pers. commun.). The female showed third-degree 
wear on all teeth in spite of her apparent youth, suggesting extremely 
gritty food. It is of interest to note that the tanga appears to have 
been associated with the male rather than with the female. 

Jar M, InajA Plain (fig. 98): The surfaces are dull tan to light brown, with 
patches of orange and red-orange and small, light-gray fire clouds. Temper is 
coarse (one grain was 1.1 cm. in diameter) but evenly distributed. A thin slip 
of the same clay as the paste was applied on the exterior, filling some of the scars 
and pits that remain on the interior but showing scattered crackle. Slight 
horizontal grooves on the exterior reveal where one coil had been joined to the 
next. These coils are 3 cm. wide. The jar is 89 cm. tall, with an exteriorly 
thickened, everted rim 54 cm. in diameter, sides sloping out to a maximum body 
diameter of 65 cm. and then curving inward to a subcorneal base. The maximum 



274 



PUREAU OF AMERICAN ETHNOLOGY 



[BDLL. 167 




Figure 98. — J-14, Mound 1, cut 1, jar M, Marajoara Phase. 



body diameter is attained 35 cm, above the base. Body wall thickness varies 
between 1.4 and 1.7 cm., rim thickness is 3.3 cm. 

The lid is Camutins Plain, with an orange-tan surface and some fire clouding. 
The interior is smooth and even, while the exterior is marred by smoothing marks 
and other irregularities. The short, outslanting sides join the curving bottom 
at an angle of 25 degrees. The base is flattened. Exterior diameter of the 
thickened rim is 52 cm., total depth is 18 cm., and wall thickness, 7 mm. 

The smaller of the two offertory bowls (pi. 66, c) was a variety of Camutins 
Plain, with a dark-tan to orange-tan, slightly fire-clouded surface and a reddish- 
brown paste. Smoothing and polishing striations are visible on all surfaces 
except the exterior of the rounded base, which remains rough and uneven. The 
bowl is 6.5 cm. deep, with a direct rim, slightly insloping sides and a rounded 
bottom. It is asymmetrical, whether seen from above or in profile, the rim diame- 
ter varying from 15.5 to 16.0 cm. Wall thickness is 9 mm. 

The larger bowl (pi. 66, d) is Inajd Plain, with a bright, reddish-orange surface 
except for a dark-gray fire cloud on the exterior. The body of the vessel was 



EVA^sf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 275 

smoothed, leaving polishing marks and a surface that remains uneven and irregular, 
apparently because the smoothing was done after the clay was too dry. The 
underside of the rim and the exterior of the neck were roughened so that temper 
particles protrude in the deep scratches. Rim diameter is 22-23 cm., flaring 
out from a short neck 4 cm. high, joined to the shoulder of the rounded body. 
Total depth is 13.5 cm., body wall thickness 1 cm. As was true of the smaller 
bowl, this one is asymmetrical. At four equidistant places along the exterior 
edge, the rim is expanded slightly to produce a bifurcated ornamental lip. 

Flanking jar L on the opposite side was jar N, also a plain ware 
vessel. It contained light-gray dirt, ash and scattered sherds; hunks 
of yellowish clay appeared at the area of maximum diameter, and 
white, sandy clay filled the bottom. Beneath a layer of miscellaneous 
large sherds were bone fragments too badly disintegrated for preserva- 
tion (jars L through O all had a concentration of sherds as large as 
15-20 cm. just above the bones). No tanga was associated. Instead 
of the usual type of lid, jar N was covered with the body of another 
jar, the base of which extended halfway into the neck of jar N. This 
second jar contained very wet dirt, sherds with ornamented rim, and 
skeletal fragments. The neck and rim were broken off and missing, 
and a large sherd had been laid over this broken top to protect the 
contents. 

Jar N, Inajd Plain (fig. 99): Surfaces light to medium orange, with a light- 
gray to black fire cloud extending over half of the neck and the shoulder. Both 
surfaces are slipped with a thin layer of the same clay as the paste. The interior 
is smoother than the exterior, but both have smoothing lines, grooves and crackle. 
Undulations on the exterior reveal coils 5 cm. wide, more easily felt than seen. 
The jar has a vertical neck 15 cm. tall, ending in an exteriorly thickened, everted 
rim with a deep groove along the exterior edge. Rim diameter is 54 cm. The 
neck joins a rounded shoulder, from which the body wall slopes outward to a 
maximum diameter of 70 cm. (21 cm. above the base) and then inward to the flat- 
tened bottom. Wall thickness varies from 1.8 to 2.1 cm. on the body and 2.4 
to 3.0 cm. on the neck. Total height of the jar is 79 cm. 

The jar fragment comprising the lid has rounded sides and bottom. Maximum 
body diameter is 48 cm., and the height of the existing fragment is also 48 cm. 
Wall thickness is 1.5 cm. 

On the southeast side of the cut, almost directly beneath jar K, 
was jar O, the base of which rested at a depth of 2.23 meters (Meggers, 
1951, fig. 7, left). Two basins, similar in shape but larger than the 
lid of jar L were superimposed right side up inside the neck. The 
dimensions and contours of the upper basin were such that it fit 
inside the lower one closely and there was little dirt between them. 
The bottom and part of one side of the upper basin were missing. 
Although the second basin was also broken when discovered, sherds 
from the bottom were found inside the main jar, indicating that it 
was complete when set in place. The exterior wall was flush with 
the interior of the outflaring rim of the jar. Inside the upper bowl 
lid, 14 cm. below the rim and upside down against the southeast side, 



276 



BUREAU OF AMERICAN ETHNOLOGY 



[BtTLI,. 167 




Figure 99. — J-14, Mound 1, cut 1, jar N, Marajoara Phase. 

was the bottom part of a small, crudely excised jar. Inside jar O 
were sherds from the broken basin lids, at least three broken bowls 
of different sizes, a red-slipped tanga and, at a depth of 68 cm. below 
the rim, the bones of a disarticulated skeleton, with the long bones 
arranged along the north-south axis. A layer of relatively large 
sherds was spread immediately above the bones. Outside the jar, 
at a depth of 1.40 meters below the surface (approximately at the 
level of the rim) were poorly-preserved fragments of another skeleton, 
with no surrounding jar but in association with a red-slipped tanga. 



MEGGERS AND 
KVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



277 



Jar 0, Tnajd Plain (fig. 100) : The surfaces are orange to tan, with small patches 
of light gray, cream and dull brown, and a large black fire cloud on the body. 
The interior surface is more even than the exterior, but the latter has a slight 
luster in spite of the irregularity. The broken edges show temper to be finely 
ground sherd. The random inclusion of vegetal material which left dark-gray 
spots and streaks is probably accidental. The bottom is flat, 21 cm. in diameter, 
the sides slope outward to the maximum diameter 24 cm. above the base and then 
slant inward until just below the everted rim. Rim diameter is 71 by 79 cm., 
the outline being oval rather than circular. Maximum body diameter is 88 cm., 
total height, 82 cm. Body wall thickness varies between 1.4 and 2.1 cm. The 
upper rim edge is adorned with two deep, equally spaced grooves. 

Upper basin, Joanes Painted (?); The paste has a gray core with scattered ashy 
spots 3 mm. or less in diameter in addition to sherd temper and small red iron 
concretions. A white slip was applied to the rim and sides on interior and exterior 
stopping just below the curve in an irregular line. The unslipped surface is 
oxblood tan except where blackened on the exterior by fire clouds. The slipped 
area is also marred on the exterior by black fire clouds and bright-orange patches, 
resulting from poorly controlled firing. Scattered crackle is present in the slip. 
Because of prominent smoothing marks remaining on the slip, the unslipped 
surfaces are smoother, particularly on the interior, which has a low polish. The 
slip is applied unevenly, varying from paper thinness to 0.5 mm. in thickness. 




Figure 100. — J-14, Mound 1, cut 1, jar O, Marajoara Phase. 

391329—57 20 



278 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

It is possible that the slipped area bore a painted design, although no trace of 
paint remains. The rim diameter is 70 cm., the reconstructed depth 32 cm. 
Body wall thickness ranges from 1.0 to 1.4 cm. The junction of the coil added 
to the rim exterior to double the rim thickness is faintly visible in the cross section. 

Lower basin, Inajd Plain: Paste poorly mixed with large air pockets. The 
interior surface is streaked with dark gray and one-third of the exterior is covered 
with black fire clouds. The remaining surface varies from light tan to light orange 
to gray. Prominent smoothing marks remain on both surfaces, running parallel 
to the rim on the upper sides and sweeping from the base upward on the bottom. 
The exterior of the flat bottom is unsmoothed. Shape is identical to that of the 
upper basin. Exterior rim diameter is 74 cm.; base diameter, 17 cm., and depth, 
30 cm. Body wall thickness ranges from 9-12 mm. 

Arari Plain Excised fragment: Surfaces reddish-tan on the exterior except for 
large black fire cloud and blackened completely on the interior. Although 
polished, the surfaces remain somewhat uneven. The design, a combination of 
incised lines and excised areas, is crudely executed with the lines crooked, unevenly 
spaced and jagged. The fragment has a flat base 20 cm. in diameter and almost 
vertical sides suggesting rare shape 2, Existing height is 8 cm. The remaining 
portions of the vessel were not encountered. 

Beginning at a depth of 1.55 meters, the soil surrounding the jars 
became a bright orange red and more sandy than previously. As 
the depth increased, the earth became increasingly dry. After the 
removal of jar O the test was continued to a depth of 3 meters. 
Immediately beneath the jar, the soil was sandy, yellowish brown 
containing orange streaks and lumps. With increasing depth the 
color became grayer, with greenish flecks. This sterile soil was 
similar in appearance to that composing the core of J-15, Mound 14, 
except for the presence of red particles of mineral origin. 

Mound 2, Monte Carmelo. — This mound lies 100 meters east of 
Mound 1, on the same side of the Rio Anajas (fig. 88). It is somewhat 
teardrop-shaped, the eastern end being considerably narrower than 
the western. At the end of the rainy season, it was flooded on all 
sides and the length was 85 meters, with the orientation running 
east-west. The width of the western part was 48 meters and of the 
eastern, 40 meters. The sides slope steeply on the north, west, and 
south, while the eastern incline is gradual. The maximum height 
is about one-third of the distance from the west end, and reaches 
2 meters. At the same distance from the east end, the elevation is 
about 1 meter. The top is a leveled area 65 by 10 meters, which 
slopes toward the east. The major portion of the site was planted 
with coconut and banana trees, and heavy grass covered the eastern 
half, making surface collection difficult (pi. 34, 6). However, parts 
of large, anthropomorphic, painted jars, excised, incised, painted 
sherds and tangas of both red-on-white and red-slipped varieties were 
sufficiently abundant to indicate that this mound functioned as a 
cemetery. 

Because of the eroded condition and extensive cultivation, it was 



EVA^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 279 

not practical to undertake any excavation. However, we were able 
to examine two complete vessels that had been recovered by local 
residents : 

Ararl Plain Excised bowl (pi. 67, b): Paste is orange with a gray core; well- 
smoothed, reddish-brown surface, showing faint smoothing marks but even and 
slick to the touch. The out-curving sides terminated in a direct rim, and the 
bottom was flattened slightly off center so that the rim sloped and the depth 
varied from 10.2 to 11.5 cm. Rim diameter was 19 cm.; base diameter, 8 cm. 
A band of excised design about 3 cm. wide ran around the middle of the exterior. 
The cutout is deeply gouged leaving a rough surface. The two incised lines that 
flank it are not evenly parallel, and tend to undulate slightly. 

Anajds Plain Incised jar (pi. 51, a): Light-orange surface with scattered gray 
fire clouds, smoothed but with many irregularities remaining. The paste has a 
gray core. The incision was done when the surface was leather hard. The lines, 
which run diagonally in both directions over the upper body, are 2 mm. wide and 
1 mm. deep, and were applied with no effort to keep them evenly spaced or closely 
parallel. The jar is 30 cm. tall, with a flattened base 9 cm. in diameter, a slightly 
depressed, globular body, a slight, constricted neck and an everted rim, with an 
exterior diameter of 17 cm. Maximum body diameter is 25 cm. 

Mound 3, Bacatal. — This mound is on the right bank of the Anajis, 
opposite the east end of Mound 1, from which it was separated in 
May 1949 by 150 meters of flooded river (fig. 88). The land on all 
sides was inundated leaving an area approximately the shape of a 
right triangle, with the arms on the south and east and the hypot- 
enuse on the northwest. The maximum north-south extent was 65 
meters, the east-west length, 75 meters. The eastern half was higher 
than the western, with a maximum elevation of 2 meters at the 
eastern part of the south edge, along the river. Here the bank was 
steep, having been subjected to yearly erosion; elsewhere, the slope 
was gradual. The highest part of the mound was occupied by a 
modern cemetery surrounded by a fence and the remainder was over- 
grown \vith trees. Its small size and the sparsity of the sherd refuse 
indicate that this mound was probably a habitation site. The sm'- 
face collection produced 1 sherd of Anajas Plain Incised, 1 of Joannes 
Painted, 8 red-slipped tanga fragments and the remainder plain ware 
(72 percent Inaja Plain, 28 percent Camutins Plain). 

SITE J-15 — OS CAMUTINS 

The group of artificial mounds along the Igarape Os Camutins, a 
smaU tributary of the upper Rio Anajas, is one of the most famous 
on Maraj6. The large mounds on the lower part of this stream were 
visited and described by Derby (1879, 1885) and later by Farabee 
(1921), both of whom made excavations. They mention that other 
sites exist along the stream, but no details are given. To provide these, 
a survey was made and 19 additional mounds were mapped, described, 
and represented by a surface collection (fig. 101). This work was 



280 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 



fO'S 

)l« Ol9 



'*"// 



(O'o 



'0^ 



'O 




FiGXJKE 101. — Plan of mounds composing J-15 — Camutins, a mound group 
of the Marajoara Phase. 

done at the climax of the rainy season in May, when the water was 
at its maximum level. All of the mounds were islands approachable 
only by boat, whereas during the dry season they are connected by 
land. The dimensions given refer to the area above the high water 
line; had they been taken during the dry months, an estimated 1 to 



EvTrs^lf^ *^° ARCHEOLOGY AT MOUTH OF AMAZON 281 

2 meters would have been added to their height and corresponding 
increases to length and width. On the basis of this survey, it was 
decided to make stratigraphic tests in two of the largest mounds, 
Mounds 14 and 17, as well as in Mound 1. 

Mound 1, Camufins. — This is on the left bank of the igarape, and 
is one of two cemeteries in the group. It has an overall length of 
255 meters and a width of 30 meters, with rounded ends (fig. 102). 
The slopes are steep, with a slant of about 45 degrees along the sides 
and 25 to 30 degrees at the northeast end. The summit is a flattened 
ridge varying between 5 and 11 meters in width and from 8.5 meters 
above the water level on the east end to 10 meters at the center 
and 9.5 meters at the west end. The entire mound is covered with 
forest, including guajard and cacao trees claimed by the present pop- 
ulace to be of Indian origin (pi. 33, a), The surface is abundantly 
littered with sherds of all types, including numerous tanga fragments, 
adornos, and plain and decorated wares. Erosion has been greatest 
on the north side, which borders on the river, and has produced such 
a concentration of surface material that the slope is literally paved 
with sherds, a high percentage of which are plain ware. 

Three excavations were made along the summit in an efTort to ob- 
tain stratigraphic information. Cut 1, 1.5 meters square, was 3 
meters from the slope on the river side and one-third of the distance 
from the east end of the ridge. Sherds were present on the surface, 
although in less profusion than on the more eroded slope. For the 
first 30 cm., the soil was very wet, dark-gray clay streaked with 
black, and containing roots and small lumps of red clay, apparently 
sunbaked. Sherds were sparser than had been expected on the basis 
of the surface condition and since there were too few to provide a 
basis for stratigraphic analysis the material was not retained by levels 
after 30 cm. As depth continued, the soil became light gray and 
more sandy. The upper edge of an Arari Red Excised vessel, jar A, 
was encountered at 75 cm., the base resting 1.10 meters below the 
surface. Sherds from a shallow bowl, white-slipped on the interior, 
red-slipped on the rim top and plain on the exterior, were found in a 
position over the jar that indicated it served as a lid. Ten centimeters 
to the north of the jar base, a small, white-slipped jar, B, was found 
upside down. Both vessels contained only wet, sandy soil. 

Jar A, Arari Red Excised (pi. 61, b): The entire exterior surface is covered with 
a red slip and decorated with a complex excised pattern. The interior is white 
slipped and both surfaces are well smoothed and even. The paste has a gray 
core. Although the rim was broken off by earth pressure, a sufficient number of 
fragments were recovered to permit its restoration. The jar was 38 cm. tall, 35 
cm. in maximum body diameter, and 38 cm. in external rim diameter. It has a 
flat bottom, 20 cm. diameter, a globular body and a short, outflaring neck 9 cm 
taU. 



282 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 




E\S^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 283 

The lid was also 38 cm. in rim diameter and had a flat bottom 11 cm. in 
diameter. 

Jar B, Joanes Painted (pi. 67, a): Exterior and interior purfaces are smoothed, 
leaving polishing tracks and many irregularities, and slipped with white. Beneath 
the slip, the surfaces are light cream or tan, and the paste has a light gray core. 
The vessel has a short, collarlike rim, 2.2 cm. tall, a globular body and a rounded 
base. The exterior surface of the collar was painted red. Maximum diameter of 
the body is 18.0 cm., of the rim 14.5 cm. Total height is 10 cm. 

Whitish sand appeared at a depth of 1.10 meters and was sterile to 1.50 
meters, where the cut was terminated. Sherds by levels totaled: 

Level .00-. 15 m.: 79 Inajd. Plain, 93 Camutins Plain, 7 Ararl Plain Excised, 

1 Arari Red Excised, 1 Anajds Red Incised, 3 Anajds 
Plain Incised, 19 Joanes Painted, 3 Carmelo Red, 4 
modeled sherds, 5 red-on-white, and 3 red-slipped 
tanga fragments. 

Level .15-.30 m.: 12 Inajd Plain, 25 Camutins Plain, 1 Ararf Plain Excised, 

2 Arari Red Excised, 5 Joanes Painted, 4 Carmelo 
Red, and 3 red-on-white tanga fragments. 

Level .30-1.00 m.: 25 Inajd Plain, 24 Camutins Plain, Arari Red Excised 
sherds from 4 different vessels, 1 Goiapi Scraped, 
Joanes Painted sherds from 3 bowls, and 2 red- 
slipped tanga fragments. 

Cut 2 was placed in the center of the mound, about 2 meters from 
the summit on the river side. It was 1.5 meters square and excavated 
in 15-cm. levels. ^^ Because of the steepness of the slope at this point, 
the first level was measured off on the uphill side and leveled out on the 
downhill side to permit the removal of an equal amount of dirt from 
all parts of the cut in succeeding levels. In the first level, the soil was 
dark-gray wet clay, roots were thick and sherds large and abundant. 
Between 15 and 45 cm., the soil became darker and the sherds less 
numerous. At 50 cm. the soil became light gray on the uphill side, 
but in the remainder of the cut it continued dark until 75 cm. From 
this point until sterile soil was reached at 2,10 meters, the sherds were 
mixed in light-gray or whitish, sandy clay streaked with light gray, and 
containing scattered charred bits of wood and small pockets of ash. 
From 2.10 to 2.55 meters the yellowish, sandy clay contained charcoal, 
but no sherds or lumps of burnt clay. 

Two unusual objects were found in cut 2. Level 0.75-0.90 meter 
produced a worked sherd of Camutins Plain (fig. 103, 6). It was 
roughly oblong, 4.8 by 3.5 cm., 2.4 mm. thick at the edges and 4.5 
mm. thick at the center, slightly convex and smooth on both surfaces. 
The edges had been rounded and there was a shallow groove in each 
end, as though to secure a string wound or tied around it. An ec- 
centrically shaped Inaja Plain object came from level 1.05-1.20 



" Because of the small sample per level, 2 levels were combined for stratigraphic analysis of the pottery 
types (Appendix, table 40). 



284 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 






J I 



3 CM 



<^ZZZZZ2i 



Figure 103. — Artifacts from J-15, Mound 1, cut 2, Marajoara Phase: 
a, Eccentric pottery object. 6, Worked sherd. 

meters (fig. 103, a). It was circular, flat on one surface and convex on 
the other. An extension at one side was drilled horizontally and 
broken off at the hole. The flat side was rough and uneven, and the 
convex one bore a design of parallel grooves curving from the center to 
either side. Width was 4.9 cm., length from the broken edge to the 
opposite end 4.7 cm., maximum thickness 1.3 cm. The object was 
crude both in form and decoration, possibly a crude pendant. 

Cut 3, also 1.5 by 1.5 meters square, was excavated 8 meters from 
the east end of the summit and 2 meters from the center on the side 
away from the river. For the first 4 levels the soil was medium brown, 
with some ash in level .45-.60 meters. In level .60-.75 meters it be- 
came darker, and between .75 and .90 meters the soil was streaked 
with fire-burnt clay. Between 0.92-1.10 meters, a bright-red band of 
burnt sand and claj'' intermixed with black ash occupied one corner of 
the cut.^^ Below 1.10 meters and continuing to the bottom of the 
test (1.50 meters) the soil was pure white, sandy clay with no ash or 
sherds. 

The slightly flattened base of a Camutins (?) Plain jar was en- 
countered at 1.00 meters. The vessel had a globular body, 45 cm. in 
maximum diameter and 50 cm. tall. The everted rim was broken off, 
but fragments were found in the fill. The red-slipped tanga inside was 
broken but all fragments were present (pi. 82, d). This last level 
(0.75-0.90 m.) also produced a number of sherds from a small, anthro- 
pomorphic, Joanes Painted jar with an insloping neck and a slightly 
flaring rim with a diameter of 18 cm. Facial features were indicated 
by low applique on the neck, outlined with red-painted lines. Another 
sherd from this vessel was a small, zoomorphic adorno. 

Burial Group 1 designates a place on the west end of the ridge, on 
the south side of the summit, where the cabodos had removed several 

" Because of the small sherd sample per 15-cm. level, 2 levels were combined for stratigrapbic analysis of 
the pottery types (Appendix, table 40). 



Bv^Nsf ^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 285 

large jars. Part of another (jar 1) was visible at the edge of the old 
excavation. Removal of the surrounding dirt showed it to be Joanes 
Painted, 70 cm. tall (rim broken off), with a flat bottom 15 cm. in 
diameter, rounded sides, a flat shoulder, and an insloping neck. The 
structural weakness of the angular shoulder and the pressure of the 
earth had broken the jar into large pieces. The interior was filled 
with wet dirt containing flecks of charcoal, miscellaneous sherds and, 
at the bottom, a complete, red-slipped tanga. The base of this jar 
was 1.15 meters below the existing surface of the mound; the rim was 
45 cm. below the surface. 

Jar 2, a globular-bodied jar 36 cm. in diameter with a low-relief 
snake twisting over the exterior, was resting in a break in the neck of 
jar 1, so that its bottom was supported by the shoulder of the latter. 
Jar 3, Inaja Plain and 30 cm. tall, was about 20 cm. to the west of jar 1 
with its rim 28 cm. below the surface. The dirt filhng it contained 
much black ash from a cremation. 

Jar 4, a second large Joanes Painted jar, was at the north side of Jar 
1, with its broken rim 60 cm. below the surface. The reconstructed 
rim diameter was 54 cm., total height 70 cm., neck height 30 cm., 
minimum neck diameter 37 cm. A shallow, broken Aran Plain 
Excised bowl, 42 cm. in rim diameter, was inverted over the top as a 
lid. The dirt inside the jar was very muddy, a fact which had con- 
tributed to the destruction of the skeletal remains. With the latter 
were scattered small vessel sherds, fragments of charred wood and 
four sherds from a single red-on-white tanga. Dr. Marshall T. New- 
man, United States National Museum (pers. commun.), made the 
following analysis of the skeletal remains : 

Three mandibular, two maxillary fragments, four teeth, one petrous, and a few 
eroded sections of long bone shaft. All the skull parts and teeth are left side, and 
belong to the same individual. This individual had erupted and worn second 
molars, but shows an unerupted third molar. Therefore, the age should be more 
than 12 but less than about 18 years. The small size of the mandibular fragments 
suggests female. 

Jar 5, Anajds White Incised, was above and between jar 1 and jar 4, 
with its rim 35 cm. below the surface. It contained bone fragments 
in a poor state of preservation, some of which Newman could identify 
as human and adult. Others were identified as crocodilian by Dr. 
Doris M. Cochran of the Department of Zoology, United States 
National Museum. 

Jar 5, Anajds White Incised (pi. 55, a): The jar was noticeably asymmetrical, 
with a cylindrical neck 24 cm. tall, a squat, rounded body 36 cm. in diameter and 
14 cm. tall, a flat bottom 14 cm. in diameter and an everted, exteiiorly thickened 
rim 31 cm. in diameter. The entire exterior was white slipped and covered with 
an incised design in which triple parallel lines outlined rectangular, hexagonal, and 
stepped panels containing simple motifs drawn with single lines. A contrasting 



286 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

color eflfect was produced by cutting the incisions through the slip into the orange 
undersurface, and was most pronounced in the small squares and triangles at the 
corners and ends of the single lines. The motif is typical of that employed on 
Pacoval Incised, but the red retouching of the lines has been omitted. 

Burial Group 2, like Burial Group 1, was an excavation begun by 
the cabodos, who had removed a large painted jar. This spot is on 
the south side of the ridge, about 25 meters southeast of cut 2. Two 
small jars, their upper edges about 38 cm. below the surface, were 
found by cutting away the sides of the earlier excavation. Both 
vessels were filled with wet soil, which contained neither bone frag- 
ments nor sherds. 

Jar A, Inajd Plain, had a flat bottom 15 cm. in diameter, from which the sides 
flared outward to the maximum diameter of 31 cm. at a height of 11 cm., and 
then curved inward to the neck, which was 18 cm. in diameter. With the rim 
missing, the existing height was 26 cm. 

Jar B, Joanes Painted (pi. 73, h), was 2 meters north of jar A. The body was 
globular, with a short, outslanting neck and an everted, exteriorly thickened rim. 
Measurements include: total height, 38 cm.; rim diameter, 35 cm.; maximum 
body diameter, 36 cm.; diameter at base of neck, 23 cm.; neck height, 9 cm. A 
white slip covered the exterior. The neck was decorated with a stylized face in 
low relief, with the background painted red. A red-painted design covered the 
body. 

The surface collection from Mound 1 does not represent a random 
sample as it does on the other Camutins mounds. Sherds were present 
in such abundance that it was felt that a better idea of the range of 
material present would be secured by picking up striking decorated 
sherds and unusual or ornamental plain ware rims. This technique 
amassed examples of the following decorated wares: Anajas Plain 
Incised, Anajas White Incised, Anajds Ked Incised, Arari Plain 
Excised, Arari Red Excised, Arari White Excised, Arari Double- 
slipped Excised, Goiapi Scraped, Guajard Incised, and red-slipped, 
and red-on-white tangas (pi. 82, a-h). Both Inajd Plain and Camu- 
tins Plain were well represented, with forms including stools, tall pot 
rests, straight-sided and flat-bottomed bowls with 4 "dimples" in the 
walls, a funnellike bowl with an open bottom, and small bowls (pi. 
67, 6) and jars of all descriptions, generally with rims ornamented by 
nubbins, adornos, notches, or applique. Also included were two 
small figurines (pi. 79, a-h), an ear plug (fig. 134, a), and 2 labret (?) 
fragments (fig. 135). Two stone objects were found: a small, flat, 
smooth stone about 5 cm. in diameter, with deep crisscross grooves 
on both surfaces produced by rubbing with a stick (fig. 138), which 
came from the vicinity of Burial Group 1, and a miniature diorite ax 
(fig. 132, a). 

Mound 2. — This is the first in the series of habitation sites scattered 
along the left bank of the Igarape Camutins upstream from Mound 1. 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



287 



It is separated from Mound 1 by approximately 2 km. One of the 
smallest of the series, it is at present 5 meters long, 50 cm. wide and 
75 cm. high. It has been badly trampled by cattle and was surrounded 
by water at the climax of the rainy season. A few plain-ware sherds 
(Appendix, table 42) were scattered in the soil to a depth of 25 cm. 
Mound S. — This mound, 75 meters upstream from Mound 2, is 
approximately oval, 32 meters long, and 8 meters wide (fig. 104). 
The north side, subject to erosion by the river, rises almost vertically, 
while the other sides slope gradually and then rise at an angle of 30 






Figure 104. — Detailed plans of J-15, Mounds 3, 4, 6, 15, 16, habitation 
mounds of the Marajoara Phase. 



288 BUREAU OF AMERICAN ETHNOLOGY [bull. 16T 

to 45 degrees to form the central part of the mound, an area 10 by 5 
meters and reaching a height of 3 meters above the high waterline. 
The soil is light-colored, sandy clay with areas of reddish, burnt sand 
and black charcoal fragments. Sherds are most abundant in the 
highest part of the mound, although a few are scattered on the flanks. 
Tests along the bank indicate that they are present to a depth of 2 
meters (Appendix, table 42). 

Mound 4} Sacrario. — This is a long, low, thin mound (fig. 104) 
curved to follow a slight bend in the river 20 meters above Mound 3. 
It is 100 meters long and varies from 5 to 8 meters in width. The 
surface is 1.25 to 1.50 meters above high water level except at a knoll 
at the east end, 25 by 7 meters, which rises to 2.50 meters. At the 
height of the rainy season the campo surrounding it on three sides is 
beneath 50 cm. of water. The soil is sandy clay, darker than that 
composing Mound 3. Predominantly plain ware sherds (Appendix, 
table 42) are distributed sparsely over the entire surface. Present 
growth includes a few large trees, but consists mainly of small, scrubby 
brush and trees. 

Mound 5, Sacacdo. — Tnis is the only mound on this part of the 
river that was constructed on the right bank. Its lower end is 
directly opposite the upper end of Mound 4. At the present time 
it measures 45 meters long by 8 meters in width and from 1.00 to 1.50 
meters in height. More than any of the other mounds surveyed, this 
one seems to have suffered from the trampling of cattle seeking refuge 
from the flooded campo and it probably was originally somewhat 
higher. The soil is light-gray, sandy clay and sherds are present 
over the entire surface (Appendix, table 42). Fifteen large and many 
smaller trees cover it. The surrounding campo is flooded, but the 
water appeared to be unusually deep near the mound, possibly the 
result of dirt for mound construction having been removed from the 
area. 

Mound 6. — This mound, 8 meters above Mound 4, was separated 
from the latter in May by a small inlet of water. The campo on the 
southeast side was muddy but not flooded. This mound is formed 
by a narrow elevation 50 by 8 meters in horizontal dimensions and 
5.00 to 6.50 meters high, and a low, circular area 1 1 meters in diameter 
and 1.75 meters high, the two sections being joined by a necklike 
construction 4 meters long and 1 meter in elevation (fig. 104). The 
river has cut away the west side so that it rises nearly vertically, while 
the protected slopes are more gradual. The surface is covered with 
thick grass, spiny palm and brush, and a few large trees. Sherds 
are not easily found, possibly partly because of the sod covering, and 
most of the sample (Appendix, table 42) came from two spots and 
from a depth of 10-15 cm., which was below the grass root mat. 



BVA^sr ^^° ARCHEOLOGY AT MOUTH OP AMAZON 289 

Mound 7, Sao Bento. — This resembles Mound 6 in general appear- 
ance and vegetation. It is separated from the upper end of the 
latter by 6 meters of flooded campo and is approachable only by 
boat during the rainy season. Ova' in outline, it measures 60 by 15 
meters, with gentle slopes from the summit to the water's edge except 
on the eroded river side. Maximum height is 4 meters, maintained 
over an area 30 by 5 meters in the center of the mound. Sherds are 
abundant, especially in the eroded bank (Appendix, table 42). 

Mound 8. — This is 45 meters long and 10 to 12 meters wide, and 
is 40 meters upstream from Mound 7. The soil is light-gray to 
whitish, sandy clay with bright fire-reddened zones. Sherds are 
moderately abundant in a layer averaging 25 cm. in depth (Appendix, 
table 42). The bank rises sharply to an elevation of 3 to 4 meters 
on the south and west sides, and slopes off gradually to the north and 
east, forming a level platform 30 by 6 meters. Trees, small bushes, 
and a little grass comprise the vegetation cover. 

Mound 9. — This is 30 meters above Mound 8, is oval, and measures 
20 by 10 meters. The sides rise steeply to a level top, 7 by 8 meters 
in diameter and 5 meters above the flooded campo. Growth includes 
one large tree, small bushes, and patches of tall grass. Sherds are 
not abundant on the surface (Appendix, table 42). 

Mound 10. — This mound is covered with forest, and is separated 
from Mound 9 by a deep, low area 25 meters wide. The site is 30 by 
15 meters, with moderately steep sides rising to a height of 3.75 meters. 
The soil is reddish-brown loam, less clayey than in the preceding sites, 
and no sherds could be seen on the surface. Random digging to 
secure a sherd sample (Appendix, table 42) uncovered a broken jar 
toward the southwest end of the summit. It had fallen toward the 
northwest and inside was a complete Inaja Plain stool, lying upside 
down (pi. 83, a). 

Mound 11. — This is 18 by 14 meters, and is 75 meters upstream 
from Mound 10. It is 1.75 meters high and pottery is abundant in 
the eroded northwest bank in a zone from 20-100 cm. below the 
surface (Appendix, table 42). Except for scattered small trees and 
a strip of wood along the southwest side, the cover is short grass 
(pi. 32, a). Soil composition is like that of Mound 10, a reddish- 
brown loam. 

Mound 12, Carmo. — This mound is 2 km. above Mound 11, on 
the same side of the igarape, which at this point was 20 meters wide 
and said to shrink to a width of 10 meters during the dry season. 
This mound is nearly circular, 12 meters in diameter and 1.25 meters 
high. A fringe of trees runs around the edge and the center is grassy. 
Soil is reddish-browD, sandy clay and sherds are abundant under the 



290 



BUREAU OF AMERICAN ETHNOLOGY 



[BDLL. 167 



sod layer (Appendix, table 42). As is typical during the rainy season, 
the mound was completely surrounded by water. 

Mound IS. — This, like the other forest-covered mounds, is camou- 
flaged by the trees so that it would be easily overlooked by a casual 
observer. It is in a bend in the stream 25 meters above Mound 12, 
and is isolated on the other side by the flooded campo. At present, 
dimensions are 18 by 10 meters and 1.50 meters high, A border of 
trees grows on the bank, but the center is grassy. Sherds are 
exposed in the eroded north end (Appendix, table 42). 

Mound 14, Inajasal. — This is 1 km. east of Mound 13 (fig. 105). 
It is the largest of the habitation sites, being 51 meters long, 25 to 35 
meters wide, and 5.75 to 6.25 meters high along the summit.^" The 
slopes are steep and covered with trees except at the center of the 
east side. The summit and the east slope are grassy (pi. 32, b). 
Because of its size and the abundance of the surface sherd refuse, 
this mound was selected for stratigraphic excavation. A cut 1.5 
meters square was begun near the center of the ridge and taken 



WATER LEVEL, 
MAY 1949 JL 



10 




20 M 



Figure 105. — Detailed plan of J-15, Mound 14 — Inajasal, a habitation 
mound of the Marajoara Phase, showing the location of cut 1. 



M Hilbert (1952, p. 10), who returned in the dry season when the base was exposed, gives the actual height 
as 7.00 to 7.25 meters. At low water the river is 75 meters away. 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



291 



112 
194 



30 
87 - 
85 - 



METERS 
00 - 



.15 - 



- .30 - 

- .43 

- .60 - 



-1.05 



- 



47 

SI - 

■ 

24 - 

84 



.75 - 

.90 - 



-1.35- 

1.50 - 

-1.65 - 

-1.80 - 



1.95 - 
2.10 - 
2.25- 



DARK-BROWN LOAM 




/~7-Z 

SHERDS a CLAY LUMPS 

/ / / / / 




GRAYISH-WHITE 
STERILE CLAY 



/ "? y- 7''""~ FHiRDs /_____ y / /_ _ 



LIGHT-GRAY WITH TRACES OF BLACK ASH 




aUNDANT 

"•SE ^ GRAYISH-WHITE 
STREAKED WITH 
ORANGE 



ASH POCKETS a STREAKS 



^;;2;;7 BASE OF LARGE JAR 
CONTAINING ASH 



WHITISH CLAY 



(n 



STERILE CLAY 



V 



FIRE-BURNT 



Figure 106. — Profile of west face of cut 1, J-15, Mound 14, Marajoara Phase. 

down in 15-cm. levels (fig. 106). Sherds were abundant in the 
dark-brown loam, which extended beneath the sod to a depth of 
about 18 cm. Then began a stratum of light orange, which became 
brighter orange with increasing depth; sherds, as well as clay lumps, 
continued numerous. At 35 cm. light, grayish-white soil appeared 
in the northwest corner aod spread over the entire cut at the beginning 
of the next level at 45 cm. From here to a depth of 70 cm. the soil 
continued sandy and whitish gray, with streaks and small pockets 
of bright orange. A pocket of black ash at the west side of the cut 
in level 0.45-0.60 m. produced the majority of the sherds from that 



292 BUREAU OF AMERICAN ETHNOLOGY [bull, m 

level. From 72 to 77 cm. was a second fire-bumt layer containing 
sherds. Another layer of light, grayish-white soil with traces 
of black ash and sherds in poor condition extended from 77 to 90 cm. 
in the northwest corner and to 1.04 meters in the rest of the cut. 
The majority of the sherds from level 0.90-1.05 meters were con- 
centrated at the bottom of this stratum, just above a third burnt- 
orange layer. The latter, somewhat thicker than those above it, 
became browner in level 1.35-1.50 m. Sherds from an ash-filled jar 
were removed from the northwest wall at this level, and an ash 
pocket was encountered on the opposite side of the cut. Sherds 
were abundant and large. Beginning at about 1.45 meters in the 
northwest corner and at 1 .50 meters in the rest of the cut was another 
light-grayish stratum. The soil was streaked with orange and 
contained pockets of black ash about 20 cm. in diameter. This 
variegated appearance, with bright orange, reddish orange, light 
tan, whitish gray and streaks and pockets of black ash continued to a 
depth of 1.95 to 2.05 meters. A large base sherd containing black 
ash was found at 1.95 meters, but sherds were generally sparse. 
Except for a pocket of burnt orange in the northwest corner, the soil 
below level 1.95-2.05 m. was moist, whitish clay, although sherds 
still appeared. At 2.12-2.15 meters there was a transition to compact, 
whitish, dry and flaky clay flecked with gray, yellow, and hard 
orange particles, which was sterOe and consistent, and comprised 
the core and foundation of the mound. 

In the sherd count by levels, the sparsity of decorated sherds is 
notable by contrast with the totals from Mound 1, cut 1 (p. 283): 

Level 0.00-0.15 m.: 75 Inajd Plain, 140 Camutins Plain, 8 Anajds Incised, 

1 Joanes Painted. 

Level 0.15-0.30 m.: 19 Inajd Plain, 93 Camutins Plain. 

Level 0.30-0.45 m.: 30 Inajd Plain, 163 Camutins Plain, 1 Joanes Painted. 

Level 0.45-0.60 m.: 8 Inajd Plain, 22 Camutins Plain. 

Level 0.60-0.75 m.: 52 Inajd Plain, 32 Camutins Plain, 3 Joanes Painted. 

Level 0.75-0.90 m.: 43 Inajd Plain, 40 Camutins Plain, 2 Joanes Painted, 

2 red-slipped tanga sherds. 

Level 0.90-1.05 m. : 20 Inajd Plain, 51 Camutins Plain, 3 Joanes Painted. 
Level 1.05-1.35 m.: 41 Inajd Plain, 55 Camutins Plain, 1 Joanes Painted. 
Level 1.35-1.50 m.: 54 Inajd Plain, 45 Camutins Plain, 2 Anajds Incised. 
Level 1.50-1.65 m.: 25 Inajd Plain, 19 Camutins Plain, 3 Joanes Painted. 
Level 1.65-1.80 m.: 18 Inajd Plain, 29 Camutins Plain, 4 Joanes Painted. 
Level 1.80-1.95 m.: 15 Inajd Plain, 8 Camutins Plain, 1 Joanes Painted. 
Level 1.95-2.10 m.: 52 Inajd Plain, 29 Camutins Plain, 1 Anajds Incised, 

2 Joanes Painted. 
Level 2.10-2.25 m.: 48 Inajd Plain. 12 Camutins Plain, 1 Anajds Incised, 

2 Joanes Painted. 

Mound 15. — This mound, 300 meters above Mound 14, is the last 
site on this part of the igarape. It is 30 by 11 meters, with the sides 
rising steeply at the south end to a knoU 8 meters in diameter and 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



293 



4.50 meters high, and sloping off gradually toward the north end 
(fig. 104). The soil is dark gray to a depth of between 25 and 50 cm., 
and then becomes light, sandy clay. Sherds are sparse except on the 
steep bank (Appendix, table 42). Heavy tree growth on the edges 
conceals the mound from the passer-by. 

Mound 16, Tesinho. — This is about 250 meters below Mound 1, on 
the same (left) bank of the igarape. It and the adjacent area are 
covered with forest, which except on the mound itself was under 
water at the time of our visit. The artificial elevation has an area 
140 by 16 meters, with the sides sloping steeply except at the north- 
east end (fig. 104). The maximum height, maintained over an area 
approximately 100 by 5 meters, is 3.20 meters. Sherds were scattered 
over the surface (Appendix, table 42) and were said to be abundant 
along the north side below the flood level. Protected from the in- 
vasion of cattle by the forest extending between it and the campo, it 
has accumulated a relatively thick undergrowth. 

Mound 17, Belem. — This mound, lying almost directly opposite 
Mound 1, is another cemetery. It measm^es 250 by 59 meters, and 
attains a height of 6.40 meters (fig. 107). At the time of our visit, 
the lower end had been cleared for cultivation but the remainder of 




■ 1949 EXCAVATIONS 

a farabee's excavations 



FiGUKE 107. — Detailed plan of J-15, Mound 17, a Marajoara Phase cemetery, 
showing location of excavations. 
391329 — 57 21 



294 BUREAU OF AMERICAN ETHNOLOGY [bull. 16t 

the site was covered with forest, although none of the trees was ex- 
ceptionally large (pi. 33, b). Many small sherds are scattered over 
the bank toward the river, where erosion has washed off the humus 
layer. The side away from the river levels down into forest, which 
is boggy but not inundated during the rainy season. Inquiries 
among the caboclos, one family of which was living on the mound, 
elicited the information that no complete jars had ever been found.^^ 
The unusually large size and the fact that decorated types were 
more abundant than on other habitations indicated the advisability 
of more extensive examination. 

A stratigraphic cut 1.5 meters square was excavated near the center 
in 15-cm. levels.-^ For the first 26 cm., the soil was dark-gray loam 
containing small but abundant sherds. From here to between 45 to 
50 cm., the soil color was light tan sprinkled with black wood ash, 
beneath which was a layer of blackish ash about 5 cm. thick. From 50 
to 70 cm., the soil became lighter tan, with a streaked appearance, and 
contained ash and burnt clay lumps. Sherds were more numerous 
than in the previous levels. For the next 10 cm., pockets of black 
appeared sporadically. At 80 cm., began a stratum of yellowish-white 
sand that contained relatively few sherds. Between 0.95 to 1.10 
meters the color became black once more. A burnt-red layer, streaked 
with black, occupied the region between 1.10 to 1.20 meters, followed 
by a thin band of black ash 1 to 2 cm. in thickness. Underlying this and 
continuing to the maximum depth tested (3.25 meters) was the sterile 
core of the mound, composed of light, yellowish-white, sandy clay, 
containing charcoal particles in the upper 40 cm., but below that free 
from refuse mixtm-e. The only pottery artifact was a spoon (pi. 
81, e), which came from level .60-.75 m. The sherd totals per level 
reveal the frequency of decorated sherds and tanga fragments typical 
of Marajoara Phase cemetery sites: 

Level .00-. 15 m.:^ 40 Inajd Plain, 113 Camutios Plain, 2 Anajds Incised, 

7 Joanes Painted, 4 red-slipped tanga fragments. 
Level .15-.30 m.: 14 Inajd Plain, 24 Camutins Plain, 3 Ararf Excised, 1 

Anajds Incised, 2 Guajard Incised, 7 Joanes Painted, 

1 red-slipped tanga fragment. 
Level .30-.45 m. : 66 Inaja Plain, 74 Camurins Plain, 5 Arari Excised, 1 

Anajds Incised, 1 Guajard. Incised, 2 Joanes Painted. 
Level .45-. 60 m.: 28 Inajd Plain, 25 Camutins Plain, 1 Arari Excised, 1 

Anajds Incised, 6 Joanes Painted, 3 red-on-white tanga 

fragments. 
Level .60-.75m.: 76 Inajd Plain, 68 Camutins Plain, 4 Arari Excised, 1 

Anajds Incised, 14 Joanes Painted, 3 Goiapi Scraped, 

14 red-slipped and 8 red-on-white tanga fragments. 



21 This is contradicted by Farabee's findings (described ou pp. 298-299). 

22 This is in the area where Farabee first dug and found only sherds (see p. 298). 

23 Because of the small totals for many of the pottery types in 15-cm. levels, 2 levels vrere combined for the 
stratigraphic analysis of pottery trends (Appendix, table 42). 



MBGGBES AND ARCHEOLOGY AT MOUTH OF AMAZON 295 

EVANS] 

Level .75-.90 m.: 95 Inajd Plain, 13 Camutins Plain, 5 Ararf Excised, 1 
Anajds Incised, 4 Joanes Painted, 3 red-slipped and 6 
red-on- white tanga fragments. 
Level .90-1.05 m.: 93 Inajd Plain, 6 Camutins Plain, 10 Joanes Painted, 

6 red-slipped tanga fragments. 
Level 1.05-1.20 m.: 25 Inajd. Plain, 7 Camutins Plain, 5 Joanes Painted, 
2 red-slipped and 2 red-on-white tanga fragments. 

Mound 18, Arraial. — This is separated from the lower end of 
Mound 17 by a strait 7 meters wide and 50 cm. deep at highest 
water. It is ahnost circular, 68 meters in diameter and sloping 
gently toward all sides from a maximum height of 3.20 meters at the 
center. Sherds (Appendix, table 42) are abundant in a path that 
runs across the center and are scattered over the whole surface. 
Forest with thick undergrowth blankets the site and the cultural 
refuse is covered with humus except at the edges where erosion has 
uncovered whitish sand. 

Mound 19.— This is a low, round mound 100 meters east of Mound 
14, and is 25 meters in diameter and 0.75 to 1.00 meters above high 
water level. It is covered with araga, high grass and a few small 
trees; the soil is black to gray to brown. No sherds are visible on 
the surface, but a test produced them just beneath the root mat. 
Too few were recovered to provide an adequate sample for seriation. 

Mound 20. — This mound, lying between Mounds 11 and 12, has 
been almost completely eroded away. Its present elevation is 50 cm. 
and its area 5 by 2 meters (at high water). A smaller nubbin about 
2 meters in diameter projecting above the water 5 meters upstream 
is probably another remnant of this same mound. These conditions 
did not permit the collection of a sherd sample. 

DATA FROM OTHER INVESTIGATIONS 

Although it has been said that "more than a hundred artificial 
mounds are now knovm" on the Island of Marajo (Mordini, 1934 a, 
p. 62; Howard, 1947, p. 47; Meggers, 1948, p. 153), it should be 
realized that the word "known" is used in a somewhat indefinite 
sense. No single individual is acquainted with all, or even with a 
fraction of this total, and three-quarters of them have never been so 
much as mentioned on any printed page, much less located and de- 
scribed. "Known" must be understood, therefore, as referring 
mainly to awareness of the existence of these sites on the part of the 
local residents smdjazenda owners. The importance attached to them 
by the owners and the prestige derived from owning a good one, as well 
as the possibility of the same site being reported by several different 
people, have perhaps increased the number "known" beyond the 
actual total figure. There is the greater probability, however, that 
far more exist than are reported, since those listed below are predom- 



296 BUREAU OF AMERICAN ETHNOLOGY [bull. 16t 

inantly cemeteries and the ratio of habitation mounds to cemeteries 
is 13 :1 for Fortaleza and 18 : 2 on the lower Camutins. 

WhOe the present census is incomplete, the making of a more ac- 
curate compilation would require the better part of a year, adequate 
facilities for transportation, and the cooperation of all of the land 
owners in the area, a set of conditions that will probably never be 
fulfilled. Fortunately, such complete knowledge does not appear to 
be necessary for the preliminary reconstruction and interpretation of 
Marajoara culture. The mounds that are known show sufficient con- 
sistency to make it almost certain that when others are recorded in 
the future they will not reveal a basically different pattern of culture. 
This section summarizes all of the scientific information available on 
Marajoara sites and in order to facilitate reference to them, the 
mounds will be described in alphabetical order. As many as possible 
have been located on the map (fig. 108), but it must be kept in mind 
that the location is usually approximate, from written description or 
from a map on which it was difficult to correlate the rivers with those 
on the air map with any assurance of accuracy. 

BACURf ALTO 

Three to four kilometers north of the Fazenda Silva, in the region 
of Cabo Maguari, is a site called Teso do Bacuri Alto. In September 
1922, it was visited by Nimuendajii, who gives the following 
description: 

The refuse deposit had a thickness of about 20 cm. and covered a somewhat 
larger area than that of Teso das Igagabas [which was 10 x 20 meters]. Though 
the owner of the Fazenda had already made excavations, great numbers of clay 
fragments were, nevertheless, brought to light. They were bigger and better 
preserved than those of the two other sites, and among the huge number of bad 
products were some fragments of really beautiful and artistic specimens. Sev- 
eral times I found fragments which, though -widely scattered, could be put 
together . . . [Ryd^n, MS.] 

Among the sherds were a large number of stool fragments with 
relief and polychrome decoration. Other objects found included 
charred tucumd nuts, sharp-edged splinters of a brittle, fiery-red 
stone, a small piece of nephi'ite, small lumps of ocher and yellow Uind, 
and a badly corroded iron nail (Ryden, MS.). All specimens are in 
the Goteborg Museum. (The nail is of more recent origin.) 

CAJUEIROS 

Couto de Magalhaes speaks of a "kind of circular fort of earth" 
on the Fazenda Cajueiros, and Ferreira Penna, in quoting him, adds 
that it "probably contains artifacts and human remains." A mound 
called Cajueiros and described as "bastante rico" is reported by the 
Barao de Marajo. The same site is mentioned by Pinto, and a bowl 
from there is illustrated by Torres. 




Figure 108. — Map of Maraj6 Island, showing the location of Marajoara Phase cemetery sites. 



Ev^Nsr ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 297 

Bibliography: Torres, 1940, pi. 49; Couto de Magalhaes, 1876, pt. 2, p. 34; 
Ferreira Penna, 1879 a, p. 48; Maraj6, 1895, p. 88; Palmatary, 1950, p. 276; 
Pinto, 1930, p. 351. 

CAMUTINS 

Although von Martius mentioned the existence of urns at Camutins 
in 1867, Ferreu-a Penna is said by Hartt (1885, p. 17) to have been 
the first to make a scientific examination and a collection, which was 
deposited in the Museu Nacional. He was followed in 1876 by Orville 
Derby, who left the first description: The main mound, about a 
league above the junction of the igarape with the Rio Anajas, was 
210 by 80 meters in base measurement and some 13 meters in height.^* 
Derby made his visit during the dry season, and was able to observe 
a large excavation on the west side of the igarape from which he de- 
duced the earth for the construction of the mounds had been taken. 
Near it was a second mound, almost as large as the first. A third 
mound, a few hundred meters below the fii'st, was lower but broader 
and was bounded by the igarape on three sides (cf. our J-15, Mounds 
1, 16, and 17). He was informed that there were a dozen mounds 
in a distance of half a league upstream from those he saw, all on the 
east side, and all but two in the narrow tree zone along the bank. 
He was also told that at times sherds were found in the campo and in 
the forest on the natural surface of the ground. On the question of 
intentional form or orientation, he concluded. 

These three mounds all extend in different directions, indicating that their 
position is without significance. They all have a more or less elliptical or oval 
form, but this seems to have been accidental, as there is no evidence that they 
were constructed according to any definite plan. [Derby, 1879, p. 226.] 

The ceramics he found to be of the same general type as those at 
Pacoval, which he had just visited, with a few differences in emphasis: 

From what I could observe, it appears that the jars are more frequently painted 
than incised, the contrarj' of what occurs at Pacoval. The predominant shape 
is large, depressed and globular, while at Pacoval smaller sub-cylindrical and 
conical forms are more common. These observations, however, are too slight 
for the establishment of distinctions, and all the principal forms are represented 
in both places. Tanga fragments are abundant, although I did not find any 
complete ones. The majority are red and without ornamentation, but I saw 
fragments painted like those from Pacoval. [In Hartt, 1885, p. 25.] 

The next visitor to leave a detailed report was W. C. Farabee 
(1916 b), who in 1916 made extensive excavations in one of the group, 
which he refers to as the "Magno Mound," and which is now called 
"Belem" (see J-15, Mound 17). He also tested our Mound 18 in 

2* It should be kept in mind in evaluating the discrepancies in the dimensions here and for the following 
sites, that some are estimates rather than measurements, and that they were made at different times of 
the year when the difference in water level changed the visible extent. In the earlier accounts, 75 years 
less erosion by rain and cattle may be partly responsible for the fact that the measurements are often larger 
than those given by later reporters. 



298 BUREAU OF AMERICAN ETHNOLOGY [bull, m 

several places to a depth of 6 feet, finding sherds to a depth of 3 to 4 
feet but no whole vessels. He attributed the location of the mounds 
to springs along the left bank of the igarape and to the conditions of 
proximity to the campo and accessibility to the breeze from the east, 
which would minimize the annoyance from insects. The owners of 
the cemetery visited by Derby (our Mound 1) refused Farabee permis- 
sion to excavate, and from the magnified dimensions he gives for it 
(1,500 ft. long), it does not seem probable that he was able even to 
make a close examination. 

Excavation was extensively undertaken on our Mound 17, as the 
diagram in his field notes shows (fig. 107). After testing the top and 
along the west side and finding only sherds, he tried the south end, 
where someone had once found a pot, "and at once found so many 
pots together it was impossible to dig without hitting one. In a space 
of six feet square there were six large pots and three small ones." 
This cut he called "plan 1." Plan 2 was excavated near the center 
of the top and plan 3 north of plan 1. An examination of Farabee's 
field notebook (1916 b), where the location of each vessel is given by 
number, leads to a feeling of frustration that the loss of the vessel 
identification should render a major part of the material useless. The 
few associations that can be recognized are extrapolated below. They 
indicate the same trend at Mound 17 as that observed in our work at 
Guajara and Camutins, Mound 1, namely, a transition from secondary 
burial with tangas in large jars to cremation without associated tangas. 

Plan 1 25 
Depth 

1 foot 29, with black ashes, inside 28; 32, engraved, containing child's 

teeth, inside 3T. 

2 feet Group of 6 jars, small to 3 feet tall, all with ashes. 

3 feet. 

4 feet. 

5 feet 24, top broken, bones inside; 27, engraved, bones; 107, large, 

painted, tanga inside. 

6 feet. 

7 feet 109, large, engraved. 

Depth ^^«^ ^ 

1 foot 12, small and round. 

2 feet 8, large, 2 tangas and bones inside, lid inverted over mouth. 

3 feet 1], large, bones inside. 

4 feet 1, 3 ft. high, plain, 4 small bowls (?) inside; 30, large, contain- 

ing bones. 

5 feet 16, large, painted, containing bones and a tanga. 



25 Vessel uumbers are those used by Farabee in his field catalog. Levels shown here without entries do 
not mean no vessels were found, but that the data is not suflicientlj' spocif5c to be of use. A complete 
listing of the vessels fi-om each plan is given by Palmatary, 1950, p. 276. 



EVANS^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 299 

Depth Plan 3 

1 foot. 

2 feet. 

3 feet 4, containing ashes. 

4 feet 1, large, decorated. 

5 feet. 

6 feet 2, painted, large, "beautiful plate" inverted over mouth as 

cover, bones inside. 

7 feet. 

8 feet 61, large, engraved. 

It is uncertain whether Farabee's description of Marajoara burial 
practices applies to both of the sites on which he dug, or to the Camu- 
tins alone. If his interpretation can be relied upon, he found primary 
urn burial: 

Many of the urns were broken from the weight of the superimposed earth, and 
when excavating it was convenient to remove these fragments before disturbing 
the earth in the interior. This method allowed us to cut down in cross sections 
and expose the outline of the bodies in profile. As the bones decomposed, silted 
earth took their places; so, by carefully cutting away the earth, we were able to 
trace out all the bones of the body. In many cases, in the early stages of de- 
composition, the head had fallen forward from the trunk and remained face up 
on the bottom of the urn. The body, no doubt, had been wrapped in cloth or 
bark and then deposited in the urn after it had been placed in a hole dug in the 
mound. The neck of the urn was sufficiently large to admit the body in this 
form. In one of the largest of the urns two adult bodies had been seated side by 
side. [Farabee, 1921, p. 148.] 

Sandoval Lage is the first to record the extent of the mound com- 
plex on the Igarap6 Camutins. He notes that their number exceeds 
40, but errs in attributing to some a height of 20 to 40 meters. He 
says of the ceramics that they are generally comparable to those from 
Pacoval, the greatest difference being a higher frequency of painting 
at Camutins (Lage, 1944, pp. 219 and 225). Lage calls attention 
particularly to a number of mounds on a tributary of the upper 
Camutins, on the Fazenda Sao Marcos, which he believes to have 
escaped previous notice because of their small size (op. cit., p. 217). 

In January, 1950, Hilbert visited Marajo as a member of a party 
from the Museu Paulista, and undertook an examination of the upper 
Camutins. He was able to discover and map 17 sites between the 
last mound we visited (Mound 15, Inajasal) and the headwaters of 
the stream. All have suffered greatly from erosion, expedited by the 
depredations of water buffalo introduced by the ranchers some 30 
years ago, with the result that all but 5 are completely inundated 
during the rainy season. The following descriptions are abbreviated 
from Hubert's notes (1952, pp. 10-15, and pers. corres.). The loca- 
tion is shown on his map (op. cit., pp. 11 and 13). 

Md. 1. Ht. 2.5 meters; covered \^ath bushes and small trees; no surface 
sherds. 



300 BUREAU OF AMERICAN ETHNOLOGY [bdll. m 

Md. 2. Similar to Md. 1, except that height is about 1.5 meters. 

Md. 3. "Pau d'Arco," height 4 meters, covered with trees and bushes. 
Sherds of domestic ware scattered over surface, especially at 
base on northwest. 

Md. 4. Sherds on bank about 1 meter above water level; no artificial ele- 
vation visible. 

Md. 5. "Ingd," height 1 meter at south end, 2.50 meters at north end; 
overgrown with Ingd, trees and bushes. Surface sherds most 
frequent on west and northwest slope. 

Md. 6. Opposite Md. 5, 2.5-3.0 meters high on river side, nearly circular, 
and covered with vegetation; sherds on surface of north slope. 

Md. 7 and 8. Sherds on river bank (cf . Md. 4) . 

Md. 9. Two small mounds, height 1.50 and 1.00 meters; covered with 
bushes and trees. 

Md. 10. Sherds on river bank (cf. Mds. 4, 7, 8). 

Md. 11. Two low mounds on left bank, 1 meter in height, covered with 
bushes and small trees. 

Md. 12. "Aratengd," two summits, south elevation 1.50 meters, north 
one 2.0-2.5 meters, separated by slight depression; many sur- 
face sherds, especially at north end. 

Md. 13. Many sherds on surface of elevated left bank, 3.00-3.50 meters 
high and ca. 300 meters long (cf. Nos. 4, 7, 8, 10). 

Md. 14. "Urubu," in forest some 50 meters from the river bank; no 
prominent elevation; many sherds on surface including typical 
ornate cemetery types. 

Md. 15. "Cuieiras," 75 meters from the river on the right bank; maximum 
height 3 meters, with gentle slope to west; many sherds on east 
slope. Excavation produced 20 anthropomorphic burial urns 
with stylized faces, protruding eyes and smaller excised jars 
with jacar^ modeling. One of the latter contained 64 cylindri- 
cal beads of a white stone flecked with black. 

Md. 16. "Tucumeira," consists of three small, round accumulations 10-15 
meters in diameter, separated from Md. 15 by a narrow low 
area. Artificial elevation barely perceptible, but surface on 
river side produced many sherds and figurine fragments. 

Md. 17. "Furinho," ca. 150 meters long by 30 meters wide; maximum 
height 3 meters at south end, decreasing to 2.5 to 2.0 meters 
going north. Surface sherds abundant. Excavation produced 
sterile soil below 1.5-2.0 meters. Secondary urn burial typical: 
jars plain, painted, or excised. One earplug, 5 tangas, mostly 
red slipped. 

Hilbert gives a general description of the burial pattern derived from 
excavations in Cuieiras and Furinho. Three main types of urns were 
encountered: (1) Painted, anthropomorphic jars, from 35 to 80 cm. in 
height (the most frequent type) ; (2) jars with globular, painted or 
plain body and cylindrical, excised neck, with a total height of 40 to 
50 cm; and (3) excised, cylindrical jars, sometimes with a slightly 
expanded base. Tangas were rare in the jars, only one being found at 
Cuieiras and 5 at Furinho, and tanga sherds were completely absent in 
the surface accumulations at Cuieiras and Tucumeira (Hilbert, 1952, 
pp. 18-19, and pcrs. corres.). There was no evidence of cremation. 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 301 

EVANS J 

Hilbert makes several observations on the quality of the decorated 
types as compared to those at Pacoval: 

The style of decoration and the manner of its execution are the same at Cuieiras 
and Furinho. All the techniques that are typical of the Marajoara Phase occur. 
Beyond this, the following observations can be made: 

I — Predominance of plain and painted pottery. 

II — The paint adheres superficially, and comes off easily. 

Ill — The decoration, whether painted, incised, or champleve, is executed more 
superficially and with less care than is usually the case. The tracing of the lines, 
as well as the disposition and distribution of complexes they form, is frequently 
arbitrary and inconsistent. The general impression is one of lack of the sure and 
uniform st}'listic sense that is typical of the Marajoara Phase. 

IV — The modelled ornamentation shows, in part, this same characteristic. 
[Hilbert, 1952, p. 20.] 

Hubert's classification of the sherds on the basis of core character- 
istics, ignoring surface decoration, gave 65.3 percent Inaja Plain of 49 
sherds from Cuieiras and 37 percent Inaj4 Plain of 27 sherds from 
Furinho (ibid.). 

Bibliography: Derby, 1879, p. 226; Farabee, 1916 b {also in Palmatary, 1950, 
pp. 275-276); Farabee, 1921, pp. 145-146; Hartt, 1871, p. 260; Hartt, 1885, pp. 
23-25; Hilbert, 1952 and Personal Corres.; Lage, 1944, pp. 215-219; Maraj6, 
1895, p. 88; Martins, 1867, p. 178; Palmatary, 1950, pp. 272, 275-277. 

CARATAT^UA 

About 8 km. southwest of Lago Guajara a patch of forest stands 
out from the treeless plain and marks the location of this site. Hilbert 
describes it as extending approximately 500 meters northeast-south- 
west, and narrowing from 150 meters near the southwestern end to- 
ward the northeast. There is no marked elevation, but it is probably 
sufficient to prevent flooding. A 1- by 1 -meter test produced few 
sherds and sterile gray sandy clay at a depth of 30 cm. Of the 44 
sherds, 40, or 90.9 percent, were on Inaja Plain; 4, or 9.1 percent, were 
on Camutins Plain. Decorated types included Anajas White In- 
cised. There was one fragment of a stool, but no evidence of tangas. 

The existence of other smaller sites of similar nature in the vicinity 
leads Hilbert to the conclusion that Caratateua represents a complex 
of sites similar to Teso dos China (pers. commun.). 

CUIEIRAS 

"Mound das Cuieiras" is one of those Usted by Barao de Maraj6 
(1895, p. 87), with no location or description. This may be the same 
as Santa Brigida, which is on the Igarape Cuieiras. 

CURUXYS 

Holdridge (1939, p. 43) mentions having excavated a burial mound 
on Fazenda Curuxys, which is located about half way between Lago 
Aran and the east coast at Soure. 



302 BUREAU OF AMERICAN ETHNOLOGY [bull. 16T 

DESTERRO 

A mound called "Ilha das Panellas" is located on the Fazenda 
Desterro, which in 1895 belonged to Francisco L. Chermont (Maraj6, 
1895, p. 88). Holdridge visited & fazenda by that name just north- 
west of Laranjeiras in 1938, and an indefinite reference to excavations 
may have been located there (1939, p. 175). Another Jazenda of the 
saro.e name is on the upper Rio MocSes. 

DIAMANTINA 

Farabee conducted excavations in what appears to be a group of 
habitation mounds east of Lago Arari: 

Went to work with 5 men but found no mounds at all, nothing but house sites, 
a dozen or more on the banks of igarap6 from Lake Arary. Here the banks 
were above high water and a good place to live, about 4 miles from the lake. . . . 
The 5th and 6th and 7th we dug these sites and found a few burial pots, all plain, 
unpainted, small; all had charred bones inside. Found only one fragment of a 
painted pot and one engraved; all (other) fragments were of that coarse undeco- 
rated ware. No fragments of animal bones, no fireplaces, only ashes mixed 
with earth in one site. . . . No place was two feet deep to original claj'. 
Saved 4 pots and 2 fragmentary plates, 1 cup, 3 stone axes found together near 
surface alone, some other pieces of stone with grooves, one a good axe. [Farabee, 
1916 b, pp. 2-5.) 

FORTALEZA 

This cemetery on the upper Rio Goiapl, southeast of Lago Ararf, 
was first reported by Barnard in 1871 as being "a mound 8-12 ft. 
high, forming an island during the annual overflow and full of vases" 
(Hartt, 1871, p. 260). Thanks to Farabee, it and the associated 
habitation mounds are the most thoroughly excavated of all Mara- 
joara sites (fig. 109). Although the ceramic identifications have been 
lost, as happened in the case of his Camutins specimens, the Fortaleza 
field notes (1914) are somewhat fuller and accompanied by numerous 
maps and diagrams. The 14 mounds are described as follows with the 
dimensions representing the extent at the end of the dry season: 

No. 1. Did not examine. 

2. 10' high, 150' in diameter, steep on the stream side, sloping gently in 

other directions. 

3. 3' high, oblong, 20' by 30'. 

4. 5' high, 30' in diameter, steep on the stream side. 

5. 7' high, 60' in diameter. 

6. 4' high, 15' Viide, 40' long, in open field by side of stream. 

7. 10'-12' high, 200 yards long, 75 yards wide, in forest by side of river. 

River 75 yards wide. Broad, flat top, great many sherds mostly same 
coarse kind as 1 and 3. 

8. 6' high, irregular in shape, 250' by 300', just across stream from 7, same 

class of sherds. 

9. At fork in the river, 6' high, 20' in diameter. 

10. Left of stream, 12' high, 70' diameter, perfect cone. 

11. On island in the middle of the river, 12' high, 300' long, 30' wide. 



MEGGERS AND 
BVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



303 



12. On island 30' west of 11. 

13. On Sta. Cruz, }i mi. below island. 

14. On Sta. Cruz, yi mi, below 13. 

Farabee made extensive excavations in 4 of the habitations and, 
although he felt poorly rewarded for his time and efforts, the results 
go far to demonstrate conclusions that would otherwise remain tenta- 
tive. In addition, he investigated Mound 7, which appears to be a 




Figure 109. — Plan of the Fortaleza Mound Group of the Marajoara Phase on 
the Eio Goiapf, (After Farabee, 1914.) 



304 BUREAU OF AMERICAN ETHNOLOGY [bull, le? 

cemetery. Since his large-scale digging led to the same interpretation 
as our limited tests, the account of his discoveries has the double inter- 
est of detailed description of the composition of the mounds and of 
demonstration of the reliability of the data derived from small tests by 
archeologists trained to recognize pertinent features. All quotations 
are from Farabee's 1914 field notes. 

Mounds 2 and S. "Nov. 25, Wed. — Sent four men to work on Mound No. 
2, there being nothing whatever in No. 3 on the sides. After cutting all to 28 ft, 
I felt justified in continuing with 20' through the center in hope of finding some- 
thing at the bottom near the middle. At noon found pot 2 in the very center on 
top, 6" below surface. 

"Nov. 26, Thurs. — Character of Md. 2 same at 30' except the narrow layer of 
ash has disappeared. The bottom ash is same, nothing in the ash. In the hard 
earth below 2)^' often is found small patches of ash with no evidence of fire. . . . 

"Nov. 27, Friday — Work continued on Mound No. 2 and 3. Nothing but 
sherds found. . . . 

"Nov. 30, Mon. — Continued work on Md. No. 2 with eight men. 

"Trench No. 1, 15' wide and 10' deep and 40' long, started from river at original 
surface of campo. Some ashes on original surface. Ashes and earth were strati- 
fied until the high water level was reached. . . . No thick stratum of ashes but 
here find numerous fire-places on burnt earth with fine charcoal in ashes and small 
or thin patches of ashes. In one of the largest fireplaces were many fragments of 
the coarse red ware with many bottoms of pots, possibly burning place. Frag- 
ments of deer jaw-bone and snake backbone . . . 

"Trench 2, 15' wide, 9' deep. Same level as other, some ashes on bottom but 
no fireplaces. Few fragments of pottery; near middle length, burnt earth as of 
fire-place. After continuing for 45' to centre of mound and finding nothing new, 
I discontinued . . . 

"Trench 3. Cut three trenches 3' wide and 30' long down to original [surface], 
lYi' apart. Nothing but very few fragments, no ashes, no fireplaces. Solid earth 
same as campo. [Trenches dug on] north side. 

"Trench 4. Cut 3 trenches 4' wide and V/z apart, 30' long, on south side 
[i. e., end]. Found immense quantities of fragments and burnt earth — sweepings 
from pottery factory. Very little ashes. Dump heap is reddish on account 
of burnt earth. Found pottery lamp ^6 and fragment of pottery mold ^s for out- 
side of pots." 

Mound 3. "Cut trench along side and cross middle, 3' wide and 20' long. 
Very few fragments and nothing more. 

Mound 4- "Dec. 1, Tues. — Put two men on Md. 4. Cut trench 10' wide, 5' 
deep, past middle. [At] 15', found some sherds and little ashes in patches. 
Stone axe near surface. 

Mound 5. "Nov. 20, 1914 — Went to work on mound No. 5 of plan as it seemed 
the most likely, being a round mound and centrally placed. Had but one man 
and my boy, but we made a good start and [I] was encouraged by finding one 
good specimen. The mound was very hard as it was near [the end of] the dry 
season. . . . 

"Nov. 21, Sat. — Continued with only one man, found nothing but sherds. 
Some white ash found but no charcoal, no fragments of hone or stone. While 
sherds are common they are not numerous and all of the thick red ware. . . . 



' These identifications are erroneous. 



MEGGEKS AND 

EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



305 



"Nov. 23, Mon. — Worked with 2 men. At 10' on bottom found white ash ^^ 
8" thick just above 3" or 4" river deposit and then 2" to 3" ashes and earth to 
top carried and placed above. The bottom is the general level of the campo and 
from 12" to 15" below high water. . . . The same stratification continues at 
20'; nothing found in the ashes. 

"Nov. 24, Tues. — Same stratification at 25'. Nothing found except fragments. 
The first 2}^ ft. of top has fragments. Next 3' nothing, next 1}^' few fragments. . . . 

"Nov. 28, Sat. — Abandoned work on Md. No. 5 at 9:00 A. M. having excavated 
well past the middle of the mound down to original soil and found practically 
nothing." 

Mound 7. — Excavation began with 2 trenches at the south end, in 
which he "dug all day and got not even sherds." Two more on the 
west side, toward the north end, were slightly more encouraging: 
trench 3 produced sherds and fragments of tangas, and trench 4, 




Figure 110. — Detailed plan of Mound 7 of the Fortaleza Group showing the 
extent of Farabee's excavations in this Marajoara Phase cemetery site (after 
Farabee 1914). 



two vessels containing ashes and bone fragments at a depth of 1 foot. 
Although a prohibition against felling any of the trees growing on the 
site meant that excavation had to be undertaken patchily, a glance 
at the diagram (fig. 110) Avill show that the testing was quite thorough. 
Among objects of an unusual nature from this mound were: 

Trench 4: 2 stone axes. 

Trench 5: Fragment of a spindle whorl. 

Trench 6: Small quartz ax 2 feet deep; 6 stools scattered. 15 inches or 

less in depth. 
Trench 8: Group of 9 clay stools piled together south of a large pot, half 

of which projected above ground. Of the stools, 3 had ashes 

inside, 2 were set over clay; all were right side up. (Farabee, 

1914.) 



-' The use of the term "ash" here and probably elsewhere in Farabee's notes is misleading. Our excavation 
showed that the core of the mounds is composed o Ja flaky, speckled white clay that he has erroneously 
identified as ash. 



306 BUREAU OF AMERICAN ETHNOLOGY [bull, m 

In a letter to the University Museum, Farabee summarizes the 
condition of the burials: 

. . . the burials were grouped. In a space of 50 feet square, we might find 
50 pots, and then another 50 without finding a thing. Once in a space 4' by 2' 
we found 7 po^s belonging to 4 different burials. 

Apparently, they practiced two methods of burial: in one, the bones after they 
had been disarticulated and, in some cases broken, were put in pots and buried. 
The other method was to burn the body and bury the ashes in a pot. . . . Very 
little was found inside these burial pots or with them. In some cases a tanga 
was inside the pot but more often it was placed near the pot on the outside, 
along with plates and dishes. [Letter of Feb. 8, 1915.] 

In the hope of providing a basis for seriating this site with those we 
excavated, we classified the collection of 746 sherds at the University 
Museum, Philadelphia, Hsted as from Fortaleza according to our 
ceramic types. The entire range of decorative techniques is repre- 
sented (Appendix, table 43). Of the decorated sherds, 376 have the 
gray core of Inaja Plain and 108 the orange core of Camutins Plain. 
Added to the plain sherds, this gives a total of 552 or 74 percent 
In&jk Plain and 194 or 26 percent Camutins Plain (Appendix, table 
41). 

Bibliography: Farabee, 1914, 1915, and 1921, pp. 144-145; Hartt, 1871, p. 260; 
Palmatary, 1950, pp. 274-275. 

guajabX 

A mound covering 4 or 5 acres is reported by Hartt (1871, p. 260) 
to be located on the campo near Lago Guajara, east of Lago Ararl. 
This mound is not to be confused with another of the same name in 
the Monte Carmelo group. 

ILHA DOS BICH08 

This cemetery was visited by J. B. Steere in the rainy season of 
1871, and the major part of his collection was deposited in the Museum 
of Anthropology, University of Michigan. Located in the campo 
near the Rio Arari, upstream from the modern settlement of Arariuna, 
it was about half an acre in area and 15 to 25 feet high. The surface 
was covered with trees, and the rains had washed deep ravines in the 
sides. Sherds were abundant on the surface and a test showed the 
refuse deposit to be of "considerable depth." Burial urns protruded 
at various levels from the eroded sides. Although these were broken, 
Steere describes them as "upright, with straight sides and with large 
covers like broad-brimmed hats. Both the urns and the covers showed 
remains of painting in various bands and figures" (Steere, 1927, 
p. 22). All traces of bones had disappeared, but several of the urns 
contained beautifully polished and ornamented tangas. As a result 
of his digging, Steere concluded that: 



EVA^Nsf ^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 307 

The mound appeared to have been built to a certain height, inhabited, and the 
dead buried beneath. Then, after a time, another layer of earth and another 
period of occupation was added. Two of these levels showed paving of burned 
clay, which was covered with ashes, charcoal and broken pottery. [Op. cit., 
pp. 22-23.] 

A small collection deposited by Steere at the Museum of Anthro- 
pology, University of Michigan, is presumably from this mound. 
It contains 31 vessels and sherds, all but three with decoration. 
This includes all of the Marajoara Phase types except Goiapl Scraped, 
Anajas Plain Incised, and Carmelo Red. A classification on the 
basis of the plain ware on which the decoration was applied gives 
(omitting 7 complete vessels) 15 or 62.5 percent Inaja Plain and 9 
or 37.5 percent Camutins Plain. 

Bibliography: Meggers, 1947; Palmatary, 1950, p. 271; Steere, 1927, pp. 22, 23. 

TLHA DOS UARCOS 

For this mound, we have nothing but a reference to its existence 
on Sifazenda belonging to Cruz Macedo & Cia. (Maraj6, 1895, p. 87). 

LARANJEIRAS 

Tocantins names this, along with Camutins and Pacoval, as typical 
of the mounds found on the campos of Maraj6, covered with luxuriant 
vegetation and containing ceramic deposits, especially burial urns. 
It also appears in the Barao de Marajo's listing, in which he describes 
it as "bastante rico." More recently, Laranjeiras was visited by 
Holdridge, who reports it to be about 15 feet high and to cover an area 
of over 2 acres. At the time he was there, the laSiiB. Jazenda house 
was situated on the summit and his description of the ravages wrought 
by many forces is typical of what has happened to many of the other 
Marajoara mounds: 

Everywhere the ground was littered with the bits of strange funeral pottery — 
the roots of the great trees had reached down among the dead, expanded in their 
vigorous life, and crushed the urns that held the bones of the men and women 
who built the very mound on which the trees grew. Cattle had stamped over the 
graveyard in the wet season, their dull feet destroying the art and dreams of 
whole generations. Ranch children had excavated, looking for dolls, and smashed 
what did not please them. Cowboys . . . had dug for gold and, in disap- 
pomtment, destroyed all they could. Several archeologists had been allowed a 
fly-by-night kind of permission to excavate which had served only to whet their 
appetites before they were asked to leave. [Holdridge, 1939, pp. 69-70.] 

Digging produced a "lovely big burial jar" with anthropomorphic 
features, associated with *'plates and dishes, fragments of dolls and 
whistles, brolien bits of tangas'* (op. cit., p. 71). On Holdridge's map, 
Larangeiras is shown about half way between Lago Arari and Cabo 
Maguarl, the eastern tip of Maraj6. 



308 BUREAU OF AMERICAN ETHNOLOGY [bdll. 167 

Bibliography: Holdridge, 1939, pp. 68-71; Maraj6, 1895, p. 88; Tocantins, 
1876, p. 55. 

MACACAO 

The Museum of the American Indian, Heye Foundation, has a 
number of specimens from this mound, which is shown on Palmatary's 
map as northeast of Lago Guajara (1950, p. 283). 

MATIN ADOS 

Holdridge is tantalizingly indefinite about this site or sites on a 
^azenda shown on his map (1939) as just northeast of Lago Guajar^. 
His use of the plural may refer to several sites here, or simply to his 
experience in digging mounds in general. It may be of Matinados 
or of Laranjeiras he is speaking when he says, "we found one burial 
urn four feet high with sculptured figures on its side, human faces 
portrayed and painted designs," with which were associated small 
vessels and tangas (Holdridge, 1933 b, p. 204). He thought he could 
distinguish three horizons of ceramics in the mounds: at the bottom, 
"a layer of simple red ware without designs, incisions, or bas-relief"; 
in the middle, "a layer occupied by exceedingly beautiful pieces 
painted with fantastic designs and incisions of an infinite variety," 
and on top, the simple red ware again (ibid.). Careful excavation 
in other mounds, however, does not produce any evidence to 
substantiate this impression. 

MBNINO DEUS 

Farabee mentions a "low mound from which round pot came" at 
Menino de Dios (1916 b, entry for Jan. 8). Mordini lists "Menino 
Deus" as one of the sites in the area enclosed by the Ganhoao and 
Cururii Rivers and the Lagos Mututi and Asapao (1934 a, p. 62). 

MONTE CARMELO 

Steere (1927, p. 23) recounts that in 1879 he "camped for several 
days on a large mound of several acres on the little river Anajas," 
which probably was one of the Monte Carmelo group. The first to 
conduct any excavation appears to be Desmond Holdridge. He de- 
scribes the site as "near the source of the Rio Anajas" (1939, p. 72), 
and although he attaches the name "Monte Carmelo" to it, his de- 
scription applies best to the mound now called "Guajard" (J-14, 
Mound 1). Either through error or exaggeration, he gives the height 
as "about 70 feet," or almost 7 times what it was in 1949, only 18 
years later. Among the products of his excavation was a large an- 
thropomorphic burial urn (now in the Brooldyn Museum) with 
modeled faces on the neck and stylized, painted delineation of the arms 
and hands on the body (op. cit., photograph opposite p. 87). In 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 309 

EVANS] 

general style it is very similar to jar L from our Guajard cut (pi. 76). 
It contained miscellaneous sherds, fragments of tangas and a secondary 
burial (op. cit., pp. 72-73). 

NAZARETH 

Our only knowledge of this site comes from the listing by the Barao 
de Maraj6 of a mound on the Fazenda Nazareth, belonging to Sr. 
Francisco L. Chermont. There is 2k fazenda by this name in the head- 
waters of the Rio Mocoes (Marajd, 1895, p. 88). 

PACOVAL 

Ilha de Pacoval, on the east shore of Lago Ararf, just above the 
mouth of the Igarap^ das Almas, is the most frequently visited mound 
on the Island of Maraj6, both because of its accessibility and because 
of the apparently inexhaustible richness of the ceramic remains. 
Even today, after 80 years of uninhibited exploitation, one can stiU 
pick up "hundreds of little items: small figurines, adornos, fancy rims, 
extremely good-looldng incised and champleve sherds or even painted 
ones" (Hilbert, pers. commun.). 

The earliest recorded visit, although not the first to be published, 
was made by Couto de Magalhaes who traversed Maraj6 in 1865. 
Even at that early date he reported that 

many artificial mounds are known ... of which one of the most notable on the 
island of Maraj6 is an artificial island in Lago Ararf. [1873, p, 410.] 

He deduced that they were built to raise the houses above the annual 
flood waters: 

Considering that the region in which they are found is inundated for many tens 
of leagues ... it is evident that the people, from the time they arrived in the 
area, began to build the mounds, without which it would be impossible to explain 
their method of existence during the rainy season in places that become true 
Mediterranean oceans. [Couto de Magalhaes, 1873, pp. 410-411]. 

Six years later, in 1871, Hartt sent his assistant, W. S. Barnard, to 
examine the site and his is the first detailed description : 

The Ilha das Pacovas lies close to the western 28 side of the lake, opposite the 
beginning of the Rio Arary, which forms the outlet to the lagoa, and just to the 
south of the mouth of the Igarap6 das Armas. It is oblong in shape, about ninety 
paces in length from north to south, and about forty paces in width. In the 
month of November, when the water was low, it was somewhat over ten feet in 
height above the level of the lake. It is for the most part covered with large forest 
trees. Situated at the northern end of the island, and separated from it by a 
narrow channel, is a little crescent-shaped islet apparently built on as an addition, 
and not so high as the main island. Both were evidently raised artificially, and 
are full of burial vases and pottery of all kinds. The vases, which are about three 
feet in height, are, in some places, buried as many as three or four above one 
another, but they are more or less scattered. The waves have worn away the edges 

" The directions are confused in several places in this account. 
391329—57 22 



310 BUREAU OF AMERICAN ETHNOLOGY [bull, igt 

of the island making a sloping shore full of broken burial jars and thickly strewn 
over with fragments of pottery. [Hartt, 1871, pp. 260-261.] 

In the following year Pacoval was visited by Steere, Derby (whose 
collection is mostly at the Peabody Museum, Cambridge) and Ferreira 
Penna (who collected for the Museums in Belem and Rio de Janeiro). 
Ferreira Penna's description agrees with that of Barnard, and he re- 
marks that the eroded northwest shore of the mound was so thickly 
covered with pottery fragments that there was hardly a spot where 
one could put his foot without stepping on a decorated sherd or part 
of a figurine. Steere had informed him that he noticed a marked 
change in the ceramics in the deposit, with the best examples in the 
lowest level and the quality deteriorating toward the top. This 
observation interested Ferreira Penna and he made an effort to check 
it: 

As soon as I arrived at Pacoval, I began to have the thick vegetation which 
covered the mound cut away, and there began shortly to appear several circles 
which were nothing less than the mouths of urns buried there and without lids. 
They were all of coarse, dark clay, and without any decoration except for some 
almost vanished traces of white paint in angular designs. They were broken and 
filled with dirt, with their own sherds and with fragments of a few small vessels 
originally placed inside them. In one of the jars was found a small pipe,^^ which 
although very crude is of interest because it is the only one that has been found in 
the Pard. mounds. 

One side of the mound was subject to erosion during the winter by the action of 
the lake waters, and the excavations executed there produced two urns, one painted 
red and yellow on a grayish background and the other with relief and painting in 
blue and red. 

While this excavation was in progress, I saw a large urn, beautifully painted and 
with modelling on its upper and lower parts, sheltered by the huge roots of a robust 
caja tree, which fell after being undermined. A few of the tree roots had pene- 
trated into the mouth of the jar and cracked it, so that the fall broke it completely 
apart. 

The work undertaken in the lower part of the mound was time-consuming and 
laborious; the earth was as if petrified by the presence of minute fragments of 
pottery which were imbedded in the clay like a coarse mosaic. The outcome of 
these efforts, if unsatisfactory in not producing a single complete specimen, gave 
me numerous fragments that were notable for their ornamentation and for their 
choice of material. I believe that it is only in this section that have been dis- 
covered those strange, triangular objects known as Tangas. . . . 

On the basis of the admittedly incomplete examination I made of this mound, I 
was able to recognize that there existed at least three levels or orders of vessels, 
the lowest level containing the most perfect ones and the upper the crudest, cor- 
responding with what Steere had previously observed. [Ferreira Penna, 1879 a, 
pp. 52-53]. 

Derby paid another visit to Pacoval in 1876, and his observations, 
as recorded by Hartt, enlarge the picture of the bm*ial pattern: 

The best-made objects are the burial urns, which were interred with particular 
care. The earth around them is frequently fine sand mixed with sherds, ashes and 

" This is probably a pottery spoon (p). 81). 



Kv^Nsf^ *^° ARCHEOLOGY AT MOUTH OF AMAZON 311 

carbon, indicating that after the urn was set in place the hole was filled with a 
special kind of dirt. Fine sand and sherds are also occasionally found inside the 
urns, mixed with the bones. Sometimes an elaborately decorated jar was put 
inside a larger plain one. They appear always to have been covered with a lid, 
but this has generally been broken and the fragments fallen into the jar. The 
bones found inside the jars are very poorly preserved, disintegrating into powder 
when exposed to the air, and in some cases completely vanished. In several 
instances I was able to tell from the bones that the complete skeleton had been 
buried, although the mouths of the jars I saw were not wide enough to admit a 
human body intact, nor was the jar large enough to hold it. It seems likely, 
therefore, that only the skeleton was buried after the decomposition of the flesh. 
It is certain also that some of the belongings of the deceased were placed with the 
body. In two cases I found tangas inside the jars, and at least one of these could 
not have been intruded accidently. The urn had been set inside a larger one and 
the tanga was in the space between the two. In one or two other burials I found 
small vessels inside the urns that seemed to have been used to hold paint or snuff. 
Stone objects are exceedingly rare. I did not find a single one, but I have seen a 
few that were said to have been found at Pacoval. [Hartt, 1885, pp. 22-23.] 

Netto, who came about the same tune, adds a few more details: 

The burial mound called "Ilha do Pacoval" is an artificial hill. . . . This hill 
is located on the east shore of Lago Arari, on the interior of Maraj6 Island, and 
having been constructed more in the lake bed than on the shore, is now an island 
and now a peninsula, depending on whether the water level is high or low. [1885, 
p. 266.] 

He also comments on the difference in the ceramics and suggests three 
possible interpretations: 

My own excavations, as well as those made by Derby and Ferreira Penna, have 
shown that along with the jars of the finest workmanship or the most delicate 
painting are found exceedingly coarse vessels which have no characteristics that 
would authorize their classification as representing the same period or would 
indicate the same source of manufacture as the former. ... I do not wish to 
say whether this crude pottery is evidence of the progressive degradation under- 
gone by the people who made the lovely jars referred to above, or whether it is 
the product of a less civilized group that by chance settled on the same spot 
deserted by the inspired and skillful makers of the most beautiful ceramics of 
South America. A third possibility is also plausible, and that is that the different 
qualities of urns were correlated with different classes of individuals: the impor- 
tance or obscurity of the deceased determining the richness or plainness of the urn 
in which he was buried. [Op. cit., p. 265.] 

Another early visitor was Tocantins, who commented on the prom- 
inence of the mound in the rainy season : 

At the time of my visit it rose 1.14 M above the level of the lake, while the 
adjacent campo was 2.28 M below the same level. This was the only point within 
a radius of several leagues that stood out from the flood waters beneath which 
the entire central basin of Marai6 was submerged. . . . The site ... is even 
today distinguished from the surrounding terrain by its thick, black layer of soil 
which is at the present time shaded by luxuriant vegetation. [Tocantins, 1876, 
p. 55.] 

Other reports about this time give the elevation as from 3 to 8 meters, 



312 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 




LOW FLAT BEACH 




• EXCAVATED JARS 



Figure 111. — Plan of the Marajoara Phase site of Pacoval made by Lange in 1913. 

depending partly on whether the water level was high or low (Ferreira 
Penna, 1879 a, p. 51; Hartt, 1885, p. 20). 

At the end of the 1913 dry season, Pacoval was visited by Algot 
Lange, A diagram he made of the site is reproduced in figm-e 111. 
He reported that: 

The general form of the island is that of a narrow parallelogram with rounded 
corners. . . . The northern section is higher than the southern, reaching, at a 
certain point, a height of four yards above the present water level. At the 
extreme of the wet season, when the greater part of Pacoval is submerged, . . . 
this highest point forms a narrow ridge just beyond reach of the choppy waves 
of the lake. This ridge slopes off to either end. . . . [Lange, 1914, p. 308 and 
photo opposite p. 318.] 

The soil throughout is a black, rich clay resembling the kind that is found 
at the bottom of the holes or ponds in the prairies. In places there is an out- 
cropping of sand. Throughout the island the earth is mixed with pottery frag- 
ments from the size of minute particles to pieces weighing as much as twenty- 
five pounds. [Op. cit., pp. 310-311, photo opposite p. 309.] 

Lange began by raking the surface of the entire site, "allowing 
no piece showing the slightest value to escape our farina baskets" 
(op, cit., p. 314). This procedure took several days, and during work 
along the western shore the mouths of a number of large m"ns were 
discovered 

sunk into the ground, forming a straight line following the littoral. I counted 
twelve in all. The first of these was broken upon reaching a depth of two feet. 
It crumbled into minute fragments when fully exposed. The mouth of the 
second jar was plainly visible amidst many broken pieces on the surface of the 
beach. This we attacked judiciously . . . [and] the result of this cautious work 
of many hours is a pyriform funeral urn of almost three feet in height. Upon 
placing this on the beach I remove a quantity of earth and greyish ashes from 
the interior, with the result that I find some human bones, all broken in pieces, 
and underneath these a most peculiar object of rare beauty and skill of 
workmanship. It is a perfect specimen of the original primitive "figleaf" for 
women. . . . 

Working our way around the old fallen mucajd palm we make some very rich 
finds. Whole vases covered with what appears to be symbolic figures of a com- 



UBGGBRS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



313 



plicated pattern, bordered by stripes of red and brown painted bands, surprise 
us and cause my sincerest admiration for the permanency of the colours which 
appear now as fresh as if painted but yesterday. I find also small platters with 
labyrinthic designs painted and incised. . . . Large and small idol heads, many 
of which are broken, lie around, some partly covered by earth and dirt. Close 
to the palm trunk I institute a series of excavations, and during many days we 
find here some rare objects of pre-historic art, such as a large richly painted and 
incised idol." . . . [Op. cit., pp. 314, 316-317.] 

In the 38 years that have elapsed since Lange's visit, the mound 
has continued to decrease in elevation. The most recent description 
of its condition is by Hilbert (1952, pp. 21-30, and pers. corres.), 
who visited the site in 1951. At that tune there was a single tree 
on the northeast end and a native house on the highest point toward 
the southwest part, where the elevation reached 1.60 meters during 
the dry season (fig. 112). As the water level in the lake rises with 




Figure 112. — Plan of the Marajoara Phase site of Pacoval made by Hilbert in 

1951. 

the rains, the mound shrinks until it is completely inimdated. Now 
that vegetation is no longer able to maintain a foothold on the mound 
and the soil is not bound by the interlaced roots, the site is being 
washed away at an accelerating pace. Its long use as a modern house 
location, its frequent submission to excavation and its annual subjec- 
tion to the erosion by the waves of Lago Arari have resulted in thoT- 

^ This white incised, red retouched (Pacoval Incised), anthropomorphic jar is in the American Museum 
of Natural History 



314 BUREAU OF AMERICAN ETHNOLOGY [bull, ler 

ough disturbance of the cultural remains, with the intrusion of modem 
materials (china, tile, and beer-bottle fragments) to a depth of 90 cm. 

Hilbert made three stratigraphic excavations. Pit 1, 1.50 by 1.50 
meters, was on the slope away from the lake, 10 meters south of the 
tree. The upper 4 cm. were occupied by humus and roots. Between 
4 to 14 cm. the largest sherds (averaging 10 cm. in diameter) were 
found. As the depth increased the sherds became smaller and in- 
creasingly eroded, often reduced to particles like coarse sand. At 
50 cm. this gave way to sterile gray clay. An expansion of the cut 
for 4 meters toward the west revealed the same conditions. Pit 2, 
at the summit of the northeast end, was 2.50 by 1.50 meters. The 
situation was the same here as in pit 1, except that the condition of 
the sherds was somewhat worse. Intrusive materials included modern 
tile and pottery at 35 cm., china at 55 cm. and bits of a green glass 
beer bottle at 90 cm. Pit 3 was a trench 1 m. wide begun at the 
bottom toward the southwest end of the east slope and extended into 
the mound for 4 meters. No new featm-es were discovered. During 
tests along the lake side, the refuse was found to extend to a depth 
of 10 to 35 cm., the higher figure pertaining to the northeast end and 
the lower one to the southwest. The surface collection of 307 sherds 
was classified by Hilbert on the basis of core color, giving 86 percent 
gray cored, or Inaja Plain, and 14 percent orange cored, or Camutins 
Plain (Hilbert, 1952, pp. 28-29 and pers. corres.). 

Specific information on the method of disposal of the dead prac- 
ticed at Pacoval is limited, but secondary burial is the most often- 
mentioned type. Hartt (1885, p. 22) concluded from the arrangement 
of the bones in one jar that the skeleton had been placed inside with 
the articulations at least partly intact. The fact that some of the 
jars were too small to hold a complete skeleton, even if disarticulated, 
led him to explore the possibility that cremation was also practiced: 

All the bones found in the urns were fragmentary. The probabilities are that 
the bodies were burned, and that only the ashes and charred bones were placed 
in the urns. An analysis of a small amount of black ash-like earth, found adher- 
ing to one of the Jars, was made for me by one of my students, and found to con- 
tain a very large percentage of phosphate of lime. [Hartt, 1871, p. 263.] 

Netto also reports the discovery of "two or thi-ee partly calcined 
skulls," but concludes that cremation was unusual (1885, p. 427). 

The question of stratigraphy at Pacoval has been of interest from 
its earliest examination in 1871, when Steere and Ferreira Penna 
agreed that at least 3 strata could be discruninated by the quality 
of the ceramics they contained. A somewhat different conclusion 
was reached by Mordini, as a result of his excavations in 1926. He 
isolated two ceramic-bearing deposits separated by a sterile layer 
48 cm. thick. The wares of the two strata were not distinguishable in 



"vA^Nsf^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 315 

style or ornamentation, but the examination by micro-emulsion showed 
the presence of cariape tempering in those of the upper level, and it 
was absent in the lower level. Mordini concludes: 

From this it is possible to maintain that the Marajoans arrived at the island 
ignorant of the custom of adding siliceous material of biological origin to the 
clay used in making their pottery. Later, they acquired this cultural element 
either by local invention or by trade with some other Amazonian tribe. [Mordini, 
1934 b, p. 15; cf. Mordini 1947, p. 640, and Linn6, 1931, p. 281.] 

As explained above, Hilbert found the site completely disturbed at 
the time of his visit in 1951, making stratigraphic analysis impossible. 
In the hope of being able to seriate this site with others and thus 
determine the relative antiquity of some of the Marajoara cemeteries, 
we classified the Pacoval collection of the Peabody Museum, Cam- 
bridge, into the ceramic types established for the Marajoara Phase. 
The full range of decorated types was present, but only 29 plain 
sherds were included in the collection. In order to get a more reliable 
picture of the ratio of Inaj^ Plain to Camutins Plain, the decorated 
sherds were classified into these two types on the basis of their paste 
color. This increased the count to 234 Inaj^ Plain and 79 Camutins 
Plain and gave a ratio of 75 percent Inaja Plain to 25 percent Camu- 
tins Plain. In computing the relative frequency of the decorated 
types, 1,039 sherds from the American Museum of Natural History 
collection from Pacoval were also included, giving a total sample of 
1,353 decorated sherds (Appendix, table 41 and 43). 

Bibliography: Derby, 1879, pp. 225-226; Derby, in Hartt, 1885, pp. 21-23; 
Ferreira Penna, 1879 a, pp. 51-53; Hartt, 1871; Hilbert, 1952, and personal 
correspondence; Lange, 1914, pp. 307-322; Marajo, 1895, p. 87; Meggers, 1948, 
pp. 153-154; Mordini, 1934a, pp. 62-63; Mordini, 1934 b, pp. 15fif.; Mordini, 1947, 
p. 640; Netto, 1885, pp. 265-268; Netto, 1890, p. 202; Sampaio, 1922, p. 849; 
Tocantins, 1876, p. 55. 

PACOVAL DO CURUEti^ 

The earliest and most detailed account of this cemetery mound is 
by Nimuendaju, who visited it at the end of 1923: 

I then went a short distance in south-western direction from the bank of the 
Cajueiro [to] the upper Rio Cururii. ... At its left bank the Fazenda Pacoval 
is situated, and 600 M further up a very interesting mound is located on a point 
of land between the Rio Cururu and its small tributary on the left. A canal 
now cuts across behind the mound so that it forms an island. The whole country, 
almost as far as the eye can see, is a tree-less plain, which is submerged nearly 
six months of the year by the high water. ... Its length is 220 M and its overall 
width 50 M. It lies in a north-south direction. The northern and central parts 
are the widest, the southern one forming a long, narrow point. The greatest 
height (5 M) is near the northern end. Then the crest slopes down to about 1.50 
M, rises again to more than 2 M and ends in the southern point mentioned. The 
alluvial soil is yellow clay, which is so intermixed with tiny fragments of pottery 
that, at a cursory glance, it gives the impression of being gravel. Fragments 



316 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

exceeding the size of a hand are rare, at any rate on the surface. The north and 
east sides are covered with sherds, as is the narrow strip of land between the 
mound and the river, whereas they are absent on the corresponding surface to 
the west. ... It is overgrown with tall trees so that it is visible from a great 
distance in the flat, tree-less surroundings. As I was forbidden to make excava- 
tions ... I had to confine myself to collecting a few fragments from the surface 
with the characteristic incisions in Maraj6 style. No traces of painting remain; 
they may, however, have been effaced by the weather. From the land-steward 
of the Fazenda I got a zoomorphous, spherical vessel without neck, with narrow 
rim and two pairs of cord-holes at the side [see Nordenskiold, 1930, pi. 14-e]. 
It is decorated in the characteristic manner and the cross-motif recurs no less 
than ten times. Further, I got a thick bead (?) of pottery. [Nimuendaju, in 
Ryd6n, MS.] 

The Goteborg Museum collection includes this excised vessel and a 
fragment of a stone ax of fine-grained, dark-green diorite (?). 

Pacoval do Cururii was visited in 1930 by Sra. Heloisa Alberto 
Torres. Lothrop, in reporting her work, describes the site as being 
under water except at low tide (1934, p. 820). This is contradicted 
by Moraes' statement that it is larger than Pacoval do Arari, reaching 
a height of 10 meters (1936, p. 34). Three small potrests are illus- 
trated by Torres (1940, pi. 47). 

Bibliography: Lothrop, 1934, p. 820; Moraes, 1936, p. 34; Nordenskiold, 1930. 
pi. 14-e; Ryd^n, MS.; Torres, 1940, pi. 47. 

PACOVAL DOS MELLO 

Hilbert reports the remains of a site in the campo 4 km. east of 
Caratateua and almost due south of Lago Guajard. The original 
contours of the mound were altered to provide a foundation for the 
headquarters of a fazenda now occupying the area, and the present 
maximum height is only 1.80 meters above flood level. A test pit 
near the northeast edge of the elevation showed the soil to be black 
to dark brown, with sherds to a depth of 1 meter. Below that was 
sterile dark gray to gray clay. Hilbert comments briefly on the 
pottery: 

The quality of this ceramic is striking. Undoubtedly one of the centers of 
the Marajoara Phase, with pottery like Pacoval do Arari or to a certain extent, 
the cemetery mound [Mound 1] of the Camutins. Anthropomorphic and zoo- 
morphic applique, so frequent at Pacoval, is nearly absent here. There are many 
sherds from shallow bowls, generally with undulating rims and two-part vessels 
like those from Fortaleza not far to the south [see Palmatary, 1950, pis. 66, 67, 
70]. Painting with polychrome designs on a lustrous white slip is common. 
[Hilbert, pers. corres.] 

In a small sample of 71 sherds, 65 were classified by Hilbert as 
gray cored and 6 orange cored, giving a ratio of 92 percent Inajd 
Plain to 8 percent Camutins Plain. Pacoval Incised appears to be 
frequent among the decorated types (ibid.). 



"v^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 317 

PANELLAS 

Farabee (1916 b) records a visit to "Ilha das Panellas" on the 
Fazenda Cacuero, where he says Rempkin had previously dug. At 
the beginning of the rainy season it was 3 feet high, 50 wide, 100 feet 
long, and surrounded by water. Since it seemed "all dug over" he 
decided further excavation would be unprofitable. 

In July, 1928, Mordini spent 10 days excavating a trench 5 by 3 
meters and 4.50 meters deep in a mound called "Panellas." The soil 
was uniformly black with an intermixture of ashes at a depth of 3.50 
meters. Although he has never pubHshed the results, a cross-sectional 
drawing of the vessels in situ is reproduced in Palmatary (1950, p. 
279), together with a Hsting of their descriptions and contents. 

SANHARAO 

No clue is given to the whereabouts of this mound, Hsted by the 
Barao de Maraj6 (1895, p. 87). 

SANTA BRIGIDA 

This site is west of Lago Arari, on the Igarape Cuieiras, a tributary 
of the Anajds-miri (Anajasinho). Palmatary, who made a brief visit, 
describes it as 12 feet high and some two acres in area. One side was 
tested and 

in the space of about two hours, the workmen removed two large, undecorated 
jars, lacking their tops, one plain shallow dish, a painted and engraved concentric 
dish, more or less complete, and fragments of several large vessels. No effort 
was made to determine stratification but, within the limits of the small area 
excavated, decorated and undecorated wares seemed buried close together and 
in confusion. [Palmatary, 1950, p. 278.] 

A sample of 123 sherds collected in 1941 by Carlos Estevao de 
Oliveira is deposited in the Museu Goeldi, Belem. It was classified 
by Hilbert on the basis of core color, giving 68 sherds or 55.2 percent 
gray cored (Inaja Plain) and 55 sherds or 44.8 percent orange cored 
(Camutins Plain). Decorated types include Anajas Plain Incised, 
Anajds White Incised, Arari Plain Excised, and Joanes Painted 
(Hilbert, pers. corres.). 

This may be the mound described to Lange (1914, p. 301) as located 
several days travel up the Anajasinho, 15 meters in height and full 
of pottery, 

SANTA IZABEL 

Ferreira Penna reports: 

In 1873 I visited this mound, situated in the campo northweft of Lago Ararf. 
It is diflBcult to find the site without a guide, because the artifacts are buried in 
an area that is level and flat like the surrounding campo. Although smaller 
and with fewer artifacts than Pacoval, it is nevertheless the only site that can 



318 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

rival the latter in the choice of material and in the perfection of the irciged and 
painted designs of the ceramics. 

It was here that I first found several Tinieiras, indispensible utensils of the 
aboriginal painters, all of them ornamented with elegant and delicate rehefs 
[probably excised designs]. One contained a good-sized lump of very fine, red- 
colored clay. . . . [1879 a, p. 51.] 

SANTO ANDRE 

Once more, there is only the mention of a mound near the Fazenda 
Santo Andre called 'Tacoval" (Marajd, 1895, p. 88). A fazenda by 
that name is located on the Rio Paracauari, which flows east to 
Soure. 



A moimd of this name is said to exist in the area bounded by the 
Rios Ganhoao and Cururii and Lagos Mututi and Asapao, which 
would place it north of Lago Ai-ari (Mordini, 1934 a, p. 62). 



No location is given for this mound, but there is a, fazenda of the 
same name about half way between Lago Ararl and the east coast. 
Holdridge excavated in this area in 1932 (1933 a; 1939, p. 105). 
Lage (1944, pp. 219-220) says the ceramics are more elaborate than 
at the Camutins. 

TESO DAS IGA5ABA8 

In 1922, Nimuendajii visited a site called "Teso das Igagabas," 
in the Cabo Maguari region, about 1 km. south of the Fazenda Boa 
Esperanga. It produced pottery in the Marajoara style. He gives 
the following brief description of the dimensions and contents : 

An old negro could still remember that in his youth several big vessels had 
been excavated here. The excavators before me seemed to have set about their 
work very seriously, for in the mound, which was only 10 by 20 M at the utmost 
with a thickness of about 40 cm, I found but insignificant fragmerts. . , . although 
most of the fragments were of inferior material and coarsely manufactured, there 
were several pieces here which were carefully executed, and above all, some with 
the characteristic painting in Maraj6 style, red and black on white ground. 
On the rims of the vessel were round projections, which had served as handles. 
One fragment shows traces of having been used for smoothing arrow shafts. 
[Nimuendajii, in Ryd^n, MS.] 

TESO DE SEVERING 

Lange describes a visit to a mound 6 miles east of Pacoval, which 
probably was Teso de Severino : 

Here we spend two whole days, returning with a canoe full of pottery, some of 
which is in fragments. The mound we visited is elevated hardly two feet above 
the level of the surrounding country, in the middle of a difficult piece of prairie, 
soggy and overrun with an uneven, coarse growth of wild cotton plants. Old 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 319 

EVANSJ 

Ludovico (the guide] indicated this spot as being likely to contain a great deal of 
pottery, particularly as no excavations have ever been undertaken there. I 
find there pottery of a totally different character from that near Ludovico's 
[Pacoval]. The clay from which this was made seems to be lighter in colour and 
weaker, or else the burning was not so thorough as the other. All the vessels I 
find are of a different shape too, and one large bowl, which we unfortunately 
break by accidentally putting the point of the pickaxe through the bottom, is a 
strange piece of work full of fine details. [Lange, 1914, pp. 330-331.] 

The pottery from tliis site in the American Museum of Natural History 
(collected by Lange) includes mainly AnajAs White Incised, with jars 
of Common shape 8 and rare shape 1 predominant. 

Another collection was made by Carlos Estevao de Oliveira in 1925, 
which IVEordini examined (1934 a, pp. 63-64). He found the ceramics 
superior to those from Pacoval in painted and incised decoration and 
in fineness of paste, firing and skill of worlmianship, but inferior to the 
latter in plastic decoration of an anthropomorphic and zoomorphic 
nature. Several pieces were partly covered with a kind of glaze. 
The tanga designs were also distinctive: 

The decoration is very carefully done, and it is noteworthy that it includes 
verj^ complex and graceful anthropomorphic stylizations, motifs unknown on 
these objects at Pacoval do Arary, where the decoration is exclusively geometric. 
The frieze characteristically found on the upper edge [at Pacoval] ... is abso- 
lutely unknown on tangas from Teso de Severino. [Op. cit., p. 64.] 

The most recent account is given by Hdbert (pers. corres.), who 
excavated at Teso de Severino in 1951. He describes it as 4 km. 
upstream from the mouth of the Igarap6 das Almas (Igarap6 do 
Severino) and some 400 meters in from the left bank. The site has 
been taken over as a foundation for a corral and cowboy barracks 
with much damage to its original condition. Hilbert reports that the 
bare patches around the buildings show many sherds and on the basis 
of their distribution estimated the site to extend approximately 75 
meters northeast-southwest by 50 meters wide. It now has an ele- 
vation of only 50 cm. Three test pits in various spots produced 
sherds to a depth of 20 to 50 cm. Of the 146 sherds collected, 125 
or 85.6 percent are gray cored and 21 or 14.4 percent are completely 
oxidised. Decorated sherds are abundant and run the gamut of the 
more elaborate types. The sample included 5 hollow rims and 20 
tanga fragments, of which 13 are white-slipped and 7 red-slipped. 

Bibhography: Hilbert, pers. correF.; Lange, 1914, pp. 330-331; Mordini, 1934 a, 
pp. 63-64. 

TESO DOS CHINA 

About 1 km. south-southeast of the south end of Lago Guajard is a 
group of 11 mounds of varying size, known collectively as "Teso dos 
China" (fig. 113). The majority are covered with large trees, which 
mark their location in the otherwise flat and treeless campo. There is 



320 BUREAU OF AMERICAN ETHNOLOGY [boll, m 

no nearby igarape, and the mounds are arranged over an area roughly 
750 by 350 meters. The average elevation was only 50 to 100 cm. 
above the flood level in February, but except in imusual instances the 
water level of the rainy season does not rise sufiiciently to inundate 
the mounds. Hilbert, who is the first to describe this group, has pro- 
vided some details of the condition of the individual sites (pers. 
corres.) : 

Mound 1: Approximately 100 by 60 meters, with an elevation *• of 1.80 meters 
at the north and 2.50 meters at the south. A 1- X 1-meter test 
excavation in the south summit produced brownish soil and sherds 
to a depth of 30 cm. 
Mound 1 A: Northernmost of the group, about the same dimensions as Mound 
1, but with a flat surface only 50 cm. in elevation. The northern 
half is covered with grass and bushes. A clearing on the south- 
ern half for cattle branding hf s exposed abundant surface sherds. 
Mound 2: About 50 meters in diameter, with a flat surface 80 cm. above the 
February water level. It is overgrown with large trees and few 
sherds are visible on the surface. 
Mound 3: About 90 X 30 meters, with a flat summit 50 cm. above water level. 

The surface is covered with large trees and many small sherds. 
Mound 4: About 20 X 30 meters and 50 cm. in elevation. Surface conditions 

duplicate those of Mound 3. 
Mound 4 A: Very small, covered with bushes, but vnth abundant surface 

sherds. 
Mound 4 B: Approximately 25 X 45 meters and 50 cm. in height; covered with 

grass except for a clearing where many sherds are visible. 
Mound 4 C: About 25 X 30 meters and 50 cm. in elevation. Surface covered 

with grass revealing few sherds. 
Mound 4 D: Very small. 
Mound 5: About 40 X 50 meters and 50 cm. in height; covered with grass 

except for clearing exposing abundant sherds. 
Mound 6: About 90 meters long by 40 meters in maximum width. Altitude at 
south end a little over 1 meter, less at north end. A test excava- 
tion in the south summit produced brown humus with frequent 
sherds to a depth of 30 cm; sparser sherds to 45 cm., and below 
that sterile whitish, sandy clay. 

Experience with other Marajoara Phase sites would suggest that 
Mounds 1, lA, and 6 might be cemeteries and the rest habitations. No 
burial vessels were encountered in the test excavations, however, and 
aU of the sites present a relatively large amount of decorated types in 
the sm'face collections. 

Surface collections were made from most of the mounds, and Hilbert 
kindly sent to us for examination those from Mounds 1, 2, 4, 4A, and 5. 
as well as the sherds from the strata cut in Mound 1 . The surfaces have 
suffered badly from erosion, but close examination showed that in 
addition to Inaja Plain and Camutins Plain, nearly all of the Mara- 
joara Phase decorated types were present. Although the samples were 

29 Figures represent amount of exposed elevation during the rainy season when the surrounding campo 
was flooded to a depth of 50 cm. Actual elevation would thus be 50 cm. greater. 



MEGQERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



321 




CUT I 







I I I I I I 



50 100 M 



CONTOUR LINES AT 50 CM 

INTERVALS. 
AFTER HILBERT, FIELD 

NOTES. 



1 

(9" 




Figure 113. — Plan of Teso dos China mound group of the Marajoara Phase. 



322 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 187 



small, none comprising more than 50 sherds, it was possible to classify 
them and to use the data for seriation by limiting the criterion of 
classification to core color and ignoring for the moment any sm"face 
decoration. This device makes it possible to see more accurately the 
relative proportions of the two plain wares, whose fluctuations provide 
the most reliable basis for seriation. This generalization gave the 
following figures (for sherd totals, see Appendix, table 41) : 



Table K. — Frequency of Inajd and Camutins Plain at Teso dos China 



Mound 



Camutins 
Plain 



1: level 0-15 cm.. 
1: level 15-30 cm 

2: surface 

4: surface 

4 A: surface 

5: surface 




TESO DO GENTIL 



Two kilometers above its mouth, the Rio Anajas-miri is joined by 
a tributary from the north, called Igarap6 do Gentil. Hilbert (pers. 
corres.) located two Marajoara mounds along its course about 1.5 
kilometers above the confluence. At present, they are separated by 
a dry arm, with the igarape passing along their eastern ends. The 
area is open campo with patches of trees and bushes. Mound 1, 50 
meters long by 20 meters wide, had a maxim.um elevation of 2 meters 
above the flood level. A caboclo house occupied the western end. A 
test pit in the summit showed reddish-brown loam producing sherds 
to extend to a depth of 40 cm.., beneath which was the sterile clay 
forming the foundation of the mound. Mound 2, about 25 meters 
north of the east end of Mound 1, was about 20 meters in diameter 
and 1.5 meters in elevation, 

A sherd sample sent to us for examination proved to contain typical 
Marajoara Phase wares. Half of the 20 sherds from Mound 1 repre- 
sented decorated types, including 5 Arari Plain Excised and 1 each 
of Pacoval Incised, Guajara Incised, Goiapi Scraped, Joanes Painted 
and Unclassified Decorated. Only 3 of the 18 sherds from IMound 2 
had any decoration, with 1 each of Pacoval Incised, Joanes Painted 
and Goiapi Scraped. Added to the difference in size, this suggests 
that Mound 1 was a cemetery and Mound 2 a habitation site. Dis- 
regarding the decoration and classifying the sherds by the plain ware 
they represent gives a ratio of 50 percent Inaja Plain to 50 percent 
Camutins Plain for Mound 1, and 61.1 percent Inaja Plain to 38.9 
percent Camutins Plain for Mound 2. 



"vA^Nsf ^ *^° ARCHEOLOGY AT MOUTH OF AMAZON 323 



TESO DOS GENTIOS 



Mordini gives no specific location for this mound, which he exca- 
vated in 1926, mentioning only that it is north of Lago Arari. Al- 
though he has not published his findings, some of the field notes are 
reproduced by Palmatary (1950, pp. 278-279). He made an excava- 
tion 5 by 3 meters in extent and 2 meters in depth. The soil was 
black to a depth of 80 cm., below which it became "black and slightly 
yellow." The diagram (op. cit., p. 279) shows the ceramics to extend 
to a depth of 1.50 meters, the final 50 cm. of the excavation presumably 
being sterile. Plain and decorated burial vessels were recovered, the 
latter including Joanes Painted and unidentifiable "engraved" types, 
which probably represent both incised and excised. Although the 
diagram is not to scale, the measurements in the text show that the 
jars in the lower level are considerably larger than those in the upper 
one. Some contained miniature vessels, fragments of bone and ashes. 
No tangas are mentioned. 



UNNAMED MOUNDS 



In addition to the mounds just listed, there are several less specific 
references to the existence of other sites on Maraj6. Barnard reported 
to Hartt that "Indian burial stations are quite numerous in the centre 
of the island" (Hartt, 1871, p. 260). Derby, in describing Pacoval, 
adds that "several other localities on the shores of the lake [Aran] 
have yielded a similar kind of pottery . . ." (1879, pp. 225-226). 
Lange reports a crescent-shaped mound at the outlet of the lake, 
which is probably a habitation site (1914, p. 332) and another cemetery: 

During the months when the waters fall very low there appears in the middle 
of the lake, some two miles to the northward, a flat clay-and-sand bank, some 
fifty feet only in circumference. Here is to be found a great deal of pottery 
similar to that of Pacoval. [Op. cit., pp. 311-312.] 

In Hartt (1885, p. 25) is the information that "there exist mounds of 
the same kind [as on the Camutins] on the Kio Mocoes, on the Igarap6 
Grande, on the Rio Camara and in various places on the campos." 
Mordini mentions a series of 7 small mounds between Cajuliros and 
Faz Cafe, which may be habitations (1934 a, p. 62). On the fazenda 
of Dr. Vincente Miranda, Farabee "located several of medium size 
and excavated fom- without finding a solitary thing of value. They 
had been used as house sites only, as was indicated by the presence of 
ashes and fragments of pottery" (1921, p. 144). Pinto (1930, p. 351) 
speaks of mounds near Soure. Statements like these support the 
conclusion that many more Marajoara sites exist than have found 
their way into the literature. It seems probable that these include 
a large number of habitations. 



324 BUREAU OF AMERICAN ETHNOLOGY [bull, le? 



CONCLUSIONS 



On the basis of these descriptions of Marajoara Phase mounds, a 
few general conclusions can be drawn: 

1. There is no intentional orientation toward any of the cardinal 
points; rather, orientation depends on the contour of the river or lake 
shore or is arbitrary. 

2. There is no intentional effort toward the production of a zoo- 
morphic shape. The majority of the mounds are oval or nearly 
circular, and those interpreted as turtle-shaped are some that have 
suffered badly from erosion. 

3. In addition to the large, ceramically rich cemeteries, there are 
numerous habitation mounds, which are typically smaller and contain 
predominantly plain pottery. 

4. The mounds continued to be enlarged after they were in use, as 
indicated by the existence of layers of sterile soil between those pro- 
ducing sherd refuse. 

ANALYSIS OF MATERIALS OF THE MARAJOARA PHASE 
PoTTEET Type Descriptions 

The classification of Marajoara Phase decorated pottery into a 
readily distinguishable and workable number of categories has been 
a difficult task, largely because of the frequent utilization of two or 
more complex and technically distinct modes of decoration on a 
single vessel. This is a situation almost without precedent in New 
World archeology. Even the advanced cultures of Mexico and Peru 
rarely employed one type of decoration on the exterior and a totally 
distinct type on the interior of the same vessel. Obvious difficulties 
arise. If the classification recognizes all the possible combinations, 
there results an involved collection of categories that is not only a 
strain on the memory, but also has the drawbacks that important 
associations between particular decorative techniques and vessel 
shapes are obscured and that the completeness of the vessel will 
influence its classification. If the classification is made on decora- 
tive technique without regard to combinations, it mil not apply to 
some specimens, but only to one or the other of their surfaces. Since 
the choice of a method of classification depends on the results it is 
expected to produce, the second alternative has been used here for 
the following reasons: (1) It is equally accurate and applicable for 
sherds and for complete vessels; (2) it permits statistical analysis of 
the relative frequency of each decorative technique at any given 
time or place; (3) it provides a workable number of easily distin- 
guishable categories. 

By the completion of the analysis, two more justifications had 



"vT^sf^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 325 

become evident. In more than 90 percent of the cases where two 
techniques were used on the same vessel, one of these is Joanes 
Painted, and of the four cemeteries included in the study, only at 
Pacoval is there a high proportion even of this association. Another 
point brought out by this classification is the high degree of con- 
sistency with which certain vessel shapes are associated with par- 
ticular decorative techniques, such as hollow-rimmed bowls with 
Pacoval Incised or Anaj^s White Incised, and flat-bottomed, cylindri- 
cal jars with Ai-ari Red Excised. 

Those specimens exhibiting more than one decorative technique 
were classified primarily according to the more complex technique 
(which was nearly always on the exterior) represented, with the fol- 
lowing order of precedence being used : double-slipped excised, double- 
slipped incised, excised and retouched, incised and retouched, excised, 
incised, painted, and scraped. However, other techniques associated 
on a single vessel are listed in each type description. 

Additional decisions had to be made regarding the classification 
of the varieties of excised and painted decoration. Up until the 
publication of the preliminary report (Evans and Meggers, 1950), 
decoration in which part of the original surface of the vessel was cut 
away, leaving the remainder in relief, was referred to as "champlev6". 
The decision was made to abandon this terminology and substitute 
the term "excised" because the standard definition describes cham- 
pleve as "having the ground engraved or cut out" and being "inlaid 
in the depressions in the ground" (Webster's Unabridged Dictionary). 
Only one variety of excised decoration, Arari Red Excised, White- 
retouched, has the excisions inlaid and has a consistently large propor- 
tion of the surface cut away. In Arari Red Excised and Ai'ari Plain 
Excised, there is wide variation in the amount of excision, in some 
cases being confined to one or two excised lines around the vessel 
at the upper and lower limits of the design area, which is predom- 
inantly incised. For the purpose of classification, aU examples with 
any excision were classified as excised because of the fact that the 
technique was the same regardless of the extent to which it was 
used on a single vessel. An alternative would have been to sub- 
divide the categories by the proportion of the surface excised, but 
because of the gradual and continuous variation between the two 
extremes, this would require constant and often subjective decisions 
that would inevitably dift'er with each classifier. By drawing the 
line between "any" and "none," there is no question as to how a 
given sherd should be classified. The validity of the inclusion of 
vessels with a minimum of excision in the excised type is substan- 
tiated by the fact that the vessel shapes are those typical of the 
excised rather than the incised types. 

S91S29— 57 23 



326 BUREAU OF AMERICAN ETHNOLOGY [bdll. m 

The classification of painted designs presented similar problems. 
A number of varieties of painting are represented on Marajoara 
Phase vessels, including red-on-white, black-on-white and three dif- 
ferent combinations of red and black on a white slip. Although a 
distinction between these would be desirable, this was not made 
because it did not seem to give reliable results. Many of the painted 
sherds have been subjected to considerable erosion, so that only a 
trace or two of the original paint remains. If this were red, it would 
not necessarily imply that black was not also present originally, and 
vice versa. To regard painting as a unit, therefore, for purposes 
of classification, seemed to be the only sound procedure. Even 
when this is done, Joanes Painted shows a significant decline in 
frequency during the Marajoara Phase. 

The classification of the decorated types that follows, therefore, 
has three points in its favor : 

1. The limits of the types are sharply defined and unmistakable. 

2. The types (with few exceptions) share characteristic design 
motifs and vessel shapes as well as the primary criterion of classifi- 
cation (slip, excision, incision, scraping, etc.). 

3. The types show clearly defined trends when viewed in temporal 
perspective. 

The following descriptions of Marajoara Phase plain and decorated 
pottery types are based on the analysis of the sherd and vessel col- 
lections at the American Museum of Natm-al History in New York, 
the Peabody Museum of Harvard University in Cambridge, Mass., 
the University Museum in Philadelphia, Pa., the Museum of Anthro- 
pology in Ann Arbor, Mich., the Museu Paraense Emilio Goeldi 
in Belem, Para, Brazil, and our excavated material. The types have 
been given names according to the binomial system of designation, 
and are arranged in alphabetical order. 

anajXs double-slipped incised 

Paste: Op Inajd Plain or Camutins Plain; see those type descriptions for details 

of paste and unslipped surface. 
Surface: 

White slip — primary slip: 

Color: White, occasionally fired cream or light tan. 
Treatment: Thick, usually well smoothed, occasionally with low luster. 
Smoothing tracks often remain. Fine to coarse crackle frequent on 
the interior. 
Hardness: 3. 
Red slip — secondary slip: 

Color: Cinnabar red, sometimes fired purplish brown, reddish brown or 

rusty red. 
Treatment: Very thin, smooth, sometimes with a low luster. 
Hardness: 3. 



HEOOSRS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



327 



Form: 

Rim: Interiorly or exteriorly thickened with a rounded lip, or everted with 

a flat top. 
Body wall thickness: 6-12 mm. 
Base: Rounded or flat. 
Common vessel shapes: 

1. Small, rounded bowls with rim slightly thickened on the interior, with 

a rounded lip. Depth, 5-9 cm. (Palmatary, 1950, pi. 7, b). Both 
surfaces white slipped, except the bottom of interior of some; 
exterior double-slipped and incised. Coils occasionally added 
around the maximum diameter on the exterior (fig. 114-1). 

2. Bowls with flattened bottom, sides sloping outward and then upward 

producing slightly carinated profile, increased on one by exterior 
thickening. Rim everted, flat or concave topped with a rounded 




I I L I I I I 
4 8 12 CM 

Vessel Scale 




.<uw4u^M«i.v»« ^^tvruL^i 



VTii^VtfJmilifi"*^ 




I ' I ' ' 

I 2 3 CM 



Rim Scale 



^^' 







.■•:':>5?«*»jw< 




•' . •..-..•»s,W»'*"-»«"'' 



Figure 114. — Rim profiles and vessel shapes of Anajds Double-slipped Incised of 

the Marajoara Phase. 



328 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

slightly thickened lip ; maximum rim diameter 36 cm. Both surfaces 
white slipped, exterior double slipped and incised (fig. 114-2). 
3. Open basins with heavy, exteriorly thickened rims. The thickened 
rim is 5-7 cm. wide, 1.5-2.2 cm. thick; body wall thickness 0.7- 
1.0 cm. White slip on both surfaces, double slip and incipion on 
rim exterior; body exterior is Anaj^s White Incised or Pacoval 
Incised (Palmatary 1950, pi. 39, i). Two other sherds from simi- 
larly shaped jars have smaller rims, double slip and incision cover- 
ing the entire exterior. Rim diameter ranges 48-50 cm. (fig. 114-3) . 
Rare vessel shapes: 

1. Jar with a vertical neck and direct rim with a rounded lip. Rim 

diameter 10 cm. Both surfaces white-slipped, exterior is double- 
slipped and incised (Palmatary 1950, pi. 35, d). 

2. Large jar with small, flattened bottom, globular body joining an 

insloping neck at the rounded shoulder, everted and exteriorly 
thickened rim. Vessel height is 92 cm. Neck exterior double- 
slipped and incised (Palmatary, 1950, pi. 96, a). 

3. Globular-bodied jar with constricted mouth and exteriorly thickened 

rim with a rounded lip. Interior rim diameter 10 cm. Exterior is 
double slipped and incised. 

4. Anthropomorphic jar with two faces modeled on opposite sides of 

a bulbous neck and separated by ears which serve for both. Vessel 
mouth has exteriorly thickened collarlike rim at the top of the 
heads; mouth diameter 4 cm. Double slipped to chin level, fine 
incised lines outlining the facial features; body is Pacoval Incised. 
Decoration (pi. 50): 

Technique: Slipped red over white and ornamented with incised lines 1 mm. 
or less in width on one-third of the examples and 2-3 mm. wide on most 
of the remainder, with a maximum width on one sherd of 4 mm. Width 
is uniform on a single sherd. Lines are generally straight and deep, always 
cutting through the red slip and on 30 percent penetrating through the 
white slip to the orange paste to some extent. 
Motif: Anajds Double-slipped Incised is not a homogeneous unit as far as the 
motif is concerned. The elements and composition of each design are 
unique, with the exception that simple spirals appear on about 25 percent 
of the sherds. Other motifs include frets, "keyholes," scallops, and areas 
filled with parallel lines. 
Associated techniques: Anajds White Incised or Pacoval Incised may occur 

on the body when the double slip is confined to the rim or neck. 
Temporal differences within the type: None. 

Chronological position of the type: Restricted to the early and middle parts 
of the Marajoara Phase sequence. 

anajXs plain incised 

Paste and surface: On Camutins Plain or Inajd Plain, see those type descrip- 
tions for details of temper, firing, color, etc. 
Form: 

Rim: Usually exteriorly thickened or everted, sometimes direct. 

Body wall thickness: 4-15 mm.; thickness above 10 mm. is limited to large, 

open bowls. 
Base: Rounded, flat or annular. 
Common vessel shapes: 

1. Large open bowls with rounded to slightly angular, outsloping sides 



MEGOBRS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



329 



and exteriorly thickened rim. Diameter 22-34 cm. Incision 

typically limited to flat or concave rim top (fig. 115-1). 
Small bowls with rounded bottom, sides curving to a direct rim. 

Mouth diameter 9-16 cm. Incision on exterior (fig. 115-2). 
Carinated bowls with rounded bottom joining slightly insloping walls 

at a rounded angle. Rim diameter 21-24 cm. Incision on wall 

exterior below the thickening (fig. 115-3). 
Small jars with flattened bottom, globular body, short, vertical or 

concave-sided neck and everted or thickened rim. Body diameter 

6.5-14.0 cm. at Pacoval, 17-26 cm. at Camutins. Incised lines on 





' ' t 1 I I I 
4 8 12 CM 

Vessel Scale 



Figure 115. — Rim profiles and vessel shapes of Anajds Plain Incised, Marajoara 

Phase. 



330 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

neck alone or neck and body; applique ribs sometimes also present 
(fig. 115-4; pi. 52, d-g; Palmatary, 1950, pis. 32, e, 47, d). 
5. Jars with flattened bottom, globular body, constricted mouth and 
everted rim. Height 17-34 cm. Incision covers the upper two- 
thirds of the body (fig. 115-5; pi. 51, a-b; Palmatary, 1950, pi. 13, 
a-b). 
Rare vessel shapes: 

1. Cylindrical jars with flat bottom, upslanting sides and exteriorly 

thickened rim (the same form as Ararl Red Excised, common shape 
6, fig. 118-6). Incision is on the exterior. 

2. Jar with rounded body, short insloping neck and direct rim. Rim 

diameter 20-32 cm. Incision on the exterior of the body or neck. 

3. Miniature jar with flat bottom, slightly irsloping sidf s and exteriorly 

thickened rim. Height 8.7 cm.; base diameter 7.8 cm. Incision 
covers the exterior. 

4. Miniature jars with large, "wing" adornos rising from the opposite 

sides of the shoulder (pi. 52, a-c). Incision covers the exterior. 

5. Stools (pi. 83, d,f, i). 

6. Figurines (pi. 79, d). 
Decoration (pi. 51, 52): 

Technique: Designs are executed with incisions averaging 1 mm. in width or 
wide incised lines averaging 2 mm. in width; the two widths of lines are 
almost never combined on the same vessel. Lines are typically sharply 
defined, but the quality of the incising and of the designs varies from 
exceedingly crudely done example.'^ where the lines are crooked, unequally 
spaced and of unequal depth to carefully laid out designs with the lines 
straight, parallel and equally spaced. This well executed type of design is 
particularly characteristic of small jars of common shape 3. 
Motif: Anajds Plain Incised designs are typically composed of large areas or 
bands containing straight, parallel lines. The monotony may be relieved 
by having the lines in one sector running at right angles or diagonally to 
those in the adjacent one. A similar alternation in direction may occur in the 
bands. Incision on bowls is frequently restricted to 2-3 parallel lines on 
the upper rim edge. Decoration on jars may be limited to a band of vertical 
or horizontal, parallel lines around the neck. A more complex treatment 
is the division of the surface into large squares which are divided by diagonal 
lines into 4 triangles containing parallel lines. Occasional jars have di- 
agonal cross hatch. Simple, predominantly rounded spirals are also 
relatively frequent. Scalloped line? and concentric triangles are rarer, and 
rectangles, ovals and arrows are limited to one or two examples. Also 
rare are intricate and complex designs resembling those of excised types, 
but without the background cut out. 
Associated techniques: Applique ribs; stylized anthropomorphic faces with the 

features in low relief outlined by incised lines. Joanes Painted occurs on the 

interior of some bowls of common shape 1. 
Temporal differences within the type: Examples from Pacoval are generally 

better done than those from the later sites. Common shape 3 shows a marked 

increase in size from the early to late part of the sequence. 
Chronological position of the type: Present at all sites of the Marajoara 

Phase, but increases in frequency in the later ones. 



HGOGBRS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



331 



anajXs red incised 

Paste and surface: On Camutins Plain or Inajd Plain; see those types for details 

of paste and unslipped surfaces. 
Slipped surface: 

Color: Cinnabar red, sometimes with orange tint. 

Treatment: Typically thinner than the slip applied to excised surfaces and 
more comparable to Carmelo Red in thickness. Smooth, often with slight 
luster, occasionally with faint smoothing marks. 
Hardness: 3-4. 
Form: 

Rims: Direct, everted or exteriorly thickened with rounded lip. 
Body wall thickness: Typically 6-11 mm., rarely 13 mm' 
Base: Rounded, flat or annular. 
Common vessel shapes: 

1. Small bowls with rounded bottom, sides curving outward or upward 

to direct rim, sometimes with scattered small adornos. Rarely the 
rim is exteriorly thickened. Rim diameter 7-18 cm. Incision is 
on the exterior (fig. 116-1, pi. 53, f-h). 

2. Large, open bowls with annular base, outsloping sides and exteriorly 

thickened rim. Rim diameter 15-28 cm. Incision usually limited 
to the rim top or rim exterior and both surfaces of annular base, 
but may cover the exterior (fig. 116-2). 

3. Bowls with rounded bottom, slightly carinated walls and everted or 

exteriorly thickened rim. Maximum diameter, 38 cm. Incision on 
the exterior (fig. 116-3). 




I 2 3 CM 
Rim Scale 





Figure 116.- 



-Rim profiles and vessel shapes of Anaj^s Red Incised, Marajoara 
Phase. 



332 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

4. Bowls with flat bottom, vertical or outsloping sides expanding and 

interiorly thickened on the rim. Diameter 14-20 cm. Circum- 
ference may be circular or heart shaped, lip level or undulating. 
Incision covers the exterior (fig. 116-4). 

5. Large, deep bowls with sUghtly outslanting sides, everted or exteriorly 

thickened rim, probably rounded or conoidal bottom (fig. 116-5; 
Meggers, 1947, pi. 2, fig. 3). 

6. Jars with rounded bodies indicated by several body sherds; too small 

to give further details of shape. 
Rare vessel shapes: 

1. Carinated bowl with flat, conoidal bottom, joining concave walls at a 

rounded angle, direct rim. Height 23.7 cm. Incision on the 
exterior walls (Palmatary, 1950, pi. 23, b). 

2. Miniature, oval-bodied vessel, small orifice at center of top with 

exteriorly thickened rim and flattened bottom. Length 7.5 cm., 
width 6.5 cm., height 6.5 cm. Incision covers the exterior. (Shape 
is similar to the excised vessel illustrated in Palmatary, 1950, pi. 
41, d.) 

3. Shallow bowl with rounded bottom and outsloping sides, latter inset 

at the junction so as to leave scalloped, horizontal flange. Depth 
2 cm. Incision covers the exterior except on the flange. 

4. Stools. Incision covers the disk. 

5. Figurines. 

6. Earspools (fig. 134). 
Decoration (pi. 53): 

Technique: Red-slipped surfaces are incised with deep, sharply defined, 
narrow (1 mm. or less in width) to wide (usually around 2 mm., occasionally 
3 mm. or more) lines, typically of uniform width and depth on a single 
specimen. The lines are rarely straight, and often are unevenly spaced and 
overlap at points of intersection. One small bowl has wide lines, broken to 
produce dashes of irregular length in conjunction with regular incision. 
Broad, deep lines reveal the underlying orange or gray paste. 
Motif: Anajds Red Incised designs show no emphasis on any particular 
motif. On the contrary, they tend to borrow motifs from various incised 
styles; e. g., double-line design featuring scallops, typical of Guajard 
Incised, and bands or areas containing evenly spaced parallel lines, typical 
of Anajds Plain Incised. Other figures include steps, concentric rectangles, 
concentric triangles, angular or rounded spirals, and "keyholes." The 
combinations are usually simple, often leaving large unincised areas. 
Patterns that cover the entire surface with closely spaced, parallel lines 
forming angular spirals are very rare (Palmatary, 1950, pi. 35, h). 
Associated techniques: Joanes Painted on some bowl interiors with small 
adornos on bowl rims. Anajds Red Incised may occur on a band just below the 
rim on cylindrical jars having Ararl Red Excised or Arari Red Excised, White- 
Retouched on the bodies. 
Temporal differences within the type: None. 

Chronological position of the type: Decreases in frequency and disappears 
before the end of the Marajoara Phase. 

ANAJAS WHJTE INCISED 

Paste and surface: Majority on Inajd Plain with the rest on Camutins Plain; 
see those type descriptions for details of paste and unslipped surfaces. 



KVANsT** ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 333 

Slipped surface: 

Color: White, occasionally fired cream or light orange. 

Treatment: Fine-textured, smooth, evenly applied, often polished producing 
a low luster. 

Hardness: 3-4. 
Form: 

Rims: Solid or hollow; exteriorly thickened, interiorly thickened or rarely 
direct, with a flat top or rounded top the most common. 

Body wall thickness: 5-12 mm. 

Base: Flat or rounded. 

Common vessel shapes: 

1. Shallow bowls with rounded bottom, outslanting sides and incurving 

rim thickened on the interior. Body wall thickness 5-10 mm.; rim 
thickness 1.1-2.0 cm.; rim diameter 12-35 cm. (Palmatary, 1950, 
pi. 15, a, 25, b, 31, a). Incision covers interior or exterior; geo- 
metric adornos common on the rim; h'p even or undulating (fig. 
117-1, pi. 55, b). 

2. Bowls of shape 1 but with hollow rim typically produced by looping 

the upper edge over into the interior, giving a contour indis- 
tinguishable from that of solid rims except in cross section. Body 
wall thickness 5-9 mm.; rim thickness 1.5-2.7 cm. (fig. 117-2; 
Palmatary, 1950, pi. 38, g). 

3. Bowls with sides curving upward and outward to angular junction 

with everted, flat-topped rim. Rim top 1.7-4.2 cm. wide. Rim 
adornos common, especially a trianguloid type with two eyelike 
eminences on the horizontal surface and terminating in 1-2 round, 
upturned knobs (Palmatary, 1950, pi. 106, b). Incision limited to 
the rim top (fig. 117-3). 

4. Bowls with flat or roimded bottom, upcurving sides and direct rim 

with rounded or flattened lip. Body wall thickness 5-8 mm.; rim 
diameter 9-14 cm. Rim adornos rare. Incision on exterior of wall. 
Rim top usually level, occasionally undulating (fig. 117-4). 

5. Bowls with flattened bottom, walls outsloping to mild carination 

then curving upward to an exteriorly thickened rim with rounded 
or angular lip. Upper wall height 3.5-6.0 cm.; rim diameter 18-30 
cm.; body wall thickness 6-10 mm.; rim thickness 1.6-1.7 cm. 
Incision on exterior (fig. 117-5). 

6. Flat-bottomed bowls with outslanting sides and exteriorly thickened 

rim with rounded or angular hp. Rim diameter 10-32 cm.; body 
wall thickness 5-12 mm.; rim thickness 1-3 cm. Rim top is level or 
undulating with occasional low relief adornos; incision covers the 
exterior (fig. 117-6; Palmatary, 1950, pi. 17, b). 

7. Deep, carinated jars with depressed conoidal bottom, insloping walls, 

and evei ted, exteriorly thickened rim. Height 21-32 cm. Incision 
on exterior, usually confined to walls; occasional small round body 
adornos (fig. 117-7; Palmatary, 1950, pi. 21, a-c). 
. Jars with bulbous or insloping neck, narrow mouth and everted or 
exteriorly thickened rim with rounded lip. Body contours are 
uncertain. Typical mouth diameter on interior 2.1-6.0 cm., 
exterior 3-8 cm. Neck occasionally connects to the body by 4 
flues instead of a single central opening (Palmatary, 1950, pi. 28, 
j and 1). Incision on exterior; rim exterior sometimes painted 
red (fig. 117-8). 



334 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 




Rim a Adorno Seal 



Figure 117. — Rim profiles and vessel shapes of Anajds White Incised, Marajoara 

Phase. 

Rare vessel shapes: 

1. Jars with two bulbous necks connected by a horizontal strap handle. 

One neck is covered over, the other has a narrow mouth (Palmatary, 
1950, pi. 28, m). Incision covers the exterior. 

2. Small, shallow bowls with rounded body and direct rim, top level or 

undulating. Diameter 9 cm. 

3. Shallow bowls with carinated profile, walls may be nearly vertical 

or widely everted, rim direct or thickened. Incision on the 
exterior or wall interior. 

4. Anthropomorphic jar with flat bottom, depressed-globular body, and 

a tall cyhndrical neck and exteriorly thickened rim. Anthropo- 
morphic face in low relief on one side of the neck. Incision covers 



JfJ^^M^f® ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 335 

the exterior. Height 21 cm. (Palmatary, 1950, pi. 25, a). This 
shape is typical of Ararf Red Excised and Aran Red Excised, 
White-Retouched and is represented here by only one example. 

5. Miniature jar with rounded body, insloping collar-like neck and 

2 zoomorphic adornos rising fiom opposite sides on the shoulder. 
Diameter 8 cm., depth 7.5 cm. Inci.'-ion covers the exterior. 

6. Miniature cyhndrical jar, flat bottom, exteriorly thickened rim. 

Height 4.8 cm. Incision covers the exterior. 

7. Figurines (Palmatary, 1950, pi. 100, i). 

8. Stools. Incision covers the disk and sometimes the exterior of the 

stand (pi. 83, c). 
Decoration (pis. 54, 55): 

Technique: Sharply defined, narrow, incised lines (1 mm. or less in width) 
on a white-slipped surface, often drawn when slip was well hardened; 
typically even and straight, sometimes cutting through the slip to reveal 
orange paste. Designs are well executed, with lines and elements evenly 
spaced. Fine lines are employed in four variations, the first three of 
which may occur alone or in combination: (1) Single, individual lines; 
(2) compound or double lines, composed of two fine lines running parallel, 
1-2 mm. apart; (3) paired lines connected with a series of closely spaced 
horizontally drawn lines giving a ladderlike effect; and (4) a wide line 
cutting through the slip to the underlying orange paste, a rare compo- 
nent except in the latter part of the Phase. Used in combination with 
lines of type 2, type 4 produces an effect similar to that of Pacoval In- 
cised with a less vivid color contrast (pi. 55, a). 
Motif: One form or another of the spiral is the major element in almost all 
designs. Variations include single spirals, independent interlocking spirals 
(sometimes triple), interlocking spirals with inner ends joined by a short 
line and double interlocking spirals with ends of one pair joined. Contour 
is typically circular, but square, triangular, lozenge and irregular forms 
occur. Interlocking arms may be of the same composition or any com- 
bination of the three types of lines. The designs thus produced are 
exceedingly light and graceful, both because of the delicacy of the incised 
lines and because of the careful spacing and symmetry of the motifs. 
Other design elements include lines with small pendant ovals, stepped 
lines, parallel lines and frets. 
Associated techniques: Joanes Painted is frequent on the interiors of bowls 
with either solid or hollow rims. Pacoval Incised may occur on necks of com- 
mon vessel shape 7 jars or on the rims of bowls of common vessel shapes 1 and 
2 (Palmatary, 1950, pi. 40, b, 37, b). All types of rim adornos are frequent 
on bowls. One stool with Anajds White Incised on the disk has Arari Red 
Excised on the exterior of the base. 
Temporal differences within the type: Anajds White Incised appears to 
exist as an important type with distinctive design motifs only in the early 
part of the Marajoara Phase. In the sites seriating after Pacoval, many of 
the designs show a close similarity to those of Pacoval Incised, differing only 
in that a wide, unretouched line that cuts through the white slip to reveal the 
underlying orange paste (technique 4) is substituted for the red-retouched line 
of Pacoval Incised. The relative frequency of this variety in the sherd sample 
is shown on table L. A similar trend is observable in Arari White Excised 
(p. 348). 



336 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

Table L. — Temporal differences in Anajds White Incised 



Technique of Incision 


Decoration 




Paco val 


Fortaleza 


Camutins 


Guajarfi 


Regular incision: techniques 1-3 - - . 


287 
1 


80 
9 


2 

1 





Technique 4, resembling Paco val Incised 


1 







Chronological position of the type: Characteristic of early sites, with a 
sharp decline in frequency in the latter part of the Marajoara Phase sequence. 

ARARf double-slipped EXCISED 

Paste: Predominantly on Inajd Plain, a few on Camutins Plain; see those types 

for details of temper, color, firing, etc. 
Surface: 

Unslipped surface — interior of jars: 

Color: Light orange to light tan to light brown. 

Treatment: Superficially smoothed, leaving uneven, rough or gritty 

surfaces because of the protrusion of temper grains. 
Hardness: 3. 
White slip — primary slip: 

Color: White, with firing variation from white to cream, salmon or 

light brown. 
Treatment: Variation from smooth and even to irregular, with prominent 
smoothing marks and medium crackle; exterior surface usually 
better smoothed than interior. 
Hardness: 2-4. 
Red slip — secondary or upper slip: 

Color: Deep, cinnabar red to purplish red to various shades of brown, 
occasionally almost black. Variation on single vessels indicates this 
to be the result of unequal conditions during firing. 
Treatment: Typically a thin film, much thinner than the underlying 
white slip; usually smooth, even, and occasionally slightly polished. 
Hardness: 2-4 on slips fired a variety of red; 5 where fired black. 
Form: 

Rim,: Direct, everted or exteriorly thickened, rounded or flattened lip. 
Body wall thickness: 4-8 mm. on bowls, 5-11 mm. on jars. 
Bases: Flat, rounded or annular on bowls, flat on jars. 
Common vessel shapes: 

1. Small, deep bowls with rounded bottom and walls curving upward 

to direct rim. Diameter 12.5 cm.; depth 5.5 cm. Both surfaces 
white slipped, exterior double slipped and excised (fig. 118-3). 

2. Small, shallow bowls with rounded bottom, outcurving sides and 

exteriorly thickened rim. Diameter 12-17 cm.; depth 3-5 cm. 
(Palmatary, 1950, pi. 98, c ?). Both surfaces are white slipped; 
exterior double slipped and excised (fig. 118-1). 

3. Flat-bottomed bowls with vertical or outsloping walls, exteriorly 

thickened or horizontal rim. Thickening on the interior at the 
junction of the wall and base transforms the sharp angle to a more 
gradual curve. Both surfaces white slipped, exterior double slipped 
and excised (fig. 118-4; Palmatary, 1950, pi. 66). 



MEOOERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



337 





Figure 118. — Rim profiles and vessel shapes of Ararf Double-slipped Excised 
and Aran Red Excised, Marajoara Phase. 



338 BUREAU OF AMERICAN ETHNOLOGY [BntL. 167 

4. Shallow, annular-based bowls with outslanting sides and exteriorly 

thickened rim. Diameter 30 cm. Both surfaces white slipped, 
exterior double slipped and excised (fig. 118-5; Palmatary, 1950, 
pi. 47, h). 

5. Cylindrical or semicylindrical jars with flat bottom, vertical or 

slightly outslanting walls, everted or exteriorly thickened rim. 
Dimensions probably comparable to Arari Red Excised, common 
shape 6. Interior white slipped or unslipped; exterior double 
slipped and excised except on the bottom. (Since this pottery 
type is represented primarily by sherds, it is possible some may be 
from jars with cylindrical necks and globular bodies like Arari 
Red Excised, common vessel shape 8). 
Rare vessel forms: 

1. Rectanguloid bowls with flat bottoms, vertical sides and direct rim. 

Length 13-19 cm.; height 4.5-10.0 cm. Both surfaces are white 
slipped; exterior double slipped, excised on the bottom and sides, 
stylized anthropomorphic faces in low relief on ends, mouth toward 
rim (Palmatary, 1950, pi. 52, b). 

2. Small jars with large "wing" adornos rising from shoulder at two 

opposite sides. Height 6.5 cm. Excision covers the exterior of the 
body (Palmatary, 1950, pi. 42, c). 

3. "Shoe-shaped" vessel with a flat bottom, rounded, ovoid body and 

cylindrical neck rising from the top of one half. Side of neck 
toward end bears white-slipped anthropomorphic face in low relief; 
remainder of surface double slipped and excised. Height 28 cm. 
(Palmatary 1950, pi. 48, c). 

Decoration (pi. 56) : 

Technique: Slipped red over white and ornamented with a combination of 

incised lines and excised lines and areas. Excision typically covers 40-60 

percent of the design surface, although on occasional sherds it is limited to 

excised lines 5-6 mm. in width dividing large rectangular or triangular 

areas containing incised designs. Excision is carefully and evenly done, 

cutting away the red slip and revealing the underlying white slip; in a few 

cases the white slip is also removed exposing the orange paste. The bottom 

of the cuts is scored horizontally to the main axis in 99 percent of the 

examples, leaving fine, parallel striations. Occasionally excision is so 

shallow that traces of the red slip remain. Technique of both excision and 

incision is uniformly excellent, with lines evenly spaced, straight, and 

sharply defined, and in general represents the acme of the excised technique. 

Motifs: A common component of these designs is a square containing a 

stylized face composed in its simplest form of two parallel lines running 

down the center and a small square in the middle of each half. Other 

typical motifs are square or rounded spirals, ovals, diamonds containing 

an incised cross, and lines ending in three prongs. 

Associated techniques: Joanes Painted may be found on the interior of open 

bowls and is always of high quality; rectangular bowls have low relief modeling 

and rounded bowls may have small rim adornos. 

Temporal differences within the type: None. 

Chronological position of the type: An early type which declines in frequency 
and disappears before the end of the Marajoara Phase sequence. 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



339 



ARARf PLAIN EXCISED 

Paste and surface: On Inajd Plain or Camutins Plain, in proportions reflecting 
those typical of the plain wares at any particular time; see those types for de- 
tails of paste, temper, firing, color, and surface treatment. The decorated 
surfaces are considerably better smoothed than is typical of either Inajd or 
Camutins Plain. 

Form: 

Rims: Great variety of shapes, running the gamut from a direct rim with 
square or rounded lip to various types and degrees of exterior thickening, 
with a flanged Jip being one of the more common. 
Body wall thickness: Range 6-26 mm.; thickness above 10 mm. is usually con- 
fined to the carination on bowls. 
Base: Rounded, flat or annular. 
Common vessel shapes: 

1. Bowls with outcurving or sHghtly angular sides and thickened rim. 
Rim diameter 12-42 cm. (Palmatary, 1950, pis. 56, b, and 68, a). 
Excision on the exterior (fig. 119-1). 





Rim Scole 



Vessel Scale 





Figure 119. — Rim profiles and vessel shapes of Arari Plain Excised, Marajoara 

Phase. 



340 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

2. Carinated bowls with rounded bottom joining short walls at a sharp 

angle, produced by marked exterior thickening. The wail is 2.3- 
3.5 cm. high, straight or slightly convex on the interior, concave to 
straight on the exterior. The nm top is flat or rounded, 5-8 mm. 
in width. Rim diameter may be slightly less, equal to or slightly 
greater than the diameter at the carination. Body wall thickness 
6-8 mm.; maximum thickness at carination 1.1-2.6 cm. Rim 
diameter 24-62 cm. (Palmatary, 1950, pi. 68, b-c). Excision 
covers exterior (fig. 119-2, pi. 57, a). 

3. Carinated bowls with rounded bottom joining concave walls at a 

pronounced angle. Rim direct with square or rounded lip, or 
everted. Wall height 5.0-8.5 cm.; wall thickness 8-10 mm.; thick- 
ness at carination 1.4-1.7 cm.; rim diameter 24-34 cm. (Meggers, 
1947, pi. 2, fig. 4). Excision covers the exterior (fig. 119-3). 

4. Small, deep bowls with rounded bottom, sides curving inward to a 

direct rim. Maximum height around 17 cm. (Palmatary, 1950, 
pis. 5, c, 46). Excision on the exterior (fig. 119-4, pi. 57, b). 

5. Annular-based, open bowls with outslanting sides and exteriorly 

thickened rim. Rim diameter 25-33 cm. Excision on the ex- 
terior (fig. 119-5). 
Rare vessel shapes: 

1. "Platter bowls" (See Joanes Painted common shape 8). 

2. Flat-bottomed bowls with slightly outsloping sides and direct rim 

(Palmatary, 1950, pi. 13, d). 

3. Globular-bodied jar with short vertical collar (Palmatary, 1950, pi. 

41, c). 

4. Stools (pi. 83, b,g). 

5. Small, short-necked jars (Palmatary, 1950, pi. 20, d). 

6. Figurines (Palmatary, 1950, pi. 4, a-b). 

7. Miniature vessels (Palmatary, 1950, pi. 4, f; 5, c-d). 

Decoration (pis. 57, 58): 

Technique: Combination of narrow incised lines with excised areas or lines 
in varying proportions from a few excised lines (Palmatary, 1950, pi. 13, 
d) to 75 percent of the surface removed by excision. In contrast with 
Arari Red Excised, the excisions are typically gouged out, rather than cut 
back evenly and scored. Margins of excisions range from straight and 
even to jagged and crooked. Floor of the excision may be deep at the 
center and sloping upward to the edges, deep and irregular, or relatively 
level, shallow and scored transversely. Incised lines are fine to 1 mm. 
wide, straight and even on the better-done specimens, wavering and with 
overlapping junctions on cruder ones. In rare cases cross hatching may 
be substituted for excision to produce the contrasting field. Another 
minor variation is the application of white to the excisions. 
Motif: The motifs of Arari Plain Excised are the same as those on the other 
excised types, except that they tend to emphasize the less complex com- 
binations. Most common elements include spirals, interlocking spirals, 
crosses in diamonds, stepped figures, T's, undulating lines, and concentric 
curvilinear or rectilinear figures. A common combination is a narrow 
band containing a line undulating between the two margins and having 
the semicircular areas between the loops filled with T or stepped figures. 

Associated techniques: Occasional bowls with Arari Plain Excised decoration 
on the exterior have Joanes Painted designs on the interior. 



MEGGERS AND 

EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



341 



Temporal differences within the type: Although the gouging out of the 
excised area is always the typical technique of decoration, a comparison of 
workmanship at Pacoval (American Museum of Natural History Collection) 
with that at Camutins { J-15, Mound 1) and Guajard (J-14, Mound 1) shows that 
gouging out becomes increasingly predominant with the passage of time. The 
relative frequency of the techniques is shown on table M. 

Table M. — Temporal differences in Arari Plain Excised decoration 



Technique of excision 



Pacoval 



Camutins and 
Quajar4 



Gouged out 

Evenly cut back and scored 
Eetouched with white 



Number 

30 

11 

1 



Percent 
71.5 
26.2 
2.3 



Number 

46 

3 

1 



Percent 
92 



Chronological position of the type: Present throughout the Marajoara 
Phase with increasing frequency. 

ARARf RED EXCISED 

Paste: On either Inajd Plain or Camutins Plain; see those type descriptions for 

details of temper, color, firing, etc. 
Surface: 

Unslipped surface: 

Color: Light orange, orange-brown, light brown to light tan. 
Treatment: Smoothed; brushmarks, coarse crackle and slight irregulari- 
ties often remain; occasionally well smoothed with a low luster. 
Hardness: 3-4. 
Red slip: 

Color: Typically cinnabar red to dark red, occasionally fired brownish- 
rust or blackish. 
Treatmeni: Thin and evenly applied, fine texture, smooth and polished, 

often with low luster. Smoothing tracks sometimes visible. 
Hardness: Typically 3-4. Rare examples attain a hardness of 5. 
Form: 

Rim: Direct or everted with rounded, pointed, or flattened lip; exteriorly 
thickened with a coil added slightly below the rim top giving a flanged 
effect. 
Bodij wall thickness: 5-10 mm. 
Bases: Rounded, flat or annular. 
Common vessel shapes: 

1. Shallow bowls with rounded bottom and outcurving sides, exteriorly 

thickened, rounded or flat-topped rim. Circumference is usually 
circular but occasionally D-shaped. Rim diameter 10-28 cm. A 
decorative coil with scalloped border is sometimes added around 
the circumference just below the rim edge (e. g., Palmatary, 1950, 
pi. 98, e). Small asymmetrically placed rim adornos are also typ- 
ical. Excision covers the exterior (fig. 118-1). 

2. Carinated bowls with rounded bottom joining concave, vertical or 

slightly outsloping sides at an obtuse angle, less pronounced than 
that on similar bowls of Arari Plain Excised because of the absence 
of exterior thickening at the carination. Upper wall slightly 
everted and terminating in a direct, rounded or pointed lip. Wall 
391329—57 24 



342 BUREAU OF AMERICAN ETHNOLOGY [bull, ist 

height 7-12 cm.; rim diameter 24-27 cm. (Palmatary, 1950, pi. 
56, a). Excision on the exterior wall, occasionally also covering 
the bottom of the exterior (fig. 118-2). 

3. Small, deep bowls with rounded bottom and outcurving sides, ex- 

teriorly thickened, rounded or flat-topped rim. Circumference 
is usually circular but occasionally D-shaped. Rim diameter 
10-28 cm. A decorative coil with scalloped border is sometimes 
added around the circumference just below the rim edge. Small 
asymmetrically placed rim adornos are also typical. Excision 
covers the exterior (fig. 118-3). 

4. Flat-bottomed bowls with vertical or slightly outslanting sides, ex- 

teriorly thickened rim with rounded lip. Occasionally the rim is 
direct. Rim diameter 23-30 cm.; wall thickness 8 mm. Thick- 
ening on the interior at the junction of the wall and base transforms 
the otherwise sharp angle to a curve. Excision on exterior of 
sides (fig. 118-4; Palmatary, 1950, pi. 31, e). 

5. Annular-based, open bowls with outslanting sides and exteriorly 

thickened rim. Rim diameter 25-33 cm. (Palmatary, 1950, pi. 
67, b). Excision covers the exterior (fig. 118-5, pi. 60, a). 

6. Cylindrical or semicylindrical jars with flat bottom and vertical or 

slightly outslanting walls, everted or exteriorly thickened rim. 
Height 20-35 cm. (Palmatary, 1950, pi. 30, b). Excision covers 
the exterior of the walls, frequently beginning 2-4 cm. below the 
upper rim edge leaving a band with incised and low-relief orna- 
ment (fig. 118-6, pi. 60, b). 

7 Jars with flattened bottoms, rounded body, insloping neck and 
everted, exteriorly thickened rim. Height 22-30 cm. (Palmatary, 
1950, pi. 59, b). Excision is limited to the exterior of the neck, 
the body being white-slipped and painted (fig. 118-7, pi. 62, a). 

8. Jars with flat bottoms, globular bodies, short vertical necks and 
widely everted rims with exterior thickening. Height, about 22-50 
cm. (Palmatary, 1950, pi. 60, 61, 63, b). Stylized anthropomorphic 
faces often occupy two opposite sides of the neck; body exterior 
covered with complex excision and often adorned with applique 
saurian motif (fig. 118-8, pis. 61, a-b, 62, b). 
Rare vessel shapes: 

1. Anthropomorphic jars with flattened, conical base, sides curving 

upward and inward and then reexpanding to simulate a head, at 
the top of which is the everted rim. The expanded area bears a 
conventionalized anthropomorphic face on one side. No anatomi- 
cal details are present on the body, which is covered with a complex 
excised design. This form is typical of Pacoval Incised and is 
represented in this type by only one vessel whose height is 56 cm. 
(Palmatary, 1950, pi. 51, d). 

2. Jars with flattened bottom, globular or depressed-globular body, 

cylindrical neck and exteriorly thickened rim. Excision covers the 
exterior. These are much more common in Arari Red Excised, 
White-retouched (Palmatary, 1950, pi. 65, e). 

3. Jars with flat conical base, sides sloping outward to maximum diameter 

about one-third of the distance from the bottom, then inward to 
just below the everted rim. Excision on the exterior. This form 
is typical of Pacoval Incised and occurs here exceedingly rarely 
(Palmatary, 1950, pi. 26, b). 



EV^Nsf^ '^ND ARCHEOLOGY AT MOUTH OF AMAZON 343 

4. Stools. Excision covers the disk top and occasionally the exterior of 

the base (Palmatary, 1950, pis. 25, c, 83, e, h). 

5. Miniature ovoid or teardrop-shaped jars. Height 6.5-8.0 cm.; 

diameter 5.5-6.0 cm. Excision covers the exterior (Palmatary, 
19.50, pi. 42, b). 

6. Miniature rounded bowls with incurving, direct rim. Depth 5 cm. 

Excision covers the exterior. 

7. Miniature jars with large "wing" adornos rising from opposite sides 

of the shoulder. Excision covers the exterior of the body 
(Palmatary, 1950, pi. 43, a). 

8. Miniature turtle-effigy vessels, produced by adding stylized head to 

one side of the body or rim of the small bowl which serves as a 
carapace. Diameter 8-10 cm. 

9. Shallow, oval vessels or spoons with perforated stem issuing from 

one end. Length, 6-11 cm. (Palmatary, 1950, pi. 42, e). Excision 
covers the exterior. 

10. "Platter-bowls" (Palmatary, 1950, pi. 55, b), 

11. Figurines. Excision covers the body (Palmatary 1950, pi. 47, a). 
Decoratton (pis. 59-62): 

Technique: Broad or narrow incised lines are combined with excised lines and 
areas in varying proportions of incision to excision on a red-slipped surface. 
Some vessels have a predominantly incised design with only a trace of 
excision; from this there is a continuous range to the opposite extreme 
where 80 percent of the original surface has been removed. The excision 
cuts through the red slip to expose the underlying orange paste. In 
contrast with Arari Plain Excised, the excised areas are typically evenly 
cut back and scored by brushing them from side to side. Depth varies 
from only enough to remove the red slip to about 1 mm. Rare sherds 
have cross hatch in place of excision. The associated incised lines tend 
to be narrow when the amount of excision is limited or the vessel is small, 
and wide (1-2 mm.) when the degree of excision is extremely high. Both 
the lines and the boundaries of the excisions are straight, sharply defined 
and carefully executed. 
Motif: The most frequently employed motif is a stepped figure, which may 
be used alone as on the walls of carinated bowls, or in combination with 
undulating lines and other figures. The exteriors of small, shallow bowls 
are usually divided into halves or quarters, filled with stepped figures or 
spirals. On larger surfaces these elements may be used in conjunction 
with undulating lines, ovals or concentric ovals containing a double crossed 
line, T's, interlocking or squared or rounded spirals, parallel straight lines, 
and other less-readily described elements. Although the exterior of a 
large jar may present the effect of symmetry, close examination shows 
that there is often considerable variation, but the pattern is so skillfully 
laid out that this is obscured by the balance of the total design. 
Associated techniques: Shallow, open bowls, platter bowls, and annular-based 
bowls may have Joanes Painted decoration on the interior. Jars of common 
vessel shape 8, with excision limited to the neck, have Joanes Painted on the 
exterior of the body. Modeling is common, in the form of geometric, anthro- 
pomorphic and zoomorphic rim adornos, and less frequently body adornos. 
An exceedingly rare combination is with Pacoval Incised, which appears on 
and just below the rim of cylindrical jars, 
Tempokal differences within the type: Ararf Red Excised shows a decline in 
the technical skill with which the design is executed, a careless gouging out of 



344 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 



the excisions replacing the eariier predominantly careful workmanship. This 
is readily perceived in a comparison of specimens from Pacoval (American 
Museum of Natural History collection) with later examples from Camutins 
(J-15, Mound 1) and Guajard (J-14, Mound 1): 

Table N. — Temporal differences in Ararl Red Excised decoration 



Technique 



Pacoval 



Camutins and 
Guajara 



Evenly cut back and scored 

Gouged out 

Cross hatched 



Number 
21 
4 
2 



Percent 

77.7 

14.8 

7.5 



Number 
7 
11 




Percent 
38.8 
61.2 



Chronological position of the type: Maximum popularity in early sites, with 
a subsequent decline in frequency throughout the Marajoara Phase. 

ARARf RED EXCISED, WHITE-RETOUCHED 

Paste: Predominantly on Inajd Plain with the remainder on Camutins Plain; 

see those type descriptions for details of temper, firing, color, etc. 
Surface: 

Unslipped surface: 

Color: Light orange to tan to rusty brown to gray brown, with the first 

two the most frequent. 
Treatment: Smoothed, frequently leaving irregularities and smoothing 

marks; temper grains occasionally protrude. 
Hardness: 3-4. 
Red slip: 

Color: Cinnabar red to dark red. 

Treatment: Thin, fine textured, smooth and sometimes polished; smooth- 
ing marks rare. 
Hardness: 3-4. 
Fobm: 

Rim: Exteriorly thickened in a variety of ways; rarely direct. 

Body wall thickness: 6-14 mm. 

Base: Flat or rounded on bowls, flat on jars. 

Common vessel shapes: 

1. Shallow bowls, flattened or rounded bottom, outsloping sides and ex- 

teriorly thickened rim with rounded lip (Palmatary, 1950, pi. 60, 
b, 98, b and f). Diameter 10-36 cm. (fig. 120-1). 

2. Flat-bottomed bowls with almost vertical or outsloping sides and 

exteriorly thickened rim with a rounded lip. Sometimes the rim 
is exaggerated to a broad, horizontal rim. The excision covers the 
exterior of the sides and occasionally the bottom. Diameter 7-30 
cm. (Palmatary, 1950, pi. 48, b). Rim may have adornos (fig. 
120-2). 

3. Rounded bowls with incurving, direct rim with a rounded lip. Mouth 

diameter, 10-20 cm. Excision covers the exterior (fig. 120-3). 

4. Round-bodied bowls or jars with short collar, direct rim and rounded 

or flattened lip. Mouth diameter 20-30 cm. (fig. 120-4; Palma- 
tary, 1950, pi. 64). 

5. Deep bowls or jars with flattened, conoidal bottom, slightly outslant- 

ing sides and exteriorly thickened rim. Height 15-30 cm. (Pal- 



MKGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 

EVANS J 



345 







FiGTTRE 120. — Rim profiles and vessel shapes of Ararf Red Excised, White- 
retouched, Marajoara Phase. 

matary, 1950, pi. 62, b). Excision covers the exterior walls (fig. 
120-5). 
6. Cylindrical or semicylindrical jars of the same type as common vessel 
shape 7 of Ararf Red Excised. Height 13-39 cm. (Palmatary, 



346 BUREAU OF AMERICAN ETHNOLOGY [buli,. 167 

1950, pis. 44, a-b, 57, b, 58). Two adornos are often attached at 
the top to opposite sides of the rim and below to the body, making 
a loop handle, with the outer surface often modeled as an anthro- 
pomorphic face (Meggers, 1947, pi. 1, fig. 2). A variant is an 
exceedingly tall and slender cylindrical jar with a flat bottom and 
widely everted, exteriorly thickened rim with a height of 37-59 cm. 
Excision covers the exterior of the body; low-relief, stylized faces 
are on opposite sides of an expanded area below the rim; relief 
saurian motif may appear on the body (fig. 120-6; Palmatary, 
1950, pis. 49, 57, a). 
7. Jars with flat bottom, globular or depressed-globular body, tall 
cylindrical or insloping neck and exteriorly thickened rim. Height 
27-42 cm. Excision covers the exterior, which typically also bears 
saurian motif (fig. 120-7; Palmatary, 1950, pis. 19, c; 51, a; 53, a; 
54, a). 
Rare vessel shapes: 

1. Carinated jar with rounded or flattened bottom, slightly inslanting 

walls and everted, exteriorly thickened rim. Height 20-30 cm. 
(Palmatary, 1960, pi. 63, a). 

2. Anthropomorphic jar with flat conoidal bottom, convex sides, hori- 

zontal shoulder from which issues a short, everted neck with exte- 
riorly thickened rim. Anthropomorphic face on neck, features in 
low relief; excision covers the body (Palmatary, 1950, pi. 92, a). 

3. Shoe-shaped vessel, flat bottom, ovoid body with mouth and exteriorly 

thickened rim occupying one end of the top. The "toe" covered 
with relief modeling; "heel" excised. Height 6 cm. (Palmatary, 
1950, pi. 48, d). 

4. Jars with flat bottom, tall, conoidal body, rounded shoulder (neck 

and rim missing) with a body height of 59 cm. Excision covers the 
exterior (Palmatary, 1950, pi. 23, c). 
Decoration (pi. 63): 

Technique: This type is characterized by the combination of incised lines 
and excised lines and areas in approximately equal proportions on a red- 
slipped surface, with the excisions evenly cut back to a depth of about 1 
mm., scored transversely and covered with a thick, white paint. The 
bond with the excised surface is usually poor and the white tends to chip 
off readily. A typical example has the incised lines and excisions straight, 
even, sharply defined and regularly spaced and the white retouch is care- 
fully applied. On some the white is smeared beyond the excision and onto 
the adjacent red-slipped surface. Some are so shallowly excised as to 
leave small patches of red in the excised areas. Incised lines are occasionally 
crooked, with overshot corners. 
Motif: Squared or rounded spirals and ovals are the most frequent, each 
diversified in a multiplicity of ways; T's, undulating lines, stepped figures, 
pronged lines, and other geometrical figures are also employed. The two 
opposite sides of jar bodies often feature a low-relief saurain figure with 
sprawling legs, the excision filling in around and between them. 
Associated techniques: Joanes Painted on bowl interiors; Pacoval Incised on 
jar rims (Palmatary, 1950, pi. 39, k); relief modeling on bowl and jar rims and 
jar bodies; Anajds White Incised on horizontal rim top of bowls. 
Temporal differences within the type: None. 

Chronological position of the type: Most frequent at early sites, but present 
throughout the Marajoara Phase sequence. 



EVANS] ARCHEOLOGY AT MOUTH OF AMAZON 347 

ARARf WHITE EXCISED 

This pottery type appears to be an experimental variety at the sites represented 
in the analysis. All of the vessel shapes and most of the design elements are more 
characteristic of Ararl Plain Excised or Arari Red Excised, except for those deco- 
rated in the red-retouch style, in which case the affinity is with shapes and motifs 
of Pacoval Incised. However, it was set up as a type rather than left unclassified 
because it represents a unique combination of white slipping with excision and 
because it may prove to be more popular at some other site in the future. 
Paste and unslipped surface: Usually on Inajd Plain, occasionally on Camutins 

Plain; see those types for details of paste and unslipped surfaces. 
Slipped surface: 
White slip: 

Color: White, sometimes cream to tan from firing diflferences. 
Treatment: Uneroded examples are fine-textured, evenly applied, smooth 
and have a low luster. Occasional crackle lines. Most commonly 
applied to the exterior surface. 
Hardness: 2-3. 
Form: 

Rim: Thickened on exterior or interior, rounded or flattened lip. 
Body wall thickness: 5-11 mm. 
Bases: Rounded on bowls, flat on jars. 

Vessel shapes {except for shape 1, these are represented by a single sherd for each 
shape) : 

1. Shallow, open bowls with rounded bottom, outflaring sides turning 

upward at the rim, which is thickened on the interior. Excision is 
on exterior, beginning at the upper rim edge; low round bosses are 
a typical component (Meggers, 1947, pi. 2, fig. 1), 

2. Bowl with outsloping sides and exteriorly thickened rim. Excision 

on the exterior beginning'^^below the rim thickening. Diameter 
20-30 cm. (Palmatary, 1950, pi. 16, e). 

3. Carinated bowl with round bottom and short walls, joined at a sharp 

angle which is produced by a marked exterior thickening. Wall, 
3.2 cm. high, straight on the interior and slightly concave on the 
exterior. Wall thickness, 8 mm. at the flat rim top and body wall 
18 mm. thick at the carination. (Cf. common vessel shape 2 of 
Ararf Plain Excised.) 

4. Bowl with flat bottom and vertical sides. Excision covers the ex- 

terior (Palmatary, 1950, pi. 37, h). 

5. Cylindrical jar with flat bottom, anthropomorphic adorno below the 

rim. 
Decoration: 

Technique: White-slipped surface ornamented with a combination of incised 
and excised lines or areas, used intermingled or to cover alternating panels. 
Excision is typically shallow, but generally removes the white slip and re- 
veals the orange paste. It may be or may not be scored. In rare cases 
the scoring is done without prior excision. Incised lines are fine to 2 mm. 
in width. 
Motif: Spirals, ovals, parallel alternating excised and incised lines are most 
common. Motifs and combinations are less complex than Arari Red 
Excised and more comparable to Arari Plain Excised. Many of the 
designs are those of Pacoval Incised, in which excision has replaced red- 
retouching to produce a contrast with the white slip. 



348 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL, le? 



Associated techniques: Occasional adornos. 

Temporal differences within the type: Pacoval is the only site where this 
type has any motif which can be called characteristically its own, although even 
here more than 40 percent of the designs are in the Pacoval Incised style, in 
which small triangles and squares at the junctions of the lines have been excised 
instead of colored red, with the cross inside a diamond the most common 
motif. At Fortaleza, Arari White Excised declines in frequency and the 
majority of the examples are in the Pacoval Incised style: 

Table O. — Temporal differences in Arari White Excised decoration 



Design 



Pacoval 



Fortaleza 



Camutins 

and 
Quajara 



Regular excised style- 
Pace val incised style- 



Chronological position of the type: Ararf White Excised is most frequent 
in the early part of the sequence and is absent at the end of the Marajoara 
Phase. 

camutins plain 
Paste: 

Method of manufacture: Coiling; coils 2-3 cm. wide visible on some large 

jars with incompletely smoothed surfaces. 
Temper: Ground sherd, with many particles quite large, ranging 1-5 mm. 
Texture: Very porous, with temper poorly mixed; temper particles easily 
rubbed out of matrix on a fresh break. Very irregular and angular cleav- 
age due to coarse temper; easy to break and very crumbly. Dull, heavy 
thud when dropped together. 
Color: Uniformly orange or reddish-orange. 

Firing: Completely oxidized; weakness of ware in part the result of "burnt- 
out" condition. 
Surface: 

Color: Typically a bright, tile orange; both surfaces may range from light 
orange to brilliant orange to reddish orange. Interiors of a few sherds 
have a light-grayish hue. 
Treatment: Exterior typically left rough and coarse textured, with wide 
finger-smoothing marks parallel to the rim. Numerous pits and protrud- 
ing temper grains contribute to the general unevenness and grittiness of 
the surface. Interior of bowls was often slipped with a thick layer of the 
same clay as the paste after the surface had been scraped or smoothed. 
Interiors of these sherds are sometimes slick and all are smoothed, though 
a few remain somewhat uneven. Crackle common on the interior and 
sometimes occurs on the exterior. 
Hardness: 2-3. 
Form: 

Rims: Typically everted and exteriorly thickened, with one common form of 
bowl having a direct or shghtly interiorly thickened rim with a rounded lip. 
Some rims have geometric, anthropomorphic or zoomorphic adornos or are 
ornamented by nubbins, scalloped lips, appliques with incisions in the form 
of crosses and nicks (pis. 64, 65). 



vvA^^f^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 349 



EVANS] 



Body wall thickness: Large vessels range 10-25 mm. with the majority 15 mm. ; 
smaller vessels range 4-10 mm., majority 6-7 mm. 

Bases: Majority flattened in one of two ways: (1) Flat base, sometimes 
thickened slightly, with the side walls attached at a prominent 45-degree 
angle (this type is the most common variety), and (2) thickened (1-3 cm.) 
base rising vertically 1-2 cm. on the exterior before joining the body wall, 
producing a slight pedestal. Diameters of both varieties range 10-20 cm., 
with the majority 12-14 cm. A few bases are rounded, and those of stools, 
pot stands and "platter-bowls" are annular. 

Common vessel shapes: 

1. Large, deep bowls with flat bottom, outcurving and upcurving sides, 

ending in a direct, vertical or slightly incurving rim. Maximum 
diameter 24-26 cm.; mouth diameter 24-32 cm. with the majority 
30 cm.; depth 12-16 cm. Majority of sherds from J-15 habitation 
mounds are of this type (fig. 121-1). 

2. Deep, basinlike carinated bowls with small flat bottom, sides outslop- 

ing to carination, then more vertical to everted, exteriorly thick- 
ened rim. Junction varies from pronounced carination to barely 
perceptible change in direction. Rim diameter 30-70 cm.; total 
depth 15-30 cm.; upper wall height 8-20 cm.; base diameter 8-17 
cm. (fig. 121-2; pi. 67, c; Palmatary, 1950, pi. 9, b-d). 

3. Bowls with rounded body, outcurving sides, and exteriorly thickened 

rim; bottom is typically rounded and occasionally flat. Rim di- 
ameter 16-40 cm. (fig. 121-3). 

4. Flat-bottomed bowls with outslanting sides, everted or exteriorly 

thickened rim. The rim is frequently ornamented with three 
large, heavy (usually solid), equally spaced, anthropomorphic, 
zoomorphic or geometric adornos (Palmatary, 1950, pi. 7, a and c; 
pi. 102, a-b, d-e; and Meggers, 1947, pi. 3, fig. 1-3). Depth of 
bowl 4-9 cm. ; exterior rim diameter usuallj' 16-25 cm., rarely 30-44 
cm. Height of adornos 5-11 cm (fig. 121-4; pis. 68, 69). 

5. Carinated jars with flattened, conoidal base, vertical or slightly' in- 

sloping sides and slightly everted, exteriorly thickened rim. 
Height, 24-43 cm.; rim diameter 20-34 cm. (fig. 122-5; Palmatary, 
1950, pis. 10, c-d, 11, a-b). Sometimes with applique spiral (op. 
cit., pi. 12 g, and i). 

6. Jars with rounded body, flat bottom and everted, slightly thickened 

rim, square or rounded lip. Rim diameter 16-36 cm. (fig. 122-6). 

7. Small jars with flat bottom, rounded body merging into insloping 

neck, everted rim. Mouth diameter 13-24 cm. (Meggers, 1947, 
pi. 1, fig. 1). Occasional vertical handles (fig. 122-7). 

8. Large jars with rounded bottom or small flat, pedestal base, walls 

curving outward to maximum diameter about one-third the dis- 
tance from the base then sloping inward to join the everted, ex- 
teriorly thickened rim. Height 80-90 cm.; rim diameter 54-76 
cm. (fig. 122-8). 

9. Large jars with small, flat bottom, sides outcurving to a maximum 

diameter of 70 cm. about one- third the distance above the base, 
then slightly inward, joining a short, vertical neck at the rounded 
shoulder. The rim is everted and exteriorly thickened. The body 
height 62-64 cm.; neck height about 15 cm.; rim diameter 32-54 
cm. (fig. 122-9). 



350 



BUREAU OF AMERICAN ETHNOLOGY 



[BOLL. 167 






Figure 121. — Rim profiles and vessel shapes of Camutins Plain and Inajd Plain 
bowls, Marajoara Phase (Appendix, tables 45 and 46). 



Less common forms : 

1. Narrow-necked jars with flat, conoidal base, globular body and 

slightl}^ everted, exteriorly thickened rim. Height 34-39 cm.; rim 
diameter 10-14 cm. (fig. 123-1; Palmatary, 1950, pi. 11, c, 12, d). 

2. Jars with flattened bottom, sides sloping outward to maximum diam- 
tliJrfSfc-^; .. . 6ter, then inward to constricted mouth with a collarlike, exteriorly 

thickened rim. Two small loop handles often occupy opposite 



fy^^s]^^^^ ARCHEOLOGY AT MOUTH OF AMAZON 351 




FiQtTRE 122. — Rim profiles and vessel shapes of Camutins Plain and Inajd Plain 
jars, Marajoara Phase (Appendix, tables 45 and 46). 



sides of the rim. Height, 10-20 cm.; diameter of orifice 1.7-4.0 
cm. (fig. 123-2; Palmatary, 1950, pi. 11, e). 

3. Flat-bottomed bowls with outslanting sides, direct rim and flat lip. 

Interior smooth, exterior with two coil lines, indicating construc- 
tion with three wide, fiat coils. Four, approximately equally 
spaced, ringlike depressions on the exterior with corresponding 
projections on the interior along the middle coil. Height 7.5-12.0 
cm.; rim diameter 10-22 cm.; diameter of depressions 4.5-6.0 cm.; 
depression depth 0.5-1.0 cm. (fig. 123-3). 

4. Bowls with flat bottom merging into rounded sides and incurving, 

direct rim with rounded or flattened lip. Rim diameter 10-17 cm. 
(fig. 123-4, pi. 66, c), 

5. Cylindrical pot stands with insloping sides and everted rim, open at 

the bottom and top. Height 15-20 cm.; diameter at the top 8-16 
cm. (Palmatary, 1950, pi. 8, a). Sometimes ornamented with 
small rim adornos or geometric wall perforations (fig. 123-5). 

6. "Platter-bowls" (cf. Joanes Painted, common vessel shape 8, fig. 

123-6). 

7. Stools (fig. 123-7; Palmatary, 1950, pi. 25, d). 



352 



BUREAU OF AMERICAN ETHNOLOGY 



[bull. 167 





9^>^ 



I I I I I I I 
4 e 12 CM 
VKiel Scale 




N^l 




1 2 3 CM 

Rim Seal* 




Figure 123. — Less common rim profiles and vessel shapes of Camutins Plain and 
Inajd Plain, Marajoara Phase. 



EVA^s]^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 353 

8. Funnels. Small, carinated bowls with an open bottom, rounded lip 

at bottom and top. Diameter of bottom opening 5-10 cm. Max- 
imum body diameter, 13-34 cm. (fig. 123-8). 

9, Figurines. 

10. Miniature bowls. 

11. Spoons. Miniature, circular or oval bowls with perforated projec- 

tion at one end. Maximum diameter 3.2-6.5 cm.; depth 2-4 
cm. (pi. 81). 
Appendages: Nubbins or small appliques may be placed on the body wall or 
on rims. Handles in the form of loops, oval or round in cross section, with 
one end attached to the rim and the other to the shoulder are found on 
small jars. Points of attachment are widened out but well tapered and 
graceful. Length ranges from small loops 3-5 cm. long up to 8-10 cm. 
Temporal differences within the type: None (Appendix, table 45). 
Chronological position of the type: Replaces Inajd Plain as the dominant 
plain pottery type in the latter part of the Marajoara Phase sequence. 

carmelo red 

Paste and unslipped surface: On Inajd. Plain or Camutins Plain; see those 
type descriptions for details of temper, color, firing and the unslipped surface. 
Slipped surface: 
Red Slip: 

Color: Cinnabar red to deep red, occasionally orange red. 
Treatment: Thin, often closer to a wash than a slip and reflecting the 
contour of the underlying surface, which is smooth and even on bowl 
interiors and irregular on the exterior of vertical-walled vessels. 
Hardness: 3. 
Form: 

Rim: Exteriorly thickened, direct, or broad, nearly horizontal and flat- 
topped with rounded lip. 
Body wall thickness: 5-13 mm. 
Bases: Probably flat. 
Vessel shapes: 

1. Large bowls with outflaring sides, often slightly thickened on the in- 

terior for several centimeters below the rim, rounded and level or 
undulating lip. Rim diameter 20-32 cm. Interior and rim top 
red slipped (fig. 124-1) . 

2. Bowls with mildly carinated walls and broad, everted rim with sloping 

top and flattened lip. Exterior rim diameter, 22-29 cm;, width 
of rim top 3.5 cm. Red slipped on the interior or exterior or 
both (fig. 124-2). 

3. Jars with insloping upper wall and everted or exteriorly thickened 

rim (based on rim sherds only). Rim diameter 22-40 cm.; red 
slipped on top and exterior (fig. 124-3). 

4. "Platter-bowls" (see common vessel shape 8 of Joanes Painted). 

5. Tangas (pi. 82, d, e). 

Decoration : Red coloring, applied to one or more surfaces, is the only ornamen- 
tation except for an occasional undulating rim edge or applique rib. 

Associated techniques: None. 

Temporal differences within the type: None. 

Chronological position of the type: Carmelo Red shows a slight increase in 
frequency during the Marajoara Phase sequence. 



354 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 





I I I I I I I 

4 8 12 CM 
Vessel Scale 





I ■ I I I 

I 2 3 CM 

Rim Scale 




Figure 124. — Rim profiles and vessel shapes of Carmelo Red, Marajoara Phase. 



HBGQERS AND 
BVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



355 



GOIAPf SCRAPED 

Paste and surface: On Inajd and Camutins Plain; at J-15 — Camutins scraping 
is used predominantly on Camutins Plain, but usually of better quality than the 
general run of that plain ware, while about 40 percent are on Inajd Plain. 
See those type descriptions for details of temper, firing, color, undecorated 
surface, etc. 
Form: 

Rim: All jar rims are everted and thickened on the exterior, with a rounded 
or flattened lip. The thickening may be gradual, forming a uniform 
curve, or abrupt and angular. One bowl rim is unthickened and flattened 
on the top. 
Body wall thickness: Range 4-7 mm.; majority 4-5 mm. 
Bases: One complete specimen from Fortaleza (University Museum, Pennsyl- 
vania, SA1870) has a small flat bottom. Since the typical shape is similar 
to jars of Guajard Incised, the bases were probably within the range of 
rounded or slightly flattened represented in that type. 
Vessel shape: 

1. About 90 percent of the rim sherds represent a form that has a con- 

tinuous variation, produced principally by widening of the neck, 
between a globular-bodied jar with a short, vertical neck and a 
slightly everted, exteriorly thickened rim and a deep bowl with a 
flattened bottom, sides that curve or slant outward to the maximum 
body diameter and then constrict slightly below the everted rim 
(fig. 126-1; Palmatary, 1950, pi. 9, a). At the jar end of the range, 
the exterior rim diameter is 18-20 cm. and at the bowl end, 30 cm. 
or more. Scraping is typically confined to a band immediately 
below the rim, corresponding to the neck. 

2. Only 2 rim sherds were from bowls, one rounded and the other with a 

slightly everted rim. Diameter 15 and 24 cm. (fig. 125-2). 
Decoration (pi. 70) : 

Technique: There was a wide variation in the tools used to produce the 
parallel striations classified here as "scraping", so that there is pronounced 
lack of uniformity from one example to another in the width and clarity 




Vessel Seal* 



Figure 125. — Rim profiles and vessel shapes of Goiapi Scraped, Marajoara Phase. 



356 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

of the marks. This type does not include those Camutins Plain and 
Inajd Plain bowls that have broad horizontal smoothing marks on the 
exterior, but is confined to those instances where scraping was applied 
as decoration after the smoothing was completed. The marks are of 
two major varieties: (1) Shallow lines made with a tool that left faint, 
fine striations in the groove, which ranges from 2.5-5.0 mm. in width 
but has small variation on a single specimen; and (2) sharply defined 
grooves, 1-4 mm. wide and lacking the striations in the trough. The tool 
used to make the hnes was usually single-edged, each line made separately. 
This results in nonuniformity not only in spacing but also in the width of 
the lines, which varj- with the angle of the tool to the surface. In some 
specimens the hues are so uniformly parallel and so close together that 
a combhke tool must have been used, making several lines simultaneously. 
The surface of the trough of the scrapings varies from even and slick to 
rough because of the temper particles dragged out and left adhering to 
the surface. 
Motif: The predominant and almost exclusive use of scraping was to orna- 
ment the neck of otherwise plain vessels. The Lines were made by dragging 
the tool vertically beginning just below the rim and stopping at the junction 
with the body or, where this is not pronounced, above the region of maxi- 
mum diameter. In some instances, this scraped band has the lines running 
horizontally., in which case additional scraping is often applied to the body, 
running diagonal!}^ or nearly perpendicularly to that on the neck. 

Associated techniques: This type of scraping appears on the necks of some jars 
with Anajds Plain Incised or Guajard Incised decoration on the bodies (Palma- 
tary, 1950, pi. 32, d and e). 

Temporal differences within the type: None. 

Chronological position of the type: Present throughout the Marajoara 
Phase sequence in increasing frequency. 

GUAJARi. incised 

Paste: Typically on Camutins Plain, occasional examples from lower levels of 
the sequence on Inajd Plain; see those type descriptions for details of temper, 
firing and color. 
Surface: 

Color: Tan, light-orange or red-orange, often with blackened, fire-clouded 

areas. 
Treatment: Both surfaces often well smoothed, particularly on smaller vessels; 
in other cases the decorated surface is smoothed but may remain somewhat 
uneven. 
Hardness: 2.5-3. 
Form: 

Rim: Typically outflaring and exteriorly thickened with rounded, blunt- 
pointed or angular lip. Some bowls have direct or expanding, rounded or 
flat-topped rim. 
Body wall thickness: Typically 6-9 mm. Rare small vessels have walls 3.5- 

4.0 ram. thick. 
Bases: Rounded or slightly flattened. 
Common vessel shapes: 

1. Small jars with bases flattened sufficiently to prevent tipping, globular 
bodies, vertical or insloping necks and everted rims. There are 
two size ranges: maximum body diameter 6-11 cm. and 20-45 cm. 
The height of the smaller group is 6.5-8.5 cm.; that of the larger 



UEGGBRS AND 
EVANS ] 



ARCHEOLOGY AT MOUTH OF AMAZON 



357 



is not exactly determinable because of the lack of complete speci- 
mens. The larger jars have an exterior rim diameter of 20-30 cm. 
Decoration may cover the body or be limited to a band extending 
one-half to two-thirds of the distance below the junction of neck 
and body. The neck is almost invariably ornamented with con- 
tiguous vertical (rarely horizontal) incised lines or scrapings 
(fig. 126-1, pi. 71, i). 
Small bowls with rounded body, slightly constricted mouth and 
everted rim. Body diameter is 6.7-11.0 cm. (Palmatary, 1950, 
pi. 4, e). Incision on the exterior (fig. 126-2, pi. 71, h, j). 




rP 



^ 






'til 

4 8 12 CM 
Vessel Scale 





I. I. I 



J I 



I 2 3CM 
Rim Scale 

Figure 126. — Rim profiles and vessel shapes of Guajard Incised, Marajoara 

Phase. 



391329—57- 



-25 



358 BUREAU OF AMERICAN ETHNOLOGY [bdll. m 

3. Shallow bowls with upcurving sides increasing in thickness toward 

the direct rim with rounded or flat lip. Rim diameter 12-16 cm. 
Decoration covers the exterior (fig. 126-3). 

4. Carinated bowls, with the rounded bottom joining almost vertical 

sides at a rounded angle. The rim is slightly everted and exteriorly 
thickened with a rounded lip. Rim diameter is 20-26 cm. The 
decoration is limited to exterior of vertical wall (fig. 126-4). 

5. Stools. Concave disk top 22 cm. in diameter; stool height 5 cm. 

Decoration covers the disk top. 
Rare vessel shapes: 

1. A jar in the form of two superimposed jars, one small jar resting in the 

mouth of another of similar shape but slightly larger (Palmatary, 
1950, pi. 11, d). Decoration covers the exterior. 

2. Miniature oval-bodied vessel with a short, constricted oval neck 

and everted rim. Ends flattened, with nubbin projecting from 
center of each. Length 10 cm., height 6.3 cm. Exterior decorated 
with undulating, triple incised lines. 
Decoration (pi. 71): 

Technique: The design is drawn with a double-pointed tool, which produces 
two evenly-parallel lines; a rare variation has a triple line. Within the 
type these vary from 0.5-2.0 mm. apart, representing variation in the 
dimensions of the tool point. On a single example, done with a single tool, 
the spacing is uniform. Incisions shallow, 0.5 mm or less in depth; width 
generally 1 mm., rarely 2 mm. The double lines are usually straight, 
although a series of double Unes is not always evenly spaced. 
Motif: The most outstanding characteristic of these designs is their repetition; 
a simple combination of straight and curved lines will form a continuous 
pattern over the entire decorated surface. A typical example is based on 
diagonally drawn lines producing a band of diamonds with their interiors 
filled by one long and two short lines drawn vertically and on triangles 
filled with similar lines drawn horizontally. One of the most frequent 
motifs is the scallop, used as the upper border, just below the neck, or as the 
basis for the whole design, dividing it into semicircular and angular fields. 
Associated techniques: Additional ornam.entation may appear on Guajard 
Incised vessels in the form of vertical applique ribs on the body, small nubbins 
on the exterior rim edge, and scraping on the neck exterior. Of these, scraping 
is by far the most common and the applique ribs rare and early, perhaps con- 
fined to Pacoval. 
Temporal differences within the type: The most important temporal dis- 
tinctions are in vessel size and in the application of the design. The later jars 
are considerably larger than the earlier ones, and more equal spacing of the 
parallel lines and a closer and more consistent approximation to true parallelism 
are characteristic of the earlier designs. The small jars are more typical of 
Pacoval and Fortaleza, while the larger variety are more common from Camutins. 
Chronological position of the type: Present throughout the Marajoara Phase 
with an increase in popularity from the early to late part of the sequence. 

iNAji. plain 

Paste : 

Method of manufacture: Coiling, coils 2-5 cm. wide, visible on some large, 

poorly smoothed jars. 
Temper: Ground sherd, very angular and often coarse, particles attaining 

5 mm. in diameter. 



JfJ'.^^af^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 359 

Texture: Very porous with temper poorly distributed. Poor mixture makes 
a granular, irregular cleavage plane. Tensile strength better than Camu- 
tins Plain. Sherds have a metallic ring when dropped together. 
Color: Cross section always has some gray core, ranging from a thin line, 
1-2 mm. wide, to 90 percent of the total cross section. Steel gray the most 
typical color. Type set up as distinct from Camutins Plain on the basis of 
this difference in core color. 
Firing: Incompletely oxidized. 
Surface: 

Color: Exterior and Interior — Range from a light orange to dull tan to grayish 
tan to grayish orange to grayish red-orange. Majority of sherds have a 
grayish hue. 
Treatment: All details of surface porosity, water bubbles, and texture indicate 
the pottery was handled when extremely wet. Crackle lines begin around 
the prominent temper particles and although found on both surfaces are 
more common on the interior. Exterior rough, coarse and irregular with 
finger tracks visible on over 50 percent of the sherds. Interior of bowls 
floated or slipped with same clay as paste and smoothed, generally leaving 
smoothing tracks. 
Hardness: 3-3.5. 
Form: All vessel shapes, rims, bases, appendages, dimensions, etc. are identical 
to Camutins Plain. See that type for profiles, vessel shapes, and descriptive 
details (also figs. 121-123; pis. 64; 65; 66, a, b, d; 67, h; 68, 69). 
Temporal differences within the type: None (Appendix, table 46). 
Chronological position of the type: This is the dominant plain ware in the 
early Marajoara Phase and declines in frequency as Camutins Plain increases. 

joanes painted 

Paste and unslipped surfaces: Details of temper, firing, color, surface treatment 
of the unslipped surfaces correspond to Inajd, Plain and Camutins Plain; see 
those types for descriptive details. 
Slipped surface: 
White slip: 

Color: Typic&lly white; firing variations include cream, light orange, 

light tan and bluish white. 
Treatment: On bowl interiors, typically smooth, even and polished, often 
producing a slight luster. Somewhat unevenly applied on jar exteriors. 
Ranges from a thin film to 1 mm. in thickness. 
Hardness: 3-4. 
Form: 

Rim: Predominantly exteriorly thickened or direct; interiorly thickened on 
some shallow bowls; hollow on some bowls with Pacoval Incised or Anajds 
White Incised exteriors. 
Body wall thickness: Range from 4 mm. on miniature bowls to 17 mm. on the 

bodies of large jars 
Bases: Flat, rounded, annular or annular pedestal. 
Common vessel shapes: 

1. Shallow, open bowls with rounded bottom, outcurving sides, ex- 
teriorly thickened rim. A thickened coil often added 1 cm. below 
the lip gives a flanged effect. Rim diameter 8-38 cm. Circum- 
ference may be circular (typical), ovoid or D-shaped. Those 
decorated with Pacoval Incised or Anajds White Incised on ex- 



360 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 





■>A':^i.:. -- ,^ 








4 


8 12 CM 


Va 


sie 


Seal* 





a 


16 24 CM 


Ve 


sse 


s 10 a 14 




I 2 3 CM 
Rim SeoK 



FiGUKE 127, — Rim profiles and vessel shapes of Joanes Painted, Marajoara 
Phase (Appendix, table 47). 



EVANS] ARCHEOLOGY AT INIOUTH OF AMAZON 361 

terior may have hollow rims. Painting covers the interior (fig. 
127-1). 

2. Bowls with slightly rounded bottom, nearly vertical sides and 

exteriorly thickened or everted rim with a flat Up. Diameter 
15-30 cm. (Palmatary, 1950, pi. 15, b; 71, d; 72, a). Painting on 
the interior or exterior (fig. 127-2), 

3. Bowls with rounded or flattened bottom, angular profile and exte- 

riorly thickened rim with flat or rounded lip. In a few the rim is 
slightly everted rather than thickened. Diameter typically 26-34 
cm., occasionally 3S-52 cm. (Palmatary, 1950, pi. 72, b-c; 78, b). 
Painting on interior or exterior (fig. 127-3). 

4. Carinated bowls with flat or rounded bottom, lower sides outcurving 

or outslanting to join upper, vertical or insloping walls at marked 
carination; direct rim with flat or rounded lip. Upper wall 
comprises one-half to two-thirds of the total height. Diameter 
16-24 cm. (Palmatary, 1950, pi. 76, d and f ; 77, a and c) . Painting 
on interior and upper exterior (fig. 127-4). 

5. Carinated, sometimes rounded, bowls with tall, annular, pedestal 

base, vertical to outslanting upper wall, everted or exteriorly 
thickened rim with rounded or flat lip. Mouth diameter 16-30 
cm. (Palmatary, 1950, pi. 76, b-c; 79, a-b). Interior painted; 
exterior plain or painted (fig. 127-5) . 

6. Rounded bowls with outcurving to nearly vertical sides, direct rim 

with flat or rounded lip. Diameter, 10-38 cm. (Palmatary, 1950, 
pi. 32, b; 69, f-g; 82, e) . Exterior and interior painted (fig. 127-6) . 

7. Bowls with rounded bottom, sides incurving to direct or sHghtly 

interiorly thickened rim with a rounded lip. Diameter 10-30 cm. 
(Palmatary, 1950, pi. 69, g; 71, e). Exterior and interior painted 
(fig. 127-7). 

8. "Platter-bowls" with flaring annular base, deep bowllike center 

inserted into the middle of a broad platter, producing a wide, 
troughlike, lateral extension terminating in an exteriorly thickened, 
often flanged, rim. Circumference circular or squared. Rim 
diameter, 25-45 cm.; central bowl diameter 5-16 cm. (approxi- 
mately one-fourth to one-third of the total diameter) . (Palmatary, 
1950, pi. 80 a-d, 81 a.) Interior painted. These vessels were used 
as burial-urn covers at Pacoval and Ilha dos Bichos (fig. 127-8). 

9. Small jars with flattened bottom, rounded bodj' and short, vertical, 

direct rim with a rounded lip. Height 6-15 cm. (Palmatary, 
1950, pi. 73, a-b). Exterior painted (fig. 127-9, pi. 67, a). 

10. Jars with fiat bottom, rounded body, pronounced shoulder, insloping 

neck and everted or exteriorly thickened rim. Height 20-80 cm. 
(Palmatary, 1950, pi. 82 a-b, 83 a, 84 b, 97 a) . Painting covers the 
exterior (fig. 127-10, pi. 73, a, c). 

11. Jars with flat bottom, globular body, pronounced shoulder, outslop- 

ing neck and everted, exteriorly thickened rim. Height 30-40 
cm. (Palmatary, 1950, pi. 86, a-b; 87, a). A rare variation has a 
carinated body (op. cit., pi. 78, a). Low relief, paired eye motifs 
on opposite side of neck are typical. Painting covers exterior 
(fig. 127-11, pi. 73, b). 

12. Jars with flat bottom, outsloping sides, rounded shoulder, short 

vertical neck, and direct rim. Height 22-32 cm. (Palmatary, 
1950, pi. 77, b). Painting on exterior, lower limit marked by 



362 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

slight ridge around the exterior about one-quarter the distance 
above the base (fig. 127-12). 

13. Jars with flat bottom, upcurving sides, rounded shoulder, short 

vertical neck and everted, exteriorly thickened rim. Height 
22-28 cm, (Palmatary, 1960, pi. 85, a). Painting covers the 
exterior, sometimes absent on the neck (fig. 127-13, pis. 73, d, 
75, a). 

14. Funerary jars with small, flat base, globular to ovoid body, rounded 

shoulder, vertical or insloping neck and widely everted, exteriorly 
thickened rim. Height 30-95 cm. (Palmatary, 1950, pis. 88, 89, 
93, 94, 95). A rare variation has a flat, horizontal shoulder. 
Painting covers the exterior. Stylized anthropomorphic faces 
modeled on two opposite sides of neck; small, grotesque or anthro- 
pomorphic figures in the round between the ears (pi. 74, a-b) , 
body painting typically includes stj'lized arms, hands and fingers 
(fig. 127-14; pis. 75, h, 76, a-h). 

15. Anthropomorphic figurines. Height 6-24 cm. (pi. 79, a-c, e; Palma- 

tary, 1950, pis. 47, b-c; 100, b-d; 101, a-d). 

16. Tangas (pi. 82, a-h; Palmatary, 1950, pis. 102, f-k, 103, 104). 
Rare vessel shapes: 

1. Carinated bowl, rounded bottom, sUghtly outslanting sides, hori- 

zontal rim with rounded lip. Painted interior of bowls of Ararl Red 
Excised, White-Retouch common vessel shape 1. 

2. Small, open bowl with outcurving sides, direct rim, with a coil added 

around the exterior generally about 1 cm. below the rim top, 
producing a decorative, often scalloped flange. Painted interior 
of occasional bowls of Arari Red Excised, common shape 4. 

3. Shallow, open bowl with annular base and exteriorly thickened rim. 

Painted interior of occasional bowls of Ararl Red Excised, common 
shape 6. 

4. Mildly carinated bowl. Painted interiors of occasional bowls of 

Anajds White Incised, rare vessel shape 2. 

5. Carinated bowls. Painted interiors of bowls of Pacoval Incised, 

common vessel shape 4. 

6. Stools (pi. 84, a). 

7. Miniature bowls. 

8. Globular bodied jars with small mouth and everted rim. Height 

17-32 cm. (Palmatary, 1950, pi. 85, b). 

9. Cylindroid jars with flat or conoidal bottom, vertical or slightly 

insloping sides, exteriorly thickened rim. Height 17-52 cm. 
(Palmatary, 1950, pis. 32, e; 69, e; 87, b). 

10. Double or multiple bowls (Palmatary, 1950, pis. 70, d; 74, a-b). 

11. Anthropomorphic vessels (Palmatary, 1950, pi. 75, b). 
Decoration (pis. 72-76): 

Technique: 

I. Red or black paint on a white-slipped surface. The color of the red 
paint may vary from red to rust, orange brown, dark reddish brown, 
or even dark brown because of uneven firing conditions or because of 
difference in the thickness of the paint, giving it greater or less trans- 
parency. At the darker end of the range, red-painted designs are not 
distinguishable from those originally painted black. On the other 
hand, the use of a true black pigment is attested by the presence of 
polychrome designs using both red and black. Black-on-white is con- 



J?J^."«o?^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 363 

siderably less frequent than Red-on- white; however both color com- 
binations occur on bowls and jars. In all painted vessels, the paint 
has a dull finish that contrasts sharply with the surface of the under- 
lying white slip when the latter is well polished. Painting is em- 
ployed in the following variations: 

A. Lines: 

1. Wide, soHd and dotted lines, width 2-5 mm. (Palmatary, 

1950, pis. 31, b, 69, a: Red-on- white) . Used on bowl in- 
teriors, the exterior being Pacoval Incised or, rarely, 
Anajds White Incised. 

2. Wide and narrow lines. In this and other categories using 

wide and narrow lines, the actual width is related to 
vessel size (i. e., the narrow lines on large vessels may 
exceed the width of wide lines on small vessels). Interior 
of bowls; exterior may be painted, incised or excised. 
(Palmatary, 1950, pi. 38, c: Black-on-white). 

B. Lines and solid areas: 

1. Narrow and wide lines and solid areas, the narrow lines 

being typically in pairs (Palmatary, 1950, pis. 82, b; 86, b: 
Red-on-white; pis. 81, a, 82, a: Black-on- white) . Interior 
of bowls or exterior of jars; bowl exteriors may be plain 
or excised. 

2. Narrow or wide lines and hatched areas. Interior of bowls; 

exterior may be plain, painted or excised (Palmatary, 
1950, pis. 77, c; 70, f ; 99, a and c: Red-on-white; pi. 69, c: 
Black-on-white) . 

3. Wide Unes and solid areas. Interior or exterior of bowls 

(Palmatary, 1950, pi. 32, c: Red-on-white). 
II. Polychrome designs combining 2 colors (red and black or red and 
orange) on a white-slipped surface. Most frequently these are a 
shade of red and one of black. In the case of filled red lines (type C), 
the colors are red or brown and light orange. The colors are combined 
in 3 ways to produce a 2-color design on a white background: 

A. Red lines accented with black. The skeleton of the design is 

drawn in red, after which pendant dots, corner "reinforce- 
ments" and the centers of small rectangles, triangles, or 
crosses are painted black. This type appears to be restricted 
to Pacoval and is^infrequent there (fig. 128, a-c). 

B. Independent use of red and black, with lines or areas of one 

color separated from those of the other color by an unpainted 
zone (Palmatary, 1950, pis. 32, f ; 71, b, e-f; 93). This is by far 
the most frequent variety and occurs on both bowls and jars 
(fig. 128, d-f). 

C. Filled red lines. The design is drawn in paired, narrow, red or 

brown lines and the area between them is painted a light 
orange or red (Palmatary, 1950, pi. 28, c). Restricted to bowl 
interiors and to tangas (fig. 129, a-e). 
Motif: The same motifs are used in all the varieties of Joanes Painted in 
about the same relative frequency. Spirals are exceedingly common, 
principally the single variety, and are usually rounded or square. Inter- 
locking spirals are somewhat less abundant. Also exceedingly common is 
a T or L (half-T), always representing an unpainted area produced by 
painting a stepped outline on the interior of a triangle or small field of 



364 



BUREAU OF AMERICAN ETHNOLOGY 



[BOLL. 167 




BLACK OR BROWN 



Figure 128. — Joanes Painted, Marajoara Phase, a-c, Polychrome Type A. 

d-f, Polychrome Type B. 



another shape. Small, pendant dots along narrow lines are another popu- 
lar motif. Other elements include undulating lines (often in combination 
with T's. and like them representing the white, unpainted background 
rather than a painted line), stylized faces (most typical on tangas), tri- 
angles, rectangles, crosses, diamonds and short wavy lines. The bodies 
of large burial urns often include an exotically stylized face along with 
geometric elements. 
Associated techniques: Anajds Red Incised, Anajds White Incised, Ararl Plain 
Excised, Ararl Red Excised, Arari Red Excised White-Retouched, Arari Double- 
Slipped Excised, and Pacoval Incised. 
Temporal differences within the ttpe: Red-on-white Variety A-1 and Poly- 
chrome Variety A are early, being found only at Pacoval. No trends are 
evident stratigraphically in vessel shape (Appendix, table 47). 



"v^Nsf^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 



365 



Chronological position of the type: Joanes Painted as a whole shows a slight 
decline in populaiity, although it is always by far the most frequent method 
of decoration in the Marajoara Phase. 








RED OR BROWN 



LIGHT ORAWGE OR RED 



Figure 129. — Joanes Painted, Marajoara Phase, a-e, Polychrome Type C. 



366 BUREAU OF AMERICAN ETHNOLOGY [bull, m 

PACOVAL INCISED 

Paste and unslipped sukface: Predominantly on Inajd Plain, occasionally on 
Camutins Plain ; see those type descriptions for details of temper, firing, surface 
finish, etc. 
Slipped surface: 
White Slip: 

Color: White; irregular firing sometimes gives a cream or orange tint. 
Treatment: Fine, smooth on bowl interiors, sometimes showing faint 
luster. On bowl and jar exteriors smoothing tends to be less perfect, 
leaving smoothing tracks and some unevenness. 
Hardness: 3-5. 
Form: 

Rim: Solid or hollow, exteriorly or interiorly thickened, rounded or flattened 

at different angles producing a faceted Up. 
Body wall thickness: 4-12 mm. The entire range is represented in all vessel 

shapes. 
Bases: Rounded or flat. 
Common vessel shapes: 

1. Shallow, open bowls with flattened bottom, outflaring sides and 

hollow rim produced by folding the upper edge toward the interior 
or by the addition of a flattened coil on the interior, changing 
the angle of the interior wall so that in some cases it becomes 
almost vertical. Thickness of hollow rim 1.5-3.0 cm.; body wall 
tliickness 5-9 mm.; maximum diameter 55 cm.; maximum depth 9 
cm. (Palmatary, 1950, pis. 38, a; 39, j). Incised design covers 
interior or exterior (fig. 130-1). 

2. Shallow, open bowls with flattened or rounded bottom and interiorly 

thickened, solid rim, usually indistinguishable from shape 1 rims 
except in cross section. Lip typically rounded, rarely flattened. 
Rim thickness 1.2-2.0 cm.; body wall thickness 5-12 tnm.; 
maximum diameter 44 cm. Incised design covers exterior and 
interior (fig. 130-2). 

3. Shallow, open bowls with rounded or flattened bottom and exteriorly 

thickened rim with rounded lip. Diameter 24-44 cm. (Palmatary, 
1950, pi. 24, b-c). Incised design covers interior, occasionally 
also the upper part of exterior (fig. 130-3). 

4. Bowls with flat bottom, outslanting sides and exteriorly thickened 

rim, sometimes slightly everted with rounded or bifurcated lip. 
Rim thickness 1.4-1.8 cm.; body wall thickness 6-10 mm.; rim 
diameter 10-29 cm.; depth 5-12 cm.; base diameter 12-16 cm. 
(Palmatary, 1950, pis, 28, f; 48, a). Incised design covers the 
exterior (fig. 130-4). 

5. Carinated bowls, the upper walls joining the lower ones so as to 

produce a marked change in direction, but usually with a more 
rounded and less pronounced angle than on Arari Plain Excised 
because of the outward slant of the upper walls and the lack of 
exterior thickening at the carination. Rim exteriorly thickened 
with a rounded lip; bottom flattened. Rim thickness 1.2-2.2 
cm.; body wall thickness 4-12 mm.; rim diameter 25-50 cm.; 
upper wall height 6-12 cm. Incised design covers the exterior 
wall (fig. 130-5). 



^vTnsT^ ^""'^ ARCHEOLOGY AT MOUTH OF AMAZON 



367 




4 8 I2CM 
Vessel Scole 



Figure 130. — Rim profiles and vessel shape? of Pacoval Incised bowls, Marajoara 

Phase. 



368 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 



6. Bowls with rounded bottoms, curved sides and broad, horizontal 

rim 2.2-3.7 cm. wide across the flat top. Diameter 22-34 cm. 
(Palmatary, 1950, pis. 37, c; 38, j). The red-retouched design is 
usually confined to the flat rim top; rarely, also found on the 
exterior (fig. 130-6). 

7. Deep, carinated bowls or jars with truncated conoidal bottom, 

vertical, slightly concave v/all and exteriorly thickened, everted 
rim with a rounded or angular lip. Wall height comprises about 
two-thirds the total height, which is 29-33 cm. (Palmatary, 1950, 
pi. 33, a-b). Incised design covers the exterior of the wall (fig. 
131-7). 

8. Jars with flattened bottom, depressed-globular body and tall neck 

terminating in a slightly everted, exteriorly thickened rim with 
a rounded or angular lip. Height 33-46 cm. (Palmatary, 1950, 
pis. 18, a and c; 33, e). Design covers the exterior (fig. 131-8, 
pi. 78, b). 

9. Jars with flat bottom, sides outsloping to maximum diam^eter about 

one-third of the distance from base and then insloping to the 
exteriorly thickened rim with a rounded or angular Up. Height 
30-60 cm. (Palmatary, 1950, pis. 22, a and d; 29, a-d, 33, c-d). 
Incised design covers the insloping walls on the exterior (fig. 
131-9). 




I E 3 c^ 




FiGUEE 131. — Rim profiles and vessel shapes of Pacoval Incised Jars, Marajoara 

Phase. 



"v^N^f^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 369 

10. Globular bodied jars with constricted mouth and exteriorly thick- 

ened rim. These are often asymmetrical with the bottom shghtly 
flattened oflf-center. Height 22-30 cm. (Palmatary, 1950, pis. 
18, b and d; 82, c). Incised design covers the exterior (fig. 131-10, 
pi. 78, c). 

11. Anthropomorphic jars with truncated conoidal base, walls insloping 

or recurved to neck, then expanding to form the head at top of 
which is jar mouth with an everted rim. Stylized anthropomorphic 
facial features; anatomical details on the body absent or suggested 
by low relief, nubbins, or small bosses. Height 36-77 cm. (Palma- 
tary, 1950, pis. 19, e; 23, d; 30, a; 34). Incised design covers the 
exterior. A less-common variety has a flat bottom and a cylin- 
drical or roimded body with slightb/ more anatomical detail. One 
example has the arms modeled in the round and raised to the 
mouth. Sex is female. Two of this type from Pacoval are 22 and 
35 cm. tall (Palmatary, 1950, pi. 27, a-b; fig. 131-11, pi. 78, a). 
Rare vessel forms; 

1. Jars with flat bottom, four-lobed body, short, vertical or anthropo- 

morphic neck and exteriorly thickened rim. Height 20-55 cm. 
(Palmatary, 1950, pis. 18, e; 19, b and d; 28, k). Neck Pacoval 
Incised with body Anajds White Incised. When the neck is not 
anthropomorphic, two opposite lobes of the body bear vertical 
applique strips flanked by two nubbins. 

2. Bottles with a narrov/ mouth 4.5-5.5 cm. in external diameter, a 

short bulbous upper section separated by a necklike constriction 
from the large body. No complete vessel of this type exists, but 
several sherds from Pacoval represent the upper part. Height 
from the neck constriction to the rim top is 8.3 cm. Similarly 
shaped jars with the bulbous upper part connected to the main 
body by four independent flues come from Teso do Severino 
(Palmatary, 1950, pi. 28, i and j). Incised design on exterior. 

3. Stools (Palmatary, 1950, pi. 28, g). 

4. Bird effigy (Palmatary, 1950, pi. 28, d). 

5. Complex jar composed of 4 figurines with intertwined arms, with a 

flat bottom and a bottle type neck (Palmatary, 1950, pi. 26, a). 
Decoration (pis. 77, 78) : 

Technique: Diagnostic feature is the presence of incised lines that have been 
colored or "retouched" with red. Lines so treated are typically wide 
(1-3 mm.) and sharply deflned. In a rare variation, the red line is formed 
by applying the pigment between two closely parallel, narrow incised lines. 
The design may be composed of (1) exclusively red-retouched lines, (2) 
red-retouched lines and broad, usually triangular or rectangular areas painted 
red, or (3) broad, red-retouched Unes and fine, unretouched lines, which are 
sharply defined and occasionally cut through the slip to the underlying 
orange surface. The fine, parallel lines are not always evenlj' spaced and 
may be broken where one stroke ended and another was begun carelessly 
so as to overlap rather than join the end of the one previously completed. 
Similar overshooting is also occasionally present at corners. The red- 
retouching was done with a thick, cinnabar-red pigment and was frequently 
applied either carelessly or with too wide a brush so that the red fine 
overlaps the edge of the incision. Rare examples have a dark-brown or 
black instead of a red-retouch, but at least in the case of the dark brown 



370 BUREAU OF AMERICAN ETHNOLOGY [boll, le: 

this could be the result of a firing difference. Red-touched lines and areas 
and unretouched lines are used in the following combinations: 

1. Retouched lines alone (Palmatary, 1950, pi. 37, b-c). On jar or bowl 

rims or to outline anthropomorphic features on jars and figurines 
(pi. 77, y, rim; 78, o). 

2. Retouched lines and retouched areas filled with lines (Palmatary, 

1950, pis. 19, d; 28, k; 38, f) . Jar necks and bowl exteriors (pi. 78, b) . 

3. Retouched lines alternating with single, paired or triple narrow 

incised lines (Palmatary, 1950, pis. 24, b and c; 32, a; 37, e). Interior 
or exterior of bowls. 

4. Retouched lines and areas alternating with single, paired or triple, 

narrow incised lines (Palmatary, 1950, pis. 18, a-d; 23, e-f; 29, a-d; 
33, a-b; 33, d-e). Typical on jar exteriors (pis. 77, e, 78, c). 

5. Alternation of paired, narrow incised lines, sometimes also ovals and 

triangles, the space between them painted red, and single, paired 
or triple, narrow incised lines (Palmatary, 1950, pis. 24, a; 35, a). 
On bowl interiors and rims (pi. 77, a-b,f). 

6. Broad, single, retouched lines dividing large rectangular or irregularly 

shaped fields filled with complex arrangements of narrow incised 
fines (Palmatary, 1950, pis. 39, j; 48, a). Exterior of shallow bowls 
(pl.77,i). 
Motif: The majority of the designs on bowls incorporate an interlocking 
spiral, one member of which is a red-retouched line and the other a single, 
paired or triple, narrow incised line. Commonly associated with this is 
a pair of interdigitating lines, composed of a red-retouched line with 
four short, vertical projections that fit between a similar number of 
corresponding projections from a paired, narrow, unretouched line. 
Predominant on jar exteriors is a continuous series of angular, generally 
_/~-shaped fields, formed by single, paired or triple, narrow incised lines 
and containing a single, red-retouched line with various numbers of tri- 
angular appendages. This red-retouched fine may be widened so as to 
cover half the surface of the field it occupies. 
Associated techniques: Pacoval Incised may be used in conjunction with one 
or more of the following decorative types: Joanes Painted, Anajds White 
Incised, Arari Red Excised White-retouch, and small adornos. When com- 
bined with Joanes Painted, the latter technique is used on the interior of bowls 
with Pacoval Incised occupying the rim or exterior, or both. Anajds White 
Incised may be found on the exterior of bowls or the body of jars with Pacoval 
Incised on the remaining surfaces. Association with Ararf Red Excised, 
White-retouch is rare, fimited to a few instances of Pacoval Incised designs 
just below the rim of a semicyUndrical Ararf Red Excised, White-retouched 
jar. 
Temporal differences within the typf,: None. 

Chronological position of the type: Red-retouching of lines to produce a color 
contrast with the wliite slip is an early technique in the Marajoara Phase 
which dies out after Fortaleza site and is absent during the latter part of the 
Phase sequence. 

unclassified decorated 

There are very few decorated sherds from the Marajoara Phase that cannot be 
classified readily into one of the decorated types if alteration of surface color by 
accidental firing differences is taken into account. Those left in this residue are 
all varieties of punctate, often combined with incised lines. All but one has an 
unslipped surface. 



^^?^^f^ *^" ARCHEOLOGY AT MOUTH OF AMAZON 371 

Punctate: 

1. Wedge-shaped punctates forming straight lines. The motif consists of 

concentric squares with the area between them divided by diagonal 
lines into 6elds filled with parallel punctated lines (similar to Anajds 
Plain Incised on pi. 52, g). Upper exterior of small six-sided jar with 
rounded bottom, constricted mouth, everted rim. Height 7.5 cm., max- 
imum body diameter 10.5 cm., mouth diameter 7.0 cm. 

2. Rows of round punctates following lightly incised guide lines. Division 

into areas, in each of which the lines are parallel to each other but diag- 
onal to those in the adjacent area. Exterior of a sherd from the 
shoulder of a small jar. 
Punctate and incision: 

1. Single, double or triple incised lines separating irregularly shaped areas 

in the manner often employed in Pacoval Incised, with these areas 
filled by oblong or wedge-shaped punctates, 2-3 mm. long. Exterior 
of a miniature jar with a flat bottom, rounded body and slightly everted 
rim. 

2. Broad, parallel incised lines alternating with a row of dotted lines, formed 

by elongated, dashlike punctates. Neck exterior of a small jar with 
a sUghtly flattened bottom, rounded body and everted rim; height 
7.5 cm., diameter 4.8 cm. 

3. Narrow, parallel incised lines in threes separated by irregular rows of 

elongated punctates. Red-slipped exterior of small, heart-shaped 
bowl with flat bottom, vertical sides and rim slightly thickened on the 
exterior. Rim diameter 14.0 by 16.2 cm, ; depth 6.0 cm. 

Ceramic and Nonceramic Artifacts 

Since only durable materials like stone and pottery are preserved 
in a tropical forest environment, it is fortunate that the Marajoara 
occasionally used pottery for other things than containers. Marajd 
Island produces no stone suitable for axes or ornaments, and such 
material had to be acquired from elsewhere. Stone artifacts are 
rare in refuse deposits, probably because, being scarce, care was 
taken not to lose them. As most of the forest-dwelling South Ameri- 
can groups have done in more recent times, it is probable that the 
Marajoara exploited the plant and animal resources to provide them- 
selves with ornaments of odd seeds and brilliant feathers. The 
occasional ear plugs or labrets of pottery are drab to the eye of the 
archeologist, and must have been so to the makers as well, although 
the latter apparently devoted no effort to making them more attrac- 
tive with painted or excised designs. 

In the descriptions of artifacts that follow, stone and ceramic 
objects of the same type have been described together because what 
was made is of more significance than the material used. Although 
the total is small and gives only a glimpse of Marajoara Phase mate- 
rial culture, what has survived is sufficiently unique to make this 
complex readily distinguishable should it ever be found elsewhere in 
South America in the future. 



372 BUREAU OF AMERICAN ETHNOLOGY [bull. i67 

Axes. — A number of people have reported finding stone axes in 
Marajoara Phase sites. One of the earUest is Derby (1879, p. 227), 
who describes them as diorite, well polished, and not distinctive in 
shape, and says they are uncommon. Netto collected a number 
from Pacoval, which he specified in one place as 10 to 12 (1885, 
p. 445) and in another as "some 20" (op. cit., p. 270). He adds 
that "Sr. Rumbelsperger, who followed me a year later, also found 
no inconsiderable number of them" (op. cit., p. 445). Lange later 
included "a quantity of stone axes and various diorite implements" 
in his collection from the same mound (1914, p. 322). 

Farabee's field notes (1914) on the Fortaleza group mention a 
stone ax from Mound 4, 2 from Mound 7, trench 4, and a small one 
of quartz from Mound 7, trench 6. An ax from Cajueiros and a 
broken specimen from Sanharao are recorded as having been pre- 
sented to the Museu Paraense by the Barao de Mara]6 (1895, p. 88). 
A fragment was found at Pacoval do Curur6 by Nimuendaju (Ryden, 
MvS.). The only specific information on size or shape comes from Bar- 
bosa Rodrigues (1876-78, fig. 57), who illustrates two examples. One 
is a fragment, for which the provenience is simply "Maraj6 Island." 
The other, from an unspecified mound, is polished diorite, somewhat 
asymmetrical, with a rounded butt, flattened blade, and notched sides. 

Our investigations produced three axes, one each from the surface 
collections at J-15, Camutins, Mound 1, and J-14, Mound 1, Guajard, 
and one associated with an upper burial (jar A) in cut 1 of the latter 
cemetery. The two surface specimens are miniatures and may be 
similar to the "little axes" Ferreira Penna found at Pacoval and 
Santa Izabel (1879 a, pp. 53-54). The one from Guajard (fig. 132, b) 
is of gneiss, well-shaped but not polished except adjacent to the bit. 
It is 4.6 cm. long, 3.2 cm. wide at the blade and 1.6 cm. thick. The 
sides taper slightly to the rounded butt, and the blade is sharp and 
nicked in the center. The Camutins miniature (fig. 132, a) is of 
greenish diorite and is almost square: 3.5 cm. long by 3.8 cm. wide, 
with a maximum thickness of 1.8 cm. The surfaces are well pol- 
ished, sloping toward the blade and sides in three facets that join at 
pronounced angles. The butt and one of the edges are battered as 
though the implement had been extensively used as a hammer in 
spite of its smallness and lightness. The blade is considerably nicked 
from use. 

The ax found outside the base of burial jar A (J-14, Mound 1, 
cut 1) is somewhat larger and has a rounded and pohshed, blunt 
blade of the type used for preparing bast fibers (fig. 133). The blade 
is convex, the sides straight and slightly tapering to the butt, which 
is concave and the only unpolished part of the implement. The 
surfaces are smooth and unworn except for a chip at one edge. 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



373 





I CM 





Figure 132. — Marajoara Phase miniature axes, a, J-15, Mound 1. 6, J-14, 

Mound 1. 



391329—57- 



-26 



374 



BUREAU OF AMERICAN ETHNOLOGY 



I BULL. 167 




CM 




Figure 133. — Marajoara Phase stone ax from J-14 , Mound 1, cut 1. 

Length is 5.5 cm., maximum width 4.7 cm. and thickness at the butt 
2 cm. The material is diorite, dark gray to gray green in surface color. 

Highly polished axes of green diorite or nephrite are mentioned by 
Holdridge (1939, p. 75) as coming from the mounds. A specimen 
from Laranjeiras, in the Museu Goeldi, is 9.5 cm. long and 4.3 cm. 
wide, with straight, flat sides, a square butt, and a well-sharpened 
bit. Thickness is 2.2 cm. The surfaces are polished, but all the 
the flaws have not been removed, and the butt is rough. The stone 
is light green with iron unpurities. 

Beads. — Very few objects that can be identified with certainty as 
beads have come from Marajoara sites. The only authenticated 
find is a recent one in which 65 cylindrical beads of a white stone 
with dark-brownish veins (nephrite?) were discovered in a burial urn 
in a cemetery on the upper Camutins. These are drilled from both 
ends toward the middle, making V-shaped holes joined at the small 
end (Hilbert, pers. commun.). 

Thirtj'^-eight animal teeth, perforated for stringing on a necklace, 
were found by Mordini in one of the upper levels of his cut at Panellas 
(Palmatary, 1950, p. 279). 

Earplugs. — Ornaments of this sort are rare, but a few have been 
recorded. Two small spools are in the Peabody Museum, Harvard 
University collection from Pacoval (fig. 134, h). Both have short, 
constricted shafts and expanded, concave ends, one of which is of 
less diameter than the other. They are circular, but otherwise crude 
and rough. The first has an orange paste with a trace of white slip 
and a fine hole pierced through the center. It is 1.3 cm. thick, 1.5 
cm. in diameter on one surface, and 1.7 cm. in diameter on the other. 



MEGOSaS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



375 



The second is Inajd Plain with a blackish surface; 1.2 cm. thick, 1.6 
and 1.8 cm. in diameter on the disks. A similar specimen is in the 
University Museum, Philadelphia, collection from Camutins. It is 
3 cm. in diameter and has smooth and polished surfaces. An exam- 
ple (fig. 134, a) from the surface collection of Camutins, Mound 1, 
is comparable in all respects, except that it is 2.5 cm. in diameter 
and slightly more ornate, having a red slip and two parallel, finely 
incised lines on the flanges. There is a perforation through the 
center that may have been used to insert a feather or a tassel. 

The ears of anthropomorphic jar L of J-14, Guajard, cut 1, con- 
tain ornaments in the lobe that appear to be earplugs of this type 
(pi. 76; fig. 147, a). They bear a painted design on the surface and 
have a pendant tassel. 

Figurines. — Stylized anthropomorphic figurines, in a complete or 
fragmentary condition, are relatively common in the refuse on ceme- 
tery mounds. The majority are small, although some are 25 cm. or 
more in height (Nordensldold, 1930, frontispiece). They may be 
unslipped, red slipped or white slipped, and undecorated, incised, 
excised, or painted. Painting is most frequent, either red-on-white 
or polychrome, in an all-over design that emphasizes the low relief 
arms and other anatomical characteristics when present. The arms 
are often absent or abbreviated and the figure is typically rendered 
seated, with the legs and body forming a U-shaped base. The legs 
or knees are rounded at the end and may have toes at their base. 
Breasts and sexual organs are sometimes shown, and where sex is 









I CM 



Figure 134. — Marajoara Phase pottery ear plugs, a, J-15, Mound 1. h, Pacoval. 



376 BUREAU OF AMERICAN ETHNOLOGY [bdll. igt 

indicated, it is to the authors* knowledge invariably female. The 
shape of the head is stylized in two basic ways: (1) sloping back from 
the forehead and up from the occipital area to form a pointed top 
(pi. 79, b), and (2) cylindrical with a rounded top, a high forehead and 
two horizontal protuberances at the back that probably represent 
the hairdress (pi. 79, c, e). Netto (1885) illustrates a large series of 
the fu-st type, and suggests that the distortion of the skull shape is 
evidence of the practice of cranial deformation. The presence of a 
deformed skull in one of the urns from J-14, Mound 1, Guajara (p. 
273) lends support to this interpretation. Facial features follow 
a standard method of representation: the eyebrows are joined to the 
nose in a Y or T and laaj continue around and down the side to form 
the ear; the eyes are low relief, either pinched up or appliqued, as is 
the mouth. The features are generally outlined by painted lines or, 
when decoration is incision, by incised ones. 

The majority of the figurines are hollow and many contain small 
pellets that make them rattle. Then* restriction to cemeteries is an 
indication that they held some ceremonial significance. 

Two small figurines were collected at the Camutins cemetery (J-15, 
Md. 1) by the 1949 expedition. One (pi. 79, 6) is somewhat pear- 
shaped, with a constriction just above the middle dividing the body 
from the head. Except that the body slopes outward in front in two 
low bosses, there is little relief indication of anatomical details. The 
face is well-modeled, with low protuberances for eyes and mouth, 
higher ones for nose and ears, and a high-peaked headdress with a 
horizontal perforation thru the tip, perhaps for suspension of the 
figurine with a cord. The surface is completely covered with a white 
slip. The eyes, nose, and mouth are outlined conventionally with 
black paint, which is also used to execute a simple design on the 
body and to depict the arms. The hair at the sides is black and the 
painted line extends over the ears. A red line runs from the forehead 
over the top of the headdress and down the back, where it joins a 
solidly painted red area at the back of the head. The bottom of the 
figurine is also painted red. Height is 9.7 cm., base 6.2 (front-back) 
by 4.5 cm. It is heavy and probably solid construction. The caboclo 
who found it said it was with a burial urn. 

The second figurine (pi. 79, a) was found on the surface near the 
top of the east end of J-15, Mound 1. It is the same height as the 
one just described, but different in execution and has a number of 
small pellets inside it that produce a loud rattle when it is shaken. 
The head, once again, is large in proportion to the body, and on top 
of it is a headdress that looks something like a modern lady's "pill- 
box" hat. The body increases in diameter from the neck to the 
base, to which the outturned legs give a semicircular outline, flat 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



377 



in front. The arms are raised to the sides of the head, and one 
joins the body noticeably higher than the other. The surface has 
suffered from exposure, but there is no indication that the facial 
features were ever as prominent as in the first figurine. Except on 
the bottom, the surface is white-slipped and painted with a pre- 
dominantly rectilinear design in red. Height is 9 cm., base 7.2 (side- 
side) by 5.4 cm. A corner of the headdress is broken off, showing 
the paste to be dark gray. 





CM 



Figure 135. — Marajoara Phase jjottery labrets from J-15, Mound 1, surface. 
Reconstruction is based on a complete stone specimen from Panellas. 

Labrets. — ^There are three objects, two from the surface of J-15, 
Camutins, Mound 1, and one from Panellas, that may have been 
labrets. The Panellas example is the only complete one, and also 
the only one of stone. It is translucent, gray-green nephrite and 
divided into two unequal parts. The longer one is a concave-sided 
cylinder, expanding toward both ends. Issuing from the greater ex- 
pansion is a slender projection with a rounded tip (Palmatary, 1950, 
pi. 105, k and p. 280). One of the ceramic specimens (fig. 135, a) 
has an identically shaped base and shows a break where the slender 
"point" is attached on the Panellas one. The existing part is 3.5 
cm. long, 1.8 cm. in diameter at the large end, and 1.5 cm. at the 



378 BUREAU OF AMERICAN ETHNOLOGY [bull. 167 

small end. The surface is light-gray, well smoothed, and ornamented 
with paired incised lines at the region of greatest diameter. 

The other specimen from Camutins, Mound 1 (fig. 135, b) is con- 
siderably larger and had no projection from the wide end. The 
sides flare out to the base, which is 2.6 cm. in diameter. The opposite 
end, which is wider, has been broken off. The surface has been super- 
ficially smoothed, and remains uneven but not rough. Length is 
4.1 cm. The manner of breakage indicates that the flanges were 
modeled with additional clay on a basic cylinder. 

Spindle whorls. — Cylindi-ical to round ceramic objects, some closely 
resembling the spindle whorls from Colombia, have been found from 
time to time, notably at Pacoval. The majority are crude, but a 
few are well made and carefully decorated. All have lengthwise per- 
forations through the center. Of 15 examples in the American 
Museum of Natural History collection from Pacoval, 1 1 are solid and 
cylindrical with straight or slightly concave sides (fig. 136, c-e, pi. 80, 
d-f). Length varies from 3.5 to 5.3 cm. and diameter from 3 to 4.2 
cm. Four of these have incised decoration and one has crudely ap- 
plied, applique bands. The remaining 4 are hollow and have the 
maximum diameter at the center, from which they taper toward both 
ends. This type is more carefully made than the solid ones and is 
decorated with delicate, incised lines (fig. 136,/-^; pi. 80, a-c). Length 
is from 3.7 to 5.4 cm. and diameter 3.5 to 5.0 cm. Two have small 
pellets inside, producing a rattle. A wimilar object, found by Hartt 
at Pacoval (1871, fig. 72, h, and p. 270), is 5.7 cm. long and incised in 
the paired-line style identified with Guajara Incised. Two of the 
concave-sided, cylindi-ical type, one plain and one with incised dec- 
oration, are in the Peabody Museum, Cambridge, collection from 
Pacoval (Palmatary, 1950, pi. 105, 1 and m). 

Another type of spindle whorl is in the shape of a ilat disk. Hartt 
describes one of these from Pacoval made from a sherd: 

It has a diameter of 8 cm. and appears to be made of the flat bottom of a 
broken vessel, reworked to give its present form and perforated after the clay 
was fired and probably after the vessel to which it belonged was broken. The 
outline is not a perfect circle and the hole slants a Uttle. [Hartt, 1885, p. 59 and 
fig. 12.1 

One surface is ornamented with an incised, interlocking spiral. 

Farabee (1921, p. 148) says he found only one object in all his ex- 
cavations that was identifiable as a spindle whorl. He probably refers 
to a specimen now in the University Museum, Philadelphia, collec- 
tion listed as from Camutins: a flat disk with smoothed but undeco- 
rated surfaces, a flat edge, and a hole pierced approximately through 
the center (pi. 80, g). The diameter is 5.2 cm. His field notes on 
the Fortalezia group, however, also mention a whorl fragment from 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



379 









11 1^ 


. ;i A 






I I 

I CM 





Figure 136.— Marajoara Phase pottery spindle whorls, a, J-15, Mound 1- 
b, J-14, Mound 1. c-g, Pacoval. 



380 BUREAU OF AMERICAN ETHNOLOGY [Bull, le? 

trench 5, Mound 7 (the cemetery). A similar specimen, found by us 
on the surface of J-15, Camutins, Mound 1 (fig. 136, a), is 5.8 cm. 
in diameter and 1.4 cm. thick at the center, tapering down toward 
the edge. 

The surface collection at J-14, Mound 1 (Guajara) produced a 
spindle whorl 5.4 cm. in diameter, but considerably thicker than the 
examples just described, measuring 2.3 cm. at the center and 1 cm. at 
the flattened edge (fig. 136, 6). The upper surface is ornamented 
with an overall application of fine, irregularly spaced punctates. The 
hole in the center is 1 cm. in diameter and was punched through from 
the ornamented side leaving a pronounced ridge around the exit on 
the opposite face. The paste is hard, compact Inaja Plain. 

Also from Guajara are the only examples of clay with cord impres- 
sion recorded from Marajoara sites (pi. 80, h-j). They consist of 
lumps of clay 2.0 to 2.5 cm. thick, with a slight curvature. Two of 
the three are smooth on the interior surface and two narrow abruptly 
at one edge to 7 mm. Junius Bird, American Museum of Natural 
History, examined the cord impressions and reported: 

Plasticene imprints . . . show what I believe is nothing more than a cord 
wrapping done while the clay was still damp. It certainly is not the impression 
of basketry or matting, nor do I think it possible to duplicate such markings with 
a cord-wrapped paddle. On the specimen where the impressions are spaced, the 
cord was slightly over 2 mm. in diameter, was of 2-ply construction, S spun and 
Z doubled, with three twists per 2 cms. Between the cord impressions, the clay 
was extruded and slightly smoothed off. 

The other specimen shows the cord impressions in contact with each other 
except at one place where they cut deeply into the soft clay. In that instance 
the cord is again 2-ply, S spun and Z doubled, with five twists per 3 cm. On 
one side the cord seems to have been over narrow leaves and failed to leave any 
impression on the clay. [Pers. corres., 1949.] 

Spoons. — There exist in the museum collections a number of objects 
of consistent shape and small size that may have been spoons or dip- 
pers (pi. 81). They are oval to circular, with a short stem at one 
end, which is pierced with a small hole. This hole is fine, but ex- 
periment showed it to be sufficienth^ large for the insertion of a stick 
strong enough to function as the handle of a spoon. Use as a pipe 
seems ruled out by the attachment of the stem at or just below the 
rim in all but two examples, which would be at or above the tobacco 
level and thus prevent a satisfactor}'' di-aw. In the 12 specimens in 
the American Museum of Natm-al History collection from Pacoval, 
the bowl ranges from 3.2 to 6.5 cm. in length and from 2 to 4 cm. in 
depth. Two have a crudely incised design, one is ornamented with 
a zigzag applique strip, and the remainder are plain. Of the four 
examples in the University Museum, Philadelphia, collection, the 
largest has a bowl 7.5 cm. long and is ornamented with incised lines 



"v^Nsf^ *^^ ARCHEOLOGY AT MOUTH OF AMAZON 381 

and a small adorno at the end opposite the stem (pi. 81, a). A plain 
specimen (pi. 81, e) was found at J-15, Mound 17, cut 1, level 60 to 
75 cm. The stem is attached at the base of the bowl, which is oval, 
5.1 cm, long and 3.4 cm. deep. 

Stools. — As unique and relatively abundant ceramic objects, stools 
are second only to tangas. The American Museum of Natural His- 
tory collection from Pacoval contains enough complete specimens to 
give some notion of the variation in shape and size. All have the same 
basic form, in which a clay disk is attached to a flaring, annular base, 
inset at the point of attachment and with a diameter typically about 
2 cm. smaller at its base than the disk (pi. 83). The majority are 
circular, but occasional specimens are oval. The disk surface ranges 
from flat to marked concavity and sometimes has a hole through the 
center, which is usually small, but may have a diameter equal to half 
that of the stool top. On 14 measurable specimens, the disk diameter 
ranges from 10.6 to 20.0 cm., but only 5 are under 15 cm. Two of the 
14 are oval. Height runs from 1.5 to 8.0 cm., and is generally cor- 
related with the size of the disk. All except 5 are m.ore than 5 cm. 
tall. The thickness of the disk is typically between 1.5 and 2.0 cm. 

At Pacoval, the majority of the stools are decorated and the tech- 
niques represented are Anajas Plain Incised, Anajas White Incised, 
Ararl Plain Excised, Arari Red Excised, and Arari Double-slipped 
Excised. One fragm.ent has an adorno at one edge of the disk, and 
another probably existed on the opposite side. A stool with painted 
decoration is in the United States National Museum (pi. 84, a). Al- 
though no provenience is given, the early date of collection makes it 
probable that it comes from Pacoval. It is 18.5 cm. in diameter and 
11.5 cm. high. The disk edge has an undulating applique band and 
small knob adornos, and there is a small perforation through the center. 
Nimuendaju found fragments of painted stools at Teso das Igagabas 
(Ryd^n, MS.). 

In his excavations in Mound 7 (cemetery) of the Fortaleza group, 
Farabee (1914) ran across a number of stools. Trench 6 produced 
half a dozen and trench 8 a group of nine. The majority are in the 
University Museum collection, and are typical in ail respects. Deco- 
ration is with simple, incised motifs on an unslipped surface. Several 
have stylized, anthropomorphic faces on the side of the base, with the 
eyebrows, nose, eyes, and mouth in low relief and outlined with an 
incised line. 

The Camutins group (J-15) is represented by 4 fragments of plain 
stools (2 with disk-edge adornos and 1 with low relief ridges on disk 
and base) and one of Anaja Plain Incised from the surface of Mound 1, 
and half of another with Guajar4 Incised decoration from the lowest 
level of Mound 1, cut 2. The latter had a disk diameter of 22 cm., a 



382 



BUREAU OF AMERICAN ETHNOLOGY 



[BULIi. 167 



height of 5 cm., and was markedly concave on top. There is a very 
small adorno at one spot on the edge of the disk. A complete stool 
(pi. 83, a) with a well smoothed surface but no decoration came from 
just below the surface at the top of Mound 10. The flat disk is 19 
cm. in diameter and the height, 5 cm. (For further illustrations, see 
Netto, 1885, pp. 395-397; Palmatary, 1950, pi. 25 c, d, e; pi. 28 g; 
pi. 77 c). 

Tangas. — Probably the most distinctive of the objects found in 
Marajoara sites are the tangas. Trianguloid, with an upper convex 
edge joining the other two concave edges in more or less sharp points, 
and mth the third intersection at the bottom broad and rounded, 
they do not notably differ from the pubic coverings used by the women 
of many of the Tropical Forest tribes today (see Levi-Strauss, 1948, 
fig. 33; Schmidt, 1942, figs. 239-242) except in the material of their 
manufacture, which is pottery. In general size, proportions, blunt- 
ness or sharpness of the points, and curvature of the edges and sur- 
faces, there is a range of variation that is probably correlated with 
differences in the anatomy of the wearers. A hole for the attachment 
of a belt cord is pierced from 1.9 to 3.4 cm. from each tip, depending 
on whether the point is slender or wide. The amount of wear produced 
by the friction of the cord on the exterior surface varies from none to 
a deep groove extending all the way from the hole to the tip (cf. 
Hartt, 1876, pp. 22-23). A tabulation of 110 tanga tip fragments 
with perforations from J-14 and J-15 gives the following percentage of 
degree of wear (Table P). 

Table P. — Differences in wear on tanga fragments 



Type 


No wear 


Slight notch 


Deep groove 


Total 


Red -slioped 


Number 
27 
18 


Percent 

< 32. 6 

66.6 


Number 
34 
8 


Percent 
40.9 
29.6 


Number 

22 

1 


Percent 
26.5 
3.8 


Number 
83 
27 


Percent 
100 


Painted R/W 


100 



This may indicate that in late Marajoara times at least the red-on- 
white type was of predominantly ceremonial significance, while the 
plainer, red-slipped tangas received greater use. 

Seven complete specimens, 6 of them red-sUpped and 1 red-on-white, 
were recovered from burial urns in J-14 and J-15 cemeteries. Five 
additional red-on-white tangas were given to us by the caboclo Uving 
on the Camutins cemetery (J-15, Mound 1), who had found them in 
his own digging in the site. Measurement of these makes it possible 
to give specific figures that illustrate the variation in size and con- 
vexity. The six standardized measurements made are shown on 
fig. 137. 



MEGGERS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



383 




FiGUBE 137. — Standardized measurements on tangas of the Marajoara Phase. 
Table Q. — Standardized measurements on tangas 



Measurement 


Red-slipped 


Red -on-white 


A-B - 


Cm. 

12.0,13.4,14.0,15.0,16.5,16.5 

10.5,11.5,12.0,12.0,12.0,12.0 

3.0,3.0,3.2,3.5,4.0,4.5 

0.5,0.7,1.1,1.2,1.2,2.0 

2.0,2.5,2.6,3.0,3.0,3.4 

2.5,2.5,2.5,3.0,3.5,3.6 

0.5 


Cm. 
13. 0, 13. 5, 14. 0, 14. 0, 14. 0, 14. 5 


C-E 


9. 5, 9. 9, 10. 0, 10. 2, 10. 5, 10. 5 


D-F... 


4.0,4.0,4.0,4.2,4.3,4.7 


G-H 


0.5.1.4,1.5,1.5,1.6,2.0 


I-E 


2.0,2.3,2.5,2.7,2.8,3.0 

2.0,2.5,2.5,2.5,2.8,2.8 

0.5 


K-a 


Thickness (all) 







The major difference between the two types is in the amount of con- 
cavo-convexity of the surfaces, which is more pronounced in the red- 
on-white examples and accounts for the differences in dimensions D-F 
and K-G. Thickness is the most standard aspect and in about 90 
percent of the sherds runs between 4 and 7 mm. The upper limit 
reaches 11 mm. in occasional examples. 

The two varieties of tangas have a time distinction as well as a dif- 
ference in shape. The red-slipped type (pi. 82, d-e) is in the minority 
in the earlier sites, but becomes equal in popularity at Camutins and 
dominant at Guajara. This transition is probably related to the 
trend toward abandonment of the more complex and precise types of 
ceramic decoration that is characteristic of the ceramic history in 
general. 

Although the red-slipped tangas are often slipped on both sm"faces, 
the painted ones are generally slipped only on the exterior or convex 
sm-face, with the slip carried over onto the interior in a band along the 
edge 4 to 9 mm. wide. The design is composed of fine, single, red 



384 BUREAU OF AMERICAN ETHNOLOGY [bull, le? 

lines or of paired red lines with the narrow, intervening space colored 
lighter red or orange. Across the top it is typical to have 2 bands or 
"friezes," each 1 to 2 cm. wide, which bear standardized motifs. 
The upper one is composed of three vertical and two diagonal lines 
placed equidistantly and separated by solidly painted triangular areas. 
This motif has been pointed out by Mordini (1934a, pp. 62-63 and 
1929) as characteristic of the majority of the painted tangas, and it 
was present on all the painted specimens from J-14 and J-15 (pi. 82, 
a-b) . There is greater variation in the second band, but here too there 
is repetition of a relatively small number of designs. It is possible 
that these designs had some symbolic significance of a social or reli- 
gious nature, wearers of the same pattern belonging to the same group. 
The remainder of the surface bears a symmetrical and graceful recti- 
linear design built upon another limited number of motifs, which are 
almost identical in the 6 painted specimens from J-15. The range of 
latitude in this part of the design appears to have diminished with the 
passage of time, since considerable variation is present on examples 
from Pacoval. 

While the function of tangas as pubic coverings by females cannot 
be proved, this conclusion fits the evidence of wear at the perforations 
and of ethnographic parallels. In most of the vessels containing a 
tanga, the sex of the individual was either unidentified or female. 
However, Newman's identification of the bones from jar M of J-14, 
Mound 1, cut 1, suggests a possible association with a male skeleton. 
Since the same vessel also contained a female, this tanga may have been 
displaced dm-ing or subsequent to the burial. If it is true that tangas 
occur only with females, then the important individual in jar L from 
the same site must be a female, which suggests an extremely high 
status for certain individuals of that sex. A further complicating 
factor is the exceedingly high percentage of fragments on burial 
mounds and their relative rarity in habitation sites. This situation 
would seem to imply a dominantly ceremonial significance for these 
objects, and has suggested to several students the possibility of a 
fertility cult (Netto, 1885, p. 436; Palmatary, 1950, p. 282; Angyone 
Costa, 1941). 

Whistles. — Tocantins (1876, p. 54) describes "a kind of whistle 10 
cm. long, hollow, mth two holes of unequal diameters, and ornamented 
■with relief spirals and other adornments," which apparently came 
from Pacoval. Holdridge (1939, p. 73) found "many little clay whis- 
tles in the form of birds . . . usually capable of three or four notes" 
at Monte Carmelo and similar objects at Laranjeiras (op. cit., p. 71). 

Miscellaneous. — Pottery polishers and net weights of stone are said 
by Ferreira Penna (1879 a, pp. 53-54) to have been found by him at 
Pacoval and Santa Izabel. A cubical piece of clay with rounded 



HEGGBRS AND 
EVANS] 



ARCHEOLOGY AT MOUTH OF AMAZON 



385 





Figure 138. 



-Marajoara Phase grooved polishing stone from J-15, Mound 1, 
surface. 



edges, perhaps a polishing tool, comes from Panellas (Palmatary, 
1950, pi. 105, j). Aflat piece of diorite (fig. 138), 1.5 cm. thick and 
roughly rounded off on the edges, with 4 grooves 5-6 mm. wide and 
worn 4-5 mm. deep on one surface, was found at the Camutins ceme- 
tery (J-15, Mound 1). A Camutins Plain sherd showing similar wear 
is in the University Museum collection. 



CERAMIC HISTORY 

StratigTaphic excavations were made in four mounds of sites J-14 
and J-15 in an effort to provide a temporal basis for the analysis of 
Marajoara Phase pottery. The problem was complicated by two 
facts: intrusive bmials distm-bed the natural sequence of deposition 
of sherd refuse in the cemetery mounds, and sherds were sparse in the 
habitation momids in spite of indications of abundance on the eroded 
slopes. After examination of the percentage distributions shown in 
the strata cuts, it was decided that the trends exhibited by cut 1 of 
J-15, Mound 14 (Inajasal) were m.ost likely to be reliable because the 
refuse accumulation was deepest and, since this is a habitation site, 
the disturbance was likely to be minimal. This stratigraphic sequence 
shows the temporal relationship of the two plain wares (fig. 139). 
Inajd Plain, a sherd-tempered, pale orange-surfaced ware with a gray 
core, is at the peak of its popularity, 76.2 percent, in the lowest level 
of the excavation. Its subsequent history is one of decline to 16.9 
percent injevel 15 to 30 cm. During the same time there was an 
increase from 19.0 percent to 82.9 percent in the frequency of Camutins 



386 



0-15 M 
.15-30 
30-45 
45-60 
,60-.75 
.75-. 90 
,90-1.05 
1.05-1.35 
1.35-1.50 
1.50-1.65 
1.65-1.90 
I.SO-1.95 
195-Z.IO 
2.10-2.25 




BUREAU OF AM 


[ERIC AN ETHNOLO 


GY 

1 

1 
■ 

■ 
■ 
1 

JOANES 
PAINTED 


[BULL. 

ANAJAS 
INCISED 


167 










■^I^I^B 


































CAMUTINS PLAIN 




1 N A JA PLAIN 











10 


20% 













Figure 139. — Ceramic stratigraphy of J-15, Mound 14, cut 1, showing trends 
in the Marajoara Phase plain wares (Appendix, table 39). 

Plain, also sherd tempered, but with a bright red-orange surface and 
core. Although the level to level occurrence of these two types is 
somewhat erratic as a result of the small sample per level (Appendix, 
table 39), the general trend is clear. 

The same decline in Inaja Plain and increase in Camutins Plain 
are shown in cut 2 of J-15, Mound 1 and cut 1 of J-15, Mound 17. 
Although these are both burial mounds, no vessels were encountered 
in the immediate excavation and so the refuse accumulation was not 
likely to have been grossly disturbed. In order to minimize dis- 
tortions resulting from the small samples per 15-cm. level, these 
were combined into divisions of 30 cm. (Appendix, table 40). When 
the two cemetery sequences are interdigitated (fig. 140), they cover 
approximately the same time span as Mound 14, cut 1. The two 
upper levels of J-14, Mound 1, cut 1 were seriated in this chart in 
order to give some indication of the relative position of this burial 
mound. The reliability of this seriation is dubious, however, because 
this strata cut produced a quantity of bm*ial m'us, and the dirt and 
sherd refuse must have been disturbed repeatedly. The only reason- 
ably reliable conclusion that can be drawn is that this cemetery is 
generally contemporary with those of J-15, but not necessarily with 
the first half of the sequence rather than the second half. 

The establishment of the decline in frequency of Inaja Plain and 
an increase in Camutins Plain as the predominant trend in Alarajoara 
Phase plain wares introduced a basis for the seriation of surface 
collections from other sites. Before this could be done, however, a 
means had to be found for reducing the potentially disturbing fac- 
tors of a small sample and selectivity for decorated sherds. Although 
the decorated types exhibit trends during Marajoara Phase history, 



CL 



CL 
CI 
CL 

cU 
cu 

CL 



CL 



CAMUTINS 
MD. I 
CUT 20-30M 



2= 30-60 
2\60-.90 
CUT I 0-.30 
CUT 3 0-30 
CUT 2 = .90-l.20 
CUT I0-.30 
CUT 2 1.20-1.50 
CUT 1^30-.60 
CUT 3 = . 30-. 60 
3 .60-. 90 
CUT I .30-60 
I .60- 90 
\ .90-1.20 



BELEM 



GUA- 
JARA 



> 



Lzi 




i / / / / A 



\/ //////// /-/-z-x 



\////////A 



////////A 



\/ / / A / A / / / y / / \ 

\///////// ///y /-r-7- 

I 

INAJA PLAIN 



V / / / / Z^ 

I 

CAMUTINS 

PLAIN 

I I I 



10 20 30% 



ARARI 
PLAIN 

EXCISED 



ca 



5 10 15% 



□ 



CZI 
D 



ARARl' ARARl' ANAJAS ANAJ^ls AN. GUAJARA GOIAPf CARMELO 
RED D-S PLAIN RED D-S INCISED SCRAPED RED 

EXCISED EXC. INCISED INCISED INC. 



Figure 140.— Ceramic seriation of the Marajoara Phase sites of J-16, Mounds 1 (Camutins, Mound 1) and 17 (Beldm), 
and J-14, Mound 1 (GuajarA) (Appendix, table 40). 



rz-?^ 



ca 



I I 

\/ / A 

V / / / 



EVANsr ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 387 

they could not be used in establishing a preliminary seriation of 
sites because the majority of the collections are too small to insure 
the inclusion of the rarer types. Furthermore, decoration is so rare 
on the pottery from habitation mounds that what is found is not 
likely to be illustrative of the total techniques in use at the time. 
The seriation had to be carried out on the basis of the plain wares, 
therefore, but this was complicated by the fact that most of the 
collections of known provenience were highly selected for decorated 
types. The analysis and classification of the decorated wares sug- 
gested a way out of this difficulty. It was impossible to discern 
any association of decoration with paste characteristics that would 
indicate a conscious selection of Inaja Plain for one kind as against 
Camutins Plain for another. The paste characteristics of each dec- 
orated type seemed to reflect instead the relative proportions of the 
plain wares in existence at the time, and altered temporally rather 
than in terms of decorated type. In other words, Pacoval Incised 
sherds are predominantly gi'ay cored, not because there was an inten- 
tional association of these features of decoration and firing by the 
potters, but because Pacoval Incised is an early type and was made 
when Inaja Plain was the predominant plain ware. Types like 
Anajas Plain Excised that extend over the entire life of the Phase 
are gray cored at early sites and orange cored at late ones, reflecting 
the rise and fall in popularity of the basic plain wares. 

Since there appears to be no correlation between type of decoration 
and the kind of plain ware on which it was placed, it seems reasonable 
to assume that the paste characteristics of the decorated sherds 
reflect the relative proportion of the plain wares being made at the 
time. On this basis, decoration and surface treatment were ignored 
and the sherds classified by their cross-sectional features — gray core 
was Inaja Plain and orange core, Camutins Plain. In addition to 
the 4 cemeteries we investigated, there are samples from 12 other 
sites that could be used for seriation (Appendix, table 41). Incor- 
poration of these gives a sequence beginning with Pacoval dos Mello, 
with 92 percent Inaja Plain and onl}^ 8 percent Camutins Plain, and 
culminating in Furinho, where Inaj4 Plain has dropped to 37 percent 
and Camutins Plain has increased to 63 percent (fig. 141). It will 
be noted that two collections from Pacoval are included, and that 
there is a difference of 11 percent in the frequency with which the 
pottery types are represented in them. Since the collections are 
almost equal in size (307 and 313 sherds), and large enough to give 
a reasonably accurate result, there is only one explanation that seems 
to account for this discrepancy. The later of the two collections 
in the seriation was made in the 1870's while the earlier was made in 
1950. It may be that the frequent looting to which Pacoval has 



388 BUREAU OF AMERICAN ETHNOLOGY [bcll. 167 



FURINHO 

J-15 MOUND 17 

J-14 MOUND I 

J-15 MOUND I 

J-14 MOUND 2 

TESO DO GENTIL, MOUND I 

SANTA BRIGIDA 

TESO DOS CHINA, MD. I p 

ILHA DOS BICHOS | 

CUIEIRAS I 

TESO DOS CHINA, MOUNO 2 
F0RTALE2A 
PACOVAL (PEABODY) 
TESO DOS CHINA, MOUND 4 
TESO OE SEVERINO 
PACOVAL (HILBERT) 
CARATATEUA 
PACOVAL DOS MELLO 




10 20 30% 




INAJA 
PLAIN 



FiGUEE 141. — Seriation of Marajoara Phase cemetery sites based on relative 
frequency of Inajd, Plain and Camutins Plain (Appendix, table 41). 

been subjected has removed a gi-eater proportion of the material 
nearer the surface and what remains comes mostly from the lower 
levels. If this is the case, the two Pacoval collections should be 
interpreted as more comparable to stratigraphy than as representative 
surface samples in the usual sense. 

If this plain-ware seriation of cemetery sites is compared with the 
stratigraphic results from the habitation mound of J-15, Mound 14, 
some conclusions can be drawn about the contemporaneity of Mara- 
joara Phase sites. The changes that occur stratigraphically (fig. 139) 
correspond to the part of the seriated sequence (fig. 142) commencing 
with Cuieiras and continuing beyond the last cemetery site, Furinho. 
The J-15, Mound 14, stratigraphic sequence is approximately the 
same as the one derived from J-15, Mound 1, and its surface collection 
seriates like that of the latter site about the middle of the stratigraphic 
sequence. If the surface collections used for the other cemetery sites 
can be assumed to summarize a similar span of time, then it can be 
concluded that J-15 was a functioning conmaunity during about half 
of the Marajoara Phase occupation of Maraj6 Island, specifically, the 
latter half. The J-15 mounds were constructed during the second 
half of the period represented by Pacoval and Fortaleza, and the two 
groups of sites were contemporaiy for a short time. When the 
seriated sequence is considered in geographical terms, it is evident 
that the earlier sites are east of Rio Ararl and the later ones west of it. 



EV^Nsf^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 389 

This constitutes a movement from the open campo in the east to the 
forest and campo west of the Arari. J-15 is at the margin between 
the campo and the solidly forested western part of Maraj6 Island. 

The 16 habitation mounds of J-15 producing a sufficiently large 
surface sample were seriated on the possibility that this might show 
whether they were in use at the same time, or whether they represent 
successive house sites (fig. 142). The analysis produced a continuous 
variation between Mound 5 with 57 percent Inajd Plain and 37 
percent Camutins Plain, and Mound 4 with 38 percent Inaja Plain 
and 60 percent Camutins Plain (Appendix, table 42). This seems to 
indicate that the mounds were generally contemporary and composed 
a village that stretched for several Idlometers along the river bank. 
The higher numbered mounds, which are those toward the upper part 
of the Igarape Camutins, tend to be in the lower half of the sequence, 
suggesting that they were abandoned slightly before those farther 
down stream. This is in accord with the seriated position of Cuieiras 
as the earliest of the Camutins cemeteries, since Cuieiras is not far 
above these habitation mounds. 

A comparison of figure 139 with figure 140 reveals distinctly the 
basic difference between habitation and cemetery refuse. Decorated 
sherds account for not more than 8 percent of the domestic pottery in 
any one level, and J-15, Mound 14 produced only painted and incised 
types. In the cemetery cuts, decorated pottery is both more abundant 
and more varied in technique. Contraiy to the impression given by 
previous ceramic collections, however, decorated types are in the 
minority even in cemetery sites. They comprise only 9 to^l5 percent 
of the sample from the levels of J-15, Mound 17, cut 1, and 8 to 35 
percent of that from J-15, Mound 1, cuts 1 to 3, with the majority of 
the levels producing about 25 percent decorated sherds. It is a rare 
level in which this is not composed of 50 percent or more Joanes 
Painted, which reduces the remaining 13 types to a very low frequency. 

Since there appeared to be no association between decorative 
technique and paste characteristics except one reflecting the relative 
popularity of the plain wares at the time of manufacture, classification 
of decorated types was made purely on the basis of decorative tech- 
nique. It was found that 7 types of decoration were employed in 
combination with one or more of 4 types of surface treatment, includ- 
ing 3 kinds of slip. Fourteen of the possible combinations^were utilized, 
giving 14 decorated pottery types. 

Joanes Painted includes all techniques of painted decoration, 
whether red-on- white, black-on-white, or red-and-black-on- white. 
In addition, any bowl sherd with a white-slipped interior or jar sherd 
with a white-slipped exterior was given this classification, even when 
no trace of paint remained. Although this may seem unwarranted, 

391329—57 27 



390 



BUREAU OF AMERICAN ETHNOLOGY 



[BDLL. 167 



a: o 


73 


< X 

UJ 


03 




fl 








03 








(^ 




•^ 








a 












h- 1 


(A O 


<4-l 


< UJ 




3 W 


>, 






2 O 


n 


< 2 


V 




p 




c 



o< 
-'a 



II 






■4^ o3 

o ^ 

cQ in 

■^ & 
to 2* 

O M 

•■:3.S 



SCO 

OS 
22 



"Sf^ ^^"^ ARCHEOLOGY AT MOUTH OF AMAZON 391 

experience derived from handling hundreds of decorated sherds indi- 
cates that white sHp is almost universally, if not always, applied as a 
foundation for a painted design, and the majority of sherds where the 
paint is no longer visible have been subjected to erosion from exposure 
to rain and sun or to wet, clayey soil. The attempt was made to dis- 
criminate between monochrome and dichrome painting in the hope 
that some time difference would emerge, but so few sherds retained 
enough paint to permit classification that the attempt was abandoned 
as impractical and not hkely to give reUable results because of the 
frequency of eroded surfaces. 

Six of the decorated types utilize incision. Guajard Incised involves 
the use of two (occasionally three) parallel lines 1 to 5 mm. apart, made 
simultaneously with a double-pointed tool (pi. 71). The surface is 
always unslipped. In another distinct incised style, the surface is 
white-slipped and incised, and then certain of the incised lines are 
painted or "retouched" with red. Since this type occurs with greatest 
frequency at Pacoval, it has been called Pacoval Incised (pis. 77-78). 
The final incised category, Anajds Incised, includes all other designs 
produced by incision. There is great variation in motif, quality of 
execution, width and depth of the lines, etc., but no design subtypes 
are sufficiently distinctive to be easily recognized, as can be done with 
Guajard and Pacoval Incised, Subcategories that are significant from 
a time standpoint are those made on the basis of slip. This results in 
the breakdown into Anajas Plain Incised (pis. 51-52), Anajds Red 
Incised (pi. 53), Anajds White Incised (pis. 54-55), and Anajds Double- 
slipped Incised (pi. 50). 

Another large and inclusive group is that containing the excised 
types. Since there is wide variation in the amount of surface cut out 
to produce the design, the term "excised" is used in preference to 
"champleve." Not all excised sherds could validly be called cham- 
pleve, and since the same technique is involved no matter what per- 
centage of the original surface is removed, all sherds on which excision 
occurs must be considered as fundamentally related (p. 325). The 
extremes of quality in execution in this category are great and easily 
distinguished, but when an attempt is made to subdivide on the basis 
of crudcness of workmanship or proportion of the area excised, the 
gradation is so complete that the residue of borderhne examples of 
uncertain classification is larger than the distinctive, classifiable group. 
As was true with Anajas Incised, the most significant breakdown in 
the excised class is by slip, giving Arari Plain Excised (pis. 57-58), 
Arari Red Excised (pis. 59-62), Arari White Excised, and Aran Double- 
slipped] Excised (pi. 56). In some red-slipped and excised examples, 
the cutout area has been filled with white, giving a further type, 
Arari Red Excised, White-retouched (pi. 63). 



392 BUREAU OF AMERICAN ETHNOLOGY [bull, ler 

Scraping as a method of decoration has been ahnost completely 
ignored by students of Marajoara ceramics, and it was somewhat of a 
surprise to find it not only present, but one of the more common 
techniques especially at the later sites. Most characteristically, it is 
used on the necks of jars as a quick method of reheving the monotony 
of the plain surface, and is done by combing or scraping vertically. 
This type has been called Goiapi Scraped (pi. 70). 

The final basic group includes sherds that have a red slip but no 
incised, painted or excised decoration. Since the slip was apparently 
applied for ornamental effect, and since red-sUpped vessels are associ- 
ated with decorated wares in cemetery sites rather than with plain 
domestic wares, Carmelo Red has been included with the decorated 
wares. 

Some idea of the relative frequency and temporal distribution of 
these decorated types can be gained from fig. 140. In order to mag- 
nify their very small percentages, the decorated types were plotted at 
twice the scale used for the plain wares. Some of the types are 
absent, and some of those present show no particular trend or change 
in frequency, notably Joanes Painted, Arari Plain Excised, and 
Ana j as Plain Incised. Ararl Red Excised undergoes a slight but 
steady decline. Anajas Double-slipped Incised seems to be early and 
Guajara Incised and Carmelo Red occur only in the upper half of the 
sequence. The remainder of the types have scattered distributions 
from which no definite conclusions can be drawn. 

In spite of this relatively indistinct picture, certain trends in the 
employment of various techniques can be recognized. The rare, 
declining or absent types are Pacoval Incised, Arari Red Excised, 
Arari Red Excised White-retouched, Arari White Excised, Arari 
Double-shpped Excised, Anajas Red Incised, Anajas White Incised, 
and Anajas Double-slipped Incised. Those that are abundant or 
increasing are Ai-arl Plain Excised, Anajas Plain Incised, GuajarS, 
Incised, and Goiapi Scraped. The first group, which is on the decline, 
is composed of types where the surface is provided with one or two 
slips before the execution of the incised or excised design. In two 
cases, this is followed by a "retouch" of the incised or excised areas 
with a contrasting color. These complex and elaborate methods of 
decoration gradually lose ground to incised and excised designs 
applied to an unslipped surface. The late types that occur with the 
greatest frequency all share the characteristic of an unslipped surface 
and the use of a single step in producing the decoration, whether this 
is incision, excision, or scraping. 

The time span represented by this cemetery and habitation stratig- 
raphy is not sufficiently long to give a good pictiu-e of the trends in 
decoration during Marajoara history, and it became desirable to 



Ev^A^Nsf^ ^^"^ ARCHEOLOGY AT MOUTH OF AMAZON 393 

carry the sequence backward by the addition of collections from 
other Marajoara Phase sites. This meant utilizing surface collections 
that had been selected for decorated tj^pes. In order to compare the 
frequency of the decorated types, a method had to be found for elimi- 
nating the distortion that would result from the inclusion of varying 
amounts of plain sherds, which had no relation to the actual propor- 
tions at the site. This was done by considering the total number of 
decorated sherds as equal to 100 percent and eliminating plain sherds 
from the total and from the percentage calculation (Appendix, table 
43). This method could be used for previous collections from only 
two cemeteries, Pacoval and Fortaleza, since the samples from others 
were too small to insure the inclusion of rare types that might have 
been present. 

The collections from J-14, Mound 1 (Guajard), and J-15, Mound 1 
(Camutins) were analyzed in the same way. When the 4 sites are 
placed in the temporal sequence indicated by the proportions of their 
plain wares (fig. 141), this enlargement of the time span (fig. 143) 
confirms and magnifies the trends revealed in the stratigraphy. The 
complex decorated wares that are rare or absent at J-14 and J-15 
are frequent to abundant at the earlier cemeteries, while the simpler 
ones are less common. Although the graph does not show it, com- 
binations of two or more techniques on a single vessel are also most 
frequent at the earlier sites. This partly accounts for the lesser per- 
centage of Joanes Painted at Pacoval, where painted decoration 
frequently occurs on vessels bearing incised or excised designs on 
another surface and was somewhat concealed by the method of classi- 
fication (see p. 325). Recognition of this leads to the conclusion 
that painted decoration is most common and best executed in the 
earlier part of the sequence. 

Examination of the details of execution of the designs gives other 
evidence of decline in ceramic quality. Excision, for example, is 
markedly more evenly done on Arari Red Excised than on Ararf 
Plain Excised. On the former, the cutout areas are typically sharply 
defined, with straight edges, and are evenly cut back and often stri- 
ated. In Arari Plain Excised, on the other hand, the excisions are 
gouged out so that the depth is uneven and the margins are ragged. 
This gradation in technique is a gradual one and there are instances 
of poorly done Arari Red Excised, but they are less common than 
the well-made examples. Changes of this kind indicate the gradual 
replacement of painstaking work with a hurriedly made and inferior 
product. 

An interesting substitution of techniques to achieve a similar effect 
with greater economy in time and labor is observable when Pacoval 
Incised is compared with certain Anajds White Incised specimens 



394 



BUREAU OF AMERICAN ETHNOLOGY 



[BULL. 167 



If 






rf «' n 
CK kJ O 



e. — 



_ _ K o 



<ds. 



m a 



— _ tE o 



5? 


n ° 




~ 


: 


m 


: 


o 



S "* 



(h 


'^ 


o 


o 

3 


cu 


cl 


,£3 


-|J 


-tJ 


.^ 




t>< 


o 


73 





PI 


o 






0( 


^ 


o, 


■n 


<3 


a> 


^— ' 


03 


vrl 


fll 


»H 


n 


ci 


+3 


eS* 


>,3 


^o 


TJ 


T-l 




fl 


(S 


c3 





a 


>m:3 


■»-> 


u 


Tl 




<U 


iji 


+-' 




a 


(11 






o 


03 


(1) 


-e 


xs 


fii 


a; 








03 


c3 


Ph 


O 


03 




s 


(in 



.2 >> 



H o 



MEGGERS AND ARCHEOLOGY AT MOUTH OF AMAZON 395 

BVANSJ 

(e. g., pi. 78, c, and pi. 55, a). In Pacoval Incised, the designs are 
typically composed of fine lines with broader ones added between 
them at intervals and colored red, producing an attractive contrast. 
The fine lines are generally employed in two's or three's to define 
rectangular or Z-shaped fields occupied by a single red-retouched 
line, or the two kinds of lines are combined in concentric spirals. 
These distinctive motifs carry on after Pacoval Incised has died out, 
the color contrast being preserved by the less time-consuming, but 
also less aesthetically effective device of cutting the single lines (for- 
merly red-retouched) through the white slip to reveal the under- 
lying orange siu-face. This variety of Anajds White Incised (which 
might occasionally be justifiably classified as Arari White Excised) 
is prominent at Fortaleza and continues at Camutins. It is exceed- 
ingly rare at Pacoval, where Pacoval Incised was an important 
decorated type. 

The major temporal changes in Marajoara Phase ceramic decoration 
can be summarized as follows: 

1. Complex wares utilizing two or more types of surface treatment, such as 
slipping or double-slipping with excision or incision, and slipping with incision 
and painting, are most abundant at the earlier sites and decline markedly with 
the passage of time. 

2. Concurrently, the technical quality of the excised designs and the amount 
of vessel surface that they cover is notably reduced. 

3. The wares showing increases in popularity are with one exception unslipped, 
and the excised, incised, or scraped decoration is applied directly to the vessel 
surface. The exception is Carmelo Red, in which there is a red slip but no 
further decoration. In short, the types on the increase, are those that require 
the least time for their execution. 

4. Painting is common in all periods, somewhat more so in the lower than in 
the upper part of the sequence represented here. There is some indication that 
complex and delicate designs are more frequent at the earlier sites. 

One can discern the same tendency toward simplification when 
other features of the pottery besides surface treatment are examined 
(fig. 144). Hollow rims are frequent on Anajas White Incised and 
Pacoval Incised vessels from Pacoval, constitutmg 1.8 percent of 
the total classified sherds. There is only one example in the Univer- 
sity Museum, Philadelphia, collection from Fortaleza, comprising 
0.13 percent, and our excavations at J-14 and J-15 produced none 
(Appendix, table 44) . Since the manufacture of hollow rims requires 
more technical skill than the making of solid ones, the loss of this 
trait can be interpreted as indicating a decrease in that skill. 

Anthropomorphic and zoomorphic rim adornos occur with vastly 
greater frequency at Pacoval than^at any of the other three cemetery 
sites, although they do not die out completely as do the hollow rims. 
Geometric rim adornos maintain a more even popularity, and this 
seems to be true also of applique body adornos, although the evidence 



396 



BUREAU OF AMERICAN ETHNOLOGY 



[BVJA.. 167 




1 



1 



3 O 




1 



■ I 



1 



w 


o 


o o 


u 


;;: ^ 


0) 


"- tr 


^ 


*~.<~> 


+3 


•i" n 


M 


^ < 


a 


o 


i-i 


o - 


05 


9< 


(£ 


a 




o 




u 




>> • 




^v 




c*^ 


P 




'^ tr 


OQ ^ 


'^^ 


^.2 


— • < 


^"3 


^2 


03 ft 


< (E 










f"'^ 




4 k 

« c3 


5 

o tf> 


5^ 


-1 2 


v-.'O 


-1 := 


o a 


o"- 


03 

M 


X 


O r^ 




^4 




O-t^ 




73 3 




« 




1^. 


o 




o 


+= S 


a. 


Its -t^ 


<n 








52 o 




Ojf^ 




>> ~ 




o^.-^ 




-JS > 




■S o 




O o 


(/> 


ft 03 


_i 


^H. 


o 
o 


03 
P 

TO 


1- 


=3 


(/) 


PI 








]- 

Jo 





o 


o 


< 




"* 


>• 


s 


«^ 


N 




-* 


o: 


^< 


z 


UJ 


-J 




UJ 


« 'I- 


- m 


_J 


< 


s 


t- 


< T 


•- T 


< 


> 




Ul 


~> -> 


3 -5 


►- 


o 


z 


< "- 


5 — 


o: 


o 


Ul 


=} 


< 


o 


< 


c^ 


o 


o 


o 


u. 


a. 





JfvA°v«f^ ^^^ ARCHEOLOGY AT MOUTH OF AMAZON 397 

is not complete. The latter constitute 1 percent of the total at 
Pacoval and 1.2 percent at J-15, Mound 1, with no figures available 
for Fortaleza and J-14, Mound 1. 

Handles, which never reach a frequency of 1 percent on the utility 
wares, also remain approximately constant, ranging from 0.8 percent 
at Pacoval to 0.7 percent at J-14, Mound 1. 

When we consider trends in vessel shape, the analysis is complicated 
by deficiencies in the data. Rim sherds were too infrequent in the 
stratigraphic excavations to permit a statistical analysis by levels 
(Appendix, tables 45-47). The J-14 and J-15 sites are too nearly 
contemporary to exhibit any significant differences when their total 
samples are compared and there are no data from Pacoval and 
Fortaleza. There are two other means, however, that can be used to 
investigate possible temporal differences. One of these is the burial 
stratigraphy revealed in J-14, Mound 1, cut 1 and the other is based 
on the temporal position of the decorated pottery types. 

Taking the latter evidence fu-st, we can make a temporal classifica- 
tion between pottery types that tend to be early and those whose main 
distribution is late. The first group includes Ararl Red Excised, 
Arari Red Excised White-retouched, Ararl White Excised, Ararl 
Double-slipped Excised, Anaj^s White Incised, Anajds Double- 
slipped Incised, and Pacoval Incised, while the second is represented 
by Arari Plain Excised, Anajds Plain Incised, Guajard Incised, and 
Goiapi Scraped. Wlien the vessel shapes associated with these two 
groups are contrasted, it is evident that the early types exhibit shapes 
that are not found with the late types. These include cylindrical and 
semicylindrical jars (Arari Red Excised White-retouched, common 
shape 6), jars with a flat bottom and concave outsloping lower wall 
(Pacoval Incised, common shapes 7, 9, 11), jars with globular body 
and tall cylindrical neck (Pacoval Incised, common shape 8); flat- 
bottomed bowls (Pacoval Incised, common shape 4); bowls with 
hollow rim (Pacoval Incised, common shape 1); and bowls with broad, 
horizontal rim (Pacoval Incised, common shape 6). None of these 
shapes occurs characteristically with the late decorated types, and most 
of them do not occur at all. Late shapes are less varied and simpler. 
Jars tend to be globular bodied with a short vertical or slightly concave 
neck (Goiapi Scraped, common shape 1) and bowls are rounded 
(Anajas Plain Incised, common shapes 1 and 2). Along with this 
decrease in variety of shapes, appears to go a decrease in size, especially 
of jars. 

This general size decrease is shown stratigraphically in J-14, Mound 
1, cut 1 (fig. 89). The deepest, and therefore the earliest, burial jars are 
considerably larger than those that were buried afterward. This 
size decrease is not necessarily related to the change in the method of 



398 BUREAU OF AMERICAN ETHNOLOGY [bdll. is? 

disposal of the dead, since secondary burial does not require much more 
jar space than cremation. 

The decreasing frequency of some of the ceramic artifacts further 
attests to the diminishing richness of the culture (fig. 144). Stools 
decline both in relative frequency and in the variety of surface decora- 
tion applied to them. At Pacoval they constitute 3.3 percent of the 
sample and may be Inajd or Camutins Plain, or Anaj^s Plain Incised, 
Anajds White Incised, Arari Plain Excised, Arari Red Excised, or 
Arari Double-slipped Incised. At Fortaleza the frequency drops to 
2 percent and decorated types include Anajds Plain Incised, Anajds 
Red Incised, Anajds White Incised, and Arari Plain Excised, At 
J-15, Mound 1, the occurrence is 1.5 percent and all are undecorated 
except for one Anajds Plain Incised. No stools were found at J-14, 
Mound 1. 

The little stemmed vessels here identified as spoons decline from 
0.8 percent to 0.5 percent to 0.3 percent to in the sequence repre- 
sented by the four cemeteries. Spindle whorls alter from biconical or 
spoolshape with incised decoration to a simple, flat, plain or punctated 
disk. 

DIAGNOSTIC FEATURES OF THE MARAJOARA PHASE 

The Marajoara Phase has a compact distribution on Marajo Island. 
A circle described on the map with a compass, its point set at Pacoval 
in Lago Arari and its radius measured to the north coast, would in- 
clude all of the recorded sites (fig. 145). The explanation lies in the 
fact that such a circle coincides approximately with the boundaries 
of the campo, the habitat the Marajoara people preferred. The largest 
concentration of sites is east of Lago Arari, where the campo is most 
open and unbroken by trees. They are typically on the shores of 
streams or lakes, circular, oval or long and narrow in outline and with 
no consistent orientation other than that dictated by the exigencies 
of the immediate location (a curve in the river, a spring, etc.). There 
is no evidence of any intention to reproduce a zoomorphic shape, as 
has been suggested by some of the earlier writers. 

The 1949 excavations do not support the interpretation that the 
same mound was used both for habitation and for burial. Rather, 
separate mounds were constructed for each purpose and are easily 
distinguishable by their contents and usually also by then* size. The 
habitation mounds are comparatively small and low: the largest of 
the Camutins group (J-15, Mound 14) is 51 meters long, 35 meters 
wide, and 6.25 meters high; the largest of the Fortaleza group is 91 
meters long and 2 meters high (although one with less area is 3.5 
meters high). Sherds are sparse and 92 to 100 percent of these repre- 
sent the undecorated, utilitarian wares, Inajd, and Camutins Plain. 



EV^Nlf^ ^^° ARCHEOLOGY AT MOUTH OF AMAZON 399 

There are some indications of house type. The stratigraphic pit 
in the habitation mound, J-15, Mound 14, showed an alternation be- 
tween layers of fire-burnt orange dirt containing relatively abundant 
sherds, and light-grayish or whitish soil with sherds sparse except 
immediately above the preceding burnt-orange layer (fig. 106; cf. 
Farabee's results at Fortaleza). It was not possible to enlarge the 
excavation so as to determine the horizontal extent of the orange 
areas, but the irregular thickness of the bands and the presence of a 
small pocket of similar character near the bottom of the cut may indi- 
cate that they are not