ANTHROPOLOGICAL RECORDS ARCHAEOLOGICAL EXCAVATIONS IN FIJI BY E. W. GIFFORD UNIVERSITY OF CALIFORNIA PRESS BERKELEY AND LOS ANGELES 1951 IIRCHAEOLOGICAL EXCAVATIONS IN FIJI BY E. W. GIFFORD ANTHROPOLOGICAL RECORDS Vol. 13, No. 3 ANTHROPOLOGICAL RECORDS Editors: E. W. Gifford, R. F. Heizer, R. H. Lowie, R. L. Olson Vol. 13, No. 3, pp. 189-288, plates 12-29, 4 figures, 7 diagrams, 3 maps Submitted by editors September 19, 1950 Issued February 23, 1951 Price, $1.50 UNIVERSITY OF CALIFORNIA PRESS BERKELEY AND LOS ANGELES CALIFORNIA CAMBRIDGE UNIVERSITY PRESS LONDON, ENGLAND MANUFACTURED IN THE UNITED STATES OF AMERICA ACKNOWLEDGMENTS A sabbatical leave granted by the Regents of the University of California gave me opportunity 'nearly six months of archaeological activity on the island of Viti Levu, Fiji. Mrs. Gifford I left San Francisco by the Norwegian motor vessel "Thor I" on February 24, 1947, and returned Pan-American plane on September 13, 1947. We were in Viti Levu from March 23 to September 9. Financial support for our expedition was generously supplied by the Viking Fund, Inc. The Government of Fiji cooperated most wholeheartedly in the enterprise, Acting Governor J. F. 11, through his staff, arranging various facilities for us. Most useful of these was the con- t help, as interpreter and aide, of Ratu Rambithi Vuikandavu Longavatu, an educated young an of chiefly rank, whose services were assigned to me by Ratu Sir Lala Sukuna, bead of the an Office. Sir Lala maintained a constant interest in our work. Giving unceasing attention to the needs of the expedition and ever solicitous of its welfare another member of Sir Lala's staff, Mr. G. Kingsley Roth, anthropologist and son of the anthro- ogist of Tasmanian fame, Mr. H. Ling Roth. District Commissioner Western Mr. Charles Harley Nott and the lieutenant governors of Ra and Provinces (Roko Tui Ra Charles Wimbledon Thomas Vuiyasawa and Roko Tui Mba George Tonganivalu), hich the excavations were conducted, made the necessary arrangements with local chiefs and idents for the smooth organization and operation of our enterprise. Various other government officials aided us in one way and another: members of the Public B Department, at Suva, Vaileka, and Lautoka, notably Mr. E. J. Jenkins at Lautoka and Mr. H. M. th at Vaileka, supplied us with and transported tools, surveyors' stakes, etc.; Mr. R. A. Der- k, principal of the Technical School had the frames made for our screens for sifting mound ma- ial. Mr. L. G. Usher, Public Relations Officer, acquainted the public with the nature of our k by press and radio broadcast in English, Fijian, and Hindustani. Many individuals, both in .government and in private life, contributed information about archaeological sites: Mr. C. S. .Reay, Commissioner of Labour; Mr. Wilfrid Wise, former Director of Public Works; Dr. H. Sil- ter Evans, who accompanied me to several sites; the late Mr. George T. Barker, Curator of the FiMuseum; Mr. T. Alport Barker, of the Fiji Times and Herald; Mr. R. H. Lester, of the Office of rian Affairs; Mr. F. R. Charlton; Dr. Lindsay Verrier; Mr. William J. Belcher, incomparable por- yer of Fijian birds in color; Mr. Patrick Costello. We are indebted to gentlemen of the staff the Colonial Sugar Refining Company for permission to excavate on their railroad right of way in Province and for the use of theit line car to inspect sites along some eighty miles of right of The gentlemen I refer to are Messrs. H. King Irving, General Manager; E. H. Griffiths, man- er of the Lautoka Mill and now General Manager; W. E. Lindsay, Manager of Penang Mill; and Mr. W. L. Cozens, railroad traffic manager. At the Fiji Museum, Mr. Barker's staff, Messrs. Batty and ipate Talei, rendered me all requested aid. To certain hotel managers, I am indebted for the gift of seventy-five packing boxes for the erseas shipment of the collection: Mrs. Aimee Monk, Hotel Singatoka; Mr. Laurence Kirk, Hotel lutoka; Mr. Gordon Johnson, Hotel Rakiraki. To the American Consul at Suva, Mr. Winfield H. Scott, and to the management of the Grand ific Hotel, we are grateful for many kindnesses. To many Fijians, among the world's most cour- ous and likable people, we are indebted for specimens generously given and for services ungrudg- %gly and efficiently rendered. I am indebted to Dr. Leo G. Hertlein and Dr. G. Dallas Hanna of the California Academy of iences for identification of molluscan species which I collected at various sites. Dr. Fenner A. Lace, Jr., Curator of Marine Invertebrates, United States National Museum, who recently worked on e Bikini crustaceans, generously identified the sorry fragments from sites 17 and 26. Dr. Henry Fowler, Curator of Fishes, The Academy of Natural Sciences of Philadelphia, has studied the fish nes recovered in the excavations. Foraminifera embodied in potsherds have been noted by Profes- ir R. M. Kleinpell, Curator of the Micropaleontological Collection, Museum of Paleontology, Uni- rsity of California. Professor J. Wyatt Durham of the same museum identified certain specimens marine animal origin. Echinoderms were kindly identified by Dr. Austin H. Clark of the United States National Museum, rals by Dr. F. M. Bayer of the same museum and by Professor John W. Wells of Cornell University, d archaeological plant fragments by Dr. A. C. Smith of the United States National Museum and Pro- assor Emanuel Fritz of the University of California. Bones of birds, other than the domestic fowl, ire identified by Dr. Alexander Wetmore, Secretary of the Smithsonian Institution, and rodent nes by Dr. David H. Johnson of the United States National Museum. Certain questions of bone ,entification were submitted to Professor Seth B. Benson, Museum of Vertebrate Zo6logy, University 'California. Dr. Hans Jenny, Professor of Soil Chemistry and Morphology, University of California, kindly amined autoptically a few of the pertinent samples of mound material and scrutinized my photo- aphs of excavated pit walls. Professor Howel Williams of the University of California Department Geological Science, Berkeley, began the study of the petrography of Fijian pottery, which Dr. G. Curtis completed. Professor Williams identified the material of our stone adzes. [iii] CONTENTS Page wledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .iii duction . . . ..... . . . . . . . ........ . . 189 vations at Navatu (site 17) . . . . . . . . . . . . . . . . . . . . . 193 ations at Vunda (site 26) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199 sition of sites 17 and 26 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202 ked bone from sites 17 and 26 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206 GIdentified fish bones .2 . . . . . . . . . . . 206 Identified mammal, bird, and turtle bones. 0 . . . . . . 8 . . . 208 Usks . .......... o ......... 213 taceans, echinoderms, and corals from sites 17 and 26 . . . . . . . . . . . . . . . . . . 218 ical specimens from sites 17 and 26 . . . . . . . . . . . . . . . . . . . . . . . . . 219 l1s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219 acts of marine calcareous materials . . . . . . . . . . . . . . . . . . . 220 artifacts . . .. . . . . . . . . . . . . . . . . . 221 ry ................ ....... . 222 Pottery from site 17. .. . . . . ... . . ...... . . ... . .. . . . .. . 225 Pottery from site 26 .... . o o . . . o . . o . . . . . . o . . . . o . o . . o . . . 230 Pottery from other sites. . . . . . . . . . . . . . . . . . . . . 231 Modern pottery vessels from Viti Levu. . o . . . . o . . . . . . o . o o o o . 232 ral and spatial relations .. . . . . . . . . . . . o . . . . . . . . o o o . . . . 235 Fijian relations .... . o . o o o . o . . . . . . . o o . . . o o . 235 Extra-Fijian relations . . . . . . . . . . . . . . . . . . . . . . 236 ndixes . o . . . o . . . . . . . . . . ... ... ... ..........................239 I. Petrography of pottery, by G. H. Curtis ... . . .... . . ..... . . ... ... 239 II. Sites visited .... . . . . o o o o . o o o o o . . o . . . . . . . 242 Zrature cited . . . 257 anation of figures 1-4 .I.. .. . . . . . 259 es o. . . o o -o . . . . . . . . . . . . . . oo o o o o . . . . . o. ............................ 265 DIAGRAMS Staked area and excavations at location A, site 17. . . . . . . . . . . . . . 197 Trench profile, location A, site 17 .198 ,Staked area and excavations at location B, site 17 .... ... . . o . . . . . . o o . . 198 Staked area and excavations at location A, site 26 .. o . o . o . o . o . 201 Strata at datum stakes at location A, site 26 . . . . . a . . . . . . . . . . . . . . . 201 Location of Tandravula, site 4 . . . . . . . . . . . . . . . . . . . . . 244 Plan of Navutolevu, site 16 . . . . . . . . . . . . . . . . . . . . . . . . 250 MAPS 'Viti Levu, showing location of sites examined . . . . . . . . . . . . . . . . 192 .Navatu (site 17) and environs .. . . . . . . . . . . . . . . . . . . . . . . . . 196 .Vunda (site 26) and environs . . . . . . .. . . . . . . . . 200 [V] V, I" il I r .1 ?Y", L It ?, II 1?. 1?;I k RCHAEOLOGICAL EXCAVATIONS IN FIJI BY E. W. GIFFORD INTRODUCTION As tropical Polynesia has yielded archaeo- Lcally only the early phase of the local cul- Pe which were flourishing at the time of dis- pry, I decided to look farther west for a wession of cultures. Fiji seemed a likely pe, and moreover I reasoned that it might t traces of early Polynesians, if they had i via Fiji. Also, I reasoned that large rnds rather than small ones would be likely produce a succession of cultural horizons; Iy colonizers would presumably be attracted large land masses rather than to small ones. qe Viti Levu, of more than 4,000 square miles F, was selected. The presence of pottery-making in Viti Levu We present day was also another considera- . I assumed that I might find evidences of early culture horizon without pottery, but I # no such material. At both sites excavated, Bherds persisted to the bottom of the cultur- leposits, indicating that the first settlers, 4iese two sites at least, brought the art of tery-making with them. This ruled out any alihood that the first settlers at these two s were Polynesians, since no pottery has m excavated in Polynesia except the sherds Tongan middens which McKern and Gifford Lected. If one were to ask, "What is the chief racteristic of Fijian archaeology?" the ter would be, "Pottery, above all else." It ionspicuous on the surface of sites and it is Kdant beneath the surface. On our expedition sw stone and shell artifacts were recovered, their ratio to sherds is less than 1 to )0. As I shall constantly refer to Late horizon Early horizon, it will be well to explain significance of these terms. The term Late Lzon is applied to the upper, shell-midden itons of the two sites excavated, viz., sites md 26. The decorated potsherds of the L1-midden deposit at site 17 were chiefly or- ited by incising. The levels below the .1 midden at site 17 were characterized by rated sherds with reli-ef design, so it is ;ifiable to speak of Late horizon incised and Early horizon relief ware. Of course 'e is also a vast amount of plain ware but irally many of the plain sherds may be frag- s of partly ornamented vessels. In general, ,sing seems to have been applied to rims and lders, relief decoration largely to bodies, ?essels. At site 26 ornamented sherds, either in- cised or relief, were s6arce and plain ware pre- dominated. There the correlation of incised sherds with the overlying shell-midden and of relief-decorated sherds with the underlying non- shell portion of the deposit is not so clear as at site 17, location A. However, the distinc- tion of shell-midden as Late horizon and the shell-less lower layers as Early horizon still hoilds. Probably a Middle horizon can be distin- guished at site 17, characterized by high per- centages of plain sherds and complementary low percentages of decorated sherds. However, this Middle horizon does not obtrude itself as do the Late and Early horizons, especially since these two are correlated respectively with pres- ence and virtual absence of shell, as well as with distinctive sherd types at site 17. Excavations at site 17 involved the screen- ing of at least 1,548 cu. ft. of material, at site 26 of 1,332 cu. ft. A brief report on the results of excava- tions at site 17 was sent to the Seventh Pacific Science Congress in New Zealand and subsequently submitted by Professor H. D. Skinner for publi- cation in the Journal of the Polynesian Society. In Fiji, with its flourishing aboriginal population of 124,000 (1948 census), the archae- ologist has one important asset almost entirely lacking in California and other parts of the United States: the Fijians have names for prac- tically all sites. In the war-troubled days of the nineteenth century it was necessary for vil- lage communities to shift to new sites quite frequently. Now the archaeologist can examine a series of sites successively occupied, thus achieving a sequence for late times at least. One disadvantage, however, is that sites oc- cupied only a short time do not have very deep deposits. All deserted village sites are called koro makawa ("old village"). An original "first" site is called a yavutu. Sites la and lb, formerly occupied by the yavusa ("tribe") Koronikalou, the villagers of Thautata, Tailevu Province, are examples of old villages, but not original sites. When we visited present Thautata, we inquired about pre- ceding villages. Thautata-l was on the site of present Thautata-4. Thautata-2 (site la) was situated a mile or so away in jungle near a man- grove swamp, apparently for protection. Thautata- 3 (site lb) was also on the edge of mangroves. 'Gifford, 1949b. [189] ANTHROPOLOGICAL RECORDS At Thautata-3 we found European as well as na- tive articles, and the site of the church was shown us. At Thautata-2 we saw no trace of European influence; the villagers, moreover, said it dated from pre-European times. Thautata illustrates a situation we found in a number of places--namely, a series of sites occupied successively during the troublous times of the nineteenth century. We selected no such sites for excavation. On the contrary, we de- cided on two (sites 17 and 26) which the natives declared had always been occupied down to their abandonment in modern times; these were the original settlements (yavutu) of certain "tribes" (yavusa). When pressed for an explanation of the unusual length of occupancy, our informants explained that these two locations, at the base of rocky crags, were so readily defensible that the inhabitants never had to abandon them, but instead withdrew to the rocky portions or behind prepared defenses. Our excavations, yielding cultural material down to almost 12 ft. at site 17 and to 10 1/2 ft. at site 26, testify to the long occupancy of these two sites. Site 26, Vunda, is reputed to be the first settlement of the ancestors of the present-day Fijians. Before we undertook excavation, we visited and made surface collections from thirty-nine sites. As a result of this surface reconnais- sance, the two sites mentioned were selected for careful excavation: site 17, Uluinavatu, or more briefly Navatu ("The Rock"), between Narewa and Vitawa villages on the north coast of Ra Prov- ince, and site 26, Vunda, on the Vunda River, in Lautoka Province, on the western side of Viti Levu. The two sites are about eighty miles apart by road; both are on the leeward or "dry" side of the island. All village sites were marked by the pres- ence of shell, since shellfish of many sorts form an extremely important part of the modern Fijian diet. Here, indeed, exists a contempo- rary shell-midden people. W. R. Steadman (see "Literature Cited") has written an illuminating article on this subject. The reefs are thor- oughly searched six days a week by the Fijians. Not only mollusks and crustaceans are taken, but also large numbers of fish. At both of the sites excavated, the hired crews were entirely Fijian. Not only were they reliable, intelligent, and conscientious workers, but they became interested in what they were do- ing. If a man of the crew of ten was unable to come, he invariably sent a substitute. Thus, we were never "let down," a circumstance which pleased us greatly, as time was limited and there was much to be accomplished. Of the in- terest manifested by the natives in the archae- ological aspect of the work, we would cite two instances. The Early relief-type pottery found in abundance at site 17 evoked praise as to its quality and wonder at its occurrence there, which none had been aware of. At site 26, the interest took another turn. Site 26, or Vunda, is reputedly Viti Levu's first settlement. Knowledge that we had obtained cultural material down to almost twelve feet at site 17 caused the Vunda crew (descendants of the ancient Vunda people) to declare their intention of beating that record. However, in this they failed, through no fault of theirs, since the cultural material was not found below ten and a half fet This competitive attitude is an example of the consistent drive on the part of chief Tui Vund' people today to establish the priority of thei forefathers in as many things as possible--for; instance, in Christianity--to the disparagemen of Mbau, the chief kingdom in the middle nine-' teenth century, on the opposite (eastern) side of the island. Before any work was commenced at either site 17 or site 26 the Fijians comprising the' crews insisted upon making and presenting yang gona (kava) to the deities and the dead at the sites. At location A of site 17 the yanggona was brewed in a pottery kava bowl. (See pi. ii c, d.) Two cupfuls were poured at the base of. shrub as an offering to two deities of the pla Mbakandroti and Ngataiwai. Thereafter we all drank. Plate 13, c shows the yanggona present tion to the dead before excavation of the firs grave at site 17. When we closed the work at site 17, yanggona was presented at graves from which we had removed bones. These graves were upslope on the site and apart from locations A4 and B, where no burials were found. On the evening of Friday, July 18, after the bones ha been removed, two spirits of the dead had suc- cessively possessed Ratu William Vuningasau, assistant pastor of the Wesleyan Church in Narewa. After these two spirits had been ap- peased by pouring yanggona on their grave, wit appropriate speech, and after new grave walls stone had been built, all trenches *and pits we' filled, except in the rock shelter, which the Narewa villagers wished to use on rainy days, At site 26 yanggona was offered at the be ginning, at the middle, and at the end of the excavation period. There it was presented to the deities of the place, especially Lutunasom basomba, rather than to the dead. Unless otherwise stated, all specimen numbers cited refer to the catalogues of the University of California Museum of Anthropolog catalogues 12 (physical anthropology) and 11 (Oceania, other than physical anthropology). T prefix 11- is omitted from specimen designatio in this paper. Excavation was by 6-in. blocks in rec- tangles which usually measured 3 by 6 ft., oc- casionally 6 by 6 ft. where conditions, suc obstructing boulders, rendered this advisable. Where the ground sloped appreciably, a top tri angle of material (TT in the depth column in tables) was removed down to the lowest corner the surface of the rectangle before excavation by 6-in. blocks was begun. All directions recorded are based on read ings from a Brunton compass which registered magnetic north. Certain native designations for modern ad ministrative officers require explanation. k Turanga-ni-koro is the title for the mayor of village (koro). Mbuli designates the officer I j C I t 'ss - 190 GIFFORD: EXCAVATIONS IN FIJI ge of a district (tikina), which comprises ral villages. Roko Tui is the title of the enant governor of a province, which com- es several districts. All of these off i- s are FiJians. "A New Fijian Dictionary," by Capell, de- l"yavusao' as follows:2 The largest kinship and social division Fijian society, consisting of the descen- ts of one originator (vui) and recognizing e kalou-vu, originating spirit (normally e father of the Vui) and one set of totems .cavuti) and war-cry (vakacaucau nI valu). e yavusa is subdivided into mataqali and i katoka. Yavutu, the village settlement origi- illy established by the Vu of a yavusa, ter which the yavusa is generally named. My informants used the word "vu" as refer- to a spirit ancestor. Sometimes they said, example in genealogies, that the first two three generations were vu, the subsequent - human. This corresponds with the defini- of vu given by Dr. Dorothy M. Spencer in eussing the matanggali (mataqali), which she ribes as "an exogamous,-patrilineal clan or- zation, the members of which consider them- es to be related by virtue of their descent one spirit-ancestor, known as the vu of the I am using as a base, a map of Viti Levu ued in March, 1939, by the Lands and Survey rtment, Suva, although this gives the old vincial boundaries. Since 1939, certain of smaller prcvinces--e.g., Mba, Lautoka, and di Provinces--have been combined, for ad- istrative purposes. These present-day en- Bed provinces are ignored for my archaeologi- -purposes. The more numerous provincial ndaries of 1939 are more convenient for re- ring to the location of my thirty-nine archae- gical sites. Map 1, Viti Levu Island (p. 192), shows the ation of the thirty-nine sites visited in my Dnnaissance. The sites are all known to the ives, who guided me to most of them. A full cription and discussion of the sites are en in Appendix II. The list of thirty-nine es (nos. la, lb, 2-38) follows. Unless erwise indicated the sites were once villages. la, lb. Thautata, District of Mbau, rovince of Tailevu. Respectively the second d third sites occupied by the yavusa Koro- ikalou, the "tribe" now occupying modern autata. This modern village is on the ite of the first and fourth settlements 'ter the legendary arrival of the ances- ors from the Kauvandra Mountains in Ra rovince. 2. Old Verata, at the base of the fthunivanua Peninsula, District of Verata, 2Capel1, 1941, p. 341. 3Spencer, 1941, p. 1. Province of Tailevu. This was the yavutu (ancestral village) of the yavusa Vunivalu of Uthunivanua village, and of the yavusa Mburetu of Naloto village, both in the Dis- trict of Verata, Province of Tailevu. 3. Nangguatamboto, near Tavua, District of Tavua, Province of Tholo North. An old village site of the yavusa Mbila, now re- siding in Tavua, Korovou, and Vatutavui, all in the same district. 4. Tandravula, formerly occupied by yavusa Navauvau, now residing in Korovou, District of Tavua, Province of Tholo North. 5. Tavua Railroad Water Tank Site. An unnamed site in front (south) of Nambuna vil- lage, District of Tavua, Province of Tholo North. 6. Vatuthere, a modern village of the yavusa Numbu, on Nandala Creek, District of Savatu, Province of Tholo North. 7. Namatakula, an open site with a rock shelter at the back, on left side of Nandala Creek, about 5 mi. from Nandarivatu, near automobile road to Navai, District of Savatu, Province of Tholo North. 8. Double rock shelter. On old govern- ment track from Vatuthere to Nandarivatu, on right side of Nandala Creek, District of Savatu, Province of Tholo North. 9. Korovatu Fortress, north of Nanda- rivatu, on south slope of Korolevu Peak, at about 2,500 ft. elevation, Province of Tholo North. 10. Navatunisala, adjacent to Ndrauniivi, District of Raviravi, Province of Ra. 11. Nasangga, about 1/2 mi. east of Indian store at Ndrauniivi, on north side of King's Road, District of Raviravi, Province of Ra. 12. West side of Nathilau Point, Dis- trict of Raviravi, Province of Ra. 13. Nathilau Point Low Tide. Potsherds were found below high-tide level. Not a vil- lage site. 14. Namotutu, on the left bank of Penang River, 1/4 mi. south of King's Road, District of Rakiraki, Province of Ra. One group of Namotutu people now lives at Rakiraki vil- lage, District of Rakiraki, a second at Narewa village, District of Raviravi, Prov- ince of Ra. 15. Tambuwai Hill, near village of Mata- wailevu, District of Nalawa, Province of Ra. 16. Navutolevu, near Rakiraki village and hotel, District of Rakiraki, Province of Ra. 17. Uluinavatu, or Navatu, north and west sides of the eminence between Narewa and Vitawa villages, District of Raviravi, Province of Ra. Excavated. See plates 12-14. 18. Embuto, adjacent to modern Nasinga- toka, District of Nasingatoka, Province of Nandronga. 19. Lomolomo, slightly upstream and to west of modern Lawai village, Singatoka Valley, District of Nokonoko, Province of Nandronga. 20. Eastern portion of sand dunes, west 191 ANTHROPOLOGICAL RECORDS I to C) x N Do U) 4-) 0 aH 0 0 0 CO 4. 0 -4 0 -c :E IZA 192 GIFFORD: EXCAVATIONS IN FIJI t mouth of Singatoka River, District of ingatoka, Province of Nandronga. See gte 14,, e, f . 21. Nanggarai, western portion of sand es, west of mouth of Singatoka River, strict of Nasingatoka, Province of Nan- nga. See plates 14, e, and 15 f. 22. Nalovo, District of Thuvu, Prov- e of Nandronga. Formerly occupied by eusa Louvatu, now in Thuvu village. ; 23. Koroinggara, cave site, formerly ocupied by yavusa Louvatu, now in Yandua ilage, District of Thuvu, Province of hdronga. One-third mi. west of Volivoli llage. See plate 15, e, f. 24. Yaro, a pit-like hole in limestone, /8 mi. inland from Queen's Road, just west ' Sila village, District of Thuvu, Prov- tce of Nandronga. 25. Natauta, a limestone burial cavern. Just west of Sila village and about 100 yds. orth of Queen's Road, District of Thuvu, ovince of Nandronga. 26. Vunda, "original and first" Fijian illage. About 1/8 mi. above confluence of runda and Varangga rivers, on left bank of runda River, District of Vunda, Province of autoka. Excavated. See plates 15, a-d. 27. Old Nasolo, on hill on left bank of first left-hand tributary of Mba River, District of Nailanga, Province of Mba. 28. Malolo Plantation of Sikituru, Prov- , ince of Nandi, on right bank of "Malolo River." 193 29. Yako Plantation, District of Momi, Province of Nandronga, just southwest of Nandi Province border. 30. Lomawai Plantation, right bank of Kumbuna River, District of Wai, Province of Nandronga. 31. Malomalo, just east of modern Malo- malo village, District of Malomalo, Province of Nandronga. 32. Mbulu, a site on west (right) side of Mbulu Creek at the mouth, District of Thonua, Province of Tholo West. 33. Sovi, a site at right side of Sovi River near the mouth, District of Thomua, Province of Tholo West. 34. Old Vatukarasa, a site on left side of mouth of Tamanua Creek, District of Thonua, Province of Tholo West. 35. Old Tangangge, adjoining modern Tangangge. On right side of Tangangge Creek near mouth, District of Korolevuiwai, Prov- ince of Tholo West. 36. Namahara or Namasara, between Votua and Korolevu Bay, District of Korolevuiwai, Province of Tholo West. 37. Korolevuiwai, 1/8 mi. west of Koro- levu Bay Rest House, District of Korolevui- wai, Province of Tholo West. 38. Nanggau, on both sides of Queen's Road and on both sides of Nanggau Creek, first small creek west of Komave, District of Komave, Province of Tholo West. EXCAVATIONS AT NAVATU (SITE 17) The conspicuous landmark of Navatu towers ft. beside the sea on the north coast of the trict of Raviravi, Province of Ra. Shown on ymaps as Navatu, "The Rock," but more fully on as Uluinavatu, "Head of the Rock," it- rves as a trigonometrical station. On the at of it flows Narewa Creek, on whose west e is modern Narewa village. Some of the vil- ers are descendants of the yavusa Naisongo- whose members dwelt on site 17. On the rtheast side of the crag is Vitawa village, so inhabited by some of the same yavusa. Site 17, on the slopes as well as at the e of the rock, is the most extensive site I w in Viti Levu, with an estimated perimeter of miles. No survey was made to determine the ct boundaries of this huge site. Mrs. S. M. Smythe shows a view of Navatu from the sea.4 also plate 12 in this paper and my published otographs.5 Brewster's map (1922) shows Nava- as a village. With Rambithi, the Mbuli Semi of Rakiraki Strict, Navitalai of Ndrauniivi village, and orge Wanggata of Narewa village, I visited the te on May 3. We found that the shell-midden posit extended up the slopes to 300 ft. eleva- ion, at the base of the cliffs. Two walls of 4Smythe, 1864, plate opp. p. 100. 5Gifford, 1951, pls. 3, b, 4J a. fortifications were observed on the slopc , the upper at 270 ft. elevation. These were of large natural boulders with smaller boulders piled between them to form walls. Besides Narewa Creek there was a small spring at the western base of the crag, which supplied water to the inhabitants; the spring was dry at the time of my visit. Living in Narewa village along with the yavusa Naisongoliku are people from Namotutu (site 14), who moved there when they abandoned site 14. These people stand in the relation of vasu (literally, privileged mother's brother's sons) to the Naisongoliku people. My informants stated that the yavutu of the Namotutu people, now living in Narewa village, was Tonga,6 at or near the site Namotutu (site 14) on the Penang River. Narewa and Vitawa villages were in the Ju- risdiction of the Mbuli Hosea of Raviravi at the time of my visit. Legend has it that the origi- nal Navatu people came directly from the Kau- vandra Mountains under the leadership of their two vu, Mbakandroti and Wanggambalambala, sons of the god Ndengei. Kingsley Roth (p. 227) re- fers to Mbakandroti as the ancestral spirit (kalou vu) of the yavusa Navatu. The Navatu people assert that they never lived anywhere 6The Fijian Government list gives Namotutu, not Tonga, as the yavutu of the Namotutu people of Narewa village. I I i0 194 ANTHROPOLOGICAL RECORDS else after their forefathers came to Navatu from the Kauvandra Mountains. The Navatu and Naisongoliku yavusa both made pottery, but the Namotutu people did not. (Three women, the sister of George Wanggata, a woman in Vitawa village, and another in Naivu- vuni village, still make pottery.) The numerous potsherds excavated at site 17 are probably largely of local manufacture. The midden deposit of site 17 was densely overgrown with vaivai, an acacia-like shrub or small tree, stated to have been introduced. It is said that, when the site was inhabited, it was clear and open, except for trees of economic value to the inhabitants. House platforms are rectangular, and were said not to have been used for burial. The land known as Uluinavatu is now owned by the matanggali Naisongoliku of the yavusa Naisongoliku, of which Ratu Ilivasi Vunimbola of Vitawa village was the senior member during the period of our excavations. The yavusa formerly occupied the now deserted yavatu site, also named Naisongoliku, until 1874, when some of its component social units (matanggali) moved to nearby Vitawa, others later to Narewa, still others to other villages. Naisongoliku and Na- valelawa were ancient settlements on our site 17 (Navatu), which thus comprises the origin places of two modern tribes: yavutu Naisongoliku, ori- gin place of the yavusa Naisongoliku; yavutu Na- valelawa, origin place of the yavusa Navatu. Our informant, when we first visited Navatu, was George Wanggata of the matanggali Navalelawa of the yavusa Navatu. Wanggata, who was later a member of our crew, was born on the site. Miss C. F. Gordon-Cumming sojourned for a time at Navatu, which she describes in a letter written there on October 10, 1876:7 This place is well described by its name. It is really Na Vatu (The Rock), being a huge rock mass, quite detached from the great Kau- vandra range of mountains, and standing alone on a level shore. The village in which I am living is on the sea-level, but a steep path up the beautiful crag leads to a lovely vil- lage, called Nai Songoliko, which consists of a number of small houses perched wherever they can find room all over the cliff, almost hidden by bread-fruit and other bowering trees, which cling to the rock-as if by magic. From this point a narrow spur runs inland, and the view from there is quite beautiful -- the bluest sea, dotted with isles and tinted by patches of coral-reef, lying outspread to right and left of the cliff. Each of these villages has a tidy well-built church. I think I have explored every corner of the great rock, and many of the tiny homes which lie so quaintly niched among the rocky boul- ders. Some of the people produced hidden treasures, which they offered me for sale; and I have bought several good things, in- cluding some stone axes. I think I must have mentioned to you that these are only just now 7Gordon-Cumming, 1883, pp. 243 ff. I passing out of common use here: they are ' brought to us tied with native string to a piece of wood shaped like a bent knee. Som times I see instances of the actual transi- tion from the stone to the iron age, when some lucky man, having got a Birmingham adz rejects his old stone celt and ties his new acquisition on to the same wooden handle. All over this crag and the neighborho there are luxuriant masses of the intensely blue clitoria, as also of a bean which is good for food, and bears white blossoms. Dating from the same year as Miss Gordon- Cumming's visit is the account of Mr. Maudsla kindly supplied me by Mr. G. Kingsley Roth:8 We [the Governor of Fiji and party] dropped anchor in Raki Raki bay opposite to Na Vatu .... we determined to land the next morning at Na Vatu, a large pile of rock rising boldly from the water's edge .... it was so late when we landed that there was only time to visit hastily the village at the foot of the rock, have a bathe, and get back to the steamer as quickly as possible.; I was sorry not to see more of the place. Its position is curious. A range of bare- looking mountains follows the line of the bay some distance inland, -- bare, that is, of trees, but grass-covered nearly to their rocky summits. A plain lies between these mountains and the sea, broken only where Na- Vatu rises abruptly at the water's edg6. Though not many hundred feet high, the rock is imposing and picturesque, both from its position and shape. It held a high place i) the old mythological stories of Fiji, as on of the jumping-off places into the world of; spirits. Now there are three villages on the rock, the one we visited at the bottom, one half way up on the land-side, and one perched almost on the top. The following observations on the crag N4 vatu were written by my assistant, Ratu Ram- bithi, following a trip up the mountain on June 22, 1947: After breakfast at about 8 A. M. I weni to Uluinavatu Mountain with Savenatha Tiko- vakayalewa and Jone Tute, two boys of Narew4 and also four boys from Vitawa, who were Iliki Seru, Inoke Thauthau, Epeli Wanggatam bu, and Emosi Naitheru. We climbed up the hill at the northeasl side. Our first place of rest was 380 feet above sea level, Vitawa village on the east. The second place of rest was 500 feet above sea level, Vitawa on the northeast. We were then at the rock Vatuveivova (stone on top of another), which looks like a human head when viewed from the King's Road. 8Extract from a letter written by Mr. Maudslay contain in Letters and Notes Written during the Disturbances in t. Highlands of Viti Levu, Fiji, 1876. Privately printed in Edinburgh, 1879. I I , GIFFORD: EXCAVATIONS IN FIJI Niundamu's crag, the highest of the four ich form the mountain called Uluinavatu is 2 feet above sea level. From the summit tawa village lies northeast, Narewa village uthwest. The length of the shelf on which Government beacon is situated is 28 feet. Next we climbed up Niumarawa's crag ich towers 620 feet above sea level, Vitawa d Narewa lying southeast and northwest re- ectively. This summit is where Ratu Sa- enatha Tikovakayalewa, Ratu Apenisa's rother, is buried. Ratu Apenisa was the hief leader of the Namotutu people who came D Navatu during a civil war at Rakiraki in re-Christian times. Ratu Apenisa was the 8u of the Navatu people. Ratu Savenatha's rave measured 14 feet by 13 feet. Earth r the grave was brought from below. Vege- tion growing on top of the crag is vaivai, iwai, ngginggila, and grasses. We climbed p from the south side. Descending from Niumarawa's crag, I tound a stone adze blade lying on the slope F5O feet above sea level; from there Narewa Ind Vitawa lay southwest and northeast re- Opectively; the King's Road was on the south. The third crag, the name of which I mould not learn, lies between Niumarawa's Crag and Mbakandroti's crag and is 550 feet high. Mbakandroti's crag is 500 feet above sea level. From this summit Narewa is southwest and Niumarawa's crag and Niunda- mu's crag are northeast by east. Vitawa vil- lage is invisible since leaving Niumarawa's ,rag. Leaving Mbakandroti's crag we descended to see Namatakona, a rock pool used for sharpening spears in heathen times. Its dimensions are 21 feet long and eight feet icross. The pool is shaped like a boat. the southwest end looks like the stern and parks from sharpening implements are still risible. This place is on the Naisongoliku lide, i.e., east side, and is 450 feet above sea level. The rock pool still contained water, but was said to be usually dry in the hot season. House mounds were indicated on some tlat places, and one under the third crag on the east side, 450 feet above sea level, is -O feet by 16 feet. Samples of potsherds md shells were collected. Thick accumula- blions of shell were found in hollow places. Tegetation was mostly vaivai with some other knds of trees interspersed. Goats had done Xgood Job in eating off some of the bush, otherwise the place would be too bushy to travel. Potsherds and shells were found trom the bottom to the summit of the moun- tain, the ascent and descent of which is risky. On our way back to Narewa I found f piece of stone like an adze on the rail- road. In 1947 we excavated at two locations, 1/4 apart: A and B in site 17. See map 2, p. 196. Location A was selected because of the presence of a huge overhanging boulder of horn- blende andesite near the foot of the talus slope. See plate 13,a,b. This boulder pretty well pro- tected the spot from material sliding down the talus slope which extended up to the base of Mbakandroti's crag. Moreover, the boulder pro- duced a rock shelter, more than 12 ft. from front to back, which was filled with shell- midden deposit. At some time in the recent past the shelter had been fenced and used as a pig pen. A nail or two and glass fragments gave evi- dence of modern use. Location A was staked in rectangles, 3 ft. x 6 ft., as shown in diagram 1. A datum stake 28 ft. north of Stake A served to locate the staked area in relation to the Narewa spur track of the Colonial Sugar Refining Company's rail- road. The nearer rail of the tracks lay 141 ft. west of the datum stake. The distance from that point on the rail to the end of the track at the King's Road was 543 ft. Surface potsherds were collected from the entire staked area. The crossed rectangles in the diagram were completely excavated to subsoil. Excavation of rectangle G3-4 H3-4 was started, but not finished because of lack of time. Rectangles, Location A, Site 17 A3-4 B3-4. In front of house mound. Blocked with boulders below 48 in. B3-4 D3-4. This constitutes a 6-ft. square. In front of, and at lower edge of, house mound. Down to 64 in. rectangle B3-4 C3-4 was dug as a test pit made to determine whether the location warranted digging. Con- sequently this portion was not dug in 6-in. blocks. Later, rectangle C3-4 D3-4 was dug in 6-in. blocks down to 64 in. The two rec- tangles were united in a 6-ft. square desig- nated B3-4 D3-4 in tables in this paper. Dug into ground water. D3-4 E3-4. On slope of house mound. Dug into ground water. E3-4 F3-4. On slope of house mound. Dug into ground water. G3-4 H3-4. On upper slope of house mound near its edge. Dug to depth of only 7.5 in. and consequently omitted from tables giving distributions by depth. H3-4 I3-4. On top of house mound. I3-4 J3-4. On top of house mound. N3-4 03-4. On top of house mound. R3-4 T3-4. In rock shelter. This con- stitutes a 6-ft. square. We first dug rec- tangl,e R3-4 S3-4, but only to 20 in. since we found a large boulder occupying much of the area, making the yield quite small. We then dug rectangle S3-4 T3-4 down to the same level. Thereafter we operated in the 6-ft. square R3-4 T3-4, but obtained very little from the R3-4 S3-4 portion of it. The 32-38 in. level was sterile yellowish vol- canic ash, wholly devoid of cultural material. T3-4 U3-4. In rock shelter. At 38 in. the floor was virtually all sterile yellow- ish volcanic ash. S4-5 T4-5. In rock shelter. T4-5 U4-5. In rock shelter. 195 ANTHROPOLOGICAL RECORDS Map 2. Navatu (Site 17) and Environs 196 Z: GIFFORD: EXCAVATIONS IN FIJI Between the rock shelter, which formed the ern end of location A, and the western edge gline A-A7, there were traces of a house torm, the western edge of which was marked tones (diagram 1). Below it (to the west) surface dropped somewhat abruptly (diagram . 198). The house platform included in our locatior A? B7 C7 D7 P? G7 A6 B6 C6 D6 E6 P6 a G6 H? I7 16 AS B5 C5 DS E5 FS GS9 H5 1 5 souzth J7 K7 L7 J6 K6 L6 KS L5 A is called Waisa ("ditch"), because there was once a small stream at the western edge of it that was used for irrigating taro planted in the low part Just west of the site; the stream bed is now crossed by the.railroad embankment for the Narewa spur track. A culvert in the embank- ,ment furnishes drainage to Narewa Creek a few n yards farther west. west A2 B2 CZ D2 E2 F2 G4 63 H4 H3 I4 I3 J4 3 X4 K3 G2 H2 12 J2 Al BI Cl DI A . ? 9t Fl G1 Hi Ii 41- I1 K2 L4 L3 LI A B C D F G I Lor?h L M4 M3 M2 N6 NS 06 05 N4 3 04/ 03 NZ NI 02 01 P6 Ps P4 Q4ft P3 P2 P1 p Q3 F77/, surface stones 77 house pZatform stones >< excavated 0 2 4 6 foot J3 V5 V4 east V3 Diagram 1. Staked area and excavations at location A, site 17. M7 M6 MS N? 07 P7 ml M N 0 197 ANTHROPOLOGICAL RECORDS weil A B C D E F G H J K east M N 0 P 0 R S T U V Contour ao Aa 3 row of stakes at ground level Contour of Ax 4 row of stakes at ground tewi -- Depth of bottom of shell mnzdden betwn Aro 3 and MA 4 roWs of stakes I Shell midden Depth of top of yellow vcanic -ash substratum at No. 3 row of stakes D-- epth of top of yellow wocanic a-sh substratum at No 4 rew of stakes * Points at which measurments were taken Litdgc of rock shelter Diagram 2. Trench profile, location A, site 17. El D C B A DI E2 D2 Cl 01.5 C2 BI B2 Al AlS A2 ral2road Zino Crossed rectangles zvcre e-'caited: BJ-2, Cl ZevZ, A1-L.; Cl-i.5 to subsopi. Diagram 3. Staked area and excavatic location B, site 17 Groups of stones associated 1 and ash gave evidence of earth ovE were encountered occasionally, as the upper 18 in. of rectangle C3-l cation A. Ash and charcoal lenses undei shelter indicated that it was perl E3 E4 used for fires and cooking, probably in pots, which would require no stones as do earth ov In addition to the line of stones mark D3 D4 the edge of a house platform on the surface location A (diagram 1), we uncovered two mor 03 C4 lines of stones which appeared to mark earli house platforms. Both were in rectangle E3- F3-4. The lower of these was found at a dep B3 B4 between 54 and 60 in. and comprised five lar round stones, wi1thin a length of 6 ft. See plate 13, e. A3 A.4 Location B, about 1/4 mi. from location was selected as a check on location A. Exca tions at B were much more limited than at A, being confined to a single pit, dug into gro water begun in rectangle Bl-2 C1-2. After reaching the 54-in.,level upon which the ral road track was laid, the excavation was con-? tinued downward to 144 in. as rectangle Al-1 Cl-1.5. See diagram 3. Location B on the shore line of site 17 chosen because of large protecting boulders the back, which presumably would minimize th slide of material from upslope, although her 2-2 torairoad ground the slope was not as steep as that behind 1o tion A. The seaward edge of the site at loc tion B had been cut away in laying the main of the Colonial Sugar Refinijg Company's rai road tracks. The tracks were only a few feel from the shore line, here fringed with mangr The coral reef offshore made the waters plao ons at Although a considerable area was staked# with a frontage of 24 ft. on the railroad tr excavations were limited, as described above.: The top of the outer rail of the track was with charcoal 9.5 in. above high-tide level. Our western ens. These stake A4 was 547 ft. east of Narewa switch, for example in measured along the inner rail of the railroad 4 D3-4 of lo- track, which curves slightly. The line of A., stakes was 6 ft. from the inner rail. The fr r the rock line of stakes (A to A4) ran 600 E of N and haps frequently W of S. See diagram 3. I eI I I @- 4 0 -- i , i i i 0 6 6 6 6 i , i | |*i* 7 I I i), .A r i t 198 A ,2 .I ?'j 14 ? -.1 I GIFFORD: EXCAVATIONS IN FIJI EXCAVATIONS AT VUNDA (SITE 26) f;We first visited this famous site on May 19. situated at a crag on the left bank of the .River, a short distance above the conflu- of the Varangga and Vunda rivers and about from the mouth of the Vunda. It proved twenty minutes' walk up the left bank of nda River from the bridge which carries *ailroad and Queen's Road across the Vunda its mouth. Although the site is inland, at ,tide the water backs up at Vunda, at which mangroves grow. The fame of Vunda rests on its reputation h "first" settlement on Viti Levu made by forefathers of the present-day Fijians, who ed on the ship "Kaunitoni" under their r Lutunasombasomba. His stone-faced house orm crowns the central crag, rising as an polis above the site of the village of Vunda. From Vunda the people moved to other places, them near-by Mereke ("America"), now de- d, on the right bank of both the Vunda and - a rivers at their confluence. Two Fijian ations of the name Mereke were given: (1) called after the ship that brought the missionaries to this region; (2) two of- a of the Wilkes Expedition were buried perhaps the two men killed on Malolo off the Vunda coast. Wilkes, however, that these victims of the Fijian assault buried on a small island that he calls Island,9 which lies slightly north of west e mouth of the Vunda. Tui Vunda (Lord of ) showed me a wooden plaque on which were ed a series of old brass buttons from States Navy Jackets, said to have been terred in 1935 with the remains of a chief nally buried at Mereke in 1875, but re- red in Viseisei village, the Tui Vunda's . Possibly the buttons are from the Wilkes dltion, since Tui Vunda said they were pre- in "1835," a bit earlier than the Wilkes iition of 1840. The crag or acropolis of Vunda (pl. 15, a, Ises to a height of 350 ft. Some twenty- house mounds were counted on 'its various levels. On its very summit (the Eorovatu) . are five house mounds, reputedly Lutuna- somba's, his temple (mbure ni kalou), his nts' house, and two residences for his Fts (mbete).- Myth states that the great p which form the Korovatu were brought and aed by Lutunasombasomba's followers, who were -not men. Each carried two huge masses. The aent Tui Vunda claims descent from Lutunasom- nmba. It is said that there were once four temples Ounda. The temple of the god Erovunavunda, ton of the yavusa Kai Vunda, was on the low ind at the northeast edge of Vunda, for the Vunda lived at the base of the crag. From- ia these people moved to Yandro, then to 9Vilkes, 1845, 3:272; also chart opp. p. 73. This p is Kandavu of modern charts, according to Mr. R. B. ron, Chieff of the Section of Geographic Names, Hydro- lglo Office, U. S. Navy. Lauwaki, their present village. A second temple was dedicated to the god Vunayawa, patron of the yavusa Tumbumbere; a third to Leka, god of the yavusa Sambutoyatoya; and the fourth to Lutuna- sombasomba himself--all male deities. These three yavusa of Vunda and their priests referred decisions to the priest of Lutunasombasomba for approval. The first three temples were in the lowland part of the village; Lutunasombasomba's was on the acropolis. At present Vunda is uninhabited. The last resident of whom I learned was one Naivoleoni ("Napoleon"), who lived in a house which once stood on the house mound Ndele ("planting place") which we excavated (pl. 15, c). Vunda descen- dants live in Lauwaki village and Viseisei vil- lage, District of Vunda, Province of Lautoka. The Lauwaki people belong to the yavusa Kai Vunda and claim ancient Vunda as their yavutu. One of the yavusa of Viseisei, namely Sambuto- yatoya, claims Vunda and Yavuna as its two ya- vutu. Two other yavusa in villages of the Dis- trict of Vunda also claim ancient Vunda as their yavutu: yavusa Viyangoisaukova of Nanganga vil- lage and yavusa Thawanisa of Lomolomo village. An eighty-year-old informant said that there was yet another yavusa at Vunda, namely Vunativi, which does not appear in the official list of yavusa. To this yavusa belonged the above- mentioned Naivoleoni, who died about 1887 and was buried in the cemetery of Lomolomo village. Our excavating crew was composed largely of descendants of the ancient inhabitants of Vunda. At Vunda (map 3, p. 100) we devoted most of our time to digging location A, which included the house mound of the aforementioned "Napoleon," and the level space in front of it which was bounded by the Vunda River on the west.10 A dirt road ran along the edge, where it sloped down into the mangroves of the riverbank. The numer- ous house mounds of site 26 were round in out- line, and it was the practice to bury the dead in them. "Napoleon's" house mound yielded several burials. Diagram 4, page 201,shows the layout of location A. Rectangles, Location A, Site 26 A4-5 B4-5. In front of house mound. Dug into ground water. B4-5 C4-5. In front of house mound. Dug into ground water. v E-El F-Fl. In front of house mound. F-Fl G-Gl. In front of house mound. A huge underground boulder extending from wall I-Il through rectangles H-Hl I-I1, G-Gl H-Hl, and into F-Fl G-G1 limited the depth of ex- cavation in these rectangles. In F-Fl G-Gl it reduced the area which could be excavated at the bottom to less than half the area of the rectangle, the boulder being in the G-G1 half. 10Actually this area had a 3-degree slope toward the river and was only 3 ft. above the high-tide level. The house-mound top was 8 ft. above high-tide level. See pl. 15, Q. 199 ANTHROPOLOGICAL RECORDS N C\} a3) V) Map 3. Vunda (Site 26) and Environs 200 GIFFORD: EXCAVATIONS IN FIJI A4 A3 AZ Al A t5 5ED4 B3 B2 B1 B Cs C4 C3 CZ Cl C ,'DS D4 D3 D2 DI Di 'E5 E4 E3 E2 El E F5 P4 F3 F2 Fl F GS G4 G3 G2 GI G 85 H4 H3 H2 H I H IS I4 13 3 2 I1 I north K5 K4 K3 K2 __K I____K L5 L4 Lp' L2 L I L MS M4 /M3 Ma M I t \ N5 N4 1N3 N12 NI N 05 04 I 03 02 01 0 P5 P4 P3 P2 P1 P 05 Q4 1Q3 Q2 QI Q0 RS R4 1R3 Ra R1 R SO S 4 Si\ s 2 S I S Crossed rectangZes were excaated. east * , . Diagram 4. Staked area and excavations at location A, site 26. G-Gl H-Hl. In front of house mound. s The boulder prevented excavation below 60 in. H-Hl I-Il. In front of house mound. The boulder prevented excavation below 60 in. G2-3 H2-3. In front of house mound. H2-3 I2-3. In front of house mound. The boulder extended into the H2 I2 end of this rectangle. N-Nl 0-01. In house mound. This rec- tangle and the three following, all in the house mound, were not dug to the bottom of the deposit as time was limited and the main purpose was skeleton excavation. Nl-2 01-2. In house mound. 0-01 P-Pl. In house mound. 01-2 P1-2. In house mound. In wall 02-P2 there was a trace of a post- ole which extended downward from the shell- midden stratum into the substratum which con- tained no shell. When the post rotted, shell- midden material worked down into the posthole making a shell-midden core. Stake A at the northwest corner of the lo- cation was about 30 ft. from the Vunda River's present edge; formerly only 10 ft., before floods deposited the additional 20 ft., accord- ing to Ratu Tevita Namuira. The river was a small affair, really a creek, and in no wise comparable to the mighty rivers elsewhere on the island, notably the Rewa, the Navua, the Singa- toka, and the Mba. Nevertheless, the natives said that the Vunda has its times of high water and that the land directly in front of location A has been deposited by the floods of recent decades. This would make it seem that the low- est levels to which we dug were once riverbank, as at line A4-A5, diagram 4. Pockets of yellow sardy clay encountered in excavating were prob- ably deposited by floods, according to Ratu Tevita Namuira. The broken-line boundary for Vunda on map 3 is only an approximation, as no survey was made to determine the exact limits of the deposit. Our location A lay west below Lutunasomba- somba's house mound on the acropolis. (See pl. 15, a.) In a horizontal plane the distance was probably not over 200 ft., for the crags of the acropolis towered up behind location A. The upper stratum of location A was shell midden, in which Arca 'was the most abundant shell. River-mussel shells found here were said to be from places upstream in the Vunda River. Diagram 5 shows the strata at site 26, lo- cation A, as exposed in trench walls and as measured at the stakes. The cultural signifi- cance of the depth of the shell-midden layer is borne out by the analysis of soil samples pre- sented in the section on "Composition of Sites 17 and 26." Stakes: M In cs v F I ' DIAlagram 5. Strata at datum stakes at location A, site 26. 201 ANTHROPOLOGICAL RECORDS About 250 ft. from location A, below the southwest face of the crags and at the edge of a sugarcane field, we began excavation of a location B, but it proved so unproductive that we decided to concentrate all of our effort on location A. At B we started two rectangles 3 x 6 ft. and 18 ft. apart. Location B was among the bushes and coconut trees between the base of the crags and the far left-hand corner of the sugarcane field in plate 15, L. It is said that ploughing has eliminated all traces of war ditches in the lowland portion of Vunda. Floods may also have contributed to their elimination. The pottery found on the surface at site 26 was said to be from Malolo Island. Plate 16, b shows a cooking pot from there, presented to me by Andi Meri, daughter of Tui Vunda. Some Vunda people, it was said, settled on Malolo, which lies off the Vunda coast. Malolo and other out- lying islands and settlements supplied Vunda with pottery, since it was a chief village. Be- fore getting pottery from Malolo, the Vunda people obtained it from Waya Island, whose in- habitants were also from Vunda. After an ancient war with the Waya people, the Vundans moved the pottery-making to Malolo Island." On May 20 Rambithi and I, accompanied by informants Ratu Tevita Namuira of Viseisei and Ratu Naivoleoni Naturanga of Lomolomo, visited modern Lomolomo and an earlier site of the same name near the coast. Modern Lomolomo is on the inland side of Queen's Road, old Lomolomo on the seaward side; the latter site was planted to sugar cane at the time of our visit. At modern Lomolomo a boulder, now enclosed by a stone wall and protected by a platform, marks the legendary spot where Lutunasombaso and his companions bade farewell to the peop]. left behind at Vunda, when he set out for th Kauvandra Mountains, only to die on the way a Mangrongro, where he was buried. As the travelers climbed the rugged ridge Tualeita o Pathway of the Shades, the people below could see them bending over the trees with their hands, and therefore named the island Viti, which is said to mean "to bend," although Ca-? pell gives the meaning "to break off small branches. ,12 This may be only a folk etymologX of the island's name. Certain of the present Lomolomo people, yavusa Thawanisa, derive from Vunda. From th they moved to the older Lomolomo, then to Nan ganga,13 then to present-day Lomolomo. At old Lomolomo I saw sherds, but collected none; I have therefore not given the site a number. Both old and new Lomolomo are in the Distriet. of Vunda, Province of Lautoka. At ol-d Lomolomo, the informants pointed three ancient dug wells. Also they indicated the site of the temple of Ngathia, an oracula god represented by a stone 5 ft. high and abo 1 1/2 ft. in diameter. The god gave oracles through his stone, which was inside the house, the priest. If it leaned to one side, the answer was affirmative. Thus, if inquiry was' made concerning the success of war against Nandi, the leaning of the stone in the direct of Nandi indicated success for Lomolomo. If stone remained vertical or did not move, this constituted a negative answer. It was also consulted for weather predictions: rain was indicated if the top of the stone shook. COMPOSITION OF SITS 17 AND 26 Analysis of the soil samples taken from pit walls confirms statistically what was strikingly apparent to the naked eye, namely, a dual stra- tification, shell midden above and earth without shell below. These two main strata mark the two main culture horizons, the Late culture in the shell midden, the Early culture in the lower layers without shell. From pits dug at locations A and B, site 17, and locations A and B, site 26, 272 samples of deposit material were taken, mostly at 6-in. intervals from certain pit walls. In taking the samples no materials were selected: stone, sherds, shell, whatever came, were bagged for the samples. The physical analysis of the soil samples was made under my direction by Mr. Albert D. Mohr and Mr. Wilhelm G. Solheim II. It con- sisted of screening the samples through a wire screen with meshes 3 mm. square. The materials intercepted by the screen were segregated: stone, charcoal, bone, potsherds, shell. The "1One of my Vunda crew, Vilikesa Musuka, said that in pottery-making in olden times turtle bone was used for pad- dles instead of wood, which is now used in the paddle-and- anvil technique. residue, consisting of soil, ash, and fine fr ments of the materials just enumerated, was served for chemical and soil analysis. The several constituents segregated were weighed. The striking thing about both sites was prevalence of shell in the upper layers and i scarcity or absence in the lower layers, exce at location B of site 17, where shell occurre down to a depth of 126 in. Samples taken at 132, 138, and 144 in. at location B showed no shell. The deep occurrence of shell at loca- tion B is probably correlated in some way wit its seaside situation. The samples collected range in weight fr 304 gm. to 925 gm. These were taken from pit. walls after excavation had been completed. Si inch intervals were established as giving a f sampling. At site 17 the highest samples were taken 6 in. below the surface of the ground; t location A, site 26, at 3 in. below the surfao The procedure followed at site 26 should have been adopted at site 17 also, since it gives a 12capell, 1941, p. 310. 13A modern village, where the yavusa Viyangoisaukava, ; formerly of Vunda, now lives. I iJ.1 L I I I i I ,i j I 'O ,4 I I j I 202 GIFFORD: EXCAVATIONS IN FIJI ter sampling of the uppermost layer near the ace. At location A, site 17, samples were taken twelve walls; at location B from one wall, to a depth of 12 ft. At location A, site nine walls were sampled. At location B, 26, two shallow pits, which were started abandoned in favor of location A, were led to only 12 and 18 in. respectively. lyses of samples from these are omitted from [e 2. Stone.--The amount of unworked stone pres- has no cultural significance, since the s were at the base of crags and stone was efore abundant. Residue.--Beyond its use for chemical [ysis of ash and powdered charcoal content, residue is also of no cultural significance. amounts of the residue and of stone repre- t the negative side of the picture and vary ersely to the amounts of charcoal, bone, [l, and potsherds isolated, these last four s being culturally significant. Charcoal.--If we discount accidental fires, presence of charcoal indicates human occu- y of the levels at which it was found. This erpretation is validated by the co-occurrence potsherds in the same levels. Judging by the ivities of our Fijian crews, building a fire the first thing a Fijian does when he settles to work or camp. The fire today serves t of all as a convenience for lighting arettes or undertaking culinary operations. The amount of charcoal isolated from all three samples was less than half of 1 per St by weight. Some fairly large chunks pre- irved might possibly be used for tree-ring Alysis, but in Fiji there is no basis for ap- ying tree-ring study to chronology. Moreover, have no information about the wood preferred ir fires. My Fijian crews seemed indiscriminate L what they burned. One sample of charcoal 010) large enough for a Carbon-14 age deter- ,nation was obtained from the bottom of rec- rgle T4-5 U4-5, location A, site 17. It came om a hearth lying directly on the sterile lcanic-ash substratum in the rock shelter. il sample has been accepted for Carbon-14 age termination by Professor James B. Griffin, Iversity of Michigan. Bone.--The presence of-bone fragments, like e presence of charcoal, is to be interpreted an indicator of human occupancy at the levels which bone is found. Again, accompanying taherds substantiate this interpretation. In ly four samples did bone constitute more than lf of 1 per cent of the weight. In general, it f far below this amount. Potsherds.--The presence of potsherds is in- rpreted as indisputable evidence of human oc- pancy at all levels at which they were found. weight, potsherds ranged from less than half 1 per cent to 17 per cent of the total sample. Shell.--All marine and freshwater shells in e,analyzed samples are certain to have arrived the deposits through human agency. This Ltement would not apply to land shells. The Fzling absence of shells in lower levels, except at location B, site 17, is unexplained. However, two speculations suggest themselves. Perhaps shells were dumped elsewhere than at the locations dug; or possibly the Early people did not eat shellfish, even though they ate vertebrate species. Possibly the Early people arrived with a farming culture and a traditional abstinence from molluscan food. Dr. Cora A. Du Bois has called my attention to the parallel in the dietary of the people of Alor Island, Netherlands East Indies, where mollusks are wholly neglected as a source of food. I have published a speculative comparison of Fijian and Californian shell-midden deposits as con- nected with aboriginal diet.14 The remarks above apply to the locations A at both sites 17 and 26, but not quite so clear- ly to location B of site 17, where shell oc- curred in considerable quantity down to 120 in., with a trace still detectable at 126 in. I have already indicated that the seashore situation of location B might-have something to do with this. Elsewhere shell is a definite criterion of the- Late culture horizon only. In tables 1 and 2 the figures are percent- ages of the total weight of each sample. The percentages are rounded off to the nearest unit, 0.5 being counted as 1. Less than one-half of 1 per cent is entered as x. These approximate percentages result in a range of total percent- ages from 98 to 101, inexact, but sufficiently accurate for our purposes. Depths are in inches. In table 1, the composition of samples from pit walls at site 17 is presented. Walls A3-4 and C3-4 were in front of the house mound; Wall D3-4 at the edge of the house mound; Walls E3-4 and F3-4 on the slope of the house mound; Walls J3-4 and 03-4 in the house mound; Walls S3-T3, T3-4, T4-5, U3-4, and U4-5 in the rock shelter. The com?osition of pit wall Cl-1.5 at location B is presented in the last column of table 1 under the caption,location B. Table 2 presents the composition of samples from pit walls at location A, site 26. All pit walls were in front of the house mound except Wall P1-2, which was near the center of the house mound. 14Gifford, 1949a. 203 ANTHROPOLOGICAL RECORDS TABLE 1 COMPOSITION OF SITE 17 BY DEPTH: PERCENTAGES OF WEIGHT St, stone; C, charcoal; B, bone; P, potsherds; Sh, shell; R, residue (x, less than half ,of 1 per cent) Depth Wall A3-4 Wall C3-4 Wall D3-4 Wall E3-4 Wall F3-4 Wall J3-4 Wall 03-4 (in.) St C B P Sh R St C B P Sh R St C B P Sh R St C B P Sh R St C B P Sh R St C B P Sh R St C BP 6... 10 x11376 5xx 2 1974 2 x 1776 14xx51962 2 xx 3 13 63 3 x x 112 84 4 x x5 2 ... 9xx3l573 Oxx 41770 13xx31570 24xx 51753 47 1 x 6 4 5x12 18... 8xx4 188 12x 11076 15 x 31666 6x 31181 8 xx 42167 8xx3 1575 3xx3 24 .. 16 x2 82 x x 2 979 5xx3 2172 2xx 21680 12xx62457 7xx28 30... 8xx2 x90 17 x9156 8 x 3 90 4xx2 1579 6 x x 7 1869 4xx11481 8xx2 36 ... 3 x 2 95 24 x 1 x75 3 xx 515 77 3 x 11326 57 7 x 1 x 92 5 xx2 42 ... 17 x 83 2 x xl 97 20Oxx 1 x79 6 x 3 190 25 xx 5 12 58 4 1 95 5 xxl 48... 6xx 3 91 11 2 87 lx x x 4 x 3xx4 x93 5xx 54... 7 1x 92 5 x x 95 li x 1 88 7 x xx 93 6 x 2 92 6 x x 61... 4 xl 95 7 x2 92 21x x 79 1 09 66... ll x xx 89 26 xx 74 7 x xl 92 5 x xl 94 2 4 94 72... 14 xl 85 6 x2 92 7 x 4 89 78... 6x l 83 4x x 96 14 86 84... 24 x 1 75 6 x xx 93 10 x x 90 90... 8 1 91 5 xiS5 80 96... 8xx2 x90 22 x 1 78 102... 40xx 60 19xx17 64 108 ... 19 x 2 79 25 xx 4 x 70 114 ... 14 x 4 82 29 x 1 70 120... 19 x x x 80 126 ... 8 x x 92 132 ... 138... 144... Wall S3-T3 Wall T3-4 I Wall T4-5 Wall U3-4 Wall u4-5 Location B St C B P Sh R I St C B P Sh R St C B P Sh R St C B P Sh R St C B P Sh R St C B P Sh R i I I 7 x x 2 14 77 11 3 x x 8 77 4 x x 8 16 72 1 x 1 25 73 3 x x 4 30 63 6 x x 11 10 73 12 x x 2 16 70 3 x x 4 11 81 4 x x 3 10 83 8 92 6 x x 1 9 84 2 x x 4 35 59 2 x x 2 20 75 1 x x 1 7 90 8 92 3 x x 6 11 12 1 x x 9 6 x x x 9 9 x x x 13 6 3 6... 12. .. 18 ... 24 ... 30 ... 36. .. 42 ... 48. .. 54 ... 60 ... 66 ... 72 ... 78 ... 84 ... 90. . . 96. .. 102 ... 108... 114 ... 120 ... 126 ... 132 ... 138 ... 144. .. 79 77 84 78 94 97 4x 5x 7x 7x 1x x x 3 9 83 x 5 19 71 x 12 82 x x 21 71 x x 25 73 100 7 16 3 10 4 16 3 17 8 29 7 17 5 31 5 16 3 21 3 29 11 21 3 9 5 12 2 11 6 11 x 49 15 28 2 7 1 2 2 4 x I 31 x x 3xx 28 x x 4xx 13 x x 6x x 4 xx 5xx 4xx 11 x x 11 x x 8x x 12 x x 8xx 29 x x 16 x 2 21 x 1 68 x x 68 x x 29 x x 55 x x x x 47 84 51 74 50 70 60 74 72 56 57 80 71 79 53 34 34 23 30 64 45 100 100 100 -I - - I 204 GIFFORD: EXCAVATIONS IN FIJI20 TABLE 2 COMPOSITION OF LOCATION A, SITE 26, BY DEPTH: PERCENTAGES OF WEIGHT St, stone; C, charcoal; B, bone; P, potsherds; Sh, shell; R, residue (x, less than half of 1 per cent) th Wall A4-5 Wall c4-5 wall B-El Wall F-Fl Wall G-Gl St C BP Sh R St C BP Sh R St C BPSh R St C BPSh R St C BPSh R 8 xx 2 5 84 4 xx 6 3 88 lO x x2 3 85 7 x xl 3 89 8 xx 2 6 83 39. 3xx 2 5 90 14 x xl 1 85 20Ox xl 6 72 15 xx 5 7 73 ll x x2 3 84 5.. lxx x 99 8 x x2 x 90 7 x x2 6 86 8 x 2 6 84 l x x 3 7 80 1.. lxx 6 93 2 x xl 98 6 x xx x 93 5 xx 2 1 92 8 x x7 6 78 7.. lxx x 99 12 x xl 87 2 x xl x 97 4 x xl 1 94 18 x x7 x 74 3.. 8x x 2 80 lx xl 99 l x xl x 98 8 x xx 92 4 1 95 9.. x xlO 89 14 x x 85 6 x x2 92 lO x x 89 8 x 1 92 l xx 1 98 17 x x2 80 2 x xx 98 8 x x 92 6 xx 2 92 2.. 2xx x 98 9 x x4 88 ll x xx 88 50ox xx 50 3 x x3 94 7.. xx x 100 lO xxl1 89 4 x xl 93 18 x x2 79 13 x xl 86 3..12 x x 88 19 x x6 75 25 xx 2 73 9 x x4 87 9.. x x 99 l x xx 99 2 x x4 94 34 x x 65 5.. 5 x 1 94 l xx x 98 x xxl1 99 6 1 x 94 1... 2 x x 97 li x x 88 28 x 1 72 l x x 99 7.. 2x x 99 15 x x 84 l x 99 2 x 98 3e.. 1llx x 89 24 x x 76 2 x 98 6 1 93 9o o 1 x 98 5 x x 95 lO x 1 89 5... x x 100 l x x 99 l x x 99 1.,o. 12 x88 x x x 100 l x x 99 7... ~~~x x x 100 x 100 3... ~~~~~~~~~~~~100 Wall H-Hl Wall G-H Wall I2-3 Wall P1-2 St C BP Sh R St C BP Sh R St C BPSh R St C BPSh R 3.. 6x x x 1 92 7 x 2 1-90 11 3 x 86 8 xx 2 x 90 9... l8 xx 5 9 68 16 xx 3 14 67 28 xx 6 x 65 7 x x2 1 90 15.. lxx 5 4 73 17 xx 5 8 69 13 x x3 84 ll x x2 5 82 1.. 7 xx 5 9 78 8 xx 2 6 85 x xl x 94 9 x xl 3 87 ,27 ... 25 x 10 x 65 -13 x x1110o67 16 x x8 75 3 x xl 2 94 .33... 8 x xl 91 17 x 3 79 l0 x xx 90 17 x x3 8 72 139... 13 x x2 86 7 x xx 93 lS x x 84 12 xx x x87 45... ~~9 x x 2 89 4 x 96 12 x xx 88 51... ~~4 x xl 95 2 98 4 96 57o.,. ~~3 xx 4 x 93 5 95 x 100 .63 ... ~~4 x 1 95 x x 100 69.... ~ ~ ~~~~~~x 100 75 .. ~~~~~~~2 x 98 1187... .2...o 205 ANTHROPOLOGICAL RECORDS UNWORKED BONE FROM SITES 17 AND 26 In the excavations at sites 17 and 26, all bone was saved from the screenings, as a clue to the diet of early Fijians. This bone material suggests some of the species of fish, reptiles, birds, and mammals eaten by the ancients. It also casts light on the problem of the time when cannibalism began and when pigs, dogs, and chickens appeared as domestic animals. The total weight of fish bones recovered is 3113.1 gm.; the total weight of bones other than fish 4950.6 gm. The first division of bone consisted of segregating fish bones from other bones. The occurrence by weight in grams of these two cate- gories of bone is as follows: Site Site Site Site 17, 17, 26, 26, location A .... location B .... location A .... location B .... Fish Bones 1200.0 1525.4 387.7 Other Bones 3192.5 280.6 1471.5 6.o The preponderance of fish bones over other bones at location B, site 17, may perhaps be at- tributed to the seaside situation of location B. In tables 3 and 4 all of the columns except the last give data on material from the rec- tangles excavated at location A, site 17. The column headed B3-4 D3-4 gives data on material from rectangle C3-4 D3-4 from 0 to 72 in. and from the enlarged rectangle B3-4 D3-4 from 72 to 84 in. (The rectangle B3-4 C3-4 from surface to 63 in. was a test pit which was not dug in 6-in. blocks, hence the results are not incorporated in the tables for site 17.) The crossbar at the bottom of each column in tables 3 to 6 indicates the depth to which the completely excavated rectangle was dug. Bones obtained from top triangular blocks of soil on sloping ground have been counted with the 0-6-in. blocks. In all tables giving depths, the lower level of the 6-in. block only is given: i.e., 6 means the 0-6-in. block, 12 the 6-12-in. block. Identification of fish bones was kindly un- dertaken by Dr. Henry W. Fowler, Curator of Fishes, The Academy of Natural Sciences of Phila- delphia. Bones other than fish were in part identified by Miss Sheilagh Thompson, a graduate student specializing in osteology, under the guidance of Professors T. D. McCown of the De- partment of Anthropology and R. A. Stirton of the Department of Paleontology, University of California. The species of rodent bones segre- gated by Miss Thompson were identified by Dr. David H. Johnson of the United States National Museum. Bird bones were similarly studied by Secretary Alexander Wetmore of the Smithsonian Institution. Strangely enough, no snake bones were found, though the keeled tree boa (Enygrus bibronii Hombron and Jacquinot) of Viti Levu was eaten.'5 No whale bones or teeth were found, although cachalot teeth were much used as tambua for cere- 15Seemann, 1862, p. 155. monial presentations in historic times, as th are now. Perhaps their absence here, as well the absence of the large white cowry shells (Ovulum ovum) used on chiefs' houses and cano has some special significance. If recency i"; not the explanation, perhaps the high esteem these objects prevented their being found in rubbish heaps at the two sites excavated. IDENTIFIED FISH BONES Dr. Henry W. Fowler has the following to say about the total assemblage of fish bones sent to him for identification: Notwithstanding that the great majorit of the fish bones are unidentifiable, very interesting evidence is afforded by those revealing upwards of 20 determinable speciea besides the several genera for which the specific identifications are not clear or are unavailable. In all these cases the identifications are only possible from the detached dental plates, or teeth, either in' the jaws or pharynx. Frequently some of. the numerous vertebrae were in a more or ' less perfected state of preservation, but nevertheless unidentifiable. One may hardly even venture to assign them to any of the various sub-orders or higher groups. It isr only possible for a very few of them to per, tain to elasmobranchs. The following are those located in the- two sites. These are mostly well known Oceanian species and are abundant at the present time in a number of the islands. Doubtless many also occur in many of the islands and areas not yet explored. I hard-, ly feel therefore that any very general con- clusions may be made. Even in the Fijian Group likely at the time the present mate- rials were discarded, many of the species were abundant Just as today, and may there- fore have contributed to a most ample food supply. Likely a hundred species may have been utilized. According to Dr. Fowler the following spe cies, represented in the archaeological collec tion, have not hitherto been reported from Fij: Serranus undulosus (Quoy and Gaimard). Though this species has not been reported from the Fiji Group, it has been found in the Solomons and at Niuafoou Island. Isurus glaucus (Mueller and Henle). Though the species does not appear to ever have been reported from Fiji, it is widely distributed in the Indo-Pacific. In Oceaniau it is known from Hawaii. Promicrops lanceolatus (Bloch). Althou this species has never been reported from Fiji, the present specimen indicates its oc- currence. Reaching a length of 12 feet (3660 mm.) it is one of the largest of the I 206 GIFFORD: EXCAVATIONS IN FIJI TABLE 3 WEIGHT OF FISH BONES AT SITE 17 (in gm.) Location A Location A3-4| B3-4 | D3-4 | E3-4 | H3-4 I3-4 N3-4 R3-4 T3-4 s S5 T4-5 B B3-4 |D3-4 |E3-4 F3-4 I3-4 J3-4 03-4 T3-4 U3-4 T4-5 U4-5 .... 7.4 1 5.2 6.7 11.3 4.9 31.2 42.7 18.3 48.4 7 50.1 .... 1.5 5 14 3.3 4 2 722 6 9 40.3 5 35 1 5.5 11.5 4.4 6.1 4 21.5 24 9 14 14.2 27 ... 4 2 4 1.2 2.3 2 1 11.8 -13.2 l12.6 7 15 7~~~~~~~~~~~~~~~~~~~~7. 14.5 .3 8.9 6.215 3 1 .5 8 17 .... 16.4 4.4 1.1 32.7 12.7 7.1 6.6 20 2 13 .... _27 4 2.8 24.5 21.2 17.3 4.2 21 , *5 8.7 4 2 16.5 1 10.3 66 .. 14.1 3.5 16.4 24.6 2.1 7 135.4 5.6 12.5 28 13 512 .4 7505 ,, ~~1.7 9 2 13.1 13 52.7 @ - ~ ~6 6.3 6.4 7 5.4 63.5 ... 22.2 47.9 24.2 1 165.6 .... ~~~3.1 17 486.5 i; . . . . ~~26.5 162.2 .... .. - ~~16 48 F ~~~~~~~15.3 10 9.1 25 20.5 28.9 '_____ 6 .5 *The crossbar in each column indicates the depth to which the rectangle was dug. TABLE 4 WEIGHT OF BONES (EXCEPT FISH) AT SITE 17 (in gm.) Location A th* A3-4 B3-4 | D3-4 I E3-4 H3-4 I3-4 N3-4 R3-4 T3-4 s4-5 T4-5 Location B3-4 D3-4 | E3-4 F3-4 I3-4 J3-4 03-4 T3-4 U3-4 T4-5 |_U4-5 27.2 21.7 29 1 1 36.3 27 30 3.9 4.9 30.6 23 11.1 .6 2.3 14.9 12.4 1.3 32 o10.4 -3 81.3 11 380.7 12 13.9 2.2 3.2 31.7 12.5 8.5 15 8.7 17.1 4 15 82.4 17.7 20 15 32.3 27.1 9.2 2 6.1 10 9.1 11.6 9.8 28, 21.5 23 15.7 31.9 6 10.7 8.5 7.4 11.7 10.8 19 24.8 15.7 21.9 56 18.9 1 26.7 20.8 38 24 16.5 14.8 2.7 7 1.9 8.9 11 2.6 4 87.8 16.8 56.5 65 135.3 10.3 8.3 18.7 6 .6 37.3 51 72 16.2 95.8 20.7 11 17 23.2 7 1 360.4 25.7 27 33.4 5 7 16 13.3 16.6 21.9 74 1 86.9 6 6 39 2 35.6 2.5 10.3 41.5 3.4 *5 42.8 *The crossbar in each column indicates the depth to which the rectangle was dug. 207 ?o o o . . a . . . . . 0. . . 'o.o o 0 . . . . -0 . . . . Is.. . . 4 0. . . 0 *. . . 0 . . . . 0 . . . . 4 . . . . 0 0 . . . ANTHROPOLOGICAL RECORDS serranids. It is widely distributed in the Indian Ocean and western Pacific, being known from the Philippines, Micronesia, and Hawaii. Girellops species? Although G. nebulosa is from Easter Island the family Girellidae has not been recorded previously from the Fiji Group. Aetobatus narinari (Euphrasen). Although the species is known from Hawaii, Samoa, Johnston Island, Tahiti, New Hebrides, the Solomons and Guam, it does not appear to have been previously reported from Fiji. Tables 7 to 9 present the species identi- fied by Dr. Fowler listed by depths (but not by rectangles) at which they were found. Apparent- ly species cannot be correlated with cultural horizons; most species found in lower levels were found in upper levels also. IDENTIFIED MAMMAL, BIRD, AND TURTLE BONES Except for fish, man was the most popular of the vertebrate animals used for food. Frag- ments of his bones, some showing evidence of fire, were found down to 108 in. at location A, and to 120 in. at location B, site 17; down to 78 in. at site 26. It seems clear that human flesh was eaten in the Early period, though per- haps not from the beginning, as well as in the Late period. For the latter, of course, there is an abundance of historical testimony. We in- cluded no bones which were known to be parts of burials. If the latter-day Fijian taboo against women eating human flesh prevailed in prehis- toric times, it means that the per capita con- sumption by males was doubled. Miss Thompson identified 146 lots of human bone fragments, distributed between sites 17 and 26, as follows: site 17, location A, 94.lots; location B, 9 lots. Site 26, location A, 42 lots; location B, 1 lot. It should be remem- bered that location B of site 26 was sampled and abandoned as unproductive. After Homo sapiens, turtle and pig seem to have been the most popular vertebrates in the Fijian dietary. The turtle bones recovered from our screens were probably all from the green turtle (Chelonia mydas Linnaeus), which Mr. Joseph R. Slevin, Curator of Herpetology, Cali- fornia Academy of Sciences, informs me is the herbivorous and shore-haunting turtle of tropi- cal seas which is commonly eaten by man. At lo- cation A of site 17 we obtained turtle bones down to 114 in.; at location B on the seashore, strangely enough, only down to 78 in., perhaps illustrating again the fortuitous aspect of archaeology. Or possibly the historic Fijian practice prevailed of the fishermen's surrender of all turtles to the chiefs and the fishermen lived at location B, while chiefs resided at lo- cation A, a shadier and pleasanter situation. It will also be remembered that fish bones ex- ceeded all other bones in weight at location B, possibly evidence of a fishing community there. At site 26 we obtained turtle bones down to 72in. Miss Thompson identified 99 lots of turt bones, distributed between sites 17 and 26 as, follows: location A, site 17, 58 lots; locati B, site 17, 7 lots; location A, site 26, 34 X The occurrence of bones of the domestic (Sus scrofa Linnaeus) in the archaeological-, posits bears on the probable time of introduc6 tion of this mammal. At location A, site 17, pig bones were recovered down to 102 in., but seaside location B only down to 24 in.; at si 26 down to 60 in. At both sites the depths indicate the p ence of the domestic pig in the Early period,~ unless it is assumed that people of the Late period dug the earth ovens down into the Earl deposits, thus introducing the pig bones into the lower deposits. If this had occurred, t should be evidence in the form of shell from overlying Late period deposits. But there i8 such evidence. Rectangle E3-4 F3-4, location site 17, yielded pig bones at 60, 90, and 10?? but shell only down to 42 in. Rectangle E-El F-Fl, location A, site 26, yielded pig bones 60 in.; H-Hl I-Il at 54 in., but shell only 27 in. Miss Thompson identified 108 lots of pig bones from sites 17 and 26 as follows: locati A, site 17, 61 lots; location B, site 17, 2 lots; location A, site 26, 45 lots. The dog (Canis familiaris Linnaeus) is represented by 22 lots of excavated bones. one of these is from site 26--a single bone (6699) found associated with a human burial (15-7563), at a depth of 24 in. in the house mound at location A. The remaining 21 lots a from location A of site 17, and range in dept] from just beneath the surface to the level 54! 60 in. (in rectangle H3-4 I3-4). Nineteen of the lots were found above 42 in., two below., This puts the bulk of the finds in the shell- iridden deposit and in the Late period. The t lots from rectangle H3-4 I3-4 came from bloce 42-48 in. and 54-60 in. deep and presumably could be regarded as dating from the Middle period. Like the dog, the domestic fowl (Gallus gallus Linnaeus) is also represented by 22 lo of bones distributed between sites 17 and 26 follows: location A of site 17, 18 lots; lo tion A of site 26, 4 lots. At both sites the were found as low as the 60 to 66-in. block i the excavations, which perhaps puts them intoi the last part of the Early period. It is pos4 sible, of course, that all or part of bones r covered may be those of wild or feral birds. Ball, reporting on the birds collected by Mr. H. Beck of the Whitney Expedition in 1924, li. jungle fowls from six Fijian islands (Koro, K Makongai, Kambara, Vanua Mbalavu, Taveuni), b none from Viti Levu.16 The presence of chicken bones in sites and 26 indicates that the prehistoric Fijians were acquainted with Gallus gallus and that t4 famous bird Turukawa, that daily awakened the,, god Ndengei by his call, could have been a 16Ball, 1933, p. 22. I II 208 GIFFORD: EXCAVATIONS IN FIJI TABLE 5 WEIGHT OF FISH BONES AT LOCATION A, SITE 26 (in gm.) A4-5 B4-5 E-Ei F-Fl G-Gi H-Hi G2-3 H2-3 N-Ni Nl-2 0-01 01-2 B4-5 c4-5 F-Fi G-Gi H-Hi I-I1 H2-3 I2-3 0-01 01-2 P-Pi Pi-2 .5 2.4 1 .7 .5 .9 2 1 ,. 8 19 2 2 .5 5 2 .3 .. 1 3 3 17 18 6.2 3 3 .5 1 1 8 4 9 1 5.8 2 6.5 1.9 10 9 1 4 1..i9.7 2 2 2 6 2 11 .. 2 2 3 4 2 2 3 11.8 3 9.6 12 24.7 22.3 4 .. 30 18 14.3 4 2.6 2 *The crossbar in each coiumn indicates the depth to which the rectangie was dug. TABLE 6 WEIGHT OF BONES (EXCEPT FISH) AT LOCATION A, SITE 26 (in gin.) th* A4-5 34-5 E-Ei F-Fi G-Gi H-Hi G2-3 H2-3 N-Ni Ni-2 O-Oi 0i-2 B) 4-5 c4-5 F-Fi G-Gi H-Hi I-li H2-3 I2-3 O-Oi 0i-2 P-Pi Pi-2 7 7 26 20.i i i7 36 32 3 5 2 i. 6.3 i4 26 9 i3 i9.8 9.i 43.2 3 1 i3 ..61 Si 2 9.5 8 54 45 33 41 i6 i3.7 40 5 40 ii 20 62 54.8 8 34 2 i 27 ...i15.5 8 8 26 i2 4 i6 7 7 4La 9 - 6.7 .. i i8 i i2 i. 8.9 i 6 i i3.9 2 1 .3 2 i 6 5.2 3 37.4 i i. 6 9.5 25.3 33 __ 2 12.8 92 .4 39.3 27 i.4 *The crossbar in each column indicates the depth to which the rectangle was dug. 209 ANTHROPOLOGICAL RECORDS rooster, as my informant at Vunda asserted. How- ever, the published accounts, with one excep- tion,17 state or imply that Turukawa was a pigeon. Seemann writes of Turukawa's cooing and quotes the poetic English version of the epic furnished by John Hunt, who died in 1848.18 Seemann does not cite the exact source or the date of publi- cation of the excerpt. In 1866, Joseph Water- house also reproduced Hunt's version: "The late 17Lester, 1941, p. 103. I suspect that Lester may have had the same infformant as I, namely, Ratu Tevita Namuira, aSSeemann, 1862, pp. 394 ff. Rev. J. Hunt furnished the following as 'bet' an imitation and a translation of the originl Basil Thomson gives a version recorded by a FiJian informant.20 In 1922 Brewster identified Turukawa as beautiful dove which dwelt in the branches of mbaka or banyan tree (Ficus indica) called ulunda, which grew at the mouth of the god Ndengei's cavern.21 Later Brewster capitalize 19waterhouse, 1866, p. 359. See p. 357 for an accoun% of the origin of the pigeon slayers. 2'Thomson, 1908, pp. 135 ff. 2'Brewster, 1922, p. TABLE 7 FISH SPECIES AT LOCATION A, SITE 17 t Depth in inches 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96 102 Species Occurrence Aetobatus narinari (Euphrasen)............. Balistes species 9......................... Cataphracti 9. ............ Cheilinus fasciatus (Bloch) .............. Cheilinus fasciatus (Bloch) ?. Coris gaimard (Quoy and Gaimard)........... Diodon hystrix Linnaeus.................... Diodon species ?............... Epibulus insidiator (Bloch) ................. Girellops species ?............... Hemigymnus melapterus (Bloch).............. Hemigymnus melapterus (Bloch) ?............ Hemigymnus species ?....................... Hemitautoga centriquadra (Lacepede)........ Hemitautoga centriquadra (Lacepede) ?.. Iniistius species ?........................ Isurus glaucus (Mueller and Henle)......... Labroid.................................... Labroid ?.................................. Lethrinus species ?........................ Lutjanus bohar (Forskael).................. Lutjanus gibbus (Forskael)................. Lutjanus gibbus (Forskael)?................ Lutjanus species ?......................... Monotaxis grandoculis (Forskael)........... Monotaxis grandoculis (Forskael) ?......... Mugil species ?............................ Percoid ?.................................. Plectropomus maculatus (Bloch)............ Promicrops lanceolatus (Bloch)............. Scaroid ? .................................. Scarus microrhinos Bleeker................. Scarus microrhinos Bleeker ? ............... Scarus species ?........................... Scorpaeniformes ?.................. Serranus summana (Forskael)................ Serranus undulosus (Quoy and Gaimard). Serranus species ?......................... Sphyraena barracuda (Walbaum).............. Sphyraena barracuda (Walbaum) ?............ Sphyraena species ?.................... Taeniura lymma (Forskael).................. Teleostomi ?............................... Terapon jarbua (Forskael).................. Tetrodon species ?......................... x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x xx x x x x x x x x x x x x x x x x x x x x .. eIf . .' . 1 x x l4 x x x x x x x x x x x x x x x x x x x x x x x -1 iII ,i 1; .,1 'I II A I .i I , - I 'I I f I .1 - w .f I I 210 GIFFORD: EXCAVATIONS IN FIJI name Ulunda as though it applied to a defi- s place, not just to a tree: "In doing so hcending to Vatukathevatheva village] we led Ulunda, where Turukawa, Nd6ng-ei' s pet e used to reside."22 A photograph of the oe called Ulunda, taken by Ratu Rambithi, h a white cross indicating the spot where Wkawa is said to have been killed, has been lished.23 Bones of birds other than those of the do- tic fowl were found in excavations. There 53 lots of these, the depths at which they e found ranging from surface down to 72 in. Locations A and B, site 17; and to 72 in. at !Ibid. , p. 270. l3ifford, 1951, pi. 4, b. location A, site 26. Very few species can be identified. Those kindly identified by Dr. Alexander Wetmore are as follows: Site 17, location A: Anas superciliosa, 6-12 in. deep; Nesoclopeus poeciloptera, 6-12, 12-18, and 24-30 in. deep; Circus approximans, 12-18 in. deep. Site 26: Anas superciliosa, 60-66 in. deep; rail, undetermined, 12-18 in. deep; Ducula sp., 60-66 in. deep. Rat bones were identified only from site 17, from which we dug 18 lots, down to a depth of 72 in. The bones were segregated by Miss Thompson and sent to the United States National Museum, where Dr. David H. Johnson, Associate Curator of the Division of Mammals, identified the species. Thirteen lots proved to be Rattus norvegicus and TABLE 8 FISH SPECIES AT LOCATION B, SITE 17 Depth in inches 6 18 24 30 36 42 48 54 6o 66 72 78 84 90 96 102 108 120 126 Species Occurrence tobatus narinari (Euphrasen) ........ x istes species ? .................... x x x x x aphracti ........................... x x x x x aphracti ? .... . . ................... . x illinus fasciatus (Bloch) x x idon hystrix Linnaeus x Ddon species ? . ........... x x x x x Ibulus insidiator (Bloch). x Listius species ?. x thrinus species ? ................... x x x x x x tJanus gibbus (Forskael) x x x x tJanus gibbus (Forskael) ? ........ 0. x tJanus species ? .................... x bodontis species ? x x Rcoid ?. x Irus microrhinos Bleeker ..... x x Lrus microrhinos Bleeker ?. x rus species? ............. x x x x xx x x x rranus undulosus (Quoy and Gaimard). x rranus species ? x x x x iyraena barracuda (Walbaum). x miura lymma (Forskael) ............. x trodon species ? ...... ......... x x x x x x x x TABLE 9 FISH SPECIES AT LOCATION A, SITE 26 Depth in inches 12 18 24 30 36 42 48 54 60 66 72 Species Occurrence A6tobatus narinari (Euphrasen) ...... x Balistes species ? x Coris gaimard (Quoy and Gaimard) .... x x x x x x Lethrinus species ? . ... x x x x x x x x Lutjanus species ? x x Monotaxis grandoculis (Forskael).... x Percoid ? .x Scarus species 9 .. x x x x x x Serranus species ? x x x x x x Sphyraena barracuda (Walbaum) x Tetrodon species ?. .. x 211 ANTHROPOLOGICAL RECORDS five lots Rattus exulans. The following quota- tion from the letter of Dr. Remington Kellogg, Director of the United States National Museum, dated April 12, 1949, makes it clear why there was no stratification of the two species within the site. Rattus exulans is a small species that was carried all over the Polynesian area by the aborigines and*is the only rodent known to have occarred in Fiji prior to contact with white civilization. Rattus norvegicus, which is a heavy-bodied burrowing rat, and Rattus rattus, a lightly-built climbing species, were introduced by European com- merce. If R. norvegicus remains are found at levels representing the period before white contact, it would be well to keep in mind the burrowing proclivities of this species. The deepest identified exulans bones came from the block 66-72 in., the deepest norvegicus from the block 54-60 in. at location A. The flying fox or frugivorous bat (Pteropus tonganus) seems to have been an occasional vic- tim of the markmanship of the prehistoric Fijians. Location A, site 17, yielded twelvi lots of bones down to a depth of 60 in. Loca tion A, site 26, yielded two lots of bones, from a 54-60-in. block, the other from a 66-i in. block. Two domestic mammals introduced by CaucaI sians are represented among the bones recover by excavation. From the rock shelter at loc, tion A, site 17, came fragments of cattle bog down to 12 in. Since these occurred only in tangle S4-5 T4-5 and are doubtless those of modern introduced species, they have not beei, cluded in table 10. Goat bones were obtained, down to 24 in. at site 26 and to 36 in. at 10 tion A, site 17. None was found at location;j Tables 10 and 11 present the occurrence mammal, bird, and turtle bones excavated at sites 17 and 26 and identified by Miss Thompsi The depth to which each rectangle was dug is dicated by the crossbar at the bottom of the column. In table 10 for site 17 all columnsi cept the last refer to location A. In table for site 26, only location A is presented. XJ these two tables, bones obtained from top tril angular blocks of soil on sloping ground have been counted with 0-6-in. blocks. BLE 10 OCCURRENCE BY DEPTH OF MAMMAL, BIRD, AND TURTLE BONES AT SITE 17 (C, chicken; D, dog; F, flying fox; G, goat; M, man; P, pig; R, rat; T, turtle) Depth* Location A |Iocatio (in.)1 A3-4 B3-4 D3-4 E3-4 H3-4 I3-4 N3-4 R3-4 T3-4 5145 Tk-54 B3-4 D3-4 E3-4 F3-4 I3-4 J3-4 03-4 T3-4 |U3-4 T4-5 U4-. DMT MPT MP M DMP MP CDFMPT DMT FMP MT CGMT MR M MT M M MPR M M MT M M FMPT MP DMP MRT M M M M M C PT DMP CDMPT MPT DPT MPRT MPT DPT CM MR FMP T MR M MT M MT MPT MT MPT MT T *The crossbar at the bottom of the PT CDMPRT MPT M DM CM DGMPT PT FGMPT CGMPT PT M DMPR MP CDMPT PT M p T C MT MP M M column indicates the depth to which the rectan-le -t-as dug; CFMPT CGMPT DGMPRT MPT M 6. 12. 18. 24. 30. 36. 42. 48. 54. 60. 66. 72. 78. 84. 80. 96. 102. 108. 114. 120. 126. 132. 138. 144. CDFGMPT DMPT DM P DMPT DFMPT CFMT MR MRT CMT R MT M CMPT FMPRT MRT PR MPT FMP FMPT MPRT p p p CM MPRT DMP M MP MT T MPT T i MRT M MT MT M M I w A A A :1 j w I 212 I GIFFORD: EXCAVATIONS IN FIJI 213 TABLE 11 OCCURRENCE BY DEPTH OF MAMMAL, BIRD, AND TURTLE BONES AT LOCATION A, SITE 26 (C, chicken; D, dog; F, flying fox; G, goat; M, man; P, pig; T, turtle) | A4-5 B4-5 G2-3 H2-3 E-Ei F-Fl G-Gl H-Hi Ml-2 N-Ni 0-01 Ni-2 01-2 | B4-5 C4-5 H2-3 12-3 F-Fl G-Gl H-Hi I-Il Ni-2 0-01 P-PF 01-2 P1-2 MP MP MP MP P MP M MPT MP M M MT PT P MPT PT PT P P P MP M .... . CMPT P MP ...... ... PT MP GP M MP MPT M P MT PT CP P GP MP D MP MP MP .... . PT P P PT PT MP T MT MT MP .... . M P M MT MPT .... . T M M .... . T TM T .... . T M T T P .... . T FT CPT T M CT T FT T S.... M *The crossbar at the bottom of the column indicates the depth to which the rectangle was dug. MOLLUSKS Mollusks from archaeological sites were nly identified by Dr. Leo G. Hertlein and G. Dallas Hanna of the California Academy Sciences 224 From the surface of various sites I select- samples of the molluscan species in evidence. the two sites excavated other species were ed as the excavations proceeded. No attempt made to collect all shell, for there was a mendous quantity and, although it would have en useful in determining the extent of utili- tion of each species, the labor did not seem rranted in view of more pressing archaeologi- 1 problems and transportation problems. There- re, we collected merely sufficient archaeo- ical shell material to indicate the species curring in the shell middens of Viti Levu. All species collected are living in Fijian ters today. The fact that certain species re found only in excavation and not on the rface Is, in my opinion, to be regarded as rely fortuitous. Shells were rare in the un- rlying deposits of the Early period. River mussels (Batissa), as well as marine ecies, were and are eaten by the Fijians. The .bsence of archaeological examples of e white cci,ry (Ovulum ovum Linnaeus) poses a oblem. This species is used extensively today 24Kuphus sp. was identified by Mr. 2. M. Bayer, U.S. F ional MuisPeu? . as ornaments on chiefs' houses, for instance on the ends of ridgepoles. These ridgepoles have extensions of black tree-fern trunks on whose exposed ends the white cowries are attached and against which they are very conspicuous. The only old specimens of this species which I ob- tained were two presented to me by the aged Na Tui na Viti Levu of Rakiraki, Ra Province, who said they had been obtained from the "cave" in the Kauvandra Mountains where the serpent deity Ndengei is supposed to reside. These two (5010, 5011) were part of a lot brought down from the god's abode to decorate a new Christian church at Navuavua, Rakiraki, Ra Province. However, another and smaller cowry, Cypraea eburnea Barnes, used as a personal ornament by individuals of rank, was found archaeologically. Waterhouse gives a mythical example of the use of these shells as the body adornment of a man who came forth from an earth oven "beautifully attired, and adorned with small white cowry- shells on his legs, arms, &c.1"25 Concerning the very spiny Murex tribulus Linnaeus, I obtained an account of an interest- ing native belief. The species is called vula valu ("eight months"), because a spine which breaks off and remains imbedded in human flesh, makes a sore that lasts for eight months. The lists of species in the following 25Waterhouse, 1866, p. 383. I : 214 ANTHROPOLOGICAL RECORDS TABLE 12 MOLLUSKS FOUND ON SURFACE OF SITES Species Occurrence by Sites* 2 3 4 5 6 7 8 9 lo 14 16 18 19 20 21 22 23 24 25 27 28 29 30 31 32 33 34 35 36 37 38_ Arca ferruginea Reeve.......... Arca cf. ferruginea............ Arca fusca Bruguiere........... Arca scapha Chemnitz........... Arca cf. scapha................ Arca (Scapharca) sp............ Arca sp........................ Avicula sp..................... Batissa obesa Hinds............ Batissa cf. obesa.............. Batissa tenebrosa Hinds........ Batissa cf. tenebrosa.......... Batissa sp..................... Cardium cf. alternatum Sowerby. Cardium unedo Linnaeus......... Chama imbricata Broderip....... Chama cf. imbricata............ Corbis fimbriata Linnaeus...... Gafrarium gibbium Linnaeus..... Gafrarium cf. gibbium.......... Gafrarium sp................... Mesodesma striatum Gmelin...... Ostrea mordax Gould............ Pedalion ephippium Linnaeus.... Periglypta puerpera Linnaeus... Spondylus cf. barbatus Reeve... Tapes punctifera Lamarck....... Tridacna noae Bolten........... Tridacna cf. noae.............. Tridacna sp.................... Bursa cf. lampas Linnaeus...... Cerithium nodulosum Bruguiere.. Conus litteratus Linnaeus...... Conus striatus Linnaeus........ Conus sulcatus Hwass........... Conus sumatrensis Hwass........ Conus tahitiensis Hwass........ Cypraea cf. annulus Linnaeus... Cypraea errones Linnaeus....... Cypraea tigris Linnaeus........ Cypraea cf. tigris............. Harpa conoidalis Lamarck....... Melania sp..................... Murex ramosus Linnaeus......... Nerita albicilla Linnaeus...... Nerita polita Linnaeus......... Oliva ?cf. funebralis Lamarck.. Pterocera lambis Linnaeus...... Pterocera cf. lambis........... Pterocera sp................... Strombus cf. floridus Lamarck.. Strombus gibberulus Linnaeus... Strombus luhuanus Linnaeus..... Strombus sp.................... Terebra maculata Linnaeus...... Thais armigera Chemnitz........ Trochus niloticus Linnaeus..... Trochus cf. niloticus.......... Trochus obeliscus Gmelin....... Trochus verrucosus Gmelin...... Turbo argyrostomus Linnaeus.... Turbo chrysostomus Linnaeus.... Turbo cf. chrysostomus......... Turbo crassus Wood............. Turbo petholatus Linnaeus...... Turbo porphyrites Martyn....... Turbo setosus Gmelin........... Vasum ceramicum Linnaeus....... Vasum turbinellum Linnaeus..... x x x x x x x I x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x ..r xi x ' x x x x x x x x x x x x x x x x: x x x x x x x x x x x *For sites 17 and 26 see tables 13-15. I : I " 4 ,I .j -,' GIFFORD: EXCAVATIONS IN FIJI es add some species to the list given by man (1941), although he gives some which I ot obtain. Possibly some of the smaller es I obtained were not eaten. Steadman that cowries are never eaten. .When collecting spider or scorpion shells ocera lambis Linnaeus) with their project- canalized fingers, the Fijians immediately off these sharp extensions of the shell, h are a nuisance in carrying.26 Concerning the extraction of the animal the shell, Steadman writes: "When boiled, .sometimes when eaten raw, the animal is ex- ted from the shell by means of a lemon thorn, rwise the shell is shattered." The archaeo- cal specimens afford plenty of testimony 0For a good picture of a scorpion shell, see Rogers, , p. 122. 215 that this is true. Comparison indicates that some species and many genera of Fijian mollusks are also found among the specimens excavated from Mesolithic and Neolithic strata in the cave of Dudumunir, West New Guinea.27 Tables 12, 13, 14, and 15, giving the oc- currence of molluscan species in the shell mid- dens, list bivalves first, in alphabetic order. then univalves. Table 12 lists surface occur- rence for sites other than 17 and 26. Tables 13 to 15 list occurrences by depth for sites 17 and 26, but disregard the rectangles in which specimens were found. In table 13 mollusks found on the surface at site 17 have been listed under location A of that site, whether actually found at locatIon A or not. I 27Benthem Jutting, 1940, p. 20. 'LE 13 MOLLUSKS AT LOCATION A, SITE 17 Depth in inches* s 6 12 18 24 30 36 42 48 60 Species Occurrence Arca cf. fusca Bruguiere . . .............. x Arca scapha Chemnitz .................... x Arca cf. scapha ......................... x x Arca sp . . ............................... x x Atrina nigra Chemnitz (perhaps) ......... x Avicula cf. lurida Gould ................ x x Avicula sp ............................... x Cardium enode Sowerby ................... x Cardium unedo Linnaeus .................. x Cardium sp .............................. x Chama imbricata Broderip ................ x Corbis fimbriata Linnaeus ............... x Gafrarium gibbium Linnaeus .............. x x x x Gafrarium cf. gibbium . .......... x x Gafrarium sp . ........................... x x Ostrea cf. circumsuta Gould ............. x Ostrea mordax Gould ..................... x x x x x x Ostrea cf. mordax ..... x Pecten radula Linnaeus x Periglypta puerpera Linnaeus ............ x Periglypta puerpera (probably) ......... x Periglypta puerpera (possibly) ......... x x Periglypta cf. puerpera . ......... x x x x Pitar cf. affinis Gmelin . . . x Septifer bilocularis Linnaeus . ...... x x Septifer cf. bilocularis ................ x Tapes cf. indica Hanleyt x Tridacna noae Bolten .................... x x Volsella hepatica Gould .................x x x *S equals "surface." tMS in Sowerby. I I C'- - I! I ANTHROPOLOGICAL RECORDS TABLE 13 (Continued) MOLLUSKS AT LOCATION A, SITE 17 Depth in inches S 6 12 18 24 30 36 42 48 60 Species Occurrence Columbella versicolor Linnaeus .......... x Conus litteratus Linnaeus .... .......... x Conus planorbis Born .................... x Cymatium pileare aquatile Reeve x Cymatium sp ............................. x Cypraea annulus Linnaeus x Cypraea cf. arenosa Gray . . . x Cypraea eburnea Barnes . . ................. x x x Cypraea errones Linnaeus ................ x x x x x x x Cypraea cf. errones ..................... x x x x Cypraea moneta Linnaeus ................. x Cypraea cf. tigris Linnaeus .............. x Cypraea sp... x Euchelus atratus Gmelin . . . x Latirus polygonus Reeve .................. x Melania cybele Gould .................... x Melania cf. cybele ...................... x Melania sp . .. x x Murex ramosus Linnaeus .................. x x Murex cf. ramosus . . .x x x Murex cf. tribulus Linnaeus x Natica mamilla Linnaeus ................. x Nerita albicilla Linnaeus ............... x Neritina chrysocolla Gould x x Neritina cornuta Reeve .................. x Neritina helvola Gould x Oliva erythrostoma Lamarck .............. x x x Oliva cf. erythrostoma x Pterocera lambis Linnaeus ............... x Pterocera sp ............................ x Terebra maculata Linnaeus ............... x Thalotia sp ............................. x ? Thalotia sp x Trochus niloticus Linnaeus .............. x Trochus verrucosus Gmelin ............... x Turbo petholatus Linnaeus (probably) x Turbo porphyrites Martyn ................ x Turbo porphyrites porcatus Reeve x Turbo porphyrites cf. porcatus .......... x TABLE 1 4 MOLLUSKS AT LOCATION B. SITE 17 Depth in inches 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96 102 108 114 120 1 Species Occurrence Arca scapha Chemnitz ............. x x Arca sp . ........... x x Batissa tenebrosa Hinds x Cardium (Hemicardium) unedo L x Cardium sp .......... ; ... x Chama imbricata Broderip ......... x x Chama sp x Corbis fimbriata Linn ............ x Gafrarium gibbium Linn ........... x x x x x x Gafrarium cf. gibbium x Kuphus sp ....X Ostrea cf. circumsuta Gld ........ x Ostrea mordax Gould . .x x x x Ostrea cf. mordax ................ x Ostrea cf. subtrigona Sby x I I i I I 216 GIFFORD: EXCAVATIONS IN FIJI 217 TABLE 14 (Continued) MOLLUSKS AT LOCATION B, SITE 17 Depth in inches 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96 102 108 114 120 138 Species Occurrence a sp ....................... X glypta cf. puerpera L ........ x wnipagus palatam Iredale ...... x inolaria elongata Lam x tifer bilocularis Linn x x x Ftifer cf. bilocularis. x x pes indica Hanley* x se punctifera Lamarck ......... x x 8 BP X rt'thium nodulosum Brug x hus litteratus Linnaeus ........ x hus striatus Linnaeus x x x iraea eburnea Barnes x x traea cf. errones Linn ......... x x x Lea pomum Linnaeus x Era episcopalis Linn ........... x bra fenestrata Lamarck ......... x rex ramosus Linnaeus ........... x rex ramosus (possibly) ......... x Osarius kieneri Anton. x isarius cf. picta Dunker x !ombus plicatus Lam ............ x ilotia sp . . x rbo chrysostomus Linn x ium cornigerum Lamarck x Pmetus sp x *In Sowerby. TABLE 15 _________________________ MOLLUSKS AT LOCATION A. SITE 26 Depth in inches* S 6 12 18 24 30 Species Occurrence pca cf. ferruginea Reeve ....... x x Pca pilula saccula Iredale x rca scapha Chemnitz ............ x x rca sp ......................... x rina sp . x itissa tenebrosa Hinds .. . x x itissa cf. tenebrosa. x itissa sp ...................... x irdium enode Sowerby . x hama cf. imbricata Broderip .... x iama p .x Lfrarium gibbium Linnaeus ...... x 8BOdesma striatum Gmelin ....... x x itrea sp ....................... x tdalion sp. x riglypta puerpera Linnaeus .... x Periglypta sp ......... x pondylus cf. barbatus Reeve.... x pondylus cf. ducalis Bolten.... x 'idacna noae Bolten ............ x x ,maea saccharina Linnaeus . x irithium cf. asper Linnaeus .... x rithium tuberculatum Linnaeus. x x nus marmoreus Linnaeus . x nus virgo Linnaeus ............ x rpraea cf. annulus Linnaeus .... x praea eburnea Barnes. x rpraea cf. eburnea. x rpraea sp ..... ................. x *S equals surface. - - - - I I - - I I . . . . . L _ _ _ _ _ _ _ _ _ _ Depth in inches* S 6 12 18 24 30 Species Occurrence Melania cf. aspirans Hinds ...... x Melania cybele Gould . . .......... x Melania cf. scipio Gould ........ x Melania tetrica Gould . ..x x Melania sp . ... ....... la. x x Mitra adusta Lamarck. . x Murex ramosus Linnaus. . x Murex cf. ramosus Linnaeus ...... x Natica cf. flemingiana Recluz x Natica mamilla Linnaeus. . x Natica virginea Recluz. . x Nerita albicilla Linnaeus ....... x Nerita polita Linnaus. . x Nerita sp . .x x Neritina crepidularia Lamarck x Oliva funebralis Lamarok ... x x Oliva sp . ............. x x x Pterocera cf. lambis Linnaeus... x Pterocera lambis (probably) x Strombus plicatus Lamarok . x x ? Thalotia sp . . x Trochus niloticus Linnaeus ...... x x ? Trochus niloticus Linnaeus x Trochus verrucosus Gmelin ...... x x Turbo chrysostomus Linnaeus ..... x x Turbo crassus Wood . . ............ x Turbo porphyrites Martyn ........ x x Turbo porphyrites Martyn var ... x Turbo setosus Gmelin . . .......... x 218 ANTHROPOLOGICAL RECORDS CRUSTACEANS, ECHINODERMS, AND CORALS FROM SITES 17 AND 26 Steadman lists several species of crusta- ceans eaten by modern FiJians 28 Scylla serrata, large mud crab; Calappa fornicata, rock crab; Panulirus penicillatus, crayfish; Thallasina anomala, mangrove lobster; Palaimon sp. (fresh- water prawns). "These together with several other kinds are boiled and eaten." Fragments of crustaceans obtained from the excavations were small and few in number. How- ever, Dr. Fenner A. Chace, Jr., of the United States National Museum, kindly consented to ex- amine these and has made the following identifi- cations of genera and species: Etisus laevimanus Randall, crab; Etisus sp.; Scylla serrata, crab; Sesarma sp., marsh crab; Calappa sp., crab. Other specimens are listed merely as crab. At location A, site 17, the lowest record for crab is 54 in. in rectangle H3-4 I3-4; at location B, Etisus laevimanus was obtained at 90 in. and Calappa sp., possibly C. hepatica (Linnaeus), at 96 in. At site 26, location A, only one species was identified: Scylla serrata from a depth of 12 in. Most occurrences of crustacean remains were in shell midden; crusta- ceans, like mollusks, were presumably missing from the diet of the Early period Fijians. Two sorts of echinoderms were identified in excavated material by Dr. Austin H. Clark of the United States National Museum: From the rock shelter at location A, site 17, came two "sand dollars" from a depth of 12 to 24 in. From loca- A8Steadman, 1941, p. 9. I tion B of the same site came a "heart urchil from a depth of 84 in. Under the caption " facts of Marine Calcareous Materials," I hay mentioned spines of Heterocentrotus mammilli With the exception of one specimen of'' (5603), identified by Dr. F. M. Bayer of the United States National Museum as Millepora cornis Linnaeus, all coral was identified b Professor John W. Wells of Cornell Universit Specimen 5603 came from a depth of 18 in. in rock shelter at location A, site 17. Table 16 gives the distribution by dept but not by rectangle, of the corals preserv samples from site 17 (locations A and B) andr cation A,site 26. Like shell, most coral waa discarded at the time of excavation. Native formants stated that the coral was brought t the sites to be calcined and used as a hair and bleaching agent. This explanation, base modern practice, may very well be true of the archaeological examples, which with a few ex. ceptions come from the Late or shell-midden horizon. Lumps of "lime" (5013) found by Ratu Ram bithi near a pool in Nakauvandra Creek in the hills south of Navatu (site 17) were of mytho logical significance; it is believed that the were left there by the goddess Naivilawasa wh she was liming her hair. This "lime" was ide tified by Professor John W. Wells of Cornell University as calcareous algal nodules (litho4 thamnion). In addition, a fragment (6110) waO excavated in rectangle E-El F-Fl, location A, site 26, depth 24-30 in. TABLE 16 CORALS AT SITES 17 AND 26 Depth in inches 6 12 18 24 30 36 42 60 66 84 132 Species Occurrence Site 17, location A Acropora species ?.................. x x x x x Millepora alcicornis (Linnaeus).... x Platygyra rustica (Dana) ........... x x Porites cf. lutea M.E. and H ....... x Porites species ? .. ... x Stylophora pistillata (Esper) x Site 17, location B Acropora species 9... ............... x x x x x x x x Fungia concinna Verrill ............ x Fungia fungites (Linnaeus) x Goniastrea rectiformis (Lamarck) x Leptastrea purpurea (Dana) x Porites species ? . .x x x Site 26, location A Acropora species ?.................. x x x Fungia concinna Verrill ...... x Platygyra rustica (Dana) ........... x x x Stylophora pistillata (Esper) ....... x I ?.l -o 11 I . I A GIFFORD: EXCAVATIONS IN FIJI BOTANICAL SPECIMENS FROM SITES 17 AND 26 The botanical material recovered from exca- .ons, aside from charcoal, was chiefly a es of fragments of shells of the candlenut, rites moluccana (L.) Willd. These were ly identified for me by Dr. A. C. Smith, tor of Phanerogams, United States National uum, a specialist on Fijian plants. The dis- ution of the archaeological nut material is ollows: Site 17, location A, rectangle B3-4 D3-4, Site 17, Site 26, 76-85 in. deep. location A, rectangle E3-4 110 to below 116 in. deep. location A, rectangle A4-5 91-97 in. deep. F3 -4, B4-5, Site 26, location A, rectangle F-Fl G-Gl, 54-60 in. deep. Presumably these are the shells of nuts uti- lized by natives during the Early period, since they come from such deep levels. A small fragment of worked wood (11564), about 3/4 in. long, was recovered from site 26, location A, rectangle E-El F-Fl, at a depth of 48-54 in. It shows a typical hardwood pore and ray structure and, according to Profeseor Emanuel Fritz, is extremely hard and perhaps slightly silicified. Its two worked faces form a right angle. Possibly it was part of a bow or a club. BURIALS Excavations in locations A and B of site 17 lded no burials. Skeletal remains were found, ever, at about 40 ft. elevation on the slope the hill above the junction of the main line the railroad and the Narewa spur track. In midden material composing the talus of the 1 slope human bones were seen protruding from bank of a tiny arroyo the rains had cut. re no fewer than five skeletons were uncovered. o of these--one, the skeleton of a child, the er that of an adult--were found in separate sitions. Then there were three adult skele- n., with their heads in juxtaposition. The skulls of all were to the east and the gs were folded under the bodies, which lay on ir backs. There had apparently been some dis- acement because of down-slope sliding of the den material. No funeral artifacts were nd with the skeletons. Potsherds of the Late nod were abundant throughout the shell-midden posit and equally so about the remains, which ld indicate that the skeletons were also Late. ese bones are catalQgued as 12-7499 to 12-7508. ee plate 13, c. No burials attributable to the Early period wre found at either site 17 or site 26. At location A of site 26 five skeletons wre taken from the circular house mound (p1. I5, c). In addition to these we found the re- iains of a child and an infant which were too fragmentary to take. There may have been others, ror only the central portion of the house mound tas trenched. Aside from a shell bracelet shown in fig. 1, e), found on a man's skeleton (12- r561), encircling the left arm above the elbow, no associated objects were found, although there were many potsherds in the soil. The details concerning the burials within the house mound at location A follow. These burials are no doubt )f the Late period, but all were rather badly lisintegrated, apparently owing to the damp soil. Skeleton 12-7560. Top of skull was highest part of burial and was 17.5 in. below the surface, north of the center of the house mound. The head lay to the southeast, the pelvis to the northwest. The skeleton was on its back, head turned to left, arms at sides, hands together at pelvis, legs folded against body on right side. The skull projected into rectangle 01-2 P1-2;. the leg bones were part- ly in rectangle N-Nl 0-01; the greater part of the skeleton was in rectangle 0-01 P-P1. Lying on the left side near the pelvis was a child's skeleton; also bones of new-born babe, all too badly disintegrated to collect. Skeleton 12-7561. The top of the skull was 23.5 in. deep. The body lay extended on the back, chin resting on backbone, arms at sides, lower leg bones in oblique position with feet crossed. The legs passed over stones of an earlier house mound. The head was to the southeast, the legs to the north- west. The head and trunk were in rectangle 0-01 P-Pl; the leg bones in N-Nl 0-01. Part of this skeleton underlay 12-7560. Skeleton 12-7562. This was a child's skeleton. It was 18 in. below the surface, with head to southeast in wall P1-2, the body extending northwest into rectangle 01-2 P1-2. It lay on its back full length, with face turned to left, arms at sides. Skeleton 12-7563. The body lay across rectangle M2-3 N2-3, mostly on the M2-N2 side. The head lay to northeast with face turned to north and 24 in. below the surface at stake N2. The body was laid out full length with feet to the southwest. A dog bone (6699) was found in adjacent rectangle Ml-2 Nl-2 and may have been associated with the burial. Skeleton 12-7564. This was only a par- tial skeleton--skull, jaw, and arms; all else was missing. It was uncovered with the aid of Dr. Lindsay Verrier. The right arm stretched to the north, the left was folded toward the right arm; the head was to the east. The skull lay at a depth of 15 in., in the middle of rectangle N2-3 02-3. It is possible the arm bones were from another body. The reason for the fragmentary condition of the skeleton was not obvious; possibly canni- balism may account for it. The skull was badly crushed laterally. I 219 ANTHROPOLOGICAL RECORDS ARTIFACTS OF MARINE CALCAREOUS MATERIALS Shell artifacts were scarce indeed and mostly damaged or fragmentary. Two thin pieces of Tridacna, broken to more or less triangular form, may have been intended for manufacture in- to adze blades, though this is by no means cer- tain and the shapes may be fortuitous. A large lump of white shell appears to have been de- rived from the hinge region of a huge Tridacna, and may be a remnant left after tool-making. Pieces had been sawed and broken off from an elongate, angled, faceted piece of Tridacna hinge (5650); this specimen comes-from rectangle D3-4 E3-4,location A, site 17, 6-12 in. deep. Shell paring knives.--Figure 1, . shows a modern breadfruit peeler presented by a native of Singatoka, Nandronga Province. It was made of a shell of Fasciolaria filamentosa Lamarck, by abrading the whorl opposite the aperture so as to make a circular opening with knifelike edge. Five smaller archaeological examples were collected on the surface of sites 20 and 21 and represent three species of the genus Turbo: 5237 from site 20, Turbo crassus Wood; four others from site 21 are 5245 and 5247 (both Turbo cras- sus), 5244 (Turbo chrysostomus Linnaeus), and v 6 (Turbo setosus Gmelin). Shell hook.--Figure 1, c shows a shell fish- hook shank. It was excavated at a depth of 24- 30 in. in the rock shelter at location A, site 17. It belongs to the Late culture horizon. In form it resembles the shank of a composite hook from the Solomon Islands, like one figured in the British Museum Handbook to the Ethnographi- cal Collections.29 The hook portion is for at- taching the line, not for hooking the fish. Beasley pictures a number of fishhooks of this type and states that the shanks are always of pearl shell.30 My specimen appears to be of pearl shell also, cut from near the hinge. Beasley in his general work on Oceanian fish- hooks, reports the type only from the Solomon Islands. Shell "stick".--A partly beveled shell "stick," from an indeterminate depth between the surface and 33 in. in test pit B3-4 C3-4, loca- tion A, site 17, is shown in figure 1, h. It is not sufficiently curved to be part of a bracelet. The broader end has a transverse groove that suggests the piece may once have been longer and perforated, the groove being the-remains of the perforation. The piece can probably be attri- buted to the Late culture horizon. Shell bracelets.--The only complete object found during all of our excavations was a shell bracelet of what appears to be pearl oyster on the upper left arm of adult male skeleton 15- 7561, excavated at a depth of 23.5 in. in the house mound at location A, site 26. Unfortunate- 29British Museum, 1910, fig. 3b, p. 9. 34Beasley, 1928, p. 68; pls. 105-107. ly, it was broken in removal. The repaired specimen is shown in figure 1, e. From the same house mound, depth 24-30 comes a fragment of a bracelet with deeply cave exterior (fig. 1, b). Location A, site 17, yielded the fragme of a bracelet of Trochus niloticus shell, 8 ing characteristic pink spots. It came from; test pit B3-4 C3-4, from somewhere between t surface and 33 in. (fig. 1, d). A fragment of shell ring, too small to bracelet except for a small child, was ident fied by Dr. Leo G. Hertlein as Conus, possibi literatus, Linnaeus. It was excavated at lo tion B, site 17, depth 84 to 90 in. (fig. 1,- Coral "tapa marker."--A "stick" of colo coral, probably Astreopora, according to Pro sor J. W. Durham, ground down to four longitt dinal sides tapered to a point, was dug at lo tion B, site 17, depth 66 to 72 in. The natl said it might have been used as a tapa marker antiquity. It is 63 mm. long, and 12.5 mm. each side at the broadest part, near the b4se which has apparently been broken off. Figure 1, J shows the piece. Sea-urchin spine.--An unmodified sea-urC spine (6395), 86 mm. long, was excavated at a depth of 78 to 84 in. at location B, site 17. Its three-faceted terminal end caused me to think it was an artifact; I questioned the Fijians about its function and they suggested that it might have been a tapa marker. Profe sor J. Wyatt Durham, Museum of Paleontology, University of California, identified its spec as Heterocentrotus mammillatus Linnaeus. A smaller spine (5620) of the same species was found in the rock shelter at location A at a depth of 18 to 24 in. in rectangle T3-4 U3-4. The piece is a portion of the proximal end, 3 long. There is a trace of what appears to ha been a transverse perforation in the distal e of the fragment, so it is to be regarded as artifact. Pot scribe or marker.--What is possibly tool for making parallel incised lines on pottery before firing is represented in figure 1, g. The specimen is a comblike bit of edge Cardium valve excavated at a depth of 18 to 24 in. at location B, site 17. Shell adze blades.--The distal end Of one adze blade of Tridacna shell is pictured in figure 1, f. The width of the cutting edge is. 33 mm. The specimen was excavated at a depth 66-72 in. in location B, site 17. The British Museum Handbook to the Ethnographical Collec-4 tions shows in its figure 10, je. a rather simil shell blade from Fiji. A broken unfinished ob ject of Tridacna, shown in fig. 1, kj 12 mm. ameter, is perhaps the distal portion of a slender pointed adze blade or chisel in proces of manufacture. It is from location B, site 1 12 to 18 in. deep-. These two specimens are the only shell adze blades obtained. 220 GIFFORD: EXCAVATIONS IN FIJI STONE ARTIFACTS Stone artifacts are rare on the surface of ri archaeological sites and equally rare around. Considering the enormous number of erds in evidence, the scarcity of stone ob- seems even greater by contrast. Apparent- enever anyone -- Fijian, European or In- -- finds a fairly perfect adze blade he a it home, just as we would an arrow point. habit helps to explain the scarcity of adze es on the surface of sites, but not their ity in excavations, from which we obtained a few, and these mostly fragments. Adze blades and chisel.--During the recon- sance of archaeological sites before ex- ting, a small series of adze blades was ac- lated by gift and purchase. Fragments were collected on the surface of some sites. gave rise to the hope that excavation would d a good series from different levels. This was not realized. Quantities of potsherds forth from the ground, but only an occa- al pitiable fragment of an adze. Besides my es of adzes from Viti Levu, I have four spe- ns from Ovalau and Taveuni islands. I am indebted to Professor Howel Williams the identification of the stone of adze es. Of the series submitted to him, shown igures 2 to 4, he writes: With the exception of the large New Cale- onian specimen composed of nephrite serpen- ne (2845), all but one of the others ap- pear to be either dense, black, non-porphy- itic basalts, or strongly porphyritic yroxene andesites. However, accurate de- ,,termination of the black, fine-grained rocks is not possible without microscopic study. The other non-volcanic specimen (6404) is 'either a diabase or gabbro. The transverse cross sections of the adzes wn in figures 2 to 4 represent the thickest rt of the implements. The upper edge of the ss section represents the front of the adze, e lower edge represents the back, or most veled part. My full view is of the back of e adze, my profile view shows the left side. terms "front" and "back" are here used as in e article by Buck and colleagues on "Termin- logy for Ground Stone Cutting-Implements in lnesia."131 My Fijian specimens show the same range of ross sections as Beyer's typical cross sections or Philippine Early Neolithic adzes. Beyer's rawings of complete Philippine Early Neolithic dzes are also reminiscent of Fiji.32 Laura Thompson classifies Lauan adzes in our main types according to their transverse ross sections: circular, elliptical, trape- Foidal, rectangular.33 The first three are rep- 31See their diagrams, p. 176. 32Beyer, 1948, fig. 8, cross sections 1-4, fig. 9; idem, 1949, p1. 6. 33Thompson, 1938a, pp. 98-101. resented in my Fijian series. The rectangular Western Polynesian type found in the Lau Islands is lacking in my Fijian series, except perhaps for my specimen (fig. 4, i) from the surface of site 17, which somewhat resembles it. However, in figure 2, b I have illustrated a Tongan adze which matches the Lauan rectangular type34 no doubt derived from Tonga. Two broken adzes with rounded cutting edge and elliptical transverse cross section are represented by specimens from Ovalau Island (fig. 2, f) and Taveuni Island (fig. 2, j). These resemble some of our illustrated pieces from both.the southern and northern parts of Viti Levu. A series of adze blades was obtained mostly by gift and purchase along the southern side of Viti Levu from the Singatoka River in Nandronga Province eastward to the vicinity of Suva. All are from the surface or obtained near the sur- face during such digging operations as tilling. Certain of the blades suggest New Caledonian types as shown in the British Museum Handbook to the Ethnographical Collections,35 some of the specimens illustrated by Sarasin,36 and the one in my figure 3, k, although this last is more lenticular (sharp-edged) than any Fijian example. New Guineea Neolithic adzes figured by Van der Hoop37look very much like those of Fiji. The amount of curvature of the cutting edge varies and it is possible to arrange my specimens in a more or less continuous, intergrading series from curved to straight cutting-edge. A meager series of adze blades was collected from the northern part of Viti Levu, from Lauto- ka Province in the west to Ra Province in the east. The sides of some of these are more near- ly parallel than those of the thinner examples (figures 2,g and 2,i) from southern Viti Levu. Specimens from a field just west of the rice mill at Tavua, Province of Tholo North, have a distinctive rectangular appearance (figs. 3, e and 3, i). The adze in figure 3, i closely matches the Early Neolithic adze from the Philip- pines pictured by Beyer (1948) in the upper left- hand corner of his figure 9. A fragmentary adze from location A, site 17, 72 to 78 in. deep, also has a rectangular appearance (fig. 3, f). A much weathered adze, from location A, site 17, depth 84 to 90 in. is oblong in appearance (fig. 4, g). On the surface of Vatulili, a hill op- posite and south of Navatu (site 17), Rambithi found one more or less parallel-sided, although the blade is rounded (fig. 4, h). An adze (fig. 4, b) picked up by Rambithi in a cane field at site 26, Vunda, Lautoka Province, is similar. Another (fig. 3, m) from the same field has a blade forming an arc; in transverse cross sec- tion the fragment is oval. One, more axelike in 3"Ibid., p. 1o4, pl. A5 and fig. 9. 35British Museum, 1910, fig. 10, b and c, p. 18. 36Sarasin, 1929, pl. 22. 37Van der Hoop, 1941, fig. 38. 221 222 ANTHROPOLOGICAL RECORDS form because of the central position of the blade, which shows resharpening, was found by Rambithi on the surface of site 17, Navatu, at 550 ft. elevation. It is of heavy black stone, but lacks the proximal end (fig. 4, i). Some- what similar is a tiny specimen from the house mound, location A, site 26, 24 to 30 in. deep, shown in figure 3, J. It is much smaller and of softer stone than tThe preceding, the cutting edge slightly oblique, the piece thicker on one side than on the other, as though not completed. In general shape the three remaining speci- mens (figs. 4, a, 4, c, 4, f) from northern Viti Levu are similar to those from the southern part of the island. All three are excavated pieces: 4a from location A, site 17, 36-42 in. deep; 4c from rock shelter at same location, 12-18 in. deep; 4f from location A, site 26, depth about 12 in., in house mound. Three specimens (figs. 4, d, 4, e, and 4, X from the Rewa Dairy on the right bank of the Rewa River, upstream from the highway bridge at Nausori and Just below the confluence of the Waimanu River with the Rewa River, Naitasiri Province, were presented by Mr. E. H. McIlwain, who found them on Waila Hill about fifty or sixty feet above the dairy, when he was digging a hole to bury a cow. The cross sections of these three adze blades suggest a Fijian type pictured in the British Museum Handbook to the Ethnographical Collections.38 Figure 4j shows the heaviest adze blade, circular in cross section. Its weight is 1 lb., 14 oz. It was found by the Reverend Desmond P. Scanlon on Ovalau Island in June, 1947. Figure 31 displays the only stone chisel obtained. It is from Korovou, Taveuni Island. Unfigured fragments of adze blades, some undiagnostic butts, all too fragmentary to de- termine the type, were found as follows: Site 17, surface: specimens 5913, 5914, 5916, 5917, 6697. Site 17, location A: rectangle C3-4 D3-4, depth 6-12 in. (6381), depth 54-60 in. (6383); rectangle D3-4 E3-4, depth 48-54 in. (6387); rectangle H3-4 I3-4, depth o-6 in. (6389), depth 36-42 in. (5932); rectangle S3-4 T3-4, depth 0-6 in. (6392); rectangle S4-5 T4-5, depth 30- 36 in. (5929). Site 17, location B: rectangle Al-1.5 Cl- 1.5, depth 90-96 in. (6396), depth 120-126 in. (6397). Site 26, surface: specimens-6400, 6401, 6402. Site 26, location A: rectangle A4-5 B4-5 depth 24-30 in. (6405); rectangle B4-5 C4-5, depth 96-102 in. (6407.) The depths at which adze fragments were cavated indicate that adzes were characterist of both Early and Late cultures. No certainN stylistic differences for these two horizons be discerned, but this may be partly because- the scantiness of the series and the fragment character of most of the specimens. Door step.--A slab door step (6439) of fine-grained hornblende andesite was brought from the base of a large stone-faced house mound on top of the Vunda crag at site 26. house mound is said to be that of Lutunasomba somba, leader of the Fijian immigrants who su posedly founded Vunda after their arrival fr overseas. The door step is 2 in. thick, 8 in at its widest part, 7 in. at its narrowest, a about 14 in. long. One long side is concave, as though it had been broken, presumably in ancient times. Both surfaces have small pits, probably pecked in shaping. The end of one flat surface has ten parallel grooves in a space of 1 5/8 in. These run from the edge t ward the center, thus paralleling the long si of the slab. The marks are 3/4 in. to 1 in. long. The grooves have no functional signifi- cance and their purpose is wholly obscure. I suspect that the whole slab is more or less in natural form, with a minimum of shaping. It i the sort of thing to be seen outside many FiJi entrances today. Hammerstone.--No doubt many unworked cobbles were used as hammerstones, especially for smashing molluscan shells to extract the meat. One (5444), of close-grained gray stone was collected by Rambithi on the slopes of Navatu east of location A, site 17. It weighs.. 2.25 'lbs., fits the hand nicely, and has an en4 bruised from use. It is about 5 in. long, but ; of variable diameter, since its surface has several natural depressions which make it more readily grasped. Pottery anvils.--Cobbles of appropriate size are commonly used for pottery anvils toda' Crew members said that the occasional cobbles found in excavating were probably used as pottery anvils. Whetstone.--A smooth flattish cobble (644 one side of which showed marks of use as a whetstone, was found near the base of the crag on site 26. It was found a short distance eas of location A. POTTERY Viti Levu was indeed a land of potters, if one may Judge by the abundance of potsherds both on t1lp surface of sites and beneath the ground. In Ra Province the term for potsherd is tem- bethi, used like the Mexican word tepelcate. In Lautoka Province, the term tembethi was unknown, but the expression tiki ni kuro ("fragment of pottery") was used. 38British Museum, 1910, fig. 10, h, p. 18. Incised decoration characterizes both the; modern pottery and the excavated Late period pottery of Ra Province. (See pls. 16, d, e an 17, a.) The Early period ware excavated has body decoration in low relief. The material e cavated at site 17 (Navatu), Ra Province, fur- nishes my criteria for classifying sherds from elsewhere. In Nandronga Province and Lautoka Province modern pottery is largely plain ware, as are also most of the sherds excavated at site 26 (Vunda), Lautoka Province. i i I' GIFFORD: EXCAVATIONS IN FIJI The finding of relief sherds in abundance he lower levels of site 17, where shell was lly absent, was the high light of our exca- .ons. Relatively few relief sherds were at site 26, and these may be trade pieces. earliest type of relief sherd has a pattern rallel waves or zigzags; this I have desig- d the "wavy" type. This was absent at site Discussion of the several relief types is ented in two special sections "Relief Sherds" 226 and 231). Some relief designs were bably impressed with a carved paddle or a k stamp. This method, one might speculate, be connected genetically with the modern ian method of block-printing tapa. Kingsley Roth has mapped the areas in which tery was still being made in 1934.39 In Ra vince it is still made on the mainland. (See . 16 and 17.) Formerly the people of Malake and, which lies off the Ra coast at Rakiraki, e compelled to supply pottery and fish to ir mainland overlords. Likewise, the Malolo Anders, and earlier the Waya islanders, both ends off the coast of Lautoka Province, pro- ed their Vunda overlords with pottery. Today, pmodern ware in Lautoka Province is still Oplied by the Malolo people. (See pl. 16, b.) In Fiji probably one of the disturbing 'tors in the vertical distribution of sherd pes in archaeological deposits is the native ptom of cooking in earth ovens,4o when these re dug in midden deposits. This is bound to 6ult in the displacement of sherds and other Ltural material. It is remarkable how, in ite of this practice, the two general types of orated sherds (incised and relief) preponder- in upper and lower levels respectively. One rat Navatu we had opportunity to observe the tual displacement of material. Members of the tanggali Vusatambua turned out en masse to 6ar their cemetery (called Kawalevu), which fKingsley Roth, 1935, map 1, p. 218. 4Thomas Williams (Williams and Calvert, 1859, p. 116) ttes: "The ovens, which are holes or pits sunk in the und, are sometimes eight or ten feet deep, and fifty in .cumference; and in one of these several pigs and turtles a large quantity of vegetables can be cooked." has been used since heathen times. The shrub- bery was cut down, the retaining walls repaired, and, most significant of all to us, shell-midden material was shoveled from the surrounding area on to the walled cemetery (p1. 14, Xji. In the sorting and cataloguing of the pot- sherd collection, I have had the aid of two graduate students. John Francis Goins made the original rough segregation into plain, incised, and relief wares. Wilhelm G. Solheim II made a more refined classification under my direction and aided me in various ways during his prepara- tion of his master's thesis on Oceanian pottery. Mr. Goins made the drawings of the specimens in figures 1-4. Sherds obtained at site 17, locations A and B, both from the surface and by excavation, total 26,033 oz. in weight.41 The comparable total for site 26, locations A and B, is 10,654 oz. All excavated material was sifted through half-inch square-mesh screens and all inter- cepted potsherds were retained. Weighing sherds seems to give more accurate statistical results than counting them. Acci- dent or the character of the paste may account for the size of, and consequently the number of, sherds in a given block excavated, so that the count is variable. But no matter how large or how small the sherds are in a given block, their total weight will remain the same. In table 17, I present a comparison of weight and number of sherds (sherd count) from one rectangle (A3-4 B3-4 at location A, site 17) which shows the different results obtained by the two methods. The average number of sherds of each of the three main types per ounce at different depths is also significant. The figures in table 18 show that at all levels the plain sherds were more numerous and therefore smaller in size on the average than incised or relief sherds. The figures clearly indicate why, in table 17, the percentages of plain ware run higher by sherd count than by weight. 41This includes 720 oz. from test pit (rectangle B3-4 C3-4, depth o-64 in.) not included in tabulations by levels (tables 21-22). TABLE 17 WEIGHT AND NUMBER OF POTSHERDS BY LEVELS IN RECTANGLE A3-4 B3-4, LOCATION A, SITE 17 epth Weight Number Percentages in.) (oz.) Plain Incised | Relief _ 1,434 9,793 Weight Number Weight Number Weight Number I~~~~~~~~ I 6..... 12..... 18..... " 24 ... 30..... .36..... 42..... 4 8.... 7 4*.... 269 177 124 208 210 195 148 56 47 1,883 1,271 801 1,510 1,710 1,392 798 269 159 79 80 81 82.5 76.5 74.5 47.5 49.5 45 89 90 91 90 89 82 60 60 66 17 16 14 10.5 10 3.5 0.5 9 7.5 4 5 4 1 0.3 4 5 7 13.5 22 52 50.5 55 2 2.5 5 5 7 16 39.5 40 34 *This block, 48-74 in., not dug in 6-in. levels. 223 ANTHROPOLOGICAL RECORDS TABLE 18 NUMBER OF POTSHERDS PER OUNCE BY LEVELS IN RECTANGLE A3-4 B3-4, LOCATION A, SITE 17 Number of Sherds per Ounce Depth (in.) Plain Incised Relief 6...... 7.5 4 5 12 ...... 8 3 4 18 ...... 7 2 6.5 24 ...... 8 3 5 30 ...... 8 3 4 36 ...... 9 4 7 42 ...... 8 4 4 48 ...... 7 4 74* ..... 6 3 *This block, 48-74 in., not dug in 6-in. levels. In testing the hardness of Fijian pottery the standards for ceramic hardness of the Uni- versity of Michigan Museum of Anthropology were used. There seem to be no differences in the hardness of Early and Late wares, nor can dif- ferent degrees of hardness be correlated with plain, incised, and relief wares at sites 17 and 26. Thus, plain, incised, and relief sherds all ranged from 2.5 to 5.0 in hardness and sometimes a single sherd would yield a different hardness at different spots on its surface. The range of color variation seemed not to be correlated with hardness. In short, the hardness tests yielded no criteria. The sherds appear to be without clay slip, with rare exceptions which may not have true slips but may merely be exceptionally well smoothed specimens. Variations in surface color can probably be attributed to different ingredi- ents and different firing temperatures and con- ditions. The following colors (approximate) were noted for sherds found at both sites 17 and 26: gray, terracotta, orange, red-brown, buffish white, greenish white. Few excavated sherds showed trace of the slip of resin glaze that is applied to some modern Fijian vessels, especially water jugs and pieces for tourist sale. It is possible that this organic glaze in time becomes obliterated by chemical action in the soil.- Even exposure to the weather causes it to lose its gloss and to change from a translucent glaze slip to an opaque whitish slip. A good example of this change is to be seen in the water jug pictured in plate 17, a, which had been left out of doors because a small hole in it made it useless for water. Some archaeological sherds evidently had this organic glaze, which has now become opaque and lost its gloss. Specimens showing this change were found down to a depth of 42 in. at location A, site 17, and to 66 in. at location A, site 26. At location B, site 17, only a single sherd with organic glaze was found, and that was in the 42-48 in. level. Nine archaeological ex-- amples of the resin coating were tested for hardness. Eight examples yielded hardness 2.5 and one 3.0. On modern vessels illustrated th resin coating registers a hardness of 2.0 and 2.5. Rarely, sherds with perforations were fo These perforations were made before firing an are in or near the rim, apparently so that co might be used for suspending the vessel. Two examples from location A, site 26, are in the rims of relief sherds with applique projecti' "ears" like one figured by MacLachlan from the New Hebrides.42 The exact provenience of my t follows: 10997, rectangle F-Fl G-Gl, 19-25 in deep; 11690, rectangle 0-01 P-Pl, 24-30 in. d Another noteworthy example is a collarlike riX sherd (5189 from site 16) with two perforation an inch apart. Presumably there were two othe perforations on the opposite side of the vesse a huge cooking-pot of Late period incised typ like the modern example shown in plate 16, d, At both site 17 and site 26 occasional i cised sherds were found at low levels below th shell-midden deposits and also in the deep she midden at site 17, location B. (See tables 20 and 25.) These incised sherds from low levels probably date from the Middle and Early perio and rnvr represent (1) an early type of incisin wholly unrelated to the Late period incising o (2) a type out of which the Late incising de- veloped, if it is not due to diffusion from th New Hebrides, a possibility suggested by the findings of MacLachlan and Surridge.43 See low- level incised sherds pictured in plates 21, i, x-z, a'-&'; 25, v, w, a'; and 27, h. Counted as relief sherds are handles and spouts of Early period water jars or bottles (pl. 23). These differ somewhat from the modexi handles and spouts pictured by Surridge and vaguely suggest ancient Peruvian stirrup handle and spouts. The only artifact made from a potsherd is+, disk shown in plate 27, d, possibly used as a gaming piece. It would seem to date from the Early period, since it was obtained at the dept of 80-86 in. at location A, site 17. 42MacLachlan, 1938, fig. 31, right. 43Idem, 1939, p. 54; Surridge, 1944, pp. 18, 21-22. U 224 GIFFORD: EXCAVATIONS IN FIJI POTTERY FROM SITE 17 The surface of Location A, after repeated inations, yielded 1,591 oz. of sherds, which aify roughly into 1,278 oz. of plain ware, oz. of incised ware, and 41 oz. of relief e. Stated in weight percentages, 80 per cent plain ware, 17 per cent is incised ware, and er cent relief ware. Of the 1,591 oz. of rds, 1,367 oz. are body sherds, 74 oz. are k sherds, and 150 oz. are rim sherds. At site 17, plain sherds are the most ndant until the lower levels at location A reached, when more than half the sherds are lief ware and less than half plain ware. How- r, in the rock shelter (rectangles R to U), is is not so, and the shallowness of the cul- al deposit, coupled with the scarcity of ly relief ware here, may indicate that the lter was formed by the fall of the overhang- boulder from the mountainside at a date sub- quent to the Early period. Otherwise it is ficult to understand the scarcity of relief e. We may be Justified in suggesting that descent of the boulder also caused the aban- ent of location A and fear of a recurrence vented the occupation of this part of site 17 a long time. Plain potsherds may be derived from com- etely plain vessels or from vessels only par- ally ornamented, as in the modern pot pictured plate 16, d, e, which shows the incised deco- tion confined to the rim and shoulders.44 An- ient vessels with relief decoration probably less plain surface that would yield plain erds, since most of the decoration was evident- y applied to the body of the vessel, as shown plate 26, a. [ Between the Late period, characterized by ncised ware, and the Early period, character- %zed by relief ware, there seems to have been a widdle period in which plain ware was the domi- Mt type. (Sherds with accidental scratches or eaf-impressions have been counted with plain re.) As the distribution tables show, this lin ware was exceedingly abundant at location and had been dominant for a long time. At lo- nation A, the plain ware is less abundant and its scarcity, I think, warrants the suggestion that throughout the Middle period location A may ave been largely unoccupied, while location B was occupied. This hypothesis is corroborated y Professor Jenny's opinion that below the phell midden at location A, soil profiles had tormed indicating a long period (perhaps several 8enturies) when the site was not occupied. Compared with the sherds from site 26 (table 23), relatively few sherds from site 17 contain calcite in the paste. The calcite in the potsherds (much more frequent at site 26) is probably organic, being the shelly material of calcitestaceous organisms, particularly mollusks and foraminifera. I suspect the calcite was in- eluded accidentally with the clay rather than used as tempering material, since there is other tempering material in the sherds. The following I4Cf. also illustrations and text in MacLachlan, 1940. list gives some examples of sherds with calcite inclusions. These are all Late period pieces taken from rectangle E3-4 F3-4, location A, site 17, which was selected as a sample rectangle: top triangle, 11 sherds; 0-6 in. deep, 13 sherds; 6-12 in. deep, 17 sherds; 12-18 in. deep, 11 sherds; 18-24 in. deep, 10 sherds; 24-30 in. deep, 9 sherds; 30-36 in. deep, 2 sherds. None was found in levels below 36 in. Ladd does not list any foraminiferous deposits in the Ra region,45 so possibly these sherds with calcite were made elsewhere. Incised potsherds are characteristic of the shell-midden layer at location A, and incising is the current style of decoration on pottery still being made in the province. The tapering, round-bottomed cooking-pot shown in plate 16, d, e, has typical incised decoration on its broad collarlike rim and shoulder. This is a medium- sized cooking-pot, purchased from its maker at Naivuvuni village, Ra Province. Its rim is 3 in. wide. A similar incised collarlike rim, col- lected at site 16 (near Rakiraki, Ra Province), is 3.50 in. wide; another from site 4 (near Tavua, Tholo North Province) i8 3.75 in. wide. The incised designs are all geometric. Parallel lines were probably made with the cre- nelated edges of such bivalve shells as Cardium. Specimen 5553 (fig. 1, E) excavated at location B, depth 18-24 in. was called a pot-marker by the natives. Such an instrument or a wooden comblike tool would serve to make parallel lines. The various types of marks are straight lines, wavy lines, arcs, gashes, gouges, crene- lations, punctations (dotlike depressions) and various combinations of these. (See pls. 20-22.) An attempt to correlate these various designs with depths at which the specimens were found failed. It is therefore impossible to establish any chronological series for Late period incised ware at site 17, and the absence of a succession of styles in incised decoration probably indi- cates the relative brevity of the Late period, which we have elsewhere estimated as lasting not more than fourteen generations.46 The weight of incised sherds excavated at location B totaled only 15 oz. Those collected on the surface weighed only 12 oz. This scar- city of incised sherds would seem to indicate limited occupancy of location B in the Late period and in modern times, as compared with lo- cation A. Below ground, the few incised sherds, except for two or three Late period specimens, had a different appearance, as did the few at low levels of location A and site 26. The small proportion of incised sherds found at location B perhaps indicates that this location was unoccupied during much of the Late period. The distribution of shell in location B (see table 1) cannot be correlated with Late period incised sherds and the assumption is that the shell accumulated during a Middle period characterized by dominance of plain ware. This Middle period was only briefly represented at location A, which I assume was unoccupied for most of that time. 46Gifford, 1951, p. 170. 225 45Ladd, 1934, pp. 102-153. ANTHROPOLOGICAL RECORDS Distribution of sherd types by depth.-- Table 19 gives the total weights of potsherds obtained in each 6-in. block in rectangles ex- cavated at locations A and B, site 17. Table 20 gives the percentages of the total sherd weight for each of the three principal types of sherds (plain, incised, and relief) in each 6-in. block in each rectangle. The correlation of incised ware with the Late or shell-midden period at lo- cation A, site 17 is obvious if table 20 is com- pared with table 1, which shows the vertical distribution of shell and other constituents at location A. In tables 19 to 22 the record for 47 oz. of sherds recovered from a depth of 48-74 in. be- tween boulders (rectangles A3-4 B3-4) is omitted because their 6-in. levels were not recorded; 45 per cent of the sherds were plain, 55 per cent relief. In tables 19 to 22 the data from rectangle G3-4 H3-4 are omitted, since this rectangle was dug to a depth of only 7.5 in. The top triangle here yielded 194 oz. of sherds, of which 79 per cent were plain, 19 per cent incised, and 3 per cent relief. The block from 0-7.5 in. yielded 193 oz., of which 81 per cent were plain, 17 per cent incised, and 2 per cent relief. The surface of location B, site 17 yielded slightly over 360 oz. of sherds collected by Fijian workmen in a single day, after cutting and burning the vegetation. The exact size the area was not recorded, but was probably 1,200 sq. ft. The quantity of sherds found would have been larger if the ground had been. repeatedly examined, especially after rain, it was at location A. The total of 360 oz. classifies roughly into 337 oz. of plain ware 12 oz. of incised ware, and 11 oz. of relieff ware. Stated in weight percentages, 94 per c is plain, 3 per cent is incised, and 3 per ce relief. Of the 360 oz., 308 oz. are body she 14 oz. neck sherds, and 38 oz. rim sherds. Relief sherds.--In stressing relief deco~ rated sherds as the important criterion of th' Early period, the different styles of relief decoration must be distinguished. Ribbed, ap plique (especially nubbin), and gouged treat- ments persist in modern Fijian pottery and ar well represented in Late period archaeologica specimens. Crossed decoration, either the rej tangle or lozenge pattern, occurs at all leve but so far as I know is not used by modern Fijian potters. I think that its occurrence i upper levels may be the result of the earth-o0 digging, or it may be that its manufacture wai continued nearly to modern times. Even the earliest style of all, which I have designated 'wavy"--usually found only in the lowest level occurs occasionally in upper levels: e.g., in rectangles D3-4 E3-4 and R3-4 T3-4. Again, th TABLE 19 WEIGHT OF POTSHERDS BY LEVELS AT SITE 17 (in oz.) Depth Location A LocatJ (in.) B A3-4 B3 -4 D3-4 E3-4 H3-4 I3-4 N3-4 | R3-4 T3-4 S4-5 T4-5 B3-4 D3-4 E3-4 F3- 4 I3-4 J3-4 03-4 T3-4 U3-4 T4-5 u4-5 * TT. 6. 12. 18. 24. 30. 36. 42. 48. 54. 60. 66. 72. 78. 84. 90. 96. 102. 108. 114. 120. 126. 132. 138. 144. 269 177 124 208 210 195 148 56 157 213 133 190 127 111 100 85 98 116 106 190 112 61 382 138 158 198 122 88 59 93 119 ' 65 99 87 95 110 106 65 42 137 161 228 192 159 264 258 140 30 137 175 111 102 122 122 127 147 146 86 32 8 180 119 108 52 110 158 230 190 160 231 165 175 41 9 171 159 136 136 173 153 138 147 91 103 135 61 16 869 605 245 175 236 95 113 67 53 17 510 297 260 186 66 175 101 125 82 28 49 316 318 248 193 112 54 353 177 183 141 112 20 *TT, "top triangle. " See Introduction, p. 190. 287 144 221 287 277 284 244 259 325 530 272 251 200 189 199 257 145 79 102 57 44 28 15 8 226 GIFFORD: EXCAVATIONS IN FIJI 0 . Lr1 U \ 1 1 CU N H CU H r-H N CO C\ 01 a i n O CO a O c o p gr: H~~~~~~~~~~~~- H 010 C O H 010 C O X m H N N N N N N M 8 8 H M H NH H 0 H O C.) O Po 4 > St UACX C co t o m H 0 C co H ) L- O C\ r-c O O I r CM O 10 1\ NC) .t p H H1 r- H (Yr) U) CM - cn (f) 04 co co 1 C t:- C\ 10 t-- xP CM O O C\ CO - L H Lr H r-4 EN ~ ~ I 1 0 Lft 1001M H N H~~~~~~4 4- F H r1i r1i r1 H CO HEH H H C') ~ ~ ~ ~ ~ ~ ~ C)V' -O m T X~~~~~~~L CC co co co co . X ~~~1C O 01 rQ C 0 E -4C C H- 4 0 01 C\O CY CO )C\O I I '.~Ln O N r- 4 0 1 H !~~~~~~Y CY P - H H rt4 C\J r1 r1i CU >3 H O4 H0 1 CY' C ~ ~~~~~~~~0 .O 'PO 1O 0 H m~~~~~~~~~~~~~~0 tt .= 10 mm pqP CM cn (YD Cy) H0 U- ) ) _s~~ ~ ~ ~~ 1 k. 01 ri C\j UA HOS S C) C I OC 0 1 - N N S w OX H Z;CO O H H H 01HC r c\o co-10 O OO- CH P4 O-' -0 o 0 CP CO C O C O - C O 0 C0 L 0 COL- 4-3 0 CCCO HY, 4 D - \ ID-- 0 - H . c ~ P4 O C O O C O 1 0 1 C) ) C - r 0 Y - a Iv < H C a c C O H> j- HoI I -c 1 0 N N o \ CtJ < < l l . S S > H O N ~~~~~~~H N CY, -4 M 8 O0 CY) CY) H r1 H r H r1 r H .) V A1 0 Q z t L t 0 0 aL H H PV4 Co 0 tc Co OY 00 t-'O 1\ LO L 10 E P N 10 H H 101 C O O C00 4 CY)~~~~~~~ C ~ .~ - .n LC 10 0 0 1 1 co 4- 4- 4-o H I I 00 CY) CM t- H LNH CO H; CY O m H H , HI C\J H rH H H (1) r1i0 C YI C O t 0 - r O w - CM L CI O 1 0 C0 CO H - P4 000 c0t--co O CocomO 10 (ONj L 0 \0 4- L10 o 4 1010 - ' \ n H C N 0 N H 1 0O 01N o _ _ > w ~~~~~~~~~~~~r1I CY) - :t 1 4- 41 U L) L15 --t 4-L U- :2; 1 1 LCv OjM;t r1 0 C\l K r- X r HH H r- m H X thI I - HC O COH N H ' N N O COCO) H H H H ( O ) L ) 0 . 0- 0J'D CY) t tD---t 3\ -COOCO1L O C \ C N 00 CO riri-t X~~~~~~~P cot- X c 00 0 0 0 \0c0 CL0 U- LC Ln n Ln L4- 4-4 w Lr) UA.f U O P 4 ---k LA4~ U' L4 C-Ot Zt \OO'fD co 0J V. j-- C\J -1=0 N U) t t ~~~~~~~~~~~C\J m (Y) CY ;e- VU) U' U'\ kD Lnz 4 Fz H 01 HO- H H0 . 'rt 1 P4 I I 1 0 H HN1 I Cz CY' w 1nS \ S 1 COC) H H HK HP4 coco\OC 1010 C O.Y O n H~ ~ ~ ~~~~ r- 0- t0 0 0 m 4 nD\ co lf\j :0 t^ 0 . NH H CO 1010 I l 10CO) ' CH N COCOi H H H H (J H rH 0 t- 0 O \J t-C P4 cO co 00COt- \04- 4'. *-'. -1 O~ 0' 01C- - O- - - 0 O 01 O.-0k 01 O 0'* - - - CO - - (1) "-iE-4 H H0r1I C O y) CY) =t - - \ .CO t- -- 00 C H0 0 r1C O C\ OC) Y4- H H H H- H- H- 227 I . ANTHROPOLOGICAL RECORDS distribution can probably be explained by the disturbance caused by native digging. Unfortu- nately, a certain number of relief sherds defied classification, usually because the marks were too indistinct, a fact which accounts for the heading "Unclassified" in tables 21 and 22. Plates 22 to 27 illustrate the several styles of relief decoration. Relief in crossed and waVy styles seems to have been essentially body rather than rim deco- ration. Early period rims were sometimes plain while the body was ornamented in relief (see pl. 26, X). Crossed relief sherds have both square and lozenge patterns and were presumably made with an incised paddle or check stamp (pl. 22). Wavy relief sherds, the earliest type of all, have two intergrading styles, angular and curvilinear (pls. 25-27). Sometimes the waves run horizontally around the vessel, at other times they are vertical. When vertical, they suggest serpents climbing to the top of the ves- sel (pl. 27, f).- Two wavy relief sherds pic- tured (pl. 27, c and i) suggest wickerwork iq press. Western Pacific parallels to Fijian cro and wavy relief are cited in the section en.: titled "Extra-Fijian Relations." I have distinguished a gouged incised s8 and a gouged relief style. In the former th4 surface beside the gouged depression is not 4 vated. In the gouged relief, on the contrarl the elevated border constitutes the "relief 2 Table 21 gives the total weight of reliL sherds excavated at each level at site 17. i 26 oz. of relief sherds recovered from a dept of 48-74 in. between boulders in rectangle A3 B3-4, omitted from tables 21 and 22, may be classified according to styles as follows: 33 per cent crossed, 5 per cent gouged, 5 pel cent ribbed, 38 per cent wavy, 19 per cent uA classified. Table 22 presents the distributi (in terms of percentage of total weight) of styles of relief sherds, including unclassifj at site 17. The data are arranged by depths rectangle. TABLE 21 WEIGHT OF RELIEF SHERDS BY LEVELS AT SITE 17 (in oz.) Depth Location A Locat2 (in.)B A3-4 B34| D3-4 E3-4 H3-4 I3-4 N3-4 R3-4 T3-4 S4-5 T4-5 B3-4 D3-4 E3-4 F3-4 I3-4 J3-4 03-4 T3-4 U3-4 T4-5 N4-5 TT 19 6 16 6.10 2 5 5 2 3 29 13 7 6 12 7 12 8 9 8 10 5 4 16 9 2 5 5 6 18 6 9 10 6 4 9 10 8 4 4 5 17 24 .22 15 13 3 2 4 4 6 1 1 8 13 30 .35 16 21 11 5 10 12 5 1 11 6 7 36 .76 26 20 12 10 8 6 1 8 4 8 42 .96 35 31 16 20 11 24 1 48 .29 36 43 10 16 23 27 5 54 47 31 57 13 24 25 4 6o 58 51 76 21 37 10 3 66 64 48 47 62 61 5 72 121 48 42 67 26 16 78 64 68 55 21 7 11 84 35 58 67 4 15 90 31 75 30 96 20 76 27 102 71 20 108 42 23 114 .. 17 17 120 4 8 126 8 132. 3 138 ...... 3 144...... 228 GIFFORD: EXCAVATIONS IN FIJI TABLE 22 DISTRIBUTION OF STYLES OF RELIEF SHERDS BY LEVELS AT SITE 17 Percentages of total weight for each level Location A A3-4 B3-4 | B3-4 D3-4 | D3-4 E3-4 j E3-4 F3-4 *C R A G W U C R A G W U C R A G W U C R A G W U 10 21 10 553 17 17 17 50 10 20 50 10 10 50 50 60 20 20 40 20 20 20 12 25 25 38 11 55 22 11 25 12 50 12 20 10 20 10 40 i..... 33 17 17 33 11 22 11 22 33 10 10 40 20 10 10 33 17 17 33 29 8 27 17 2 17 40 20 7 20 13 15 15 23 46 33 33 33 39 11 10 15 3 22 56 13 6 6 19 29 19 9 19 24 18 27 36 9 9 48 7 8 7 30 65 12 15 8 40 10 10 5 10 25 17 8 41 17 17 ... 47 7 2 8 14 22 57 6 11 6 3 17 52 6 6 629 31 6 13 13 631 49 7 1 8 20 15 78 3 8 11 51 7 9 923 40 10 20 30 42 13 2 9 15 19 51 13 10 323 28 4 6 10 18 34 50 14 3 12 21 37 16 12 14 22 41 4 1 17 11 26 41 16 3 14 26 33 6 210 29 19 43 6 6 9 13 23 30 7 3 2 40 18 33 6 2 2 54 2 43 12 2 14 14 14 26 8 5 36 25 34 7 1 43 15 38 13 5 22 22 17 20 6 43 14 23 3 2 65 7 38 6 7 25 24 32 3 55 10 33 6 5 1 28 26 25 40 5 15 15 30 5 11 37 17 41 4 3 10 23 20 29 21 5 24 21 41 24 12 24 25 25 50 *.@@ *C, crossed; R, ribbed; A, applique; G, gouged; W, wavy; U, unclassified. TABLE 22 (Continued) Location A H -4 I3-4 IR-4 j3-4 N-4 o3-4 RR-4 TR-4 _ C R A G W U C R A G W U C R A G W U C R A G W U s:4 .. 050 50 33 33 33 7 3 17 34 38 23 15 15 8 8 31 O... 0 40 20 40- 50 50 6 6 19 63 6 11 67 22 .k.000-1.25 25 25 25 11 22 22 33 11 40 10 40 10 12 13 12 38 12 13 50 50 25 25 50 25 25 25 25 50 33 17 , 20 40 40 60 20 20 8 67 17 7 40 40 20 . . 0 10 10 30 20 30 12 12 25 25 25 33 17 17 33 .... 10 5 65 5 15 18 9 27 18 27 21 50 8 4 8 8 6 31 44 19 26 17 13 26 17 26 26 26 22 O...O 15 8 15 31 31 46 8 4 12 821 36 32 12 12 8 , .-.. 24 5 14 24 5 28 46 5 5 5 11 27 20 50 20 10 O,O .. 53 2 5 8 13 19 49 3 2 5 16 25 , .... 52 7 110 6 22 35 4 8 23 31 ..... 57 10 10 24 14 57 29 ,00.. 00 75 25 , .... 229 230 ANTHROPOLOGICAL RECORDS TABLE 22 (Continued) I Depth Location A Location B (in.) I T3-4 U3-4 | S4-5 T4-5 | T4-5 U4-5 C R A G W U C R A G W U C R A G W U C R A G W U TT ....44 44 12 6....... 29 29 29 14 17 33 17 17 17 8 8 75 8 71 29 12 50 50 20 40 40 100 50 50 18.... 25 25 25 25 75 25 60 40 6 47 47 24 100 100 12 12 38 12 25 62 38 30 100 27 18 18 9 27 17 17 32 17 17 71 29 36...... 100 25 25 12 37 25 25.50 87 13 42 .... x 100 48 ...... 20 6o 20 54.... 25 25 25 25 6o..... 33 33 34 66 ..... 40 20 40 72 ..... 25 62 13 78 .... 45 18 9 27 84 ..... 13 7 40 40 90 40 7 10 737 96 ...... 48 15 7 4 11 15 102 40 5 5 30 20 108 39 9 4 443 114 ..... 29 24 6 18 24 120.... 38 13 12 12 25 126 .... 50 12 13 25 132 .... 33 67 138 33 67 144 . . 0 0 0 0 0 0 POTTERY FROM SITE 26 Plain sherds were dominant at all depths in the excavations at site 26. Incised and relief sherds were scarce, as table 25 shows, and, al- though some were perhaps made locally, these types never achieved the same popularity as at site 17. Local manufacture is apparently indicated by the presence of calcite in the form of shell and foraminifera in the paste of great numbers of sherds. Modern pottery used in the Vunda region is imported from Malolo Island and is characterized by the calcite in the paste. How- ever, it should not be assumed from this that all of the ancient pottery came from that island also, since Ladd indicates foraminiferous de- posits near the Sambeto River,47 only a few miles south of site 26. It may well be, therefore, that much of the clay with calcite used came from this or other deposits on Viti Levu itself. Table 23 indicates depths to which sherds containing calcite were found in the excavations at location A, site 26. At location B, sherds showed similar inclusions. Since these observa- tions are based upon examination with the naked eye aided by a hand magnifying glass, not a microscope, they are inadequate and may refer chiefly to shell rather than the microscopic foraminifera. Two surface sherds from location 47Ladd, 1934, station 130, fig. 10 (map of Vitilevu), p. 141. A were examined by Professor R. M. Kleinpell, who reported Orbitoid foraminifera (probably Miogypsina) in one and numerous Rotalid fora-, minifera (apparently Rotalia) in the other. (See App. I.) A single examination of the surface of 10 cation B (area, 810 sq. ft.) of site 26 yield slightly over 84 oz. of sherds. These classif roughly into 82 oz. of plain ware, 2 oz. (a single sherd) having both incised and relief decoration, and a trace of incised ware. Stat in weight percentages, 98 per cent is plain w 2 per cent is incised and relief ware. Of the 84 oz., 61 oz. are body sherds, 8 oz. neck sherds, 13 oz. rim sherds, and 2 oz. rim and neck combined (this last is the piece combini incised and relief decoration). At location A the surface sherds were not removed before excavation commenced, but were counted in the 0-6-in. block of each pit. By excavation at site 26 only 343 oz. of relief-decorated and 163 oz. of incised-deco- rated sherds were obtained. This is in marked contrast to the 3,118 oz. of relief-decorated sherds and 1,783 oz. of incised-decorated sher excavated at site 17. At site 26, 10,064 oz. plain ware were obtained by excavation; at site 17, 19,337 oz. The figures for site 17 include sherds from the test pit made in rectangle B3- 4 C3-4, depth o-64 in. Plate 21, x-z, a'-g , shows Early period incised sherds from location A, site 26, all -du below 48 in. A few incised sherds from upper levels look like Late and modern Ra Province I f GIFFORD: EXCAVATIONS IN FIJI and are probably trade pieces. However, does not imply that they were necessarily at site 17 or even in Ra Province since lar sherds were found on the surface at us geographically intermediate sites, the rnmost being site 27, some miles up the Mba See table 26 and map 1. Distribution of sherd types by depth.-- 24 gives the total weights of potsherds $ned in the 6-in. blocks in the various rec- Les excavated at location A, site 26. Table Ives the percentages of each of the three cipal types of sherds (plain, incised, and ef) in each 6-in. block in each rectangle. ,-The depths to which shell occurred are n in table 2(p. 205), which thus clearly in- tes the extent of the Late or shell-midden tum of site 26, where, largely because of scarcity of incised sherds, the correlation hell with incised ware is not so marked as Location A, site 17. Relief sherds.-- The total weight of relief- ~rated sherds excavated at site 26 was only oz. as compared with 3,118 oz. at site 17. we disregard lots weighing less than 1/2 oz., relief sherds may be classified by styles as Lows: crossed, 63 oz.; ribbed, 14 oz.; ap- que, 34 oz.; gouged, 195 oz.; unclassified, Oz.; wavy, none. Making allowance for the llness of the samples, we find that the verti- distribution of the four types found at site (crossed, ribbed, applique, and gouged) was h the same as at site 17, namely, more or s throughout the deposit. TABL] As the maximum quantity of relief-decorated sherds obtained at any one level in site 26 was only 15 oz. and the next largest lot only 10 oz., it seems unrewarding to give a detailed tabula- tion of styles according to the depths at which they occurred, as was done for site 17 (table 22). The potsherds in wavy relief, so abundant in the lower levels at site 17, were entirely lacking at site 26. Possibly the absence of wavy relief sherds at site 26 indicates that Vunda was founded later than Navatu. Otherwise, if the two settlements had been established simultaneously, this style, as well as the other relief styles, would appear at both settlements. POTTERY FROM OTHER-SITES None of the sites included in this section was excavated because none looked as promising as sites 17 and 26. Thus all the specimens re- ferred to in this section are from the surface. Only plain sherds were found on the surface at- sites 6, 15, 23, 24, 29, 36, and 38. Plain sherds were found at all the sttes listed in table 26. For descriptions of sites other than 17 and 26, see Appendix II. A single sherd of crossed relief ware is pictured by Laura Thompson from Kambara Island, Lau Group.48 She also shows incised sherds from Kambara and Mothe islands. *8Thompson, 1938b, pl. A, fig. le, opp. p. 113. .E 23 OCCURRENCE OF CALCITE IN POTSHERDS AT LOCATION A, SITE 26 pth* Rectangles *) | ~A4-5 |B4-5 |E-El F-Fl G-Gl H-Hl G2-3 |H2-3 |N-Nl Nl-2 |0-01 |01-2 |B4-5 |C4-5 |F-Fl G-Gl H-Hl I-I1 H2-3 |I2-3 |0-01 |01-2 |P-P1 Pl-2 S.......... x x x x x x x x x x x x .......... x x x x x x x x x x x x 3 . x x x x x x x x x x x . x x x x x x x x x x p ......x x x x x x x x F4.. . .. .. .... x x x x x x x x x 0 .. ........ . x x x x x x x h ........x ' x x x x x x x i4......... x x x xX x 6 x 2. x x _ 8 x 4.......... x x 0.......... x x 6 ....... . . ....... 8.......... D. . . . . x 6 ..... *The crossbar in each column indicates the depth to which the rectangle was dug. 231 ANTHROPOLOGICAL RECORDS No wavy relief sherds were found at these sites; wavy relief ware occurred only at site 17. Site 21, situated in a swale in the sand dunes west of the mouth of the Singatoka River on the south coast of Viti Levu, yielded a con- siderable number of plain sherds, many of which were badly sand-blasted. Indeed, the prevailing southeast wind had blown most of the sand out of the swale, leaving the potsherds lying on the brown earth. (See description of this site in App. II; also pl. 14, e, f.) After our return to California, Dr. Lindsay Verrier sent us a choice collection of pieces from this site, the most interesting of which are pictured in plates 19 and 20. He also found there a fragment of "flint" similar to that in the unique blade found by Mr. G. T. Barker at site 15. Professor Howel Williams has identified Dr. Verrier's fragment as chalcedony. Certain of Dr. Verrier's sherds require special comment. Plate 20, a-c shows leaf and mat impressions. The sherd in plate 19, c is roulette-marked and appears to be part of a shallow, flat-bottomed dish. Plate 19, d shows a curvilinear loop pattern quite unlike anything we found elsewhere. MODERN POTTERY VESSELS FROM VITI LEVU MacLachlan (1940) pictures fifty-one Fijian vessels and one large incised sherd; also one Solomon Island vessel with incised decoration which can be matched by some of our incised sherds. The fact that all of the Fijian vet are complete probably indicates that all art ethnological specimens. Not a single one sk either crossed or wavy relief designs that t might be attributed to the Early horizon. Ni precise provenience within Fiji is given fotA any of the vessels. The specimen in MacLaci figure 4 is apparently not a container, but conical pot rest like those I illustrate frol Rewa Province (my p1. 18, c). MacLachlan's1 figures show well the modern use of incisinrg and relief decoration, especially applique r lief. Margaret N. Surridge discusses the typei ornamentation on water jars in her article et titled "Decoration of Fiji Water-jars." My plates 16-19 display various types of vessel$ collected in Viti Levu in 1947, some of whiei are water jars. Kingsley Roth's excellent article on "PC tery Making in Fiji" deals in some detail wil the work of potters of Ra Province, especial] those of the yavusa Navatu,49 formerly inhabil our site 17. He mentions the use of a Cardi} shell for making incised decoration and also attaching of applique ornaments.5 A paddle and an anvil obtained from a p ter at Naivuvuni, Ra Province, were similar 1 pieces illustrated in Kingsley Roth's paper.5 49Kingsley Roth (1935, p. 227) refers to Mbakandroti the ancestral spirit (kalou vu) of yavusa Navatu. 5OIbid ., p. 224 . 51Ibid ., pl . 11 , fig TABLE 24 WEIGHT OF POTSHERDS BY LEVELS AT LOCATION A, SITE 26 (in oz.) Depth A4-5 B4-5 E-Ei F-Fl G-Gi H-Hi G2-3 H2-3 N-Ni Nl-2 0-01 0o (in.) B4-5 c4-5 F-Fi G-Gi H-Hi I-Il H2-3 I2-3 0-O1 01-2 P-Pl P: 6.......... 174 128 122 128 io6 95 137 164 16 41 47 12 .......... 83 59 123 125 166 136 137 138 48 75 iii 18 .......... 72 53 182 14o 144 169 192 110 127 72 79 24 .......... 40 57 69 83 121 83 174 99 81 90 120 : 30 .......... 57 33 41 65 60 67 98 70 94 77 121 36 .......... 47 48 35 28 59 62 99 42 80 42 .......... 155 57 34 41 82 62 87 32 70 48 .......... 67 116 43 77 78 19 39 26 54 .......... 49 129 116 112 103 10 16 8 6o .......... 22 156 316 147 56 2 6 66 .......... 19 11 284 115 6 72 .......... 24 15 198 84 x* 78 .......... 21 19 39 20 84 .......... 31 5 29 9 90 .......... 42 17 79 20 96 .......... 29 16 64 15 102 .......... 15 11 67 1o8 .......... 15 16 20 114 ........... 14 6 16 120 .......... 8 8 126 .......... 2 *Less than 1/2 oz. 232 94 GIFFORD: EXCAVATIONS IN FIJI TABLE 25 PLAIN (P), INCISED (I), AND RELIEF (R) SHERDS BY LEVELS AT LOCATION A, SITE 26 PERCENTAGES OF WEIGHT (x, less than half of 1 per cent) B4-5 C4-5 E-El F-Fl F-Fl G-Gi G-Gi H-Hi I H-Hi I-Il P I R I P I R P I R FP I R I FI R I FI R 92 95 91 74 73 83 95 95 99 95 91 100 95 80 94 81 82 100 100 5 2 4 x 3 6 2 4 x 1 x x x x 6 x x 3 3 6 26 24 10 4 1 1 4 9 x 5 20 6 13 18 x 97 97 94 91 88 94 97 98 96 99 97 98 95 97 99 95 93 90 81 88 2 3 2 1 2 x 3 2 4 1 1 1 x x x x x x x 98 95 96 96 97 89 90 99 97 99 100 100 90 100 90 93 1 x 4 7 10 6 x x x 2 1 5 3 1 5 7 10 19 12 100 x x 2 1 x 1 4 x x 1 x x x x x x x 2 3 2 4 2 7 10 1 2 1 x x 10 x 10 7 97 97 94 94 92 97 99 99 100 100 2 2 2 2 5 x x x x x 1 1 4 3 3 3 1 1 x x 95 97 95 89 94 100 95 95 100 100 1 2 2 5 A A A A 3 1 4 6 6 x 5 5 x TABLE 25 (Continued) (x, less than half of 1 per cent) rth G2-3 H2-3 H2-3 I2-3 N-Ni 0-01 Nl-2 01-2 0-01 P-PF 01-2 P1-2 _ _ FP I R P I -R P I R P I R P I R P I R 98 1 1 98 2 x 100 x 95 2 2 96 2 2 94 2 5 95 1 4 97 2 1 98 2 x 97 x 3 95 3 2 97 2 1 ....... 94 2 4 91 3 6 96 3 1 96 x 4 100 x x 95 1 4 93 2 6 94 1 5 96 1 2 97 2 1 98 1 2 93 3 4 93 2 5 97 x 3 93 x 7 96 1 3 96 2 2 97 3 x 99 1 x 95 x 5 96 x 4 98 x 2 ....... .. 95 1 3 94 x 6 94 1 4 98 x 2 100 x x 81 19 93 x 7 94 6 38 62 88 13 ... -50 50 ....... 67 33 x x A4-5 B4-5 1 2 13 8 7 x 6 3 x 94 92 86 90 89 94 92 97 96 100 100 92 86 84 90 83 87 80 79 88 5 6 1 2 4 6 3 x 4 x x x x 3 x x x x 7 12 x 8 14 13 10 17 13 20 14 I - - a - - I - - I - - - I - - i - - - - I 233 4 ANTHROPOLOGICAL RECORDS This potter had about twenty cooking-pots for sale around the walls of her house. The one shown is of medium size. The method of cooking with this type of Jar, with tapering rounded bottom (aryballoid, in the terminology of Peru- vian archaeology), was observed at Vathuthere (site 6), where one was seen lying on its side on an oblong stone-edged hearth. A hearth of this kind was also seen at site 9. Wilkes, Miss Gordon-Cumming, and Kingsley Roth give pictures of such vessels in use,52 the illustrations show- ing them resting on the side, not the bottom. In western Viti Levu the cooking-pots are globose in shape (aryballoid in the terminology of Greek archaeology), as shown in plate 16, b, which pictures one from Malolo Island presented by Andi Meri, daughter of the Tui Vunda (Lord of Vunda). When in use, such vessels were sup- ported in a vertical position, contrasting with the horizontal position of the cooking-pots of northeastern Viti Levu (Ra and Tholo North provinces) mentioned above. A globular water bottle, made at Vunitongo- loa, Ra Province, is shown in plate 17, a, a pottery kava bowl in plate 17, c, d. The bowl was made in Laselase, Nandronga Province, chased by people from Narewa, Ra Province, 1942, and sold to me in 1947. Lester illus a fine water bottle, with elaborate incised tern suggesting clouds, and two pottery kav4 bowls,53 one of the bowls from Namosi Provina The way in which water Jugs are carrie depicted by Wilkes.54 A story related by my formant tells of a goddess' water Jug beingH smashed because it had no carrying device.i myth thus accounts for the presence of potsb on the surface at site 11. Wilkes shows the mode of drinking from Jugs,56 which I observe also among the members of our excavation crS Ratu Sir Lala Sukuna and Mr. G. Kingsl1 Roth generously had made to order for me mod of pots commonly used in the delta region of Rewa Province. Some of these are shown in X plate 18. Models, instead of full-sized vei were made so as to reduce the bulk in shippt The hardness of the modern FiJian vesse shown in my plates 16-19 falls within the ra of hardness of the ancient potsherds. The f' that the resin glaze is never as hard as the pottery itself is interesting. 52Wilkes, 1845, p. 349; Gordon-Cumning, 1883, opp. p. 208; K. Roth, pl. 15, fig. 1. 53Lester, 1941, pl. 3. 54Wilkes, 1845, p. 224. 55Gifford, 1951, p.s 56Wilkes, 1845, p.3 TABLE 26 DECORATED SHERDS FROM SURFACE OF VARIOUS SITES Site Occurrence Incised Relief Ribbed | Crossed | Applique ' Gouged la... x x x lb ... x 2.... x x x x x 3.... x 4.... x x x x 5.... x x x 7.... x 8.... x 9.... x 10 ... x x x 11... x 12 ... x x 13... x 14 ... x x x x 16 ... x x x x 18... x x 19... x x 20... x x x x 21... x x x x 22... x x 25... x 27... x x x 28... x 30... x 31 ... x 32... x 33... x 34... x x 35 ... x x 37... x x x I ,, 234 GIFFORD: EXCAVATIONS IN FIJI TEMPORAL AND SPATIAL RELATIONS FIJIAN RELATIONS It is indubitable that, at sites 17 and 26, underlying cultural deposits which have no 11 are earlier than the upper cultural de- its where shell occurs. According to Profes- . Hans Jenny the cultural deposits underlying shell middens at locations A of both sites and 26 appear to be mature soil profiles, ch indicate a long interval, perhaps centur- during which the locations were not occu- ld, until their re-occupation by the shell- den people. Dr. G. H. Curtis, in his petro- aphic study of twenty-five potsherds, dis- ered four sherds from site 17 and one from te 26 (all from the underlying deposits with- shell) which showed alteration of certain stituents since burial. In Appendix I he ex- sses the opinion that such alteration "is not rapid process, and may be of the order of dreds of years." The question of the age of the underlying Itural deposits remains unanswered. My guess, it is only a guess, is that the bottom of be deposits at site 17, may date from about two ousand years ago. The occurrence of Late period incised erds in shell midden at location A, site 17, eems to offer a nice temporal correlation. reover, the depth of the shell midden corres- nds approximately to that at site 26, about ghty miles away, where a few incised sherds of e same Late period type as those at location site 17, occur in the shell-midden layer al- ough they are much scarcer here and irregular ndistribution. (See tables 1, 2, 20, and 25.) us it seems reasonable to suggest the likeli- od of contemporaneity of the shell midden at hese two sites, especially since both places re occupied in modern times. I have therefore pplied the term Late horizon or Late period to e shell-midden layer at location A, site 17, d at site 26. If an anthropologist may ven- re to accept legendary tales at face value, ese shell-midden deposits might be attributed the descendants of the immigrants arriving der the leadership of Lutunasombasomba on the ship "Kaunitoni." The longest count of genera- tions I obtained at Vunda records fourteen gen- .rations from Lutunasombasomba to the present ETui Vunda. This is not a very long list by Polynesian standards, but for that very reason it is perhaps these generations that are to be correlated chronologically with the relatively ghallow Late culture (shell-midden) deposits in ,'location A at both sites 17 and 26. This view, already set forth in my published paper (1951), would date the beginning of these shell middens 'from about 1600 A.D. The shell midden at location B of site 17 does not fit this picture, for it extends down- ward to a much greater depth (table 1) and yielded only an insignificant lot of incised sherds (table 20), which were very rare below the surface. I have already suggested that lo- cation B was unoccupied during most of the Late period. Relief sherds are considerably more abundant at location B than incised, but the really dominant material is plain ware. There may have been an immigration between the Early and Late periods which accounted for a Middle period, apparently indicated at site 17 by the abundance of plain pottery between the lower levels (Early) with relief-decorated pot- tery and the upper levels (Late period) with incised-decorated pottery. Table 20 presents these data bearing upon the possibility of a Middle period. I do not feel Justified in at- tempting lines of demarcation in the table to separate three periods; rather, it is a matter of transition in abundance from one type of sherd to another. At location B the weight of incised sherds at any level does not exceed 1 per cent of the total weight for that level (see table 20). I am not sure of the significance of this, but have suggested that it may mean that, during a Middle period characterized by plain pottery, location B was continuously occupied, while lo- cation A was largely unoccupied. Earlier, I ad- vanced the hypothesis that the sliding down of the huge boulder that farms the rock shelter at location A may have ended the occupancy of that area for a long time, that is, for most of the Middle period. It would also be during this Middle period that the mature soil profiles at unoccupied location A would have had opportunity to form. It may be that, when location A was again occupied, it was by the Late or shell- midden people, and that the shell deposit in the rock shelter accumulated. (See tables 1 and 20.) Upon this assumption the presence of the few Early wavy relief sherds found there (table 22) could be explained by disturbances due to native digging or by deposition before the formation of the rock shelter. The complete absence of wavy relief sherds from site 26 may mean that site 17, where they are abundant in the lower levels, is an older site; or it may mean that, if the two sites were contemporaneous, no wavy relief sherds reached site 26 as trade articles. But since crossed relief sherds were found at site 26 as well as site 17--and are presumably trade pieces at site 26--the hypothesis of greater age for the lower levels of site 17 is perhaps correct. The detailed tables 20 and 22 for site 17 show a small amount of relief sherds in the shell-midden levels (Late period) of location A. The decorative style of some of these sherds is Late, especially the ribbed, applique, and gouged styles, which also occur in modern pot- tery. The presence in the shell midden of occa- sional wavy relief sherds is probably due to earth-oven digging. Table 25, for location A, site 26, gives detailed distribution (in terms of weight) of incised, plain, and relief sherds in the various levels of the rectangles. There was only one level in which plain sherds weighed less than 40 per cent of the total; usually they weighed 90 per cent or more. Relief sherds were next in 235 236 ANTHROPOLOGICAL RECORDS order of abundance, but were not as numerous as at site 17 (cf. table 20). Incised sherds were the rarest, their weight exceeding 10 per cent in only one level. Thus the sherd materials from sites 17 and 26 indicate minor differences in the ancient cultures of these two sites which are about eighty miles apart by modern roads, though, of course, closer in an air line. There were local cultural differences in historic times, as there are today. House platforms at the two sites are unlike, rectangular at site 17, round at site 26. At site 26 burials were made in house mounds, at site 17 they were not. Modern pottery in the two regions differs, cooking-pots being deco- rated with incising in Ra Province where site 17 is situated and plain ware being used in Lautoka Province where site 26 is situated. Ra cooking- pots have rounded, tapering bottoms and are laid on their sides in cooking, whereas Lautoka pots have broad, rounded bottoms and are used upright. Lastly, there are considerable linguistic dif- ferences, mutually unintelligible dialects being spoken in the two regions today. EXTRA-FIJIAN RELATIONS Turning to relationships between the Fijian archaeological material and similar artifacts elsewhere, we find that resemblances are with Melanesia and other regions to the westward, none with Polynesia to the east. The single shell hook (fig. 1, c) recovered at site 17 matches a Solomon Island type. Stone adzes (figs. 2-4) resemble some of those from New Cale- donia, New Guinea, and the Philippine Early Neo- lithic, and differ from the adzes prevalent in Polynesia. Fijian pottery, both prehistoric and modern, exhibits some kinship with that elsewhere in Melanesia, as Margarete Schurig's seven plates show.57 MacLachlan has pointed to certain resem- blances between the decoration of New Hebridean pottery and Fijian.58 The resemblances, however, are to the Late period and modern Fijian pottery, not to the Early horizon relief ware. Surridge, writing on the decoration of Fijian water jars, also emphasizes New Hebridean relationships, and cites three types of decoration in support of her thesis:59 rope design, incised designs, and knob-motif (which I have called nubbins). The incised decorations on potsherds from the Northern Solomons, figured by MacLachlan,6 are even more reminiscent of the Late period in- cised sherds from our site 17. A Bishop Museum specimen from the western end of New Britain (shown in my plate 16, a) displays an incised design. The roulette-incised design of one sherd from site 21 (p1. 19, c) suggests the 5TSchurig, 1930, pl. 1-7. 58MacLachlan, 1939, P. 54 . 59Surridge, 1944, pp. 21-22. 6?Maclachlan, 1938, figs. 1-29. I style of decoration on certain sherds excava on the Isle of Pines, south of New Caledonia," and in Tonga.62 The Tongan potsherds described by McKern probably represent a diffusion from Fiji, not the reverse. A few are incised, the balance, plain. Their connection seems to be with the Late period in Fiji. Some of them have a seg mented, or roulette-impressed, incised design' like the one from site 21 mentioned in the p ceding paragraph. The paddle-and-anvil technique and the ii cised decoration of certain Papuan pots64 are gestive of Fijian methods and of certain of t markings of Late horizon and modern incised tery in Viti Levu. However, the paddle-and- anvil technique by itself is too widespread it both the Old World and the New to be worth muc in any attempt to establish connections betwe Viti Levu and some specific region to the wes This is also true of relief decoration, which occurs in Malaysia, Indo-China, Korea, Japan, eastern North America, etc. Moreover, the ap.. plication of relief pattern by means of check stamp, carved paddle, or wrapped paddle is bo to bring generally similar results wherever ut lized, whether in Borneo or the eastern United States. Nearer to Fiji than Borneo is the modern lief ware of the islands of Bougainville Strai in the Solomons. Wheeler (1928) describes for. Alu Island the use of a carved wooden paddle t impress a relief design upon the vessel. One such paddle from Shortland Island is shown in plate 16, c. This produces the general, but the specific, type of relief pattern character istic of the Early horizon in Fiji. The probable appearance of certain Fijian. crossed relief vessels is perhaps simulated by an example from the Skaran (Skarang) River, southwestern Sarawak, Borneo, pictured by H. L.' Roth,65 who describes the cross lines as being relief and the enclosed spaces depressed, pre-, cisely as in our sherds from Viti Levu. Such pattern was made with a carved paddle or malle like the one illustrated by Nieuwenhuis66from the Bahau and Kenya of Borneo. The body of thda vessel, not the neck and rim, was decorated, .asl was presumably the usual practice in Viti LevuwA Van der Hoop figures67 a sherd from Kerintji, west coast of Sumatra, with crossed relief in lozenge pattern. Also of this general type are, vessels and sherds from Indo-China, figured by", Janse;68 from Perak, figured by P. V. Stein Cal-' 6~~~~~~~~~~~~~~~~~ 61Lenormand, 1948, pis. racing p. 56. 62McKern, 1929, pl.. 6A. 63Ibid*, pp. 115-119. 6cC. 0. Finsch, 1903, pp. 329-334. 65H. L. Roth, 1896, p. 391. 66Nieuwenhuis, 1907, pl. 51, fig. 19. 67Van der Hoop, 1941, p1. 87, fig. J. 68Janse, 1947, pl. 151, and others. i r 4 I GIFFORD: EXCAVATIONS IN FIJI els69 and I. H. N. Evans; and from the Hong- region.70 Various other examples might be d, but the above suffice to show the occur- e of crossed relief decoration in the Malay- and southeast Asiatic regions. Angular relief decoration occurs in Indo-China,71 is apparently less common. It is of interest to record certain of the essions of Professor H. Otley Beyer (letter ugust 2, 1949), eminent authority in Philip- anthropology, upon examining sample Fijian s which I sent him. First, as to the 12 incised specimens om your Late Horizon, I can say at once at - apart from certain minor differences texture of the body ware, etc. - they ld have come from any one of several sites Luzon and other parts of the Philippines. is general type of incised decoration - e of our specimens being practically iden- cal with yours - is found in sites of three e groups in the Philippines, which we date follows: , Prehistoric Iron-Age (2nd or 3rd century B.C. to 9th century A.D.). Protohistoric Early Porcelain-Age (9th to 14th centuries A.D.). Historic Later Porcelain-Age (15th to 19th ,centuries). Plate 20, d-p reproduces a picture of Lippine incised sherds serat me by Professor er. Their resemblance to site 17 sherds is Lking. If there is any genetic connection Ieen Fijian and Philippine incised sherds, it tia intermediate island groups, e.g., New tain, the Solomons, the New Hebrides. The tural associations are different: in Fiji Lithic, in the Philippines Iron Age. But ob- sly Iron Age incised ware might well diffuse lOut the accompanying iron. As already dis- Ied, Fijian incised ware has its geographi- .y nearest relative in the New Hebrides, ch might well be the penultimate stepping e in eastward diffusion. Crossed relief sherds were sent to Profes- Beyer, who says: Some of these suggest to me certain tate Bronze-Age wares of South China, Formo- a, and parts of Indo-China (probably all of ate Chou or early Han date - say 500-150 ..) .... We do have in certain Lake District Fites, Rizal Province, especially, certain tather thick pottery fragments covered with I grid of raised network; but I am not cer- tain that this pottery is of native origin, s it may have been imported from South China 6r Siam. Had your fragments come from one of iur sites I would be inclined to think that the hard semi-stoneware had been imported 69Stein Callenfels and Evans, 1928, pl. 70, fig. 8; p. 157. 7OSee, for instance, Finn, 1932, pls. 33-34. 71See, for instance, Janse, 1947, pl. 102, fig. 1; guy, 1902, pl. 9. from the continent, and that the softer red pottery was a local and somewhat cruder imi- tation of the imported wares. Referring to certain Fijian wavy relief sherds, Professor Beyer writes: These angular lines or chevrons, inter- spersed with diamond-shaped figures, are also common on pottery and stoneware from South China, Indo-China, etc., particularly of Bronze-Age times and after. And of the curvilinear wavy designs, he says that he-has never encountered anything just like them before. See my plates 25-27 for the angu- lar and curvilinear wavy designs. Summarizing, Professor Beyer states: Judging by the potsherds alone, then, your Fiji sites show two, or more probably three, periods of ancient contact with South- east Asia and Malaysia. The Fiji wares show no relationship to any authenticated Neolithic pottery with which I am familiar - but they do show decided affinities with the Bronze- Age and Iron-Age wares and decorative patterns of this region. The continued use of stone implements, and the lack of metal tools or weapons, is not especially remarkable - as the Late Neolithic and Bronze periods were con- temporary over practically all of Southeast Asia and Malaysia, while in Western New Guinea both Early and Late Neolithic implements have survived side-by-side down to modern times; and this despite the fact that the presence of considerable quantities of ancient porce- lains, glass ornaments and beads clearly in- dicate sporadic contacts with outside civili- zations for a thousand years or more. On the other hand, most of your sherds show a certain degree of deterioration of the original imported designs - in some cases mild, but in others so marked that it is only with difficulty that the original pattern can be made out - and this would seem to indicate a rather long period of local re-copying of slowly deteriorating designs, without the stimulation of freshly imported material. If the two general decorative styles of Fiji are to be accounted for by the theory of diffusion instead of local invention, it would appear that the Late period incised ware con- nects with the Iron Age incised pottery of the Philippines via New Britain, the Solomons, and the New Hebrides, the crossed relief ware with that of the Bronze Age of southeastern Asia via Borneo and the Solomons, and the wavy relief ware also with the Bronze Age of southeastern Asia with no intermediate occurrence yet dis- covered. Undoubtedly, the tracing of most of Fiji's material culture will lead to the west, as the archaeological evidence suggests, in spite of the powerful Polynesian social influences that have assailed Fiji from the east. Ethnography and archaeology are the twin disciplines for the in- vestigation of relationships of the material culture in Melanesia and Malaysia. 237 APPENDIXES APPENDIX I PETROGRAPHY OF POTTERY By G. H. CURTIS Twenty-five selected specimens of pottery examined in thin section under the polariz- microscope. In the preparation of thin sec- of such friable material, it is almost ntial that they be first impregnated with binding material to hold them together dur- the grinding process. In this instance a id plastic monomer named Castolite was suc- sfully used, to which first had been added a 11 amount of hardener. The specimens were aced in a castolite bath, and the whole placed a vacuum Jar in which the pressure was re- ced to the boiling point of the castolite at temperature. After this pressure had been intained for several minutes, or until all air been expelled from the pores, the vacuum was eased, allowing air pressure to force the tolite into the pore openings. The specimens e then polymerized at the proper temperature an oven until hard. They were then ready for tioning, and no difficulty was encountered in ir preparation. From the first it was apparent that, al- ugh there were small compositional and textu- differences between the various sherds, be- quite marked in a few, for the most part se were so slight that it became necessary to pile all data in table A for rapid comparison individual sherds. ; A casual glance at table A reveals at once remarkably uniform mineralogy of the sherds both sites 17 and 26. It would appear cer- Fin that the source rock, from which, by chemi- 1 weathering, the clay, as well as the mineral t ins, was derived is essentially the same for Fth sites. The one striking mineralogic dif- brence between the two sites, the presence of * tite in sherds from site 26 and its absence sherds from site 17, may or may not be sig- ? icant. Biotite is rather scarce in the few herds (pl. 29, a) in which it does occur, and t enough sherds were examined to say definitely hat it does not occur in the source material of ite 17. Other variations in mineralogy may have ignificance but are probably to be expected, wing to local variations in clay and temper. The ultimate source of most of the material n these sherds is andesite of somewhat variable Dmposition. This is clear, not only from the pmposition of the grains but also from the fact hat actual fragments of andesite are often resent in the sections (pl. 29, b). A certain Mount of what appears to be deep-seated igneous ock of granitic composition is present (pl. 29, ), which may have been brought up from the epths during extrusion of the andesitic mate- ial - a not uncommon phenomenon in many regions. Shell fragments and foraminifera, listed as cal- cite in table A and present in three of the slides, are, of course, organic and of marine origin, as determined by Professor R. M. Klein- pell, who was able to identify, tentatively, a genus of Orbitoid foraminifer, probably Miogypsina (pl. 29, c). As for comparing sherds from sites 17 and 26 with the idea of contemporaneity or identity of origin of similar levels, it can be stated that no valid criteria for such conclusions were found. However, some sherds from the two sites are so similar texturally and mineralogically, or differ so slightly, that it is not impossible that they had a common maker or source of mate- rial, but with source material of the two sites being quite similar, coincidences in this re- spect might be expected. Such similarities have been noted in table A and will not be discussed here. The sampling is too small to reach any conclusions on valid differences between plain, incised, and relief sherds. Several lines of evidence lead to the con- clusion that in all, or nearly all, of the spe- cimens examined the clay has been tempered with sand. Further, this tempering has been done ac- cording to a rather general ratio of two parts clay to one part sand by volume. Percentage of clay in each section is shown in table A, and it may be seen-that in most sections this percent- age does not vary far from the overall average of 67 per cent. The strongest reason for believing that the sand was added is the freshness of the grains in most slides, especially of the feldspars. Natural clay is formed from the breakdown of aluminum silicate minerals by chemical weather- ing. Therefore, evidence of transitional stages to clay might be expected in grains in an incom- pletely altered rock. Such grains, showing pro- found but incomplete alteration, are present in many of the slides and have been classed as clay in percentage measurements. However, in most sections the grains are strikingly fresh, even the feldspars, which are generally the first to decompose. It is safe to assume then, that sand, from which all soft altered material had been re- moved by abrasion, either by stream transport or surf action, was added as temper to the clay. In support of this, many of the grains show the effects of abrasion in being slightly rounded. Crushing of the .temper in a mortar would only have made them more angular. Indeed, in those sections containing foraminifera, the tests are almost invariably whole, which eliminates this possibility entirely for these sherds, and sug- gests beach sand as the temper. [239] ANTHROPOLOGICAL RECORDS TABLE A PETROGRAPHY OF POTTERY Site 17, 8287-8795; Site 26, 10334-10727 4, abundant; 3, common; 2, scarce; 1,, trace U) U) Cl) 4 O) U) U) ~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 U) F! z U) co U) 4 bo c) Cr Cl . H U) C)A C)b crS-. <-4H Cd ) 4)U)U U bO .H H H U) a) U) 0 - U) U) ) bO H 4 4 4) U) .H ?4 S P N 00 4 U - 4) 4 4.H.H H 4 0 00 H 4) "H 0 4) "- lH H 4H U)4 $ 0 ) o .H rHO D h zF- bO 4 H U 4) r- Cdo 0 Id ) r-i r-- U) Cd Cd Cd 4 O 4 ) 'HO bS S H 'H Cd 00 0 Cd C Ud rH 1 rH W4 ; 0 'H r-iH CdH gHl al H O O? o . m 0 c o I P4 Q o : m o 0-' x 1 0 I 0 - P - = coI-- 4 2 4 4 3 4 4 1 4 2 4 3 3 4 2 4 3 2 4 1 3 4 1 4 3 3 4 3 4 4 3 1 3 3 4 1 3 1 4 2 1 3 1 2 1 1 3 4 3 1 3 2 3 3 2 1 1 4 1 3 3 3 1 3 1 2 2 2 1 2 2 1 1 1 3 2 1 3 3 2 3 1 3 2 3 2 1 3 2 4 4 3 2 3 3 1 3 3 1 1 1 3 1 3 2 4 4 4 1 3 3 3 2 2 2 1 3 1 3 1 1 1 1 1 3 3 3 3 1 3 3 1 3 3 2 2 1 2 1 3 4 1 3 3 3 3 4 3 1 3 1 4 4 3 2 3 3 3 3 1 3 3 2 4 2 4 4 3 3 3 3 3 3 4 3 *Shown in plate 29. *8287 ... 8334 ... 8416 ... 8464. .. 8532 ... 8554... 8589... 8666 ... 8682. .. 8725 ... 8787 ... *8794 ... 8795 ... 10334 ... 10369... *10392. . . 10417. . . 10444 ... 10454 ... 10475. . . 10490... 10510 ... *10516 ... 10526 ... *10532 ... 10727 ... 57 68 64 75 72 71 79 62 68 68 73 65 68 73 72 81 68 68 61 85 72 85 65 68 62 64 1 3 4 4 2 4 2 3 3 3 2 2 3 1 2 2 2 3 3 2 1 3 2 2 2 3 3 3 2 2 3 4 3 2 4 3 4 1 4 2 2 4 4 4 1 2 2 1 3 2 1 1 4 3 3 240 GIFFORD: EXCAVATIONS IN FIJI Each section shows a rather uniform grain ,which varies from section to section. This ksily explained by current action sorting band (pl. 29 a, d, e, f). In many of the is a particular mineral predominates, as is Mn in sands of all types in which a selec- ,,sorting has been effected by current action Von differences in density of the various lituents. Specimen 8287 illustrates this (pl. 23, c, e), being composed principally gite grains of uniform size. A few of the sections contain sand grains i, though almost certainly added material, theless show a marked degree of alteration; !eldspars particularly are altered (pl. 29, Alteration can be seen in many of these grains to be progressing inward uniformly from the present periphery of the grain. These grains are too fresh to be considered part of the clay matrix, so it is thought that they must have begun to weather subsequent to the burial of the sherds. It is perhaps significant that the two specimens showing this subsequent alteration most strongly, namely 8794 (pl. 29, f) and 8795, were taken from the deepest part of site 17. Just how long it has taken to effect this degree of alteration there is no way of de- termining since so many variable factors are in- volved, but under the most favorable conditions it is not a rapid process, and may be of the order of hundreds of years. NOTES TO TABLE A These notes give provenience, pottery types, and memoranda on the sectioned sherds. RECTANGLE E3-4 F3-4, LOCATION A, SITE 17 8287. Top triangle of sloping mound mate- , Late incised type, containing plainly Ile calcite specks. Plate 29 c, e. Very orm grain size and relatively high propor- of sand to clay. Not similar, however, to r fossil-bearing slides. 8334. Depth o-6 in. Crossed relief type. 8416. Depth 12-18 in. Relief, type inde- nate. 8464. Depth 24-30 in. Late incised type. F almost no similarity to other fossil-bear- glides. 8532. Depth 36-42 in. Plain type. 8554. Depth 42-48 in. Crossed relief type. 1e grains show uralitization. Compare with 8787, 8795, all wavy relief type sherds, 4ominant early type of ornamented ware at 17. 8589. Depth 54-60 in. Ribbed relief type. 8666. Depth 72-78 in. Wavy relief type. slide, as many sand grains lost in prepara- . Augite shows uralitization. Almost tical mineralogically and texturally to 8554 8787. 8682. Depth 78-84 in. Early incised type. Igrains weathered, but not as severely as in , which, though more reddish in color, this pables in texture. 8725. Depth 90-96 in. Crossed relief type. 8787. Depth 108-114 in. Wavy relief type. We altered in part to uralite, but not as as in 8795, though the slides otherwise are s similar. Even more similar to 8554 and jthough darker in color. 8794. Depth 114-120 in. Plain type. e 29, f. Sand grains show marked degree of bering. The evidence indicates that this oc- gd after burial of the pottery. 8795. Depth 114-120 in. Wavy relief type. Lte altered in part to uralite; also feldspar hered. In many respects it shows similarity ther slides containing uralitized augite, v 8554, 8666, 8787. RECTANGLE A4-5 B4-5, LOCATION A, SITE 26 10334. Depth 0-6 in. Plain type, contain- ing plainly visible calcite specks. Absence of chlorite and presence of a trace of augite may not have much significance. 10369. Depth 12-18 in. Incised type. 10392. Depth 24-30 in. Incised type. Plate 29, d. Sand grains show uniform size, but are finer than in any other slide. Gray color of clay is unlike all other slides. 10417. Depth 36-42 in. Crossed relief type. 10444. Depth 54-60 in. Plain type. Con- tains biotite, and is mineralogically similar to 10490 and 10532, but differs somewhat in general texture. 10454. Depth 66-72 in. Plain type. Sand grains show alteration, probably from weathering after burial; texturally similar to 8794. 10475. Depth 78-84 in. Ribbed relief type. Color similar to 10516, but otherwise not com- parable. 10490. Depth 84-go in. Crossed relief type. Contains biotite, and is similar in mineralogy and clay color to 10532, but sand grains are smaller and the ratio of sand to clay is smaller. See also 10444. 10510. Depth 96-102 in. Crossed relief type. 10516. Depth 102-108 in. Plain type. Plate 29, a. Very similar to 10475 in its red color, but mineralogically and texturally quite different. 10526. Depth 108-114 in. Incised type. 10532. Depth 114-120 in. Plain type. Plate 29, b. One of few slides containing bio- tite, but is similar to only one of them, 10490. Grain size, however is much coarser than in 10490. See also 10444 which is somewhat similar. RECTANGLE B4-5 C4-5, LOCATION A, SITE 26 10727. Depth 106-113 in. Plain type. Very poor slide. Clay percentage probably not accu- rate. Minerals show alteration from weathering. 241 APPENDIX II SITES VISITED The following pages present descriptions of sites and observations made in the course of reconnaissanc-e prior to selecting two sites for excavation. These two, sites 17 and 26, are discussed in the body of the paper. As far as possible I have incorporated in the descrip- tions ethnological material bearing upon the sites, as, for example, the name of the "tribe" (yavusa) formerly inhabiting each. Some of the sites are the legendary original habitations of tribes and are therefore designated by the term yavutu. Any deserted village site is called koro makawa, "old village." The locations of the thirty-nine sites visited are shown on map 1. (Sites la and lb bring the total to thirty-nine.) THAUTATA (SITES la, lb) The site of the modern Thautata and the first Thautata is the same; no specimens were collected here. The second and third sites, labeled la and lb respectively, yielded speci- mens, which have been catalogued with these site designations. These sites are in the District of Mbau, Province of Tailevu. See map 1. On April 5, accompanied by Dr. H. Silvester Evans and his Fijian servant, I went to Thautata, to which I had been invited by Mr. Ponipate Talei, Fijian member of the Fiji Museum staff and resident of Thautata. An account of the day's doings follows. We traveled by bus from Suva to Nausori on the Rewa River, thence by another bus to the crossroad leading to Thautata. We then walked one and a half miles over a road which was soggy after an all-night rain. After tea and kava in Talei's house we went to the house Naseuvou, the assembly house of the village, where kava was served ceremonially. No women were present. Four young men, naked to the waist, oiled, and decorated with waist ornaments, armbands of green leaf, etc., sat near the bowl. The one who strained the kava had no ornaments above the waist. The other three did. The cord of the bowl faced Evans and myself, who were seated at a table. The bowl was very large but lacked an Ovula ovulum shell like that attached to the cord of the bowl in Talei's house, where the kava had been brought in already mixed. At both places hands were clapped when the kava was drunk. At the house Naseuvou a man strained the kava with hibiscus fibers to the accompaniment of a monotonous bass chant sung by all the seated men. It was strained about twelve times (I failed to keep exact count). There were two strainers. Each time the strainer was removed with kava particles in it the singing stopped instantly, to be resumed when the alternate strainer was introduced into the bowl. Each full strainer, after being wrung out, was handed to one of the young men sitting behind the $ who strained. He shook out the kava fragmen The two young men sitting at the sides of ti4 bowl did nothing. Each time, when the man h finished straining, he folded the empty strl er in his left hand and struck it two resou1 ing blows with his right palm. Twro or three7 times during the process he also struck the strainer in this way. The water for the kava was poured from: globular pottery jar, as in Lester's illustr tion.1 I am not sure, however, that the inc ing on the jar was the same. Green leaves (sugar cane?) protruded from the jar's mouthVi decoration, recalling similar decorations po trayed in ancient Egyptian table ware. Thes were removed before pouring. Evans received the first cup, a man (31 years old) who spoke excellent English receiv. the second cup. I received the third, then others were served, the chief, I believe, be" fourth. Evans had presented a small package of verized kava. A whale's tooth (tambua) on a sinnet braid was presented to me at the end di the ceremony. It lay on the floor meanwhile, being handed to me finally by the English-spea ing man. All present, except the four cere- monialists at the kava bowl, were wearing shii Evans made a speech in Fijian., I made a brie, one in English, translated by the English-spg ing native. Considerable kava was left in t1 bowl. Some of this was drunk after our retun from visiting two old sites (la and lb) but without ceremony other than hand-clapping. 1 After the formal kava ceremony we set oul to visit the ancient sites (koro makawa). Thl first was at a large tree in the village, whel an exceptionally large stone adze blade, in F Museum, had been found. Four settlements al- together were occupied successively by the Thautata people after their migration from th Kauvandra Mountains. The first and fourth vi. lages were at the present site of the modern Thautata, though the first was called Naseuvo ("beginning"), not Thautata. The second (sit la), third (site lb), and fourth settlements were called Thautata. The large tree represe ing the first was the yavu ("foundation") of temple. The yavu of the assembly house and o the cemetery of this settlement are not known The present assembly house is called Naseuvou memory of the first settlement. The present lage is on a hilltop, as was the first settle ment. The other two (la, lb) were on lowland When the ancestors arrived from the Kauv dra Mountains, there were two "ladies" (spiri at Naseuvou. One of these was married by the leader. The Thautata people belong to yavusa ronikalou; they name Nakauvandra as their yav 'Lester, 1941, pl. 3b. [242] GIFFORD: EXCAVATIONS IN FIJI Site la was about a mile away from the ent Thautata and buried in jungle. A stream ater flowed through it. The rectangular forms of the houses are outlined by stones. uropean objects were found there, but there a few plain potsherds. There were 'bivalve e shells in the soil there as at the other Be, but they were less abundant than at site here they were very numerous. These shells, and Chione-like, were the remains of food. ailed to take samples.) Small, quite fresh -er-of-pearl shells, also the remains of food, seen in various places. Site lb was close to the mangroves and t half a mile from the present Thautata. of the older men of today were born there, o'bf them being now fifty years old. No one iving who was born at site la, which must been deserted long ago. At site lb there the yavu of a church--nothing was to be seen but the place was pointed out. An up- limestone slab about 3 ft. long and 18 in. marked a former playground. Fragments of pean ceramics and glass bottles were in evi- e. Marine bivalves were very abundant. At place, the community was said to have split four "tribes," each with a separate loca- for its houses, but all collectively called tata. About 1912 this site was abandoned the village returned to the original (pres- site and was called Thautata (not Naseuvou). eason could be elicited for the successive ts. Potsherds were found at sites la and lb; .were said to be trade pieces, as no pottery iade by the Thautata people. Those from lb were examined by the old men upon our n to the present Thautata. The thin in- fragments were said to have been obtained the Soso people of Mbau, who made them; were from vessels used for cooking fish. thick sherds were from large vessels ob- ed from Nggelikuro2 near Londoni,3 District akasa, Tailevu Province. In addition to berds and European ceramics and glass, I col- d at site lb a small cylindrical stone, looked like a natural pebble but had been as a hammer, so the natives said. These jaid to be quite common. OLD VERATA (SITE 2) On April 11, I went to Uthunivanua with Dr. . Mr. Roth arranged for the Roko Tui tvu (Joeli Ravai) to meet me at Mbau Island ~for us to proceed by sailing canoe to Uthu- aua. Leaving Suva on the 8 A.M. bus to ri, we connected there with the 9 A.M. bus au. The bus terminus is at the embarcadero e mainland where the telephone line from reaches the mainland. From there the d was reached by rowboat. From Mbau we eli, "clay"; kuro, "pot." ondoni, "London," is the embarcation village for the to Levuka, Ovalau Island. went by sailing canoe (thamakau) to Uthunivanua, where we spent the night in the acting chief's house. The canoe had a matting sail and looked like one pictured by Thomson.4 After presentation of a tambua and some rather informal yanggona (kava) drinking without chanting, we set out to visit the deserted ya- vutu Verata, about a mile away, famous as giving rise to both the Rewa and Mbau kingdoms, accord- ing to Ratu Sir Lala Sukuna. Nine "tribes" (yavusa) claim Verata as their yavutu. The place was still occupied when the acting chief, a man of seventy, was born. However, the site proved densely overgrown and I saw no trace of European articles. Shell was abundant, as also at Uthunivanua and Mbau. As we walked along the beach on the north side of the peninsula we saw several rectangular slabs that had fallen down from the Catholic cemetery on the edge of the bluff. Similar limestone slabs formed the facing of a house platform at Verata. The mound they were on was perhaps 3 ft. high and the form rectangular. There was another small mound near by, also a small low platform of small stones, about 3 ft. square, said to mark the house's gateway. There were traces of a moat and an earthwork. The acting chief at Uthunivanua said the potsherds found at Verata were not made there, that the Verata people got their pottery from the fisherfolk of Nggoma Island. The Verata site was only a few feet above sea level, and there were mangroves close by on the north side of the peninsula. At Uthunivanua I saw one stone adze blade from the Verata site. It was of medium size and quite heavy. Evans said it was a western Viti Levu type. The modern village of Uthunivanua. contains three yavusa: Vunivalu with yavutu at Verata (site 2); Mathoi, of unrecorded yavutu; and Yatusawa with yavutu at Nanduivosavosa, Verata. Both Uthunivanua and Verata are in the District of Verata, Province of Tailevu. According to the legends of the yavusa Vunivalu, the fore- fathers of the tribe landed at Vunda in Lautoka Province on the west coast of Viti Levu and pro- ceeded to Verata (site 2). Visible on the beach, it being low tide, were tree roots which revealed red bark when cut. These were said to be the roots of a tree growing in the Kauvandra Mountains. Other parts of its roots are to be seen at the base of the Tualeita Ridge near Lomolomo, Lautoka Province, up which Lutunasombasomba and his followers climbed when they set out for the Kauvandra Mountains .5 On April 12, when the tide was high enough, we sailed back to Mbau. At Mbau we saw much shell and black earth at the landing-place facing the mainland. The desirability of excavating Mbau is dubious, since much of the soil has been scraped down from the high part of the island. We saw two house platforms, each with two ter- races, occupied by houses. *Thomson, 1908, opp. p. 290. 5Gifford, 1951, p. 169, pl. 2, a. I i 1? 1, ??i h, 243 ANTHROPOLOGICAL RECORDS Kumbuna, on the mainland near and slightly to the north of Mbau, was the last place occu- pied by the yavusa Kumbuna before the tribe moved to Mbau Island. Joeli Ravai said slips i the cliff have exposed burials at Kumbuna. NANGGUATAMBOTO (SITE 3) On April 18 I visited the site of Nanggua- tamboto, about two miles north of Tavua Fijian village; in low land near the edge of the man- groves. Several house mounds without stone borders were seen. We saw some shell, and col- lected sherds. The older informants said that the pottery was made by their ancestors but tha pottery-making had now ceased. They asserted that the sherds found were of local manufacture not imported pieces. The Tavua people lived in Mbila in Chris- tian times; Mbila was on the site of the presen school at the Tavua Fijian village. The people had lived at Nangguatamboto until Christianity came, when they moved away, splitting up into three hamlets: Vanuakula, Wainitavua, and Nalan all between Nangguatamboto and Tavua in the Dis trict of Tavua, Province of Tholo North. Later these united to form the village of Mbila-Tavua Mbila is the name of the yavusa of the modern Tavua village, and of the yavutu, presumably at the schoolhouse site. The site of Vanuakula is near Nangguatam- boto, in a grove of trees (presumably vaivai, Serianthes myriadenia), and is marked by abun- dant shell and potsherds. TANDRAVULA (SITE 4) On April 19 I visited the deserted site Tandravula, which like certain others was lo- cated in the mangroves, apparently for protec- tion. The site lies about a half-mile west of Korovou, the present village of the yavusa West / East Korovoi Villag Road to Nandarivatu Diagram 6. Location of Tandravula, site 4. Navauvau. The Navauvau people were already Christians when they moved to Tandravula froi Navauvau, their yavutu, which was near Nangui Ln tamboto (site 3). Two people now living in Korovou were born in Tandravula. Some peopli yavusa Mbila also live in modern Korovou. aB Korovou and Tandravula are in the District oi Tavua, Province of Tholo North. The pottery found at Tandravula was made locally by thei. - ancestresses, the Korovou people said. Several house mounds were seen among thA mangroves at Tandravula. The site is inter-; sected by an abandoned government road, now a - placed by the King's Road. it TAVUA RAILROAD WATER TANK (SITE 5) On April 19 and 21 we collected potsher at site 5, which lies just north of the Kingl Lt Road about 400 yds. south of Nambuna villagei near the Tavua railroad water tank, which is the souith side of the highway. The site is about 250 yds. east of the railroad and highF ii, bridges across the Nasivi River. Here, in al ploughed field, was a low mound about 18 in.- above the surrounding surface, with consider4 L. shell and sherds in the soil. A government X parallel to the present east-west highway onE ran along the south edge of the site, whichi in the general area called Yalondro, in the ministrative District of Tavua, Province of Tholo North. This was evidently a forgotten site, fog the oldest man in Nambuna village knew no na4 for it as a village site, nor knew of anyone having lived there; he regarded it only as at farming place. The people of Vanuakula vill4 now abandoned, formerly had a garden there. VATUTHERE (SITE 6) This occupied site is on the south side the divide which separates the short streams the north coast of Tholo North Province froiu9 long streams of the interior which flow tol south coast of Viti Levu. Vatuthere is on of these streams, Nandala Creek, an affluentA the great Singatoka River. The elevation off; Vatuthere is probably around 2,000 ft. It i the District of Savatu, Province of Tholo No about one and one-half miles from the head ol Nandala Creek near Nandarivatu. The yavusa occupying Vatuthere village, Numbu. The chief of yavusa Numbu is Tui Nw who resides at the inland mountain village o Nangatangata, District of Savatu, Province o Tholo North. The yavutu is Namoriti, nearl drau, District of Nandrau, Province of Thol& North. From Namoriti some of the villagers moved to Namatakula (site 7), thence to Vat (site 6). Part of the modern village of Vatuthere been deserted. One section of the site was' cupied from 1913 to 1938, when the people mol further inland to Navai where a lumber mill operates. We examined the rubbish heap of tm 244 GIFFORD: EXCAVATIONS IN FIJI 1938 site and found cans, bottle fragments, marine and freshwater shells. The marine Is, mostly bivalves, were brought up from Tavua coast, said to be only a half-day's ney. The marine mollusks were of the same sie as I had seen at lowland sites. The r mussels were said to come from the Mba r. I was given, besides shell samples, a .bent of pottery from Rakiraki, Ra Province. oking-pot from Nandronga Province was lying ts side on two rows of stones over a fire in a small kitchen outhouse. It was used in position6 for boiling food and was much kened by fire. No pottery is made by the there people. e,Houses were on round mounds, but were them- es square and small, with pyramidal roof ported by a short center king post which ted on two crossed tie beams. These beams to the tops of four posts, one in the middle ach of the four walls of the house. Inside, llar of twilled basketry work supported the ers near the apex of the pyramid. NAMATAKULA (SITE 7) Namatakula ("the red spring"), on the east e of Nandala Creek, is about 5 mi. south of darivatu on the automobile road to Navai. On iL 23 we visited Namatakula, a former site of yavusa Numbu people of Vatuthere village. is in the District of Savatu, Province of Lo North. No sherds were found on the open e, but there were bottles and tin cans left the American troops. Even though the troops e there till 1944, the place was already much rgrown. Sherds were found in the rock shel- at the rear of the site. Behind the rock Lter rose a steep 1,000-ft. hill capped by fortress Navatunivitolo ("the rock from ch look down"), to which the people retreated time of war danger. Before living at Namatakula the yavusa bu lived at Namoriti, where the ancestor (vu) settled when he arrived from the Kauvandra tains. At Namoriti, the vu married a Ra an of Narawarawa and from them sprang the Kai u (people of the yavusa Numbu). The vu be- nine male children, who became the heads of e matanggali (lineages) of the yavusa (tribe). wht were listed: Navukula, Namoriti, Sukama- Nangerengere, Ratuniovalau, Nambulo, Sau- Wa, Nandola. In the overgrown Namatakula site in front the cave two or three rounded (oval or circu- ') house mounds could be discerned in the tekets. The rock shelter at the back of the Namata- la site was under a huge boulder which lay tride Waindungudungu Creek; the creek bed was y on the day of our visit. The opening of the elter faces northwest and downstream. Up- ream, the opening under the boulder was partly 6For pictures of pots on their sides for cooking, see kdon-Cumming, 1883, opp. p. 208; Wilkes, 1845, p. 349; Roth, 1935, pl. 15, fig. 1. blocked with a wall of small boulders, said by the Mbuli of Savatu District to be artificial. At the outer edge of the overhang of the boulder the height from floor of the shelter was 8.5 ft. The width of the main opening was 25 ft. There was another recess, 12 ft. deep, with raised floor at the northeast side, behind small boulders that supported or were wedged under the huge roof boulder. The distance from front to back of the cave was about 28 ft., not counting a low, narrow, 10-ft. recess beyond that. On the south side of the entrance the roof boulder projected 15 ft. in mid-air, while on the north side its outer edge rests on a smaller boulder. Evidences of human use of the cave included tin cans, one marine bivalve shell, traces of fire, smoked rock, reeds for bedding material in the small raised recess, and a few potsherds (declared by the natives to be from Ra Province). The floor was mostly of small stones; at times the creek runs through, as well as around, the rock shelter. The shelter was used by the Nama- takula women and children as a hiding-place dur- ing war. The fortified summit behind the cave was a more secure refuge. It was reached on the side toward Namatakula, by a ladder of vines and wooden rungs, which could be drawn up. On the opposite side access was easier and there war ditches were dug for protection. DOUBLE ROCK SHELTER (SITE 8) On the walk back to Nandarivatu after visit- ing site 7 on April 23 the Mbuli of Savatu Dis- trict took me to a double rock shelter. This was near an abandoned government track from Nan- darivatu to the interior and proved to be fifty minutes' walk from Nandarivatu. I estimated the distance as 2.5 mi. from Nandarivatu Rest House and about 3/4 mi. from Vatuthere village. The rock shelter was upslope on the right side of Nandala Creek, probably less than 1/8 mi. air line from the creek, and downslope from the old government track. The steep slope was so dense- ly clothed with vegetation that a trail had to be cut 50 to 75 ft. down to the shelter. The double rock shelter was formed by a tremendous triangular boulder imbedded, apex down, in the soil and overhanging both uphill and downhill. On both sides of this enormous boulder are rock shelters with ample clearance at the outer edge. On the downhill side there was ample evidence of occupancy--smoked ceiling, ash beds, and a trace of an earth oven. On the uphill side, apparently now the resort of wild pigs, there was less evidence, although hearths, one sherd, and two or three marine shells were found. On April 26 I returned with Rambithi, the Constable Samuela Mbatindreu, and the Postal Clerk of Nandarivatu to site 8 and dug (1) a foot- deep pit and (2) a 20-in.-deep pit, both in the downslope shelter. The shelter is 46 ft. long along the base of the rock by 22.5 ft. deep, measured from the outer edge of the rock at the center. The outer, overhanging edge forms a great arc. The inner 245 ANTHROPOLOGICAL RECORDS edge, 46 ft. long, runs nearly straight N by E and S by W. Most of the ashes were in the N-by- E half. The shelter is open to the west. Its roof edge on the western (downhill) side is 13 ft. 3 in. high. The floor ends at this point also and there is an abrupt drop of 10 to 12 ft. to a streamlet in a narrow gully, sufficient to supply sojourners in the cave. This streamlet runs down toward Nandala Creek. We dug a trench, 3 ft. by 6 ft., in a cen- tral position. The upper 6 in. was largely blackened by fire and yielded cultural items in- cluding a button and part of an umbrella rib, as well as potsherds and shells. The second 6 in. was red clay soil, hard to dig, and with no cul- tural material, not even shells. The north half of the cave, particularly toward the back, had an inch-thick layer of ashes, beneath which there was soft brown soil (altered ash?). A small pit was dug into this 20 in. deep. At 15 in. two sherds were encountered; above this level were three cobbles, which had apparently been used as pot rests. One or two small samples of oven stones were included in my collection. These were from an earth oven near the outer edge of the shelter and about the middle of that edge. We did not excavate the oven. KOROVATU FORTRESS (SITE 9) There are many sites in Viti Levu desig- nated Korovatu ("village of rock"). This par- ticular one lies on a ridge to the north of Nan- darivatu, separated from it by the deep canyon of Waikumbukumbu Creek. It is on the slope of a peak called Korolevu, a trigonometric station at 2,764 ft. elevation in the Province of Tholo North. The fortress was at one time occupied by the matanggali Numbuiluva of the yavusa Numbu, now residing in the modern village of Waikumbu- kumbu,7 at the base of the escarpment upon which Nandarivatu is perched. On April 24, Eremasi Naranga, a member of matanggali Numbuiluva and an important man of Waikumbukumbu, took Rambithi and me to Na Korovatu, "The Rock Village." After vacating Korovatu, the matanggali Numbuiluva occupied successively Natungere, Yandonu, and Waikumbukumbu. Another group of people, of the same matanggali, lived at Ndrion- drio, then moved to Rarakavindi, then to Waikum- bukumbu, where their descendants are today. Eremasi Naranga's father came from Ndriondrio. Eremasi named two other matanggali,namely, Nan- dala and Wana, which also live in modern Waikum- bukumbu. He also pointed out from the Nandarivatu Rest House a rock shelter called Matanawesi, in the canyon below and upstream from the Rest House, which was used as a camp place. The occupants of Korovatu, still heathen, had moved out of the interior, precisely why, the informant did not know. The yavutu of the matanggali Numbuiluva, Nandala, and Wana are lo- cated on Nandala Creek. That of Numbuiluva is 7Shown on Brewster's map, 1922. downstream a short distance from Vatuthere v lage, and is a large, flat, brown, tablelike rock in the middle of Nandala Creek, suppor, it is said, on a postlike stone. The tabula rock is irregularly oval, about 10 by 6 ft. When I saw it, on April 22, it was about 18 below the surface of the creek waters. The is also called Numbuiluva and is reputed to the yavutu8 of the snake (ngata) god Tuyando' as well as of the matanggali Numbuiluva. Fo kava (yanggona) offerings were poured into b creek on top of the rock. At my visit, kava made and drunk in a house at near-by Vatuthe before visiting the stone, over which some w poured. The informant said that there were other places, besides the flat rock in the Na dala Creek, which were sacred to the vu of th Waikumbukumbu villagers. One is a conical earthen mound on the hill slope west of the creek. Naranga said that all the Waikumbukumbu people were of the tribe Numbu and that the chief Tui Numbu lived at Nangatangata. Wai bukumbu has been inhabited only in modern ti since the people have become Christians. The ascent from Waikumbukumbu to Korovat was steep and took two hours. I judged it to a 2,000-ft. climb, and the altitude to be abo 2,500 ft. Korovatu was admirably situated fo defense, but very inconvenient for farming an fishing, the two principal sources of sustena People took refuge there only in time of peril from enemies. As we approached Korovatu, we first en- countered a stone house platform on top of a slightly lower hill, the steep slope of which curved around from west to north in a great co cave arc to Korovatu. This stone platform was in plain view of the fortress and served as t residence of lookouts, who could warn of enemy approach over the grassy hills. The lookout'8s| house platform was oval in form, the diameters being 19.5 by 12.5 ft. Its height was 1.5 to 2 ft. The top as well as sides were of unwork but selected, cobbles and slabs. A second sin lar lookout platform was situated about 100 yd farther away to the west. From the nearer loo out Korovatu bore north by east across the gre open chasm formed by the steep curving mountai, slope. By trail, curving level along the stee slope, the distance was half a mile. In air line from lookout to fortress a shout or trumpi blast could be heard, as the distance was much less. Korovatu lies on a series of transverse benches or terraces on the lower part of a nort south knife-edge ridge culminating in Mt. Koro- levu, 2,764 ft. high. From Korovatu, the Rest House at Nandarivatu bore south by east. To th eastward, cliffs and a precipitous slope gave way to a broad valley lying between Korovatu an the Kauvandra Mountains. To the westward the land dropped abruptly to the foothills and beyo to the flat land toward Tavua. Waikumbukumbu 8Thi8 is the only example I recorded of the applicatior of the term yavutu to an object which was not an original village site. 246 GIFFORD: EXCAVATIONS IN FIJI knot visible from Korovatu on account of the vening hills, but Korovatu was visible from modern highway both above and below Waikum- u. Korovatu is probably from 300 to 500 ower in elevation than Nandarivatu, which iot 2,900 ft. Xorovatu seemed practically impregnable to ce without firearms. The east, west, and h sides of the fortress could not be scaled he face of a defending force. The south ,-as will be described, was defended by a and terraces. The upper or north end of village terminated in a citadel-like rock, ng vertically 15 ft. and extending complete- cross the knife-edge ridge. Its height had further increased by a stone wall around northern and eastern edges. In this stone I a tree once stood, or perhaps it is better ay that the wall was built up to each side the tree, now only a burned stump. In the ices of the stump were calcined fragments of limb bones, said to be cannibal feast rem- s. Another uncalcined, but decaying, human r was buried near by in the earth atop an- r part of this wall. Proceeding from the south along the knife- ridge into the fortress, we came first to a L1 stone house platform for a guard's hut. followed, astride the ridge and about 100 apart, three low stone walls 4 or 5 ft. high. ye these to the north followed seven terraces varying height, partly made by cutting away )ry outcrops to make their faces vertical. The races were of natural origin, but had been ified to make them unscalable. Some of the .away stone or the talus at their bases had used to build walls to an even height, their edges, making a total height of 10 Narrow steps and openings gave access from race to terrace. The total length of Koro- from the first transverse wall to the north- edge of the citadel was less than 1/8 mi. here was the crest of the ridge over 50 ft. e. From the crest the view was unobstructed both the east and west slopes. There were several small oval or circular se platforms, some perched on the edge of the br cliff which fell away to the eastward in ,ical drops of more than 100 ft. The rock ap- red to be black lava. Water and wood were ob- ed from a small wooded gully to the north- ;, above which now runs the native track from iwbukumbu to Nasaiyani and Nananu in the far tern part of the District of Tavua, near the Province border. The trail to the sources of d and water was difficult indeed. Small vel spots on the ridge were said to have been led for planting. Sherds were fairly numerous; Eremasi Naranga iid pots were not locally manufactured, but had e*n imported from Ra Province, from Tavua, and oi Nandronga Province. There were marine bi- lives from the Tavua coast, but not so many as k,coast sites. Several stone fireplaces were en. These were oblong, about a foot wide and ft. long, with rather thin, flat stones stood ledge to form the sides and ends. On these, Swas said, large cooking-pots were laid on their sides, as I have recorded for Vatuthere (site 6). Because of heavy rain we obtained no photo- graphs of site 9. Its general character, how- ever, is similar to the ridge fort pictured by Weckler9 from a photograph by J. F. G. Stokes of a fortification on Rapa Island, Tubuai group. NAVATUNISALA (SITE 10) Navatunisala, adjacent to modern Ndrauniivi ("leaf of the Tahitian chestnut"), is on the edge of the mangrove swamp. Although the site is no longer occupied, on the day of our visit, April 29, the Ndrauniivi people were engaged there in making salt by boiling down salt water obtained from pits dug in the saline ground. Ndrauniivi and Navatunisala (site 10) have been occupied by the same group. The sites are on the coast in the northwest corner of Raviravi District, Ra Province. Our informant on this day was Navitalai, a man of Ndrauniivi. Vatukaloko, in the Kauvandra Mountains, was the yavutu of the yavusa Nakumbuti, one of the three yavusa listed for modern Ndrauniivi vil- lage. At Vatukaloko there were said to be three vertical stones of natural origin, in a row, about "132 ft." high, which served as a sun dial. Tradition says that the inhabitants of Navatuni- sala (site 10) came direct from Vatukaloko in the Kauvandra Mountains. Site 10 was Qccupied for three or four generations, and it was there that the people became Christians. After Ces- sion in 1874 they were exiled to Kandavu Island because of the activities of the messianic pro- phet Navosavakandua.10 From Kandavu, they were moved to Nanukuloa, the modern seat of govern- ment of Ra Province, and thence to the present village of Ndrauniivi. Near Navatunisala was one of the tradition- al landing-places of the famous immigrant ship ''Kaunitoni.''11 NASANGGA (SITE 11) Also on April 29 we visited site 11, while walking from Ndrauniivi to Nathilau Point, Dis- trict of Raviravi, Province of Ra. Nasangga is on the north side of the King's Road, about 1/2 mi. east of the Indian store at Ndrauniivi. In- formant Navitalai stopped to show me the knee marks of the god Rokomoutul2on two boulders at this place. He did not think of it as an ancient habitation site, but potsherds made it clear that it probably was once occupied, though it has long since been forgotten. Navitalai had a different explanation of the sherds, which is given below. 9Weckler, 1943, p1. 8. 10See Thomson, 1908, p. 14o; Brewster, 1922, pp. 256 ff., on Tuka, The Life Immortal. "Gifford, 1951, p. 170. 12Thomas Williams (1859, p. 171), says that Rokomoutu was the son of the god Ndengei's sister, born from her elbow. Waterhouse (1866, p. 357) uses the name Rokomautu, apparently for the same god. 247 ANTHROPOLOGICAL RECORDS Nasangga is a low knoll projecting into the coastal mangrove swamp. On top were two boul- ders of blackish stone common to the region, about 14 yds. apart. The upper one had one shallow depression on its upper surface, the lower one had three; all of these were called "knee marks" by the informant, but they appeared to be grinding depressions or querns. The story has it that here the god Rokomoutu from Verata (site 2) knelt to draw his bow to shoot a gar fish in the near-by water. The name Nasangga means "the water jar." I collected sample sherds, which were in general smaller and markedly more weathered than those found on most sites, apparently indicating a long deserted site. No house mounds were in evi- dence. The top of the knoll was about 15 ft. above high-tide level. The ground was littered with small stones and some potsherds and did not look particularly suitable for planting. The name Nasangga refers to a myth concern- ing the goddess Naivilawasa, who, came down from the Kauvandra Mountains with her water jar, but without the container in which to carry it. She dropped her jar and it shattered into innumer- able pieces. The Mbuli Semi of Rakiraki, who was with us, said that Naivilawasa was the vu ("ancestress") of the yavusa Namotutu to which he belonged. A story about the broken water jar of the goddess Li maui tukituki is cited by Schurig for Efate Island, New Hebrides."3 NATHILAU POINT (SITE 12) On April 29, after our stop at site 11, we proceeded to sites 12 and 13, which perhaps should be treated as one. Nathilau Point is a northward projecting promontory in the District of Raviravi, Province of Ra. Its tip is said to have been the residence of the god Wanggambalam- bala, whose house mound is thought to be dis- cerned in the numerous boulders, a few apparent- ly piled, which clutter the headland. The god, human in form, was the patron deity of canoe builders, who constructed canoes on the sandy flat below and behind the headland. Piles of stones there, separated by stretches of sand, are said to indicate shares of food for the car- penters, who built the canoes for the chiefs of the Kauvandra Mountains. All stones piled on the elevated tip of the headland and on the sandy flat below are said to have been piled by vu (originating gods or spirits), not by men. The informants knew of no archaeological sites on Nathilau Point. Nev.ertheless, we found potsherds in the road along the west side of the promontory; also broken shell and black soil just above the beach level. This stretch of road, perhaps 150 yds., constitutes site 12. The word Nathilau means "passage through reef," in reference to the opening through the barrier reef offshore from the headland. Navitalai of Ndrauniivi and the Mbuli Semi were our inform- ants. 13Schurig, 1930, P. 35. Joseph Waterhouse makes the following statement concerning Nathilau Point:14 The birds at Nacilau made too great a clam at night to permit [the god] Degei to sleepy with comfort. The god sent Naqai, who orde them to sleep elsewhere. This is the reaso why the birds leave that point after sunset; and return after sunrise to spend the day there. NATHILAU POINT, LOW TIDE (SITE 13) A small gully runs down to the sea at the western base of Nathilau Point. A tiny strea empties into the sea here. The sand, gravel, and mud banks exposed at low tide yielded pot- sherds. This place I have called site 13, eve though it should probably be regarded as a southerly extension of site 12. Site 13 is reached via a steep and rocky road running fr King's Road down the gully to the beach. NAMOTUTU (SITE 14) On April 30, we visited the site said to have been inhabited by two branches of a singi yavusa, which had split in ancient times. The two groups arising from the separation are called by natives Namotutu-l and Namotutu-2. People of the former now reside at Navuavua in modern Rakiraki village in the District of Rak raki; people of the latter in modern Narewa vi lage in the District of Raviravi, Province of The Native Lands Commission list gives Namotut as both the yavusa name and the yavutu name fo both these groups, not distinguishing them by and 2 as my informant did. My informant was the Mbuli Semi of the Di trict of Rakiraki. He said that Yavunuku ("sandy banks") was the yavutu for the first group; Tonga, a spot near or at site 14, for t second group. He said that "tonga" means brother in the Rakiraki dialect and refers to the identical origins of the two Namotutu yavu Site 14, now an Indian farm site, occupie by a Telugu family, was about 1/4 mi. south of King's Road on the left bank of Penang River (also called Namotutu Creek). The people move from there to Navutolevu (site 16) to escape t floods from Namotutu Creek. The southwestern part of the site rises in a knoll 20 or 30 ft. high, on which the chief Tui Namotutu lived, th commoners living below. At the northeastern edge of the site is the house mound Yavunuku, or 6 ft. high, but without stone facing. Two flat stones imbedded in the soil on the east side of this house mound were put there by the vu to mark the place if the site should be inur dated by flood waters. The stones bore no special names, nor did they have any other use The house on Yavunuku house mound was the place. to which offerings for the god Ndengei were brought by various surrounding communities. "4Waterhouse, 1866, p. 362. I II 1 I 248 GIFFORD: EXCAVATIONS IN FIJI these offerings were taken to Ndengei's j the Kauvandra Mountains. 7Te vu (ancestral god), human in form, was Naisemanivitilevu. The name refers to the of the vu in settling disputes when the tribes become embroiled. The residence of was the site of the present Rakiraki The people were still heathen when they from site 14, but became Christian while at Navutolevu (site 16). Potsherds were collected at site 14, but nformant stated that they were not manufac- at the site, as the Namotutu people ob- d all their pottery from the people of e Island, lying just north of Rakiraki. The people served as fishermen as well as rs for their mainland overlords. The coronation stone, formerly at Namotutu 14) but moved to present Rakiraki, is a ral irregular pillar 30 to 36 in. long, set ght in the ground. The new ruler sits on stone when he is "crowned." He is the re- ent of offerings for Ndengei in the Kau- a Mountains. This coronation stone was nally set up on tiny Vatuturanga Island in l Levu Bay. The stone is called Vatuniveim- When the people were driven from Viti Bay, they took the stone to Namotutu i 4). TAMBUWAI HILL (SITE 15) On May 1, Dr. H. Silvester Evans, Mr. Poni- W Talei of the Fiji Museum, Rambithi, and I .ted Tambuwai Hill near the village of Mata- ~Ievu, District of Nalawa, at the head of Viti 1 Bay, Ra Province. Mr. G. T. Barker, Cura- .of the Fiji Museum, had given us special in- ctions how to 'find the site from which, e digging a latrine hole in 1920, he ob- d a large flaked "flint" knife. We care- y searched through the pile of stones at the ine site, but saw no similar stone. Along vtrail no such stone was exposed; all stone was soft "mudstone." The latrine site was kin an area apparently once protected by a ich and earthwork; the trench had been re- bd to a shallow depression at the time of our Lt; the weather, cattle, and vegetation had ,but obliterated it. Two potsherds, said by brmants to be Malake Island ware, were the v artifacts found and they were not in the Fine rubble pile. b Mr. George T. Barker, Curator of the Fiji eum, supplied me with the following notes on 1920 find: Fist axe or coup de poing, found at ambuwai, Tova Estate, Viti Levu Bay, by . T. Barker, 1920. Site of a long-aban- oned hill fort, elevation 800 feet, over- soking Viti Levu Bay. To the westward, on his same ridge, I found the three monoliths n Thulumanu Peak and other ancient features. Description:-- Made of flint core of bheulean type, more elongated or narrower han typical fist axe of Acheulean classifi- cation, but excellently fitted to fit the hand. It appears to be a hand tool and never hafted. The blade is rounded and thin, the head pointed and thicker than the blade and was almost a working end. The blade is not twisted from the horizontal plane of the implement. The more rounded or back of the tool has some evidence of crude retouching along the edges. The axe is deeply pati- nated -- there is a central band around the axe which is probably the original surface. It has a chalky appearing surface common to many European palaeoliths. The flaked sur- face is stained with orange brown patination. It is important to note that there are no attempts at polishing. Found six feet below surface in loose mudstone (soapstone). Another axe was found by the Reverend Fr. Demarien at Namosi. The material of which these "palaeoliths" are made is apparently a buff chalcedony like a fragment (3290) found at site 21 by Dr. Lindsay Verrier who sent it to me after I returned to California. NAVUTOLEVU (SITE 16) On May 2 Rambithi, the Mbuli Semi of Raki- raki, Navitalai, and I visited Navutolevu, the site to which some of the people of Namotutu (site 14) moved and where they became Christians. It was a fortified site on flat land near pres- ent-day mangroves to the north of Rakiraki vil- lage, District of Rakiraki, Province of Ra. The ditch of the fortification is now very shallow. Some of the village was within the ditch en- closure, a semicircle with two ends at the beach; the beach is now grown up to mangroves, but was formerly clear. The site is now heavily clothed with many sorts of trees and bushes, where for- merly there were only coconuts and Tahitian chestnuts. The earthworks are said to have once been surmounted by a growing fence of wiriwiriniviti, which, when grown full size, was called mbanin- dakai ("shield for the garrison"). Two littoral species of wiriwiri trees are listed by B. E. V. Parham: Gyrocarpus jacquinii and Thespesia populnea; the latter being called wiriwiri in Nandi Province. The name wiriwiriniviti pre- sumably applies to Gyrocarpus jacquinii, and the designation ni viti, "of FiJi," evidently serves to distinguish it from the introduced Physic Nut, Jatropha curcas, also called Wiriwiri'5 and mbanindakai.16 There were 24 divisions of the village in all, as shown in the sketch (diagram 7) made by Rambithi from information given by the Mbuli Semi of Rakiraki. Each division had a ditch and hedge around it. The pottery found on the sur- face at this site is said to have come from Malake Island. It was Late period incised ware. 15Parham, 1942, p. 45. 16Ibid., p. 1. 249 ANTHROPOLOGICAL RECORDS Y?2&and ji dideo Diagram 7. Plan of Navutolevu, site 16. The area called Naiwase was the site of the Christian church at Navutolevu. Two other areas were the sites of earlier and later heathen temples to Naisemanivitilevu, the vu, human in form, who came from the Kauvandra Mountains and made his residence at the site of the present Rakiraki Hotel, which is only about a quarter of a mile from site 16. He was the son of the first, or indigenous Ndengei,17 who lived in a cave in the Kauvandra Mountains. Sarosaro was the title of the priest of the vu Naisemani- vitilevu. Naisemanivitilevu was also the-title of the chief at the successively occupied sites 14 and 16. Two chiefs were buried at site 16, the earlier at the early temple of the vu of the same name. About 25 yds. from this temple was the freshwater bathing pool of the priest Saro- saro. The last chief with the name Naisemaniviti- levu became a Christian at site 16. His grave received special treatment in time of drought, when water was poured over it to bring rain, ac- cording to his instructions. This is said to be still done. The chief was buried extended, supine, wrapped in mats; no vault was used. Onisevoru Lethia, mother's father of the informant Mbuli Semi of Rakiraki, was sent by the priest Sarosaro of site 16 to the god Ndengei's cave. This was in Christian times, and the purpose was to beg white cowries (Ovulum ovum) from Ndengei to decorate the church at Navuavua. Onisevoru used six bundles of reeds for light when he entered the cave. He told Ndengei what he had come for and Ndengei told him to look to the right where he would find two bundles of fifty cowries each. He returned to Navutolevu the same day, being led by a heathen god. When the old church was torn down at Na- vuavua, a relative of the present Na Tui Na Viti Levu appropriated the shells. Two were given to me by the present Tui. I7Gifrord, 1951, pp. 170, 17'+. EMBUTO (SITE 18) Embuto, adjacent to modern Nasingatoka vi lage and used now as its graveyard, is in Dis- trict of Nasingatoka, Province of Nandronga,. ol the west bank of the Singatoka River, perhaps two or three miles from the mouth. We visited it on May 12. Embuto was occupied for three generations, the people having come there from upriver Nambumoli near present-day Lawai villa District of Nokonoko, Province of Nandronga. From Embuto the people moved to Nasingatoka. T yavusa which occupied these three sites con- secutively is yavusa Ketenamasi. A creek on one side and a war ditch, be- ginning and ending at the creek, encircled the village site. The war ditch was 15 ft. wide a filled with water at the time of our visit. T site is located about 1/8 mi. south of the Jun tion of the roads forking to Lawangga and to Singatoka. All sherds were collected on the west side of Queen's Road, between it and the railroad tracks. Informants were the Mbuli Nasingatoka Iliesa Sanimo and Akei Neinotha, chief of yavu Nandrukumboto of Laselase village. The Mbuli: called a meeting of elders of the District of Nasingatoka for 4 p.m. About fifty men attend Five stone adze blades were presented to me. Discussion revealed no knowledge of an ancient village site in the sand hills west of the mou of the river (see sites 20 and 21). LOMOLOMO (SITE 19) On May 13 we visited Lomolomo, which is t yavutu of the yavusa Louvatu, now living in Thuvu and Yandua villages, District of Thuvu, Province of Nandronga. Lomolomo, however, is several miles up the Singatoka River from its mouth, near Lawai village in the District of Nokonoko. From Lomolomo the yavusa Louvatu moved to Nasama, where they divided into two groups, one remaining in Nasama, the other gol to Thuvu. At Thuvu they occupy the site Navak 250 GIFFORD: EXCAVATIONS IN FIJI Nambuambua Creek. Another yavusa from Lomo- was Numbu. The vu (ancestor) of the Lomolomo people "ffrom Yanutha Island, lying close to the nland near Thuvu village, separated from the nland by a fordable channel. The name of the was Wakanimolikula, who was in human form. zrteen generations ago he drifted in a bowl mete) from the Kauvandra Mountains to Kandavu -and, thence to Yanutha Island near Thuvu. tomb is near the site of old Lomolomo near ern Lawai. (Earlier the people of Lawai ed about 5 mi. farther upstream on the Singa- a River than they do at present.) His tumu- or circular conical grave mound is about 10 i high and 25 ft. in diameter, at the base of ow hill on the west side of the Singatoka er and about 1/8 mi. northwest of Lomolomo per. The lower 3 ft. of the mound are faced h stones; it has no burial vault. The mound, losed by a barbed wire fence, is weeded every r months. On the occasion of my visit kava poured at the foot of a small wi tree ndias dulcis) growing at the base of the The mound is named Nandronga. Not only was a tomb, it was also the coronation place of animolikula as Tui Nandronga. He was the y one of the line who was "crowned" at that ce and also buried there. He had two brothers accompanied him from the Kauvandra Mountains Kandavu Island. These two went to "foreign" ntries, while Wakanimolikula went to Yanutha land at Thuvu. Amani Nasinga, a man of the yavusa Louvatu, companied us to Lomolomo since that site is e yavutu of his yavusa. Lomolomo is a fairly tensive site, occupying both flat land and a a1l isolated hill rising in its center. The L1 is slightly upstream from modern Lawai vil- se and farther back from the river. Both wai and Lomolomo are on the west side of the ngatoka River. The hill is the place where kanimolikula was to have been crowned as Tui dronga, but because of the noise in these roundings, he was crowned at the site of his rave . The highway along the west side of the rver passes through Lomolomo site, and quanti- ~es of river-mussel shells can be seen in the ad cuts. Traces of moats which once encircled rts of the village were visible. A few uni- alve marine shells appeared in the cultural de- sit, but apparently river mussels were the hief molluscan food, as they are today in Lawai. be informants said that river mussels were set redge in hot ashes or coals and twigs were aid over them and ignited to cook them. Pre- 4sely this method is employed in California. Fortunately, the vegetation on the hill at molomo site had recently been burned over, so $was easy to see the potsherds, which were ery abundant, as was also shell. The pottery s said to have been of local manufacture. aces of house platforms were visible, as indi- ited by the stones with which they were faced. ~irteen generations ago, tradition says, Waka- imolikula founded the village, yet the deposits, 251 from the surface at least, looked relatively shallow. At Lawai I saw a woman mixing clay and sand for pottery with her hands. Then she threw lumps of the mixture down hard on a mat. Next, standing, she kneaded it with her right heel, stamping hard.'8 A round-bottomed pottery kava bowl, from which kava was served at Lawai before we went to Lomolomo, was said to be a local product.19 Although only three years old, the bowl had a pronounced kava coating on the inside, a decorative effect which is appreciated by the owners. EASTERN SINGATOKA SAND DUNES (SITE 20) West of the mouth of the Singatoka River, District of Singatoka, Province of Nandronga, the shore for about two miles has a breaking surf, as there is no protecting coral reef. My opinion is that the absence of coral is due to the current of fresh water from the river mouth. High sand dunes fringe the shore from the river mouth westward for several miles. The more easterly of these tend to travel inland, for they have very scanty vegetation and the south- east wind blows strongly. Plate 14, e, f shows two excellent views of the shore side and the inland side of the dunes near Kulukulu village, by Mr. H. H. Storck of Singatoka and reproduced by his kind permission. Even the force of the prevailing southeast wind is indicated by the foliage of the pandanus tree in one picture. Site 20 comprises the seaward side of the dunes from the vicinity of Kulukulu village eastward toward the river mouth. Potsherds, human bone fragments, and the mandible of a horse were found in scattered spots over a stretch of a quarter to half a mile. The bones were very friable and bleached white. The pot- sherds, badly sand-blasted by the wind-driven sand, were said to be all of Nandronga type, in other words of local origin. An elaborate piece is shown in plate 19, g. My informants and com- panions were Rambithi, Nemani Tumbou of Volivoli village, and the Mbuli of Singatoka. We visited the dunes on the afternoon of May 13, when a strong southeastern wind was blowing. NANGGARAI, WESTERN SINGATOKA SAND DUNES (SITE 21) On May 15, accompanied by the same compan- ions as at site 20, I visited a site called Nanggarai, in the sand dunes somewhat to the southwest of Volivoli village, District of Singa- toka, Province of Nandronga. We were taken by taxi to a point on the railroad where there is a siding for loading freight cars with sand from the dunes. This was 3 1/2 mi. by road from Singatoka Hotel. Here the dunes are dissected 18Cf. K. Roth, 1935, pl. 11, fig. 2. 19cf. p1. 17,c, d, which shows a kava bowl from this region. I I ANTHROPOLOGICAL RECORDS by broad swales, with a southeast-northwest trend, from which the wind has scoured out the sand down to the underlying earth in some places. This site is in the dunes fronting the reefless shore Just east of the reef visible in the back- ground in plate 14, e. In two swales, many river mussels and pot- sherds were seen, but no bones were visible. In the easterly swale there were clusters of sherds lying on the brown soil from which the sand had been blown. Their position suggested house sites or spots where pottery had been fired. Possibly these were ancient kilns, so placed on account of the strong winds from the southeast. Although the informants gave the name Nang- garai for this place, they did not know whether it was a village site or merely a temporary camp site, nor did they have any story to explain who the former occupants were. Personally, I think it a temporary camp site, where people brought river mussels to eat, and perhaps fired pottery. Water could have been brought from the inland side of the dunes, which are not very wide. Again, the potsherds were declared to be of lo- cal type, although they were badly sand-blasted. Many of the potsherds are quite thin, thus re- sembling certain vessels of the region, e.g., the modern cooking-pot (3166) shown in plate 19, a. Plates 19 and 20 show sherds collected at this site by Dr. Lindsay Verrier and kindly for- warded to me in California. NALOVO (SITE 22) On May 15 we visited the deserted site of Nalovo, District of Thuvu, Province of Nandron- ga, which was occupied until Christian times by the people whose descendants now live in Thuvu village just east of the creek which separates the two sites. The site of Nalovo fronted the bight in which lies Yanutha Island. Today it is separated from the shore by Queen's Road and the railroad tracks, which both skirt the shoreline. The creek, unnamed on the map, enters t-he sea opposite the east end of Yanutha Island, its mouth being crossed by the railroad and highway bridges. The yavusa formerly inhabiting Nalovo and now Thuvu is Louvatu. The yavutu is Lomo- lomo (site 19). Nalovo is in jungle now at the base of a hill, which rises on its northwest side. The temple site was near the former shore edge and was unimpressively small. The "god" was Waka- nimolikula, whose tomb is at site 19. The soil was full of shell from the adjacent sea. Al- though potsherds were found, the informants said that pottery was not made at Nalovo because there was no potter's clay in the vicinity. Pottery was imported from Singatoka. A well about 8 ft. deep at present, with a stone coping a few inches high, is said to be used for bathing at present, but anciently it supplied drinking water. Although the creek is near by, its banks are fringed with mangroves owing to the seawater which enters it at high tide, so presumably it was not a suitable source of drinking water. There were no defensive moats at Nalovc Informants explained their absence by the so ment that Nalovo was always strongly defende by the best fighting men, since it was Tui dronga's "capital." I KOROINGGARA (SITE 23) We visited on May 16 a site on top of a flat-topped hill (pl. 15, e) about a third o mile west of Volivoli village and, like it, the north side of Queen's Road and the railr tracks. The site includes a small cave whic slopes down into the hill from the more or 1 flat summit, hence the name, which means "c village." The mouth of the cave is in a de sion behind the escarpment forming the south east faces of the hill. Across from the sou face are the sand dunes described in connect with sites 20 and 21. The dunes are separat from the escarpment by a narrow defile (p1. f), but nevertheless have encroached on the southwestern portion of the escarpment and ascent easy. The cave within the village precincts id said to have been used for refuge only. Pot sherds and shells, but no human remains, wer found within the cave, which has an extremel uneven floor because of boulders. Similar sherds and shells were seen about the villag site. The downward slope of the cave floor about 30 degrees, and the roof rapidly becom9 too low for one to stand upright. The altitO of the cave entrance is 350 ft. above sea lev that of the village area outside 375 ft. It is said that pottery was not made lo1 ly, but came from Singatoka. The Yandua vil lagers, who moved from Koroinggara, do not m pottery. Just east of Koroinggara is Volivo Where pottery is made, but its inhabitants m there from near-by Naruku. Some Yandua peop are of yavusa Louvatu, with their yavutu at Lomolomo (site 19), like the people of Thuvu- Others are of yavusa Koronikula. The Volivo people belong to two yavusa, Nalimoulevu and Ekumbu. Yandua village is in District of Th Volivoli village in District of Nasingatoka, Province of Nandronga. YARO (SITE 24) Yaro is not a koro makawa ("old village but merely a hole, 8 or 10 ft. in diameter, the coral limestone rock of southwestern Vit Levu, open to the sky. Two sloping bottom s faces of earth extended downward under a low-7 roof. The natives explained it as a cooking- place. From this "pit" a few sherds were ob tained, but no human bones. The location of Yaro is 1/8 mi. inland f the Queen's Road, Just west of the new vill of Sila (not on the map), which, in turn is west of Naevuevu. Sila is on the seashore beX tween the highway and the sea, in the Districi of Thuvu, Province of Nandronga. I I -i 7. . iI I,j IA ...A ?i A I r w i 252 GIFFORDs EXCAVATIONS IN FIJI NATAUTA (SITE 25) Natauta is not a koro makawa ("old village") Ier, but is a limestone burial cavern about | yds. north of a point on Queen's Road slight- mest of the new village of Sila, District of vu, Province of Nandronga. It is a hummock coral limestone containing a small cavern P. the entrance on the north side. This en- oce is Just large enough to squeeze through, I the interior of the cave is so low that a cannot stand in it comfortably. The irregu- t natural chamber is about 20 ft. in one di- er and 15 ft. in the other. Bones of at least five persons were scat- d about the floor. A sixth skull had been ed by a Suva dentist (one of the sons of present Tui Nandronga) and taken to New Zea- The local people spoke of Natauta as a n site, not a village site, and attributed .potsherds found inside the cave as well as op of the hummock to the people who came e to plant. The name Natauta applied to the 1 cavern as well as to the surrounding land. aneroid barometer gave a reading of 50 ft. e cave. Four fragmentary skulls and eight femora collected. These were the best available, were in poor conditiQn. Apparently corpses been laid on the floor of the cave without ent. Informants explained the presence of bones as due to two sources: (1) strangled ns; (2) persons whose souls had been stolen tches and their bodies deposited there. MALOLO PLANTATION (SITE 28) On May 23 the Colonial Sugar Refining Com- pany courteously supplied us with a line car (track automobile) and an Indian driver. We traveled from Lautoka to Singatoka and four miles up the east side of the Singatoka River in this conveyance. No cuts exposing archaeologi- cal material were seen, but four ploughed fields along the right of way contained archaeological sites (28-31). As we had no Fijian informants, we learned nothing of the former occupants. Our first site was at the Company's Malolo Cane Plantation on the right bank of a stream that the driver called the "Malolo River," though the name does not appear on the map. This stream was south of the Nandi River. Here a ploughed field on the seaward side of the rail- road tracks yielded a few tiny sherds and shells. The site is in the District of Sikituru, Province of Nandi. YAKO PLANTATION (SITE 29) Like site 28, this site was.in a ploughed field beside the mainline railroad tracks in the Yako Cane Plantation of the Colonial Sugar Re- fining Company. Its black soil yielded sherds and shells. It is on the shore side of a man- grove-filled lagoon, on the inland side of which is the modern village of Yako, a short distance southwest of the border of Nandi Province and in the northern tip of Nandronga Province. The site is in the District of Momi. OLD NASOLO (SITE 27) On May 22 Rambithi and I visited Nasolo and, ld by the turanga-ni-koro, Marika Natoto II, a second man named Semi Nale (born at the ent Nasolo), we proceeded to old Nasolo. iste is about 250 ft. in elevation, and is ed after a walk of a mile and a half from lt Nasolo on the road leading inland up the Valley and then over a trail to the left. lite was on a hill slope on the left side of tt tributary of the Mba River below the con- ate of four creeks which form this tributary. rour creeks are Varathiva, Korotambua, Nan- and Numbuniveto, but no name appears on pap for the short stretch between their ion and the Mba River, past our site 27. The present Nasolo has been occupied fifty Before that the people lived at Nailanga, that at Old Nasolo, and before that at bu about eight miles upstream on the bank of the Mba River. This earliest site planted to sugar cane. either informant knew whether pottery we was formerly made locally. The decorated found had incised designs like modern Ra ce pottery. e Fijian Office list gives Namathuku as e of the yavusa inhabiting Nasolo, and Kavuwa as the yavutu. Our informants did ,w the location of the yavutu. Both old w Nasolo are in the District of Nailanga, ce of Mba. LOMAWAI PLANTATION (SITE 30) This is another ploughed-field site, dif- fering from others in having light brown soil without trace of charcoal. Most midden sites have black or gray soil showing evidence of fire. The site had an abundance of potsherds, but shell was relatively scarce. The site is on the Lomawai Cane Plantation of the Colonial Sugar Refining Company, on the right bank of the Kum- buna River downstream from Kumbuna village, and is cut by the railroad tracks. It is in the District of Wai, Province of Nandronga. MALOMALO (SITE 31) This site is in a cane field just east of Malomalo village and on the inland side of the railroad tracks, which here lie close to the sea. I have designated it by the name of the neigh- boring village, which is in the District of Malomalo, Province of Nandronga. The deposit was a typical shell midden. MBULU (SITE 32) This site is on the west side of Mbulu Creek, near its mouth, downstream from the Queen's Road bridge and cut by the road. We 253 ANTHROPOLOGICAL RECORDS visited the site on May 26. It was formerly oc- cupied by the inhabitants of modern Malevu vil- lage, District of Thonua, Province of Tholo West. Our informant was Ilisavani Remoa of Malevu vil- lage, who said that the people of Malevu, Nawa- mangi, and Naroro villages, District of Thonua, belonged to the yavusa Noi Tonga and traced their descent from one Finau Maile Latumai, a political refugee from Tongalevu (Tongatapu). Finau Maile is said to have lived at Yalaki (Pelehake, probably), about eleven miles east of modern Nukualofa. The present chief of the ya- vusa Noi Tonga is Elbert Nanggasima of Nawamangi village, District of Thonua. The people today are said to represent the eighth generation, from Finau Maile, their ancestor. This would seem to indicate that his flight from Tonga took place in the troublous times about the end of the eighteenth century. The Malevu people lived once at Tavuni near Naroro; there they became Christians. From there they moved to Mbulu (site 32), then to Korotongo. About 1937 they moved to Malevu be- cause of a disagreement with the Korotongo people. Potsherds found at Mbulu were said to have been made in Singatoka, not locally. On May 27 we visited Nawamangi, interviewed Elbert Nanggasima, and with him visited the koro makawa Mahevu, about half a mile upstream from Nawamangi and on the east side of the Singatoka River. At Mahevu we found no sherds because of the dense covering of grass and the lateness of the hour; hence the site has not been given a number. We did not visit either Tavuni or Serua hill, where Finau Maile Latumai, the original Tongan settler, is buried. The yavusa Noi Tonga lived at Mahevu in pre-Christian times, thence moved to Tavuni, where they became Christians. His Fijian de- scendants assert that Finau Maile, whose real name was said to be Maile Latumai, was the Tui Pelehake, but Ratu Sir Lala Sukuna informs me that Queen Charlotte of Tonga says that no Tui Pelehake fled Tonga. It is likely therefore that his descendants' claim of his high rank is fictitious. As the Tongan fugitive sailed along the south coast of Viti Levu, he first stopped at Serua Island, District of Serua, Province of Serua, for firewood. Latumai became exasperated at the slowness of the men whom he sent ashore, so he sailed on without them. He landed at Vunindilokula, a particular spot at Korotongo, District of Thonua, Province of Tholo West. He had his first house built a little way inland at a place called Waya, while he was looking about the country. The house was not satisfactory, so he chased the carpenters away upon his return. He married two Fijian women, the first wife be- ing from Narata, a present-day village up the Singatoka Valley in the District of Thonua, on the east side of the river. Later he moved in- land to a little hill, which he called Serua after the island where he had stopped. This hill is about a mile south of Nawamangi and about 300 yds. east of the road and toward a range of hills. Latumai's tomb is on the hill. It is of soil only, with no stones; it is still cared for and flowers are planted there. Latumai's death his sons founded Korokune. At the old site Mahevu, I was shown t mound reputed to be that occupied by Latuma house when he resided there. It, like oth was round in outline. Informants said this; the old style. At Tavuni they said there both round and rectangular house mounds, t latter being the modern style, which, as f I saw, has replaced the earlier type in the' region. The letter h in Mahevu, represents the local dialectic sound replacing the letter,' the official Mbauan dialect. The same soun shift is exemplified in the name of site 36, Namahara. SOVI (SITE 33) This site, visited on May 27, is on the right bank of the Sovi River between the Qu Road and the ocean shore. The people were S heathen when living here. They had occupied spries of places after the traditional depa of their ancestors from the Kauvandra Mounta First was Noikoro, a district far up the Si toka River in the northeast part of the Prov of Tholo West. Then they moved to Yalava ne Nawamangi, then to Korotongo, then to Sovi. From Sovi they moved to old Vatukarasa (site where they became Christian. Finally they m to modern Vatukarasa, where their descendant now live. The modern people speak the Nand dialect like others in the region, substitut4 h for s. Few potsherds were found at Sovi, but t may be partly because of flooding by the Sov River, of which there was ample evidence. O informant, Vamosi Misiwani of Vatukarasa vil who was fifty-seven years old, did not know whether pottery was ever made at Sovi. Hous mounds were visible. A ditch at the site wa said to be for drainage, not for protection against attack. No actual rubbish heap was found, but there were shell and pottery here: and there. The pottery was said to be from Singatoka. OLD VATUKARASA (SITE 34) This site, visited on May 27, was the pl to which the Sovi (site 33) people moved. L Sovi it is in the District of Thonua, Provind of Tholo West, on the left side of Tamanua C at its mouth. This creek has a lagoon and n reef immediately offshore; hence the breakers come directly on to the beach at old Vatukara Informant Vamosi Misiwani showed us the; sites of the old church and the minister's ho The informant was born in old Vatukarasa, and was seven years old when the village moved to modern Vatukarasa. He said that the yavutu wi Louvatu, near Nawamangi. The Government list gives Nambili and Namboka as the two yavusa o4 cupying modern Vatukarasa. Potsherds were said to be of Singatoka 254 GIFFORD: EXCAVATIONS IN FIJI since pottery was not made locally. The informant spoke of Sovi as being the akawa ('old village") of the Vatukarasa e. OLD TANGANGGE (SITE 35) This village, visited on May 28, is in the ict of Korolevuiwai, Province of Tholo West. ravusa inhabiting the village is named tukia; its yavutu bears the same name and Ocated in the District of Koroinasau, Prov- of Tholo West. Legend derives the Ndavu- a from the Kauvandra Mountains originally. Tangangge lies east of the modern village extends to the right bank of Tangangge Creek. ld temple site, as well as sites of the Ch and of the minister's house, were shown n old Tangangge. Our informant was Ratu Natomba Alipate awalu, who was a captain in the First Bat- on of the Fijian Infantry regiment which saw on in the Solomon Islands in World War II. aid that pottery was made in Tangangge, so ay regard the sherds locally manufactured er than imported from Singatoka. At Namanda, village west of Tangangge, no pottery was ,,all pottery being imported from Singatoka. r-mussel shells collected at site 35 were to be from the Singatoka River. An old woman in Votua village later told us pottery-making in Tangangge was explained the marriage of a Singatoka woman into the lage. NAMAHARA (SITE 36) Namahara (Namasara in Mbauan dialect), 3/8 E west of Korolevu Bay Rest House, between * Rest House and the village of Votua, was erly occupied by the people of modern Votua ~lage, District of Korolevuiwai, Province of lo West. The Queen's Road cuts through the site. The inhabitants became Christians lie living at Namahara; the site of the first rch was pointed out on the beach side of the hway. Before the move to Namahara the people red at the village of Korolevuiwai (site 37). p people of Votua still bury their dead in the pinity of the site of the old -church at Nama- ra. Pottery found at Namahara was said to be Singatoka origin. A man at Votua, sixty years of age, said at Votua was already occupied when he was ung. Today there are five yavusa represented ,men at Votua: Noi Thambe, Noi Tumbai, Noi bu, Ndavutukia, and Wai. The same five yavusa re said to have been represented in ancient rolevuiwai and in Namahara. All were supposed have originated in the Kauvandra Mountains. ewomen of the village are from various yavusa. emarried-in woman is from Thaukandrove yavusa Vanua Levu Island. KOROLEVUIWAI (SITE 37) Korolevuiwai, visited on May 29, lies 1/8 mi. west of Korolevu Bay Rest House and on the inland side of Queen's Road opposite Mr. Free- min's house, which is on the shore side. The site is in the District of Korolevuiwai, Prov- ince of Tholo West. Under site 36 I have al- ready named the yavusa which formerly lived at site 37. The house mounds were poorly preserved. A waterfilled moat or war ditch encircled the site, but there was midden material outside it as well as inside. The shell midden looked to be about 3 ft. deep at the ditch edge. The pottery found abundantly here was said to have been locally made. After the people moved to Namahara (site 36) they abandoned pot- tery-making. NANGGAU (SITE 38) This site was visited on May 30. We were taken there by two brothers from Komave, Dis- trict of Komave, Province of Tholo West. These brothers were Juliano W. and Leone W. Mbatiko- likoli, descendants of the last chief of Nanggau, Mbatikolikoli Nambou, who was their great-great- grandfather. He became a Roman Catholic, along with his subjects, at Nanggau, being converted by Father Deniau. (House mounds of three Catho- lic churches are to be seen at Komave.) The site is on both sides of Nanggau Creek, a half-mile west of Komave, and on both sides of the Queen's Road. Nanggau was the former resi- dence of the Komave people, their koro makawa or old village. Before Nanggau they lived at Na- vatusa, about four miles inland, on a ridge, now surrounded by jungle. Still earlier they lived in the Noikoro District, Province of Tholo West, where their first village was established after they left the legendary home in the Kauvandra Mountains. The informants said the two yavusa in Komave village are Vusu and Nasikawa, though the Government list gives only the former. Potsherds collected at Nanggau were re- garded by informants as deriving from Noikoro, rather than Singatoka. In Noikoro suitable clay was available at Kuronivusu, where clay was ob- tained to make pots for the yavusa Vusu. A round-bottom pottery kava bowl in use at Komave during our conference came from Singatoka, whence pottery is regularly imported nowadays. Some house platforms at Nanggau were round, others oblong in shape. The house of the chief Mbatikolikoli was round, with diameters of 46 and 41 ft. measured from surviving house posts. The house platform was slightly larger and had rounded "corners." Small coral stones marked the edge of the platform, which was evidently low. Black rocks forming the surface of an earth oven were seen. The name of the chief's house was Narongi ("very good"). The latter part of the owner's 255 ANTHROPOLOGICAL RECORDS name, Nambou, was said to mean "to carry heavy things," in reference to the heavy house posts in his house. The house was in the area sur- rounded by a moat or war ditch; the chief's ten wives lived within this enclosed area. The war ditch was broad and shallow when I saw it, but formerly was more of an obstacle to negotiate; it had three crossings by which people had ac- cess to the village entrances. Nanggau means "to wade or swim," since strangers ran the risk of being killed if they passed through the village, so they waded i swam in the sea. The chief's name was said analyze as follows: Mbati, "chief"; kolikw "striped or marked." It was thought that name might refer to the chief's varied dieb which included human flesh. Mbatikolikoli'l eldest son was Ratu Luke Letewale, who weni Levuka on the occasion of the cession of P to Britain, but arrived too late to sign. I became the first Mbuli of Komave District. 256 J k LITERATURE CITED 1, Stanley C. 1933. Jungle Fowls from Pacific Islands. Bernice P. Bishop Museum, Bull. 108. pley, H. G. 1928. Pacific Island Records: Fish Hooks. Seeley, Service and Co., Ltd., London. them Jutting, W. S. S. van 1940. Molluskschalen von Prahistorischen Mahlzeitresten aus der Hohle Dudu- munir in West Neu Guinea. Nova Guinea, n.s. 4:11-30. er, H. Otley 1948. Philippine and East Asian Archae- ology, and its Relation to the Origin of the Pacific Islands Popu- lation. Nat. Res. Council of the Philippines, Bull. No. 29. 1949. Outline Review of Philippine Archae- ology by Islands and Provinces. Philippine Journal of Science 77 (nos. 3 and 4, July-August, 1947): 205-3741 Nster, A. B. 1922. The Hill Tribes of Fiji. J. B. Lip- pincott Co., Philadelphia. tish Museum 1910. Handbook to the Ethnographical Col- lections. , Peter H., Kenneth P. Emory, H. D. Skinner, and John F. G. Stokes 1930. Terminology for Ground Stone Cutting- Implements in Polynesia. Jour. Polynesian Soc., 39: 174-180. 11, A. 1941. A New Fijian Dictionary. Compiled for the Government of Fiji. Austra- lasian Medical Publishing Company Limited, Sydney. .-and R. H. Lester 1941. Local Divisions and Movements in Fiji. Oceania, 11: 313-341; 12: 21-48. , D. J. 1932. ch, 0. 1903. Archaeological Finds on Lamma Island near Hong Kong. Pt. 1. Hong Kong Naturalist, 3:226-246. Papua-Topferei. Aus dem Wiegenalter der Keramik, Globus, 84:329-334. Gifford, E. 1949a. 1949b . 1951. W. Diet and the Age of Californian Shellmounds. American Antiquity, 14: 223-224. Excavations in Viti Levu. Jour. Polynesian Soc.; 58:83-90. Mythology, Legends, and Archaeology in Fiji. Univ. Calif. Publ. Sem., Philol., 11:167-177. Gordon-Cumming, C. F. 1883. At Home in Fiji. (2d ed., complete in 1 vol.) A. C. Armstrong and Son, New York. Janse, Olov 19417. Ladd, Harry 1934. R. T. Archaeological Research in Indo- China. Vol. 1. (Harvard-Yenching Institute Monograph Series, Vol. 7.) Harvard University Press. S. Geology of Vitilevu, Fiji. Bernice P. Bishop Museum, Bull. 119. Lenormand, M. H. 1948. Decouverte d'un Gisement de Poteries Indigenes a l'Ile des Pins. Etudes Melanesiennes, n.s. (1st year), no. 3, pp. 54-58. Noumea, New Caledonia. Lester, R. 1941. H. Kava Drinking in Vitilevu, Fiji. Oceania, 12:97-121, 226-254. MacLachlan, R. R. C. 1938. Native Pottery from Central and Southern Melanesia and Western Polynesia. Jour. Polynesian Soc., 47:64-89. 1939. Native Pottery of the New Hebrides. Jour. Polynesian Soc., 48:32-55. 1940. The Native Pottery of the Fiji Islands. Jour. Polynesian Soc., 49:243-271. Mansuy, H. 1902. Stations Prehistoriques de Somron- seng et de Longprao (Cambodge). F.-H. Schneider, Imprimeur-Editeur, Hanoi. McKern, W. C. 1929. Archaeology of Tonga. Bernice P. Bishop Museum, Bull. 60. Nieuwenhuis, A. W. 1907. Quer durch Borneo. Zweiter Teil. E. J. Brill, Leiden. [257] ANTHROPOLOGICAL RECORDS Parham, B. E. V. 1942. Fijian Plant Names. Department of Agriculture, Fiji. By Authority: F. W. Smith, Government Printer, Suva. Rogers, Julia Ellen 1942. The Shell Book. Doubleday, Doran and Co., Inc., New York. Roth, Henry Ling 1896. The Natives of Sarawak and British North Borneo. Vol. 1. Truslove and Hanson, London. Roth, Kingsley 1935. Pottery Making in Fiji. Jour. Royal Anthro. Inst., 65:217-233. Sarasin, Fritz 1929. Ethnologie der Neu-Caledonier und Loyalty-Insulaner. C. W. Kreidel's Verlag, Munich. Schurig, Margarete 1930. Die SUdseet6pferei. Druckerei der Werkgemeinschaft, Leipzig. Seemann, Berthold 1862. Viti: an Account of a Government Mission to the Vitian or Fijian Islands in the Years 1860-61. Macmillan and Co., Cambridge. Smythe, Mrs. S. M. B. 1864. Ten Months in the Fiji Islands. John Henry and James Parker, Oxford and London. Stein Callenfels, P. V., and I. H. N. Evans', 1928. Report on Cave Excavations in Pe Jour. Federated Malay States Mus 12:145-159. Surridge, Margaret N. 1944. Decoration of Fiji Water-jars. Jour. Polynesian Soc., 53:17-36. Thompson, Laura 1938a. Adzes from the Lau Islands, Fiji Jour. Polynesian Soc., 47:97-lo8. 1938b. The Pottery of the Lau Islands, Fiji. Jour. Polynesian Soc. , 471 109-113. Thomson, Basil 1908. The Fijians. William Heinemann, London. Van der Hoop, A. N. J. Th. a Th. 1941. Catalogus der Praehistorische Ver zameling. Koninklijk Bataviaasch` Genootschap van Kunsten en Weten-. schappen. Waterhouse, Joseph 1866. The King and People of Fiji. Wesleyan Conference Office, Londoi Weckler, Jr., J. E. 1943. Polynesians, Explorers of the Pacific. Smithsonian Inst., War Background Studies No. 6. Wheeler, G. 1928. C. On Some Pottery from Alu, Bougain4 ville Strait, Solomon Islands. Man, 28:38-40. Spencer, Dorothy M. 1941. Disease, Religion, and Society in the Fiji Island. Monographs of the Amer. Ethnol. Soc., No. 2. Steadman, W. R. 1941. Edible Shell Fish in the Fiji Islands. The South Australian Naturalist, vol. 21, no. 2, pp. 8 and 9 (December 22, 1941). Adelaide. Wilkes, Charles 1845. Narrative of the United States Ex ploring Expedition during the Yeav 1838, 1839, 1840, 1841, 1842. Vol. 3. Lea and Blanchard, Philadelphia. Williams, Thomas, and James Calvert 1859. FiJi and the Fijians. D. Appleton and Co., New York. i F 4. A I "I ? I I 4 1? m p 258 EXPLANATION OF FIGURES 1-4 -Fig. 1. Shell and coral artifacts. a. Modern breadfruit peeler, made of shell of Fasciolaria entosa Lamarck; 5005; length, 153 mm.; Singatoka, Nandronga Province, Viti Levu. b. Fragment ell bracelet; 6422; width, 9 mm.; site 26, location A, rectangle 01-2 P1-2; depth 24-30 in. ell fishhook shank; 5936; length, 66 mm.; site 17, location A, rectangle T4-5 U4-5 (rock er); depth 24-30 in. d. Fragment of shell bracelet of Trochus niloticus Linnaeus; 5939; , 7 mm.; site 17, location A, test pit B3-4 C3-4; depth 0-33 in. e. Shell bracelet, probably arl oyster; 6423; outer diameter, 93 mm.; site 26, location A, rectangle 0-01 P-Pl; depth t 25 in. f. Adze blade of-Tridacna shell; 5937; width, 41 mm.; site 17, location B, rectangle .5 Cl-1.5; depth 66-72 in. g. Pot scribe or marker (?) of Cardium shell; 5553; length 24 mm.; 17, location B, rectangle Bl-2 Cl-2; depth 18-24 in. h. Shell "stick"; 5940; length 34 mm.; 17, location A, test pit B3-4 C3-4; depth 0-33 in. i. Fragment of shell ring of Conus,pos- y literatus,Linnaeus; 5935; outer diameter, 65 mm.; site 17, location B, rectangle Al-1.5 .5; depth 84-90 in. J. Coral "tapa marker"; 5934; length, 63 mm.; site 17, location B, rec- le Al-1.5 Cl-1.5; depth 66-72 in. k. Unfinished adze (?) blade of Tridacna shell; 5938; length, mm.; site 17, location B, rectangle Bl-2 Cl-2; depth 12-18 in. Fig. 2. Fijian and Tongan adzes. a. Black volcanic stone; 5020; length, 66 mm.; from hotel- en, Navua, Serua Province, Viti Levu; collector, Mr. W. J. Belcher. b. Dense basaltic lava (?); length, 90 mm.; from native living halfway between Houma and Haamonga-a-Maui, Tongatapu and, Tonga; gift of collectors, Mr. and Mrs. John S. Hampel. c. Gray volcanic stone; 5016; th, 73 mm.; Taveuni Island, Fiji; collector, Mr. W. J. Belcher. d. Coarsely porphyritic xene andesite with abundant phenocrysts of feldspar; 5023; length, 95 mm.; Korovisilou, old e near Laselase, Nandronga Province, Viti Levu; gift of Mr. Akei Neinotha of Nasingatoka village. Dense basalt; 5027; length, 92 mm.; Nasama, Nandronga Province, Viti Levu; gift of Mr. Aseri awave of Nasingatoka village. f. Dense basaltic lava (?); 3149; length, 130 mm.; Ovalau Island, i; gift of collector, Rev. Desmond P. Scanlon, June, 1947. g. Black volcanic stone; 5025; gth, 97 mm.; Nasama, Nandronga Province, Viti Levu Island; gift of Mr. Aseri Kunawave of Nasinga- a village. h. Porphyritic andesite with abundant phenocrysts of feldspar; 5028; length, 134 mm.; gi, Nandronga Province, Viti Levu; gift of Mr. Joseva Tonawai of Lawai village. i. Porphyritic a-andesite, containing phenocrysts of feldspar largely altered to epidote; 5024; length, 178 mm.; m Yavulo, adjoining Nasingatoka village, Nandronga Province, Viti Levu Island; gift of Mr. Semi amUsu of Nasingatoka village. Fig. 3. Fijian and New Caledonian adzes; chisel. a-k, m, adzes. a. Porphyritic (?) pyrozene lesite; 5018; length 81 mm.; found at quarry, Lami, Rewa Province, Viti Levu; collector, Mr. W. "Belcher. b. Andesitic crystal tuff with abundant broken chips of feldspar in a dense matrix; 31; length, 83 mm.; surface of site 38, Nanggau, Tholo West Province, Viti Levu. c. Volcanic ne; 3153; length, 85 mm.; Tamavua, Naitasiri Province, Viti Levu; gift of Mr. W. T. Coster- tards. d. Dense (?) basalt; 5029; length, 103 mm.; from Old Tangangge, Tholo West Province, Viti ru; gift of collector, Captain R. N. A. Vitukawalu. e. Dark gray volcanic stone; 5021; length, Emm.; from ploughed field just west of rice mill at Tavua, Tholo North Province, Viti Levu. Pur- IBed from an Indian. f. Dark gray volcanic stone; 6380; length, 46 mm.; site 17, location A, jtangle B3-4 D3-4; depth 72-78 in. g. Altered porphyritic pyroxene andesite with some hornblende; 26; length, 105 mm.; Nasama, Nandronga Province, Viti Levu; gift of Mr. Aseri Kunawave of Nasinga- a village. h. Black volcanic stone; 5017; length, 72 mm.; Bay of Islands, Rewa Province, Viti M; collector, Mr. W. J. Belcher. i. Dark gray volcanic stone; 5022; length, 48 mm.; from 3ughed field just west of rice mill at Tavua, Tholo North Province, Viti Levu; purchased from an hian. J. Gray volcanic stone; 11613; length, 29 mm.; site 26, location A, rectangle N-Nl 0-01; ?th 24-30 in. k. Nephrite serpentine and talc; 2845; length, 174 mm.; Houailou, New Caledonia; 't of collectors, Mr. and Mrs. John S. Hampel. 1. Chisel of gray volcanic stone; 5019; length, ,m.; Korovou, Taveuni Island; collector W. J. Belcher. m. Coarse augite diabase or gabbro; Af; length, 53 mm.; from sugarcane field to west of Vunda crag, site 26, Lautoka Province, Viti [259] 260 ANTHROPOLOGICAL RECORDS Fig. 4. Fijian adzes. a. Gray volcanic stone; 5931; length, 92 mm.; site 17, location A, rectangle H3-4 I3-4; depth 36-42 in. b. Black volcanic stone; 6403; length, 75 mm.; from-sugar field to west of Vunda crag, site 26, Lautoka Province, Viti Levu. c. Gray volcanic stone; 593 length, 76 mm.; site 17, location A, rectangle S4-5 T4-5; depth 12-18 in. d. Dense basalt (?) length, 138 mm.; Waila Hill, on Witherow property, upslope from Rewa Dairy about 50-60 ft. abov highway on west side of Rewa River between highway bridge and Waimanu River, Naitasiri Province Viti Levu; gift of collector, Mr. E. H. McIlwain. e. Dense basalt (?); 3151; length, 140 mm.; from same place as d(3150); gift of collector, Mr. E. H. McIlwain. f. Dense andesite or basalt: 6406; length, 66 mm.; site 26, location A, rectangle N2-3 02-3; depth about 12 in. (in house mo L. Dense andesite or basalt; 5930 length, 71 mm.; site 17, location A, rectangle D3-4 E3-4; dep 84-90 in. IL. Porphyritic (?) pyroxene andesite; 5015; length, 84 mm.; surface at Vatulili, a h opposite and south of Navatu (site .17), Ra Province, Viti Levu. i. Dense basalt; 5915; length, 106 mm. from surface of Navatu (site 17) at 550 ft. elevation on Niumarawa's crag, Ra Province, Viti Levu. J. Dense ultrabasic rock (possibly peridotite) carrying sporadic grams of pyrite; 3 length, 179 mm.; Ovalau Island. Like fig. 3, ?f (3149) it was found in June, 1947, on road to Catholic headquarters; gift of collector, Rev. Desmond P. Scanlon. k. Dense basalt (?); 3152; length, 157 mm.; from same place as d (3150); gift of collector, Mr. E. H. McIlwain. GIFFORD: EXCAVATIONS IN FIJI a ',Il If I'l d '1d O I l AI p 1p I 'liii, I''1 II Ij pp I I j Fig. 1. Shell and coral artifacts. c (774 h i I p?' k 261 b . . N . k?- I. . -b 262 X li ! j%4jl65; a ,I~ ~ ~ ~ ~ ~ ~ ~ ~~I 11 I~~ ANTHROPOLOGICAL RECORDS i ~~b3 I 1, /I ,,,/ , i q 1, ?? il ;. 11 Ilil,'I. I 11I 'i i.1 - I18} -0~ ~~~~~ !,I h Fig. 2. Fijian and Tongan adzes. I. . I : ( p c :Ii e 9 I .I.I I O iij 'l, il; I I I'! 1? 1.1 IIII jlj? I., I I I Ii!I V i GIFFORDs EXCAVATIONS IN FIJI a d j I~~~~~~~~~~~~~~~~~~~~~~~~~~~~I K' Fig. 3. Fijian and New Caledonian adzes; chisel. I I: .? . 1. 1. I I II a~F I' -c-vXD / I 1;?. 1. J:; ? , I I 263 II ANTHROPOLOGICAL REC ORDS a b e d i . I I A h i ) I- ~k Fig. 4. FiJian adzes. I~ 9 Il N l I 264 I ?,i- I I I ;; .: A ?!? .?- c {- I0 V ,I 'k ,J?i II I i? . I I I i PLATES EXPLANATION OF PLATES PLATE 12 Navatu, site 17. a. Navatu or Uluinavatu viewed from King's Road on its south side. b. Na tu viewed from the west (in Narewa village). Talus slope on left of main crag is shell midden covered with a growth of vaivai. c. Navatu viewed from the northwest on the inland side of the main railroad tracks near the shore. d. View from the upper fortification wall (270 ft. elevati on Navatu, looking west by north. Narewa village in middle; King's Road at left; railroad with:' train at right near shore; tidal flats partly exposed by ebbing tide; Tovu Island in distance at far right. PLATE 13 Excavations at Navatu, site 17. a. View into rock shelter from stake J3, location A. From left to right: Kalaveti, Ratu Rambithi, George Wanggata. June 20, 1947. b. Looking west from t of boulder which forms rock shelter at location A. Some crew members picking sherds from screen and piling them on piece of corrugated iron. June 27, 1947. _. Presenting yanggona to the dead before removing bones from burial no. 1, site 17, July 17, 1947. Photo by Rambithi. d. Wall U. in rock shelter, location A. Shell midden overlying sterile yellowish volcanic ash. July 5, 19 e. Rectangle E3-4 F3-4 at back of cut, location A, in process of excavation; two rows of stones two earlier buried house mounds, the forward, lower row being in Wall E3-4. (The third and late on the surface; see diagram 1.) July 11, 1947. f. Wall B4-D4, location A; tape hangs from stake C4; upper portion of wall contains shell, indicated by white spots. June 26, 1947. PLATE 14 Excavation at site 17 and views of other sites. a. The cut for rectangle Bl-2 Cl-2, locati B, site 17; Isikeli Ndonu (left) and Ratu Rambithi standing in front. The opening of rectangle Al-1.5 Cl-1.5 is behind their feet. Chalk line passing behind their shoulders is 4.5 ft. above railroad ground level. July 15, 1947. b. Looking west by north toward Tovu Island from locationr B, site 17, at low tide. At high tide water.extends into mangroves in foreground. Strip of wat in background at left is course of Narewa Creek through tide flat at low tide. c. Stones of un- known significance marking legendary house mound of the Siamese twins Thirikaumoli and Kalusamba slayers of Ndengei's pigeon Turukawa. At northern base of Mt. Ulunda, Kauvandra Mts., Ra Provin Photo by Rambithi. d. Cemetery Kawalevu of the matanggali Vusatambua, after clearing and repairi by members of the matanggali on July 15, 1947. Site 17, Navatu. Width of rectangle 12 ft.; leng including second part (in background with man standing at far end), about 130 ft. Shell-midden material around and on cemetery, which has been in use since heathen times. This cemetery was situated between Narewa railroad switch and location B and a few yards inland from the main rail-* road tracks. e. Shore side of Singatoka sand dunes, looking west to coral reef in background. area in the foreground is site 20. Site 21 is in two swales opening on the beach with breakers front of dunes in the background. f. Inland side of Singatoka sand dunes near site 20. Kulukulu village is about 100 yds. behind photographer. Both photos of the sand dunes by Mr. H. H. Storck Singatoka, and published by his kind permission. PLATE 15 Site 26 (Vunda) and site 23. a. View of the acropolis (Korovatu) of Vunda as seen from the west; the picture was taken from the left bank of the Vunda River. Location A lies in low trees base of crag and below the deep scar or undercut in the crag in the left half of the picture. b. View of the acropolis of Vunda as seen from the southwest. Location B lies among the bushes coconut trees between the base of the crag and the left-hand part of the sugarcane field at the dark bush. c. View of part of location A, site 26, showing round house mound in rear. Partially' excavated rectangles show on level ground in front of house mound. The black background is dense vegetation. August 11, 1947. d. View from top of Vunda acropolis, looking down Vunda River at high tide. Crew pausing in sugarcane field on way to river to wash for luncheon. August 13, 194 e. View from east of flat-topped hill on top of which is site 23, with its cave. Koroinggara, th name of site 23, means "village of the cave." Modern Volivoli village lies a short distance to t east on the Queen's Road, Nandronga Province. f. View eastward into defile from escarpment of f1 r r L266] i GIFFORD: EXCAVATIONS IN FIJI .d hill on which site 23 is situated. Hills at right are sand dunes with vegetation. Sites 20 21 are on the shore side of these dunes. One end of the pond shown in e is visible in this ure near Queen's Road. PLATE 16 2 Modern pots and carved paddle. a. Modern pot from western New Britain, Bishop Museum no. 5; collected by H. C. Votaw, August, 1944; greatest diameter 290 mm., height 289 mm. b. Cook- pot (5304) from Malolo Island, off Lautoka and Nandi provinces, source of much modern pottery oestern Viti Levu; white specks shell or calcite; greatest diameter 4.00 mm., height 358 mm.; 55 3.5. c. Carved paddle from Shortland Island, Bishop Museum no. 8377; for producing relief gn on pottery; photo from Bishop Museum. d, e. Two views of medium-sized cooking pot (5007) in Naivuvuni, Ra Province; Modern and Late period incised decoration; upper part partially n-coated; greatest diameter 399 mm., height 489 mm.; hardness 4.0; hardness of resin slip 2.5. PLATE 17 Modern pottery from Ra and Nandronga provinces. a. 5009; water Jar made in Vunitongoloa, Ra vince; resin coating now opaque whitish and partly weathered off, since the vessel was left out- Irs,B a hole rendering it useless; greatest diameter 364 mm., height 357 mm.; hardness 4.0; hard- off resin slip 2.5. b. 5030; very thick-walled cooking pot made in Vunitongoloa, Ra Province; tbably a degenerate modern type; resin coating on upper part; greatest diameter 212 mm.; height Bmm. ; hardness 3.0; hardness of resin slip 2.0. C, d. 5008; pottery yanggona (kava) bowl made iLaselase, Nandronga Province; purchased by Narewa, Ra, people in 1942 and sold to the author in 47; diameter of outer edge of rim 431 mm., diameter of inner edge 367 mm., height 200 mm.; hard- w 4.0; hardness of resin slip 2.0. e. 5031; flower holder made in Vunitongoloa, Ra Province; a cern type for sale to Europeans; front resin-coated; length 262 mm.; hardness 2.5; hardness of win slip 2.0. PLATE 18 Models of cooking pots and water Jars made in Nasilai Village, Rewa Province, and presented by tu Sir Lala Sukuna and Mr. G. Kingsley Roth of the Fijian Office, Suva. a. 3165; model of chief's sin-coated pottery water Jar called sangga tambua; greatest diameter (between shoulder tips) 4 mm., height 204 mm.; hardness 2.5; hardness of resin slip 2.0. b. 3156; model of cooking pot r vegetables called kuro; greatest diameter 203 mm., height 315 mm.; hardness 2.5. c. 3154; del of pot for boiling fish called i vakariri; greatest diameter 177 mm., height 132 mm., hard- as 2.5. Three pottery pot rests, called sue (3162, 3161, 3163), support the vessel. Measure- rts of these pot rests are as follows: left (3162), greatest diameter 69 mm., height 125 mm.; ddle (3161), greatest diameter 72 mm., height 120 mm. Hardness, 3.0, 3.0, 2.0, respectively. 3164; model of resin-coated pottery water Jar called sangga ndina; greatest diameter 189 mm., ight 226 mm.; hardness 4.0; hardness of resin slip 2.0. PLATE 19 Cooking pot from Nasingatoka Village and sherds from sites 20 and 21 in sand dunes west of uth of Singatoka River, Nandronga Province. b-f, h-J, gift of collector, Dr. Lindsay Verrier. 3166; thin ware cooking pot from Nasingatoka Village, Nandronga Province; greatest diameter 5 mm., height 401 mm.; hardness 4.0; gift of Messrs. Sukuna and Roth. b. 3268; rim sherd, tched; site 21; hardness 3.5. c. 3269; rim sherd and base, apparently of shallow flat-bottomed oh; site 21; roulette type incising; hardness 4.5. d. 3271; sherd with incised design of con- oted loops; site 21; one end sand-blasted, exposing white particles of shell or calcite; possibly rt of a flat-bottomed vessel; greatest length 127 mm.; hardness 4.0. 2. 3267; potsherd rim with tches apparently made by pressure with a round stick; site 21, hardness 4.0. f. 3273; sherd with aged and incised design; site 21; greatest length 109 mm.; hardness 2.5. g. 5241; potsherd rim th wavy edge and deeply incised shoulder; site 20; greatest length 134 mm., hardness 2.5. h. J6; sherd with gouges; site 21; hardness 4.0. i. 3278; sherd with cross-hatched pattern; site 21; rdness 4.0. j. 3274; sherd with nubbins made by pressing out with fingers; site 21; hardness 4.5. ale: b and c. same scale as d; h-L, same scale as f. PLATE 20 Sherds with leaf or matting impression, Philippine, and Late incised sherds. a-c, from site gift of Dr. Lindsay Verrier; _-p, Philippine incised sherds; g-x, Late period incised sherds Dm location A, site 17. a. 3275; sherd with leaf impression, probably part of bottom of dish; 267 ANTHROPOLOGICAL RECORDS length 93 mm.; hardness 4.5. b. 3280; sherd with matting impression, probably part of bottom dish; hardness 4.0. c. 3284; sherd with matting impression; hardness 4.0. d Philippine int sherds; reproduced by courtesy of Professor H. Otley Beyer. q. 8840; rectangle G3-4 H3-4, o-T. deep; hardness 4.0. r. 6702a; surface; hardness 2.5. s. 6703; surface; hardness 4.0. t. 722 rectangle A3-4 B3-4, 0-6 in. deep; hardness 4.0. u. 6702b; surface; hardness 3.0. v. 6708; 8 face; greatest length 101 mm.; hardness 4.0. w. 6705; surface; hardness 4.0. x. 6713; surfac hardness 2.5. Scale: b and c, same scale as a -u, w x same scale as v. PLATE 21 Incised potsherds from site 17, Navatu, and site 26, Vunda; a-w, site 17, x-z, at-gL', site a. 8912; rectangle H3-4 I3-4, location A, 6-12 in. deep; hardness 3.0. b. 6907; rectangle C3 D3-4, location A, 0-6 in. deep; hardness 4.0. c. 7963; rectangle B1-2 C1-2, location B, 12-18 deep; hardness 4.0. d. 7998; rectangle Bl-2 Cl-2, location B, 18-24 in. deep; hardness 4.0. 6706; location A, surface; hardness 4.0. f. 8320; rectangle E3-4 F3-4, location A, o-6 in.. dee hardness 3.0. g. 6741; location A, surface; hardness 4.5. h. 7984; rectangle Bl-2 Cl-2, locat B, 18-24 in. deep; ribbed relief and incising; hardness 4.5. i. 8107; rectangle A1-1.5 C1-.5, location B, 54-60 in. deep; hardness 4.0. 8018; rectangle B1-2 C1-2, location B, 30-36 in. deep; hardness 4.0. k. 6712a; location A, surface; hardness 4.0. 1. 7329; rectangle A3-4 B3-4 location A, 18-21 in. deep; hardness 4.0. m. 7894; rectangle B1-2 C1-2, location B, surface tr angle; hardness 4.0. n. 7955; rectangle Bl-2 Cl-2, location B, 12-18 in. deep; hardness 4.5. a. 6712b; location A, surface, hardness 4.5. p. 7374; rectangle A3-4 B3-4, location A, 21-27 i deep; hardness 4.0. q. 6887; rectangle B3-4 C3-4, location A, 64-70 in. deep; rim sherd with i cising extending to rim edge; hardness 4.0. r. 7972; rectangle B1-2 Cl-2, location B, 12-18 in deep; rim; hardness 4.0. s. 7971; rectangle B1-2 C1-2, location B, 12-18 in. deep; rim, hardne 4.0. t. 7994; rectangle Bl-2 Cl-2, location B, 18-24 in. deep; water Jar rim; hardness 4.0. u 6763; location A, surface; hardness 3.0. v. 8925; rectangle H3-4 I3-4, location A, 6-12 in. de w. 8116; rectangle Al-1.5 C1-1.5, location B, 54-60 in. deep; rim; greatest length 115 mun.; har ness 4.0. x-z, a' -. Early period incised sherds from location A, site 26, Vunda, Lautoka Pro ince: x. 10447; rectangle A4-5 B4-5, 54-60 in. deep; notched rim; hardness 2.0. a. 10826; rec- tangle E-E1 F-Fl, 48-54 in. deep; hardness 4.0. z. 10682; rectangle B4-5 C4-5, 71-77 in. deep; hardness 2.5. a'2. 10877; rectangle E-El F-Fl, 76-82 in. deep; hardness 4.0. b'. 10865; recta E-E1 F-Fl, 66-70 in. deep; greatest length 82 mm.; hardness 4.0. c'. 10704; rectangle B4-5 C4- 88-94 in. deep; hardness 2.5. d'. 10933; rectangle E-E1 F-Fl, 112-118 in. deep; hardness 2.0. e'. 10724; rectangle B4-5 c4-5, 100-106 in. deep; hardness 3.0. f'. 10534; rectangle A4-5 B4-5 115-119 in. deep; hardness 2.0. g. 10660; rectangle B4-5 C4-5, 53-59 in. deep;-hardness 3.0. Scale: a, b, e-g, k, 1, o-q, u, same scale as v; c, d, h-m, m, n, r-t, same scale as w; x-z, a' c'-g', same scale as b'. PLATE 22 Cross relief and incised sherds from site 17, Navatu. All from location A except d, _f, and from location B; d-f, i-r, incised sherds; p-s, v-i, rim sherds. a. 7174; rectangle C3-4 D3-4, j 64-70 in. deep; hardness 4.0. b. 7117b; rectangle C3-4 D3-4, 48-52 in. deep; hardness 4.0. c. rectangle C3-4 D3-4, 24-30 in. deep; hardness 4.0. dc. 7984; rectangle B1-2 Cl-2, 18-24 in. deep;1 incised "tree" design; relief decoration applique; hardness 4.0. e. 7903; rectangle Bl-2 C1-2, triangle; relief decoration ribbing; hardness 5.0. I. 9254; rectangle I3-4 J3-4, 12-18 in. deep relief decoration ribbing and applique nubbins; greatest length 58 mm.; hardness 4.0. g. 9139; rectangle H3-4 I3-4, 66-73 in. deep; finger-print impressions in "gouges" on inner as well as out surface; hardness 3.5. h. 7117a; rectangle C3-4 D3-4, 48-52 in. deep; greatest length 110 mm.; hardness 3.5. i. 6742; surface;- relief decoration applique nubbins; hardness 4.0. 4. 7963; rec tangle Bl-2 C1-2, 12-18 in. deep; relief decoration applique; hardness 4.0. k. 8854; rectangle:c G3-4 H3-4, 0-7.5 in. deep; incised crescents bordered by relief ribs. L. 6736; surface; relief decoration raised band with punctations; hardness 5.0. m. 8821; rectangle G3-4 H3-4, top triangb of earth on sloping surface; relief decoration "rope" design; hardness 4.5. n. 7155; rectangle A C3-4 D3-4, 58-64 in. deep; greatest length 136 mm.; hardness 4.0. > 6889; rectangle B3-4 C3-4, - 63-70 in. deep; hardness 4.5. p. 7167; rectangle C3-4 D3-4, 58-64 in. deep; "pie crust"type of applique rim; cross-relief on body very indistinct (see also pl. 24, c, which classified as verl cal ribbing); greatest length 121 mm.; hardness 3.5. q. 8715b; rectangle E3-4 F3-4, 86-92 in. d i hardness 4.0. r. 8715a; rectangle E3-4 F3-4, 86-92 in. deep; hardness 4.0. s. 7382; rectangle A3-4 B3-4, 21-27 in. deep; notched rim edge; hardness 4.0. t. 7097; rectangle C3-4 D3-4, 42-48 i' deep; hardness 4.0. u. 7076; rectangle C3-4 D3-4, 36-42 in. deep; hardness 4.0. v. 7836; rec- tangle D3-4 E3-4, 78-84 in. deep; hardness 3.0. w. 7742; rectangle D3-4 E3-4, 48-54 in. deep; hardness 3.0. x. 6896; rectangle B3-4 C3-4, 64-70 in. deep; gouges Just below rim edge; hardnes6 3.5. y. 7185; rectangle C3-4 D3-4,64-70 in. deep; hardness 4.0. z. 8764; rectangle E3-4 F3-4, 98-104 in. deep; gouged design between edge of rim and cross relief; hardness 3.0. Scale: a-c, A same scale as h; d, e, i-m, same scale as f; o, t, u, same scale as n; q-s, w-z, same scale as p 268 GIFFORD: EXCAVATIONS IN FIJI PLATE 23 Early period pottery spouts and handles from location A, site 17, Navatu: f-k , Fiji Museum lamens dug by author. a. 7151; rectangle C3-4 D3-4, 52-58 in. deep; edge of orifice visible on ir left side of handle; hardness 4.5. b. 7216; rectangle B3-4 D3-4, 76-82 in. deep; handle and t of body with two spout openings; hardness 4.5. c. 6882; rectangle B3-4 C3-4, 33-63 in. deep; Ulple orifices indicated by inserted chalks; greatest length 97 mm.; hardness 2.5. d. 7187;- 'tangle C3-4 D3-4, 64-70 in. deep; handle with spout opening; hardness 4.5. e. 7111; rectangle * D3-4, 42-48 in. deep; handle with spout opening; hardness 3.5. f. From wall c4-D4, 86-92 in. p; greatest length 50 mm. g. From wall C4-D4, 86-92 in. deep; greatest length 60 mm. h. From 1 C3-D3, 86-92 in. deep; greatest length 50 mm. i-k. Three views of one specimen; from wall Di, 86-92 in. deep; greatest length 100 mm. i. Side view showing spouts on two sides of handle. Pettom view showing inner orifices of the two spouts. k. End view of perfect spout. Scale: d, d same scale as c. PLATE 24 Relief potsherds from site 17, Navatu. All from location A except i from location B. a. 6971; -tangle C3-4 D3-4, 6-12 in. deep; ribbed; crenelated incising on rim, small gashes below ribbing; latest length 94 mm.; hardness 4.0. b. 7138; rectangle C3-4 D3-4, 52-58 in. deep; hardness 4.0. 7165; rectangle C3-4 D3-4, 58-64 in. deep; rim sherd with applique band applied to outer surface trimmed like a pie-crust edge; vertical ribbing below rim (see also p1. 22, p); greatest length mm.; hardness 4.0. di. 9036b; rectangle H3-4 I3-4, 36-42 in. deep; irregular applique; hardness 4, e. 9036a; rectangle H3-4 I3-4, 36-42 in. deep; applique nubbins; hardness 4.0. f. 7116a; tangle C3-4 D3-4, 48-52 in. deep; widely spaced ribs or raised lines; hardness 4.0. g. 7361; itangle A3-4 B3-4, 21-27 in. deep; ribbed and applique nubbins; hardness 4.0. h. 9524; rectangle 4 03-4, 6-13 in. deep; hardness 4.0. i. 8008; rectangle Bl-2 Cl-2, 24-30 in. deep; applique ges; hardness 4.0. J. 9132; rectangle H3-4 I3-4, 66-73 in. deep; applique ridge with gouges oining; hardness 3.0. k. 7118; rectangle C3-4 D3-4, 48-52 in. deep; applique nubbin; hardness i. 1. 9668; rectangle N3-4 03-4, location A, 38-44 in. deep; hardness 4.0. m. 7116b; rectangle 1 D3-4, 48-52 in. deep; hardness 4.0. n. 6837; rectangle B3-4 C3-4, from test pit, 0-33 in. p; reflexed rim with applique supports; hardness 4.0. o. 7140; rectangle C3-4 D3-4, 52-58 in. p; hardness 4.0. p. 7175; rectangle C3-4 D3-4, 64-70 in. deep; hardness 4.0. a. 7023; rec- gle C3-4 D3-4, 18-24 in. deep; rim sherd with gouged design below shoulder; rim is at written Oer; hardness 4.0. r. 9132; rectangle H3-4 I3-4, 66-73 in. deep; notched applique perhaps in- 4ed for rope design; hardness 4.0. S. 9541; rectangle N3-4 03-4, 6-13 in. deep; rim sherd with ohed applique, rope (?) design reaching to edge of rim; probably Late horizon (cf. pl. 25, g) tatest length 93 mm.; hardness 3.5. t. 7098; rectangle C3-4 D3-4, 42-48 in. deep; ribs separated deep gouges or grooves, crossed design on ridges; hardness 4.0. u. 7156; rectangle C3-4 D3-4, 664 in. deep; hardness 4.0. Scale: b, f-h, 1, m, same scale as a; e, d, i-k, n, same scale as c; ! t, u, same scale as s. PLATE 25 i Early period potsherds from site 17 and site 26: k and E) from site 26, all others from loca- A, site 17. a. 7130; rectangle C3-4 D3-4, 48-52 in. deep; rim; hardness 2.5. k. 7203; rec- LeB3-4 D3-4, 70-76 in. deep; rim; greatest length 75 mm.; hardness 3.0. c. 7168; rectangle D3-4, 58-64 in. deep; rim; hardness 4.0. d. 8767; rectangle E3-4 F3-4, 98-104 in. deep; I,ed sherd; greatest length 78 mm.; hardness 3.5. e. 7130a; rectangle C3-4 D3-4, 48-52 in. deep; hardness 2.5. f. 6897; rectangle B3-4 C3-4, 63-70 in. deep; rim; hardness 3.5. ,. 8633; rec- le E3-4 F3-4, 61-68 in. deep; notched applique running up to rim (cf. pl. 24, s); hardness 4.0. 201; rectangle B3-4 D3-4, 70-76 in. deep; rim; angular wavy relief design rather than gouging; Bess 3.0. i. 7130b; rectangle C3-4 D3-4, 48-52 in. deep; hardness 4.0. S. 7168; rectangle ,D3-4, 58-64 in. deep; rim; hardness 4.0. k. 10531; rectangle A4-5 B4-5, location A, site 26, 115 in. deep; rim; hardness 2.5. 1. 8658; rectangle E3-4 F3-4, 68-74 in. deep; rim; keeled, gouges below, incised lines above; hardness 3.0. m. 7109; rectangle C3-4 D3-4, 42-48 in. ; rim; hardness 4.0. n. 7863; rectangle D3-4 E3-4, 93-96 in. deep; rim; hardness 2.0. o. 7166; tangle C3-4 D3-4, 58-64 in. deep; rim; hardness 4.0. p. 10506, rectangle A4-5, B4-5, location site 26, 91-97 in. deep; gouges run up to rim edge; hardness 2.0. q. 7157; rectangle C3-4 D3-4, 64 in. deep; angular wavy style; hardness 3.5. r. 7120; rectangle C3-4 D3-4, 48-52 in. deep; !ed wavy style; hardness 2.5. s. 8656; rectan4e E3-4 F3-4, 68-74 in. deep; longitudinal gouges aw rim edge; trace of cross relief design on shoulder; greatest length 72 mm.; hardness 2.5. B579; rectangle E3-4 F3-4, 48-54 in. deep; applique nubbins in two rows below rim edge. u. 7208; tangle B3-4 D3-4, 76-82 in. deep; angular wavy style; hardness 4.5. v. 8762; rectangle E3-4 4, 98-104 in. deep; "pitted" design bordered by deep incised diagonal line; rim edge down; hard- 0 3.5. w. 7848; rectangle D3-4 E3-4, 84-90 in. deep; punctate design on rim edge; hardness 3.0. 269 ANTHROPOLOGICAL RECORDS x. 8765; rectangle E3-4 F3-4, 98-104 in. deep; rim; nubbins, apparently rubbed down after appli tion; hardness 3.5. -y. 7141; rectangle C3-4 D3-4, 52-58 in. deep; curved wavy style; hardness .z... 7182; rectangle C3-4 D3-4, 64-70 in. deep; plain portion showing number is apparently near r angular wavy style; greatest length 97 mm.; hardness 2.5. a'. 8781; rectangle E3-4 F3-4, 104- in. deep; rim; vertical ribbed design on body; hardness 4.0. b'. 8737; rectangle E3-4 F3-4, 9 in. deep; incised design; rim edge at bottom of picture; hardness 2.5. Scale: a, e, f, i, M same scale as b; c, h, k 1, , E r, same scale as r v-x, a', same scale as s. PLATE 26 Early period wavy relief potsherds from location A, site 17. a. 7186; rectangle C3-4 D3-4s, 64-70 in. deep; angular wavy relief; plain rim on relief decorated body; greatest length 246 mm.' hardness 4.0. b. 7192; rectangle B3-4 D3-4, 70-76 in. deep; curved wavy style; greatest length 135 mm.; hardness 4.0. c., 6891; rectangle B3-4 C3-4, 63-70 in. deep; curved wavy style; hardnes 3.5. d. 7157; rectangle C3-4 D3-4, 58-64 in. deep; angular wavy style; hardness 4.0. Scale: q! same scale as b. PLATE 27 Potsherds, handles and spouts from various sites. a-g, i, 1-s, w, x, from location A, site 17. - i, Early period, a. 8799; rectangle E3-4 F3-4, below 116 in. deep; plain rim on wavy relief body; greatest length 94 mm.; hardness 4.0. b.. 8717; rectangle E3-4 F3-4,.86-92 in. deep pie-crust type applique rim overlying wavy relief body; hardness 2.5. 5a 7154; rectangle C3-4 , D3-4, 58-64 in. deep; "wicker" type wavy relief; hardness 2.5. d. 8697; rectangle E3-4 F3-4, 80-86 in. deep; potsherd ground to disk shape; hardness 2.5. e. 7762; rectangle D3-4 E3-4, 54-6 in. deep; incurved rim; hardness 3.5. f. 7150; rectangle C3-4 D3-4, 52-58 in. deep; relief up t rim edge; hardness 3.5. g. 7837; rectangle D3-4 E3-4, 78-84 in. deep; outer and inner edges-of 4 notched; hardness 4.5. h. 10434; rectangle A4-5 B4-5, location A, site 26, 43-48 in. deep; incii sherd with notched rim; hardness 2.5. i. 6892; rectangle B3-4 C3-4, 63-70 in. deep; "wicker" war relief; greatest length 101 mm.; hardness 2.5. S. 7869; from surface, east slope of Navatu; Late, period incised; hardness 2.0. k. 5077; from surface, site 5, near Nambuna Village, District of Tavua, Tholo North Province; unusual style of applique relief; hardness 3.0. 1. 6880; rectangle: B3-4 C3-4, 33-63 in. deep; unusual style of Early gouged relief; hardness 3.0. m. 6922; rectang4 C3-4 D3-4, o-6 in. deep; portion of collarlike rim; Late period incised; greatest length 93 mm.;- hardness 4.0. n. 7318; rectangle A3-4 B3-4, 12-18 in. deep; portion of collarlike rim; Late pert incised; hardness 2.5. o. 7352; rectangle A3-4 B3-4, 21-27 in. deep; Late period incised; greatel length 139 mm.; hardness 2.5. -. 9856; rectangle S3-4 T3-4, 6-12 in. deep; bottle orifice with radial applique relief, applique bearing incised decoration; Late period; hardness 3.0. g. 7260, rectangle A3-4 B3-4, o-6 in. deep; bottle orifice with relief decoration on rim edge; Late period hardness 4.5. r. 7225; rectangle A3-4 B3-4, 0-6in. deep; Late period incised; hardness 4.0. . 6787; surface; handle or leg (?); hardness 2.5. t. 7867; from surface, east slope of Navatu, Ra Province; Late period incised; hardness 2.5. u. 7878; from surface, east slope of Navatu, Ra Province; relief and incised decoration; probably Late period; hardness 4.0. v. 11415; rectangle H2-3 I2-3, location A, site 26; 24-31 in. deep; bottle top with horizontal perforation and adjoin ing vertical spout openings on two sides; hardness 2.5. w. 9507; rectangle N3-4 03-4, o-6 in. de nipplelike spout; hardness 2.5. x. 6841; rectangle B3-4 C3-4, 0-33 in. deep; ring handle with traces of spouts on two sides; hardness 4.0. Scale: b, e-g, same scale as a c, d, h, same scale as 1, 3-i, r-t, same scale as o; n, p, SL u-x, same scale as m. PLATE 28 Fragment of food dish with rim base. 5920; from 200 ft. elevation on Navatu, found under boulder below a gateway of lower fortification or war fence; diameter of rim 110 mm.; hardness 3. a. Top. b. Bottom. PLATE 29 Thin sections of potsherds magnified 15 diameters. a. 10516; rectangle A4-5 B4-5, location~ site 26, 102-108 in. deep; shows slight rounding of the grains and rather uniform sorting; feldsp quartz, augite, and acid igneous rock chips present; small square flake of biotite (dark gray) visible near the center of the photograph. b. 10532; rectangle A4-5 B4-5, location A, site 26, 114-120 in. deep; large plutonic rock fragment composed of several grains of quartz at far right edge of photograph; right of center and also at bottom of photograph andesitic rock fragments 270 GIFFORD: EXCAVATIONS IN FIJI taining phenocrysts of plagioclase in microcrystalline ground masses. c. 8287; rectangle E3-4 4, location A, site 17, top triangle of mound material; rotalid foraminifer visible at center of tograph; other grains of magnetite (black) and augite (gray); (see also e). d. 10392; rectangle 5 B4-5, location A, site 26, 24-30 in. deep; average grain size in this section finer than in other; note uniform sorting as to size; quartz (white), feldspar (dark gray), and augite riated) fresh and readily discernible. e. 8287 (see c) grains in this section almost entirely oh augite of uniform size (dark gray and striated). f. 8794; rectangle E3-4 F3-4, location A, e 17, 114-120 in. deep; feldspars (dark gray) all show considerable alteration to clay; note roaching alteration in feldspar grain at top right; quartz (white) is not affected, and augite riated) only slightly. # ?. .? a c d PLATE 12. NAVATU, SITE 17 b 271 272 ANTHROPOLOGICAL RECORDS a c a e PLATE 13. EXCAVATIONS AT SITE 17 b f i GIFFORD: EXCAVATIONS IN FIJI 273 )t d C /2 PLATE 14. EXCAVATION AT SITE 17 AND VIEWS OF OTHER SITES a c 274 ANTHROPOLOGICAL REC ORDS a C d PLATE 15. VUNDA (SITE 26) AND SITE 23 - I , a , _- - .. w.f tt4 St C f7 .I I I I I 4 .1 t GIFFORD: EXCAVATIONS IN FIJI PLATfE 16. MODERN POTS AND CARVED PADDLE 275 ANTHROPOLOGICAL RECORDS d PLATE 17. MODERN POTTERY FROM RA AND NANDRONGA PROVINCES 276 GIFFORD: EXCAVATIONS IN FIJI a b c, d MODELS OF COOKING-POTS AND WATER JARS MADE IN NASILAI VILLAGE,REWA PROVINCE 277 PLATE 18. 278 ANTHROPOLOGICAL RECORDS a PLATE 19. COOKING-POT FROM NASINGATOKA VILLAGE AND SHERDS FROM SITES 20 AND 21 GIFFORD: EXCAVATIONS IN FIJI w X PLATE 20. SHERDS WITH LEAF OR MATTING IMPRESSION; PHILIPPINE INCISED SHERDS; LATE PERIOD INCISED SHERDS 279 ANTHROPOLOGICAL REC ORDS yI PLATE 21. INCISED POTSHERDS FROM SITES 17 AND 26 280 GIFFORD: EXCAVATIONS IN FIJI PLATE 22. CROSS RELIEF AND INCISED SHERDS FROM SITE 17 281 ANTHROPOLOGICAL RECORDS b L PLATE 23. EARLY PERIOD POTTERY SPOUTS AND HANDLES FROM LOCATION A, SITE 17 282 GIFFORD: EXCAVATIONS IN FIJI n PLATE 24. RELIEF POTSHERDS FROM SITE 17 283 ANTHROPOLOGICAL RECORDS PLATE 25. EARLY PERIOD POTSHERDS FROM SITES 17 AND 26 284 GIFFORD: EXCAVATIONS IN FIJI PLATE 26. EARLY PERIOD WAVY RELIEF POTSHERDS 285 286 ANTHROPOLOGICAL RECORDS d PLATE 27. POTSHERDS, HANDLES, AND SPOUTS FROM VARIOUS SITES 287 a b PLATE 28. FRAGMENT OF FOOD DISH WITH RIM BASE V' t - I -1 , I 7 L e-~- . 4 z it LO N~ A. 288 .t 40 3 v Q E, t It ik