572.05 FA N.S. no. 22-27 199^-96 iiti;s^Ma?a;.i^m^S!»»^' .t^. V7 Anthropology NEW SERIES, NO. 23 Archaeological Research at Tumatumani, Juli, Peru Charles Stanish Lee Steadman With a contribution by: Matthew T. Seddon August 31, 1994 Publication 1457 PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY Information for Contributors to Fieldiana jiisidcrcd ai. space pcmiius. > per printed page or fractii. I . ossmg, which reduces the publication lime. Contributions IV kred for publication regardless of ability to pay page charges, uthors of unsolicited manuscripts. 'ITiree complete copies of the text (including tit! Mons should be submitted (one original copy plus two review copies \v1iich may b nanuscripts will be considt ir submitted to reviewers before all materials ar >.^ .lands of the Scientific lidjto s should be submitted to Scientif ■ ield Museum of Natural I i ago, lllinoi Manuscripi landard-weight, 8V4- by 11-inch paper v. typed on an liiM-compatibie computer using MS-DOS, also submit text on SVi-mch diskette 2, or 5.0, MuItiMatc, Displaywrite 2, 3 & 4, Wang PC, Samna, Microsoft Word. Volksyvriter. or UoroSiar programs or ASCII). F(ir papers over 100 manusci, , ^ , authors are requested tosubmit a "Table of Contents," a "List of IIU.,. ,„..„„.„ Liiid a I ist of Tables" immediately following title page. In most cases, the text should be preceded by an "Abstract" and should conclude with "Acknowledgments" (if any) and "Literature Cited." All measurements should be in the metric system (periods are not used after abbreviated measurements). The format and style of headings should follow that of recent issues of Fieldiana. For more detailed style information, see The Chicago Manual of Style (13th cd.), published if Chicago Press, and also recent i.ssues of fjWrf/a/ja. References: In "Literature Cited," book and journal titles should be given in full. Where abbrc ._,.„ ,„. c (e.g., in citation of synonymies), authors consistently should follow Botanico-Periodicum-Huniianum and TL-2 TaxonomU Literature by F. A. Stafleu & R. S. Cowan (1976 et seq.) (botanical papers) or Serial Sources for the Biosis Data Base (1983; published by the BioSciences Information Service. Names of botanical authors should follow the "Draft Index of Author Abbreviations, Royal Botanic Gardens, Kew," 1984 edition, or TL-2. References should be typed in the following form: Croat, T. B. 1978. Flora of Barro Colorado Island. Stanford University Press, Stanford, Calif r..,,uK n J J ^ Lloyd, and T. D. Pennington. 1963. A comparison of montane and lowland ram lorcM in i coaoor '. The forest structure, physiognomy, and floristics. Journal of Ecology, 51: 567-601. ix;ni;u.!n, 1 ■ . J. M. 1979. Yage among the Siona: Cultural patterns in visions, pp. 63-80. In Browrnan. D. I Schwarz, eds.. Spirits, Shamans, and Stars. Mouton Publishers, The Hague, Netherlands. ^f.I ra, J. 1946. The historic tribes of Ecuador, pp. 785-821. In Steward, J. H., ed., Handbook of South American Indians. Vol. 2, 'ITte Andean Civilizations. Bulletin 143, Bureau of American Ethnolop;'', Smifhsoniar Institution, Washington, D.C. Stoize, R. G. 1981. Ferns and fern allies of Guatemala. Part II Polypodiaceae. Fieldiana: Dot. Illustrations: Illustrati not as "plates"). Figures mu some indication of scale, normally a rclerence bar Statenicnls m tigurc captions alone, such as "x 0.8," are not acceptable. (]!aptions should be typed double-spaced and consecutively. See recent issues of Fieldiana for details of style. All illustrations should be marked on the reverse with authors name, figure number(s), and "top." Figures as submitted should, whenever practicable, be 8Vi x 11 inches (22 x 28 cm), and may not c\ s (30 X 42 cm). Illustrations should be mounted on boards in the arrangement to be obtained in the printed work. Ihis al set should be suitable for transmission to the printer as follows: Pen and ink drawings may be originals (preferred) itostais; shaded drawings must be originals, but within the size limitation; and photostats must be high-quality, glossy, and-whilc rrints. Oricjn.'i' ill ustiai inns will ho re turned tn the rnrresof^ndinp; ^uihor nr-x^n mil-ilirniMn nn]i^.;<; n(h/»ni'i^;e i. , ! ^1 ,iu vith the Scientific Editor ' emploj-s a two-step correction system. The corresponding author will noi iiially receive a copy which deletions, additions, and changes can be made and queries answered. Only one set of cd corrections of type must be made on the single set of page proofs. Changes in page ' aic \'crv expensive. Aiilhor-i^enerated chnnfTcs in nnyc. nmofs can onlv be m.uie if the h'er rrfeets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Archaeological Research at Tumatumani, Juli, Peru Fig. 1 . Map of south central Ancles. FIELDIANA Anthropology NEW SERIES, NO. 23 Archaeological Research at Tumatumani, Juli, Peru Charles Stanish Lee Steadman Associate Curator Department of Anthropology Field Museum of Natural History Chicago. Illinois 60605-2496 Anthropology Department University of California 232 Kroeber Hall Berkeley, California 94720 With a contribution by: Matthew T. Seddon - Anthropology Department University of Chicago 1126 East 59th Street Chicago, Illinois 60637 - Accepted April 2, 1993 Published August 31, 1994 Publication 1457 PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY © 1994 Field Museum of Natural History Library of Congress Catalog Card Number: 93-73815 ISSN 0071-4739 PRINTED IN THE UNITED STATES OF AMERICA This book is dedicated to the memory of Victor Barua Remy. Victor was the director of the Cuajone area of Southern Peru Copper Corporation where I spent my first season in Andean archaeology, in 1 982. Victor and his wife Lucy treated me as one of their family. My first trip to Lake Titicaca was as their guest on one of their rare vacations in July of that year. Five years later I returned to Juli to begin the research reported on in this book. Victor was a strong supporter of the archaeology of Moquegua and a dear friend. His passing is a great loss to us all. C.S.S. Table of Contents Preface and Acknowledgments ix Part 1 . Background to the Tumatumani Investigations 1 Charles Stanish and Lee Steadman A Brief Chronological Background 2 Previous Research in the Juli Area 3 The Lupaqa Project 4 Research Etesign of the Regional Survey . . 5 The Sillumocco Occupation in the Juli- Pomata Region 6 The Tiwanaku Occupation in the Juli-Po- mata Region 9 Research at the Site of Tumatumani .... 10 Part 2. Analyses of Artifacts from Tumatumani 19 Ceramic Artifacts 19 Lee Steadman LiTHic Artifacts 65 Matthew T. Seddon Part 3. Summary and Conclusions 72 Charles Stanish Literature Cited 75 9. Reconstruction of the site of Tumatu- mani 13 10. Site of Tumatumani 13 11. Five excavation units on the west mound 16 1 2. Arbitrary levels of excavation units 1 7 13. Natural levels of Unit 1, west wall 18 14-165. Ceramic drawings 79 166. Complete hoe 104 167. Complete adze 104 168. Hoe and adze dimensions 105 169. Hoe fragment 105 1 70. Adze fragment 1 06 171. Discoids 107 172. Triangular projectile points 108 173. Concave-base projectile points ... . 109 174. Comer-notched projectile points . . 109 175. Scrapers 110 176. Cumulative percentage of tools by component Ill List of Tables List of Illustrations 1 . Map of south central Andes Frontispiece 2. Lake Titicaca region 1 3. Chronology of sites in the Titicaca re- gion 3 4. Juli and the site of Tumatumani . . 5 5. Sillumocco period sites in the Juli- Pomata survey region 6 6. Tiwanaku period sites in the Juli- Pomata survey region 7 7. Distribution of ceramics at Tumatu- mani 11 8. Contour map of Tumatumani show- ing location of platform faces 12 1 . Total number of sherds in Unit 1 exca- vations 20 2. Percentage of paste types by construction episode 21 3. Tumatumani paste distribution by period 22 4. Finish categories in Unit 1 excavations 25 5. Surface color and finish of Paste 1 spec- imens in Unit 1 excavations 34 6. Surface color and finish of Paste 3 spec- imens in Unit 1 excavations 44 7. Tumatumani lithic type collection .... 68 8. "Other" category 69 9. Type collection with "hoe/adze frag- ment" category omitted 69 10. Type collection raw materials by com- ponent 70 11. Debitage raw material composition by component 71 Tumatumani is located near the modern town ofJuli, Peru, on the shores of Lake Titicaca. It is one of the principal Formative and Tiwanaku period sites in the Juli-Pomata region of the southwestern Titicaca Basin. This report pre- sents the results of intensive surface mapping, controlled sur- face collections, and excavations conducted at the site. These data serve to define the Formative period culture of Sillu- mocco that dominated the region prior to the expansion of Tiwanaku into the area. We also describe a major Tiwanaku site type in the area and offer a detailed analysis of ceramic and lit hie artifacts from the site. Preface and Acknowledgments In the spring of 1988, the Juli Project began in- tensive archaeological investigations in the Juli region of southern Peru. Juli is located in the pro- to-historic and historic Lupaqa zone in the south- western Titicaca Basin. The Juli Project mapped, systematically collected, and excavated two sites. The first site is known as Pucara Juli. It dates to the Late Intermediate or Altiplano period, be- tween approximately a.d. 1 000 and 1 450. The sec- ond site is named Tumatumani and is composed of two large mounds located near the lake edge. Previous research by John Hyslop and Elias Mu- jica in the 1 970s indicated that Tumatumani dated principally to the Formative and Tiwanaku peri- ods of the region. The surface collections, exca- vations, and detailed surface mapping reported here provide the most systematic data on the site to date. The 1988 season was funded by the Wenner- Gren Foundation for Anthropological Research, the Montgomery Fund of the Field Museum of Natural History, and Ms. Patricia Dodson. These excavations were conducted in cooperation with, and under the direction of, the National Institute of Culture in Lima and Puno. By 1990, the Juli Project had evolved into the Lupaqa Project, fund- ed by the National Science Foundation (BNS 9008181), the Field Museum of Natural History, the Heinz Foundation, and Ms. Patricia Dodson. Steadman's research was funded by a grant from the Fulbright-Hays Doctoral Dissertation Re- search Abroad program (PO22A90021). Seddon's research was supported in part by a grant from the Scholarship Committee of the Field Museum. We offer special thanks to officials of the National In- stitute of Culture, including Dr. Elias Mujica, Lie. Oscar Castillo, Lie. Oscar Ayca, and Dr. Luis Lumbreras. The project was assisted by the an- thropological faculty of the Universidad Nacional del Altiplano, including its director, Lie. Felix Pa- lacios, Juan Bautista Carpio Torres, and Lie. Abel Torres Comejo. Dr. Percy Che-Piu Salazar, Dean of Social Sciences, Sr. Julio Cesar Gomez Ga- mona, director of the Instituto Superior Tecnolo- gico in Juli, and Ec. Luis Salas Arones, rector of the university, were very supportive of our project. We also gratefully acknowledge the support of Dr. Luis Watanabe M., Dr. Francisco Iriarte B., Dr. Fernando Cabieses, Dr. Michael Moseley, Dr. Don S. Rice, Dr. Alan Kolata, and Dr. Walter G. Tapia Bueno, director of the National Institute of Cul- ture in Puno; Ms. Patricia Dodson of Chicago; Sr. Benedicto Valdez C, mayor of Juh; Sr. Jorge M. Remond Alvarado, provincial subprefect; and Sr. Alberto Miranda Arenas, Sr. Vicente Quispe Cho- quegonza, and Sr. Manuel Encinas, governors of Chuquito Province from 1988 to 1991. We also thank Sr. Hugo Teodoro Chavez Talavera, pro- vincial mayor of Yunguyu, Sr. Arturo Arce Ter- razas, subprefect of Yunguyu Province, Sr. Andres Carbajal and family, Sr. Percy Heman Calizaya Chuquimia and family, Sra. Fresia Gandarillas Sardon, Sr. Moises Sardon Pelags and family, Srta. Zenaida Espezua Bustinza, and the people of Juli, Yacari-Tuntachawi, Sillucani, and Inca Pucara. Professional help was provided by Edmundo de la Vega M., Cecilia Chavez J., Kirk Lawrence Frye, Edwin Castillo, Adan Umire Alvarez, Lisa Ci- polla, Javier Ticona, and Luperio David Onofre Mamani. Student participants included Alberto Sosa B., Elsa Chuquimia Paredes, Clemente Caxi Maquera, David Benjamin Antezana Bustinza, Teodora Atahuachi Chuquimia, Julie Realmuto, Rodolfo Carita Ancco, Jose Castillo Velo, Lisa Shogren, and Alfredo Curo Chipana. We also gratefully acknowledge the input from IX several readers, including Drs. David Browman, matumani as well. We also thank Dr. Harold Vo- Joyce Marcus, Katharina Schreiber, James Ma- ris, editor of Fieldiana. A very special thanks to thews, Brian Bauer, Paul Goldstein, Margaret Goes, Marjorie Pannell, Managing Editor of Fieldiana, and anonymous reviewers of Fieldiana. Drs. Luis for her infinite patience. Errors, of course, are purely Lumbreras, John Hyslop, and Luis Watanabe pro- the responsibility of the authors, vided important insights into the nature of Tu- Part 1 Background to the Tumatumani Investigations Charles Stanish Lee Steadman For almost two millennia, the Titicaca Basin has been the economic, political, and cultural center of the southern region of the central Andes {fron- tispiece). The Titicaca region has been home to one pre-Hispanic empire and countless smaller political and ethnic groups throughout its prehis- tory. Without question, the world's highest navi- gable lake ranks as one of the great centers of na- tive American civilization. In this report, we present excavation and surface • Qaluyu • Pucara o S" Huancane ( Vilquechico Hatuncolla , • Omiryo I ^ Paucarcolla Puno, Santiago De Huata 0 10 20 30 40 50 km Zeplta DesaguaderoNV Fig. 2. Lake Titicaca region. data from the site of Tumatumani, located in the southwestern Titicaca Basin of southern Peru (Fig. 2). Tumatumani was known for years as poten- tially one of the most important sites in the west- em Titicaca region and constituted one of the prin- cipal sites of the Tiwanaku and Antecedents macropattem proposed by John Hyslop in 1976. The twin mound complex at Tumatumani covers an area of more than 5 ha and was occupied from at least 500 B.C. to the present day. One of the two artificial mounds was rebuilt at least twice by dif- ferent complex pre-Hispanic polities that con- trolled the Juli region, attesting to the importance of the site throughout history. This report provides data on the occupational history, the architecture, the ceramic assemblage, and the lithic assemblage from this important Titicaca Basin site. These data help refine the ceramic styles and chronology of the region, define a new Formative period culture that we call Sillumocco, provide insight into the nature of Tiwanaku expansion into the south- western Titicaca region, and suggest that the Up- per Formative period in the region was of greater complexity than previously reported. A Brief Chronological Background Naturalists and archaeologists have conducted sporadic research in the Titicaca Basin since the 1 9th century. This research has provided a broad chronological framework for describing the cul- tural history of the region. Following Lumbreras ( 1 974), we divide the Formative into three phases: Upper, Middle, and Lower (Fig. 3). The first, sed- entary, agropastoral societies in the Titicaca Basin define the Lower Formative period (ca. 2000-1 300 B.C.). This early period is very poorly known at the present time. Currently, the Middle Forma- tive/Upper Formative transition dates to the mid- dle of the first millennium b.c. and represents a significant cultural change in the Titicaca Basin. The Middle Formative cultures are early Chiripa (Chiripa Condori and Chiripa Llusco, ca. 1300- 650 B.C.), located in the southern Titicaca Basin, and Qaluyu (ca. 1300-500 b.c), located in the north. The Upper Formative cultures include Pu- cara (ca. 500 b.c.-a.d. 400), late Chiripa (i.e., the Mamani phase, ca. 650-100 b.c). Early Tiwanaku (ca. 300 b.c.-a.d. 200), and Qeya (also known as Tiwanaku III), which dates to around a.d. 200- 400.' ' The dates for Pucara include the Cusipata Phase of Mujica (1987), which dates to approximately 500-200 The end of the Upper Formative in the Titicaca Basin is marked by the expansion of Tiwanaku out of its core Pacajes territory in the southern Titicaca Basin (Fig. 3). The development of the expansive Tiwanaku polity is archaeologically rec- ognizable by the manufacture of a new pottery style that replaced Qeya in the south and Pucara in the north. This new Tiwanaku ceramic assem- blage is known as Tiwanaku Classic or Tiwanaku IV in its earlier phase, and as Decadent Tiwanaku or Tiwanaku V in its later one. These periods date to approximately a.d. 400-750 and a.d 750-1000, respectively. This chronology is based on the work of Wendell Bennett (1934: 445-459) and Carlos Ponce Sangines (1981b: Table I ). We borrow from Bermann (1990: 55) and refer to this as the Ben- nett-Ponce chronology, recognizing that problems still exist with this framework. By around a.d. 600 or so, the Tiwanaku state most likely controlled the entire southwestern side of the Titicaca Basin, including the Juli-Pomata region. Data from outside the Titicaca Basin area indicate that by at least a.d. 800, and probably earlier, Tiwanaku had expanded throughout the south central Andes (Berenguer 1978; Bermann 1990; Goldstein 1989; Kolata 1983). In this pe- riod, Tiwanaku, along with Wari, constituted one of the two great Middle Horizon states of the cen- tral Andes. We concur with Schreiber (1992: 275- 276) and see Wari as an empire. Likewise, we see Tiwanaku as an expansive, most likely imperial state system. The end of the Tiwanaku state (corresponding to the end of the Tiwanaku V period) dates to about A.D. 1 100, perhaps slightly earlier. Through- out the Titicaca region, the collapse or fragmen- tation of Tiwanaku gave rise to the proto-historic Aymara senorios, sometimes improperly translat- ed as "kingdoms." Polities such as the Lupaqa, Pacajes, Colla, and others flourished around the Titicaca Basin until the Inca conquest (Hyslop 1976). These Aymara-speaking groups have been discussed by Andean historians such as Cieza de Leon, Bemabe Cobo, and Guaman Poma. They paint a picture of intense conflict between these groups, particularly the Lupaqa and Colla, in the time immediately prior to the conquest of the re- gion by the Inca state. Our best evidence suggests that the Inca con- quered the region toward the end of the 1 4th cen- tury A.D. (Julien 1983: 244). The Titicaca region B.C. We extend the traditional date of Pucara by several centuries, based on our interpretation of the Tumatu- mani data presented in Part 2. Part One South Titicaca North Titicaca Juli-Pomata 1500 Inca Inca Inca Altiplano Altiplano Altiplano 1000 Tiwanaku V Tiwanaku V Tiwanaku V 500 Tiwanaku IV Tiwanaku IV Tiwanaku IV Qeya T (Tiwanaku III) 1 AD/BC t Qalasasaya Pucara Late L Chiripa Cusipata Sillumocco 500 Mamani Chiripa Eariy Llusco Qaluyu Sillumocco 1000 Chiripa Condon 1500 Fig. 3. Chronology of sites in the Titicaca region. was viewed by the Inca as one of the most im- portant of its provinces. The site of Hatuncolla was established as one of the largest provincial centers of the empire (Julien 1983: 79), and the Inca built one of its most impressive religious cen- ters on the Island of the Sun. The Juli-Pomata region was part of Lupaqa Province. The elite of the Lupaqa state apparently enjoyed a privileged position in the imperial bureaucracy. This privi- lege most likely derived from an "alliance" be- tween the Inca and the Lupaqa against their com- mon foe, the Colla (e.g., see Cieza de Leon n.d.: Bk. 100). Previous Research in the Juli Area There has been relatively little research in the Juli area in spite of its importance in the region. The celebrated pioneer archaeologist Alfred Kid- der conducted a limited reconnaissance in the northern and western Titicaca Basin in the 1 940s. He discovered a number of sites with Pucara and Tiwanaku pottery on the surface. Kidder's 1943 report describes two sites that are located close to the Juli area. These sites are Asiruni and Sarapa (1943: 10), located in the Pampa Have several ki- lometers northwest of the modem town of Juli. Asiruni is described as a slight elevation of the pampa near the Have River. Kidder noted that a broken stela and large, cut sandstone blocks were found on the ground. These features strongly sug- gest the existence of an important structure with a cut stone semi subterranean court. Sarapa is sim- ilar to Asiruni in having several mounds and cut sandstone blocks on the surface. The ceramic frag- ments associated with these sites indicate a Ti- wanaku and possibly Upper Formative occupa- tion. Asiruni and Sarapa are the southernmost Tiwanaku and possibly pre-Tiwanaku sites dis- cussed by Kidder. Kidder never discussed Tu- matumani, and we can reasonably assume that he was not aware of the site. One extensive reconnaissance of the Lupaqa re- gion was conducted by John Hyslop in the 1970s. Tumatumani is one of the principal sites in the Tiwanaku and Antecedents macropattem pro- posed by Hyslop (1976) for the Lupaqa zone. A macropattem is a synthesis of the general archae- ological characteristics of sites of a single, long time period. Hyslop defined three of these macro- patterns for the Lupaqa area. The Tiwanaku and Antecedents macropattem includes all sites of the Tiwanaku and Formative periods. This macro- pattem is characterized by undefended sites on low plains near the lake. Several of the larger sites of this macropattem, such as Tumatumani, have substantial corporate architecture. The importance of the site of Tumatumani for understanding the very complex Formative and Tiwanaku period Titicaca Basin prehistory cannot be overstated. Previous research on Tumatumani (Hyslop 1976; Mujica 1985: 124-127) has indi- cated that many of the major ceramic styles from the Formative and Tiwanaku periods are found at Background to the Tumatumani Investigations the site. Geographically, the site is located between the two traditional centers of political power in the Titicaca Basin— the northern Collao region, where Pucara developed, and the southern Pacajes zone in Bolivia. Pacajes is the ancestral home of the Tiwanaku empire. Pacajes is probably the heartland of Qeya pottery manufacture, the style associated with the immediate predecessor of the expansive Tiwanaku state. Mujica's map (1985: 1 24) of Pucara site distribution includes 34 sites, ranging from Batan Urqo, near Cuzco, in the north to Tumatumani in the south. Mujica also listed 20 sites with Qeya ceramics that are distributed throughout the southern side of the basin (Mujica 1985: 127). In this map of Qeya site distribution, Tumatumani appears as the northernmost site along the southwestern edge of the lake. With one exception, Pucara and Qeya site distributions do not overlap in Mujica's map. That one exception is Tumatumani. The Lupaqa Project The Lupaqa Project is a research program of systematic survey, controlled surface collections, and excavations of archaeological sites in the Lu- paqa area of the Titicaca Basin. The project is headquartered in Juli, one of the major Colonial and modem towns on the Peruvian side of Lake Titicaca. Juli is located in the province of Chucuito at approximately Km 79 on the Puno-Desaguade- ro highway (Fig. 4). Known informally as the "Lit- tle Rome of Peru," Juli was the center of Domin- ican and Jesuit activity in the region during the Colonial period. It was home to four huge church- es, a fact that attests to its importance in early Colonial times. Juli is listed as a major town in the Visita of Garci Diez de San Miguel in 1564. It is also mentioned as a major Inca site, a fact corroborated by archaeological research (Hyslop 1976: 399; Stanish et al. 1992). Juli continues to be the judicial "capital" of the province of Chu- cuito. While the population of the Have Peninsula has grown substantially larger than Juli's, Juli con- tinues to be one of the most important centers in the western Titicaca region. The geography and ecology of the Titicaca re- gion have profoundly affected the development of political and economic systems throughout history and prehistory. The classic work of Pulgar Vidal (n.d.) divides the region into two broad ecological zones, the puna and the suni. The suni is located between 3500 and 4000 meters above sea level (m.a.s.l.). The suni represents the upper limit of plant agriculture, while the puna is a grazing zone for the extensive camelid herds owned by many Titicaca Basin peoples. The suni includes the lake edge and the area a few kilometers away from the lake. It is the area of greatest economic potential. Pulgar Vidal notes the rich variety of agricultural products that grow in the suni. These include many varieties of tubers, legumes, and chenopods (n.d.: 95-98). Not sur- prisingly the suni is the major area of human set- tlement today, especially at the base of the hills below 4100 m.a.s.l. Different ecological studies have suggested a more complex mosaic of resource zones other than the suni/puna distinction. Tosi (1960), for instance, lists eight zones for the Ti- ticaca region, a classification based on the Hold- ridge system. For our purposes here, however, Pul- gar Vidal's distinction between the agricultural and pastoral zones is the one that is the most culturally relevant. The suni is the region where pre-Hispanic raised fields were located (Erickson 1988: ii; Kolata et al. 1987; Smith et al. 1968). Raised fields are feasible along the lake shore and along rivers where flat topography permits. They are possible only in the low suni zones. Likewise, Flores Ochoa and Paz Flores (1983) have documented the use of qoc ha in the suni zone. These are small, water-filled de- pressions or lakes in the Titicaca Basin used in- tensively by modem farmers to grow altiplano crops. It is likely that they were used by pre-His- panic populations as well. The lake itself provides an important additional economic resource base on a scale unique in the Andes. The highest density of Formative and Tiwanaku sites is found in the suni zone as well. Tumatumani is a good example of a settlement choice located to ensure access to several resource areas. The puna is defined as that zone between 4000 and 4800 m.a.s.l. (Pulgar Vidal n.d.: 104). The major plant agricultural product is the potato, which can be grown throughout the puna. While potatoes can be grown up to the snow line (Pulgar Vidal n.d.: Ill), the most optimal yields occur in the warmer suni zones and in the lower puna. Overall, by far the most important economic prod- uct of the puna is the camelid, particularly the llama and the alpaca. Camelids provide wool and meat and serve as pack animals. The virtually unique capacity of the Titicaca Basin to support large camelid herds has contributed to its position as a major center of civilization in the Americas. Part One Fig. 4. Juli and the site of Tumatumani. Research Design of the Regional Survey Initial work on the Juli Project concentrated on ceramic analysis and reconnaissance around the Juli area. In 1990-1991, we completed a system- atic survey of the area north of modem Juli to just southwest of Pomata (Fig. 2). Our survey meth- odology was a "full-coverage regional survey" (Fish and Kowalewski 1990: 2). We conducted a 100% pedestrian survey of 360 km- in the Juli-Pomata region. The survey used a standard methodology for semi-arid lands. Teams of three to four ar- chaeologists walked between 10 and 50 m apart. The distance varied according to terrain. A site was defined as any substantial concentration of artifactual remains and/or architecture. By far the most common diagnostic surface artifact was ce- ramic fragments. Usually, a concentration of more than five sherds per square meter over at least 25 Background to the Tumatumani Investigations ••I 441 # o LAKE TITICACA ^—^001 022/ -^ 133 / \__^'''^l^l^ / ^~<^%^[ JUU ^ I 228 #0220 V -^/ ) \^ 2<0«« ^ L A, N^^ 208 \ ^"^ n ^ A '% \ ^"•r^n ~1 \500W A 333 ^ \ o 365# 0=' ^ Fig. 5. Sillumocco period sites in the Juli-Pomata survey region. sq m was necessary to establish a location as a site. Occasional and isolated occurrences of a few sherds, particularly post-Tiwanaku ones, were not treated as sites. Two loci were considered separate sites if they were separated by more than about 100 m. When a site was found, we recorded information on a standard form. If the site was small, it was uniformly surface-collected with a conscious at- tempt to achieve a representative sample. If there were discrete areal concentrations or functionally distinct areas in the site (e.g., cemeteries), then separate collections were made. The research design called for a concentration of the survey along the lake shore with additional coverage in the puna zone away from the lake. This strategy was based partially on Hyslop's re- sults, which showed a lakeside concentration of pre-Hispanic settlement. This strategy ensured that the study region included representative sections of both puna and sunt environments (Figs. 5 and 6). In total, we surveyed almost 40 km of lake shore area and moved into the puna at distances that varied between 2 and 22 km. At present, almost 500 sites representing more than 800 occupations have been discovered. The vast majority of the sites are post-Tiwanaku in date. We were able to achieve complete coverage throughout the area, with the exception of one small military base near Pomata. The survey covered all geographic zones from lake level (3810 m.a.s.l.) to 4600 m.a.s.l. The survey also included three areas of relict raised fields. More than 70 Sillumocco and Tiwanaku occupations were discovered in the Juli-Pomata area (Figs. 5 and 6). The Sillumocco Occupation in the Juli-Pomata Region One major result of the Juli-Pomata survey and Tumatumani excavations is the definition of the local Formative period occupation. We have Part One ^^ ^,„ ^^ LAKETmCACA 1^ A- 133 / v^ /^'*^%^r~^*" ^ • *r I6OVJ (UU -.^ 120 136 1 2ia«« C ^ u \. 208 /^ y 261 \ -•r<^ W 'k''^ < 284 \ ,0 ^•o C^ ^ Fig. 6. Tiwanaku period sites in the Juli-Pomata survey region. named this occupation Sillumocco, after a type site (158) located several kilometers due west of Juli (Fig. 5). The term Sillumocco refers to the associated ceramic assemblage and culture in the Juli-Pomata area during the Formative period. Its usage is similar to the use of the terms "Pucara," "Qalasasaya," and "Chiripa."^ The term Sillumocco was created because the ceramic assemblage from Tumatumani and other contemporary sites shows sufficient local variation that it cannot be equated directly with any pre- viously defined sequence. Sillumocco ceramics are clearly a local product and are distinctive enough to merit their own cultural designation. The archaeological excavation data from Tu- matumani allowed us some stratigraphic control for dividing the Sillumocco period. It is important to state at the outset that we do not have stratified midden levels. Rather, the Tumatumani excava- - "Pucara" is also spelled "Pukara." "Qalasasaya" is also spelled "Kalasasaya." tions encountered two distinct construction levels clearly separated by a prepared clay floor. The lower construction level dates to the Late Sillu- mocco period, ca. 200 B.C. to a.d. 400, while the upper, second construction dates to the Tiwanaku IV period, ca. a.d. 400 to 750. The division between Early Sillumocco and Late Sillumocco results in two distinct assemblages, one fiber-tempered and one without fiber temper. The period of transition which must have existed be- tween the two assemblages and the amount of overlap between them could not be defined with the archaeological evidence from these excava- tions. Some overlap does exist, particularly in the decorated wares. However, the bulk of the ceramic assemblage shows a marked change in paste com- position as well as in vessel shape and finish be- tween the two periods. The transition between Early Sillumocco and Late Sillumocco is believed to be contemporary with the end of Chiripa Mamani and the beginning of the Tiwanaku I or Qalasasaya period. Again, Background to the Tumatumani Investigations this conclusion must remain tentative since we do not know if there is an exact correlation between the Juli-Pomata area and Bolivian sequences. We see the beginning of Qalasasaya (Tiwanaku I) as later than reported by Ponce Sangines (1981b: 137), who hinted at a date sometime between 1 580 and 600 B.C. We place the beginning of Qalasasaya sometime after 300 B.C., based on an analysis of Ponce's data (1981b: 137) and Hermann's work at Lukurmata (Hermann 1990: 55). The Early Sillumocco period spans the end of the Middle Formative and the early part of the Upper Formative. It is roughly contemporary with Late Qaluyu, Cusipata (Mujica 1987: 22), Chiripa Llusco, and Chiripa Mamani (Browman 1978b: 808-809; Mohr 1966: 163). It therefore dates to approximately 800-200 b.c. The Early Sillumocco period is characterized by fiber-tempered utilitarian wares with strong similarities to Chiripa ceramics, fiber-tempered incised and painted ceramics with similarities to Chiripa, and non-fiber-tempered (or specimens produced in a paste with very little fi- ber) incised and painted ceramics with similarities to Qaluyu. The Qaluyu-related ceramics can be placed in the Middle Formative, contemporary with Early Sillumocco. In fact, comparisons with similar ceramics from Chavez's Marcavalle se- quence would place them at approximately 800 B.C., which probably dates the earliest occupation of most Sillumocco sites in the Juli-Pomata area. The Late Sillumocco occupies the later part of the Upper Formative in the Juli-Pomata area. It is roughly contemporary with Classic Pucara, Qa- lasasaya (Tiwanaku I), and Qeya (Tiwanaku III), and dates to approximately 200 B.C. to a.d. 400. The Late Sillumocco period is characterized by sand-tempered utilitarian and decorated wares. The use of fiber temper persists in this period only on a limited scale and occurs only in minor quantities in one of the pastes, used for the manufacture of a small number of utilitarian wares and Tiwanaku IV decorated ceramics. We anticipate being able to subdivide the Late Sillumocco period in the future, since it is evident that some of the speci- mens assigned to this period bear more of a re- semblance to Qalasasaya ceramics and some are more clearly related to the later Qeya assemblage. The Sillumocco settlement pattern is shown in Figure 5. It is immediately apparent that the set- tlement is heavily concentrated on the lake shore. Few sites are as high as 4 100 m, and there are no Sillumocco sites found above 4200 m in the survey area. It is significant that the largest cluster of elite/ceremonial Sillumocco sites is on low hills in or on the periphery of the Moyopampa raised field system in the northern section of the survey (Fig. 5). There is a distinctive site size hierarchy with sites ranging from small mounds to the large mounded site of Tumatumani (001) and the im- pressive terraced hill sites of Sillumocco (1 58) and Palermo (2 1 2). These three sites in particular are characterized by the presence of corporate archi- tecture. These sites almost certainly represent the elite/ceremonial centers of the Sillumocco polity. It is significant that we did not find similar sites in any area outside of the Juli-Moyopampa zone, which suggests that this latter area was a regional center. It is likely that additional regional centers occur near the Have River and east of Pomata. This is suggested by the work of Kidder (1943), who discovered the sites of Asiruni and Sarapa on the Have pampa, and the work of Hyslop (1976), who reported the large site of Quenuani in the Pomata pampa. It is also possible that the sites in the Challapampa area may correspond to a re- gional center further east, perhaps one associated with the Quenuani site. Such propositions await future testing. It is surprising that there are few high-altitude sites in the Sillumocco period. Excavations at the sites of Tumatumani (001) and Palermo (212) in- dicate heavy use of camelid meat, as evidenced by relatively large quantities of butchered bone in the middens. We can hypothesize that the pastur- ing of camelids did not involve any formal settle- ment system utilizing special activity sites. It is more likely that access to camelid herds involved informal systems of control by individual house- holds or larger kin groups. Certainly there is no archaeological evidence of corrals or special high- puna sites involved in camelid raising. The survey data strongly suggest that raised field agriculture was an important component of the Sillumocco economy. As we said earlier, sites clus- ter near the raised field section in the Moyopampa and are found along the periphery of the fields up to the northwestern end of the survey zone. The site of Palermo (2 1 2) is located directly between two aqueducts that feed the raised fields. The first substantial occupation of Palermo was in the Sillu- mocco period, as evidenced by analysis of data from controlled surface collections and excava- tions. Control of these freshwater sources may have been a major settlement determinant for Palermo, at least in the Late Sillumocco period. Raised fields Part One are also found in the Pomata area and were prob- ably constructed in the Huancane zone due west of Pomata (Fig. 6). There is a substantial Sillu- mocco settlement in the Huancane region. The Pomata area was not adequately surveyed to de- termine the settlement pattern, but informal sur- vey by one of us (Stanish) indicates that there is a significant number of Sillumocco sites in this sector. In summary, the Sillumocco settlement pattern is characterized by a lakeside settlement focus, the absence of fortified settlements, clustering of sites near raised field areas, and a lack of any archaeo- logically visible formal system of camelid pastur- ing. The Juli area is hypothesized to be a Sillu- mocco regional center as represented by the sites of Tumatumani, Sillumocco, and Palermo. The Tiwanaku Occupation in the Juli- Pomata Region A total of 4 1 Tiwanaku sites were discovered in the study area (Fig. 6). Tiwanaku sites are iden- tified by the presence of Tiwanaku pottery on the surface. Tiwanaku ceramics in the Juli-Pomata region are recognized by very distinctive pottery shapes and design motifs. Certain Tiwanaku fine- ware pastes are also distinctive and are probably imports into the region. The best Tiwanaku mark- er, however, was pottery in the form ofkero, tazon (flaring-sided bowl), and incense burner shapes. We also were able to identify sites by the presence of typical Tiwanaku design motifs on keros, ta- zones, bottles, and jars. Most of the Tiwanaku pottery from the study area would be classified as Tiwanaku IV or Ti- wanaku V, or Classic and Decadent respectively in the traditional Bennett-Ponce sequence. Un- fortunately, there is an insufficient number of dec- orated fragments from the survey to accurately phase the sites into Tiwanaku IV and V. The most common shapes are keros and tazones. Keros are found with and without bands around the body. Incense burners are also a common shape in the Tiwanaku ceramic assemblage in the region. The first manifestation of Tiwanaku influence in the region dates to the Classic phase (ca. a.d. 400- 750). There are no Chiripa or early Tiwanaku (i.e., Tiwanaku I or III) sites in the region, although an exchange network that involved Chiripa finewares certainly existed. It is critical to note that the Ti- wanaku occupation at Tumatumani, as with the other Tiwanaku sites in the Juli-Pomata area, be- gan in the Tiwanaku IV period. The Tiwanaku settlement pattern is shown in Figure 6. One of the most striking features of the Tiwanaku settlement pattern is its similarity to the Sillumocco one. A significant number of Sillu- mocco habitation sites continued to be occupied in the Tiwanaku period. Most of the sites that were abandoned were hamlets or small villages, while the large centers continued to be occupied. More than 20 new sites were founded in the Tiwanaku period, including Tumuku (121), an elite/cere- monial center located in the Kajje area between the Chocasuyu and El Molino areas. These data indicate a strong continuity between the Sillu- mocco and Tiwanaku period settlement patterns. It is most likely that the Tiwanaku state expanded into an already complex political and economic system dominated by the Sillumocco peoples. Another very significant addition to the settle- ment pattern in the Tiwanaku period is the found- ing of sites south of the Moyopampa region in the puna. The distribution of sites 271 and 411 is suggestive of either a road or a settlement system designed to control the Pasiri region water re- sources. The Pasiri drainage ultimately feeds the raised field sections in Moyopampa. If these sites follow a road, it is most likely that they led to the low, swampy areas in the Pasiri region. This area is an ideal camelid grazing zone at the present time. This aspect of the Tiwanaku settlement pat- tern could therefore be the first indication of a formal system of camelid production in the Juli- Pomata region. It is also significant that the raised field areas continued to be utilized by the Tiwa- naku peoples. In fact, the abandonment of sites 200 and 228 could have resulted from expansion of the raised field areas into these former Sillu- mocco habitation zones. Tumatumani represents one of at least six major elite/ceremonial centers in the region during the Sillumocco and Tiwanaku periods. In our classi- fication of site types (Stanish et al. 1 992), Tuma- tumani is a classic example of a large, artificial mound with corporate construction (Type 1 site). The other elite site type (Type 3) is characterized by sunken courts on low hills with surrounding domestic terraces. There are three Type 1 sites in the survey area— Tumatumani (001), Huaquina (210), and an un- named site, no. 444 (Fig. 6). The sites of Asiruni and Sarapa, located just north of the Juli-Pomata Background to the Tumatumani Investigations 9 survey area and briefly described by Kidder ( 1 943: 10), would fit into the Type 1 category as well. In our survey area, only Tumatumani and Huaquina have surface artifacts indicative of a Sillumocco period occupation. Tumatumani is therefore also significant in that it represents one of three Type 1 sites in the Juli-Pomata area and one of a few such large sites with large mounds in the lake re- gion. In summary, the Tiwanaku settlement pattern is characterized by a lakeside settlement focus and the absence of fortified settlements, patterns sim- ilar to the earlier Sillumocco pattern. Also like the earlier pattern, sites clustered near raised field ar- eas. All major Sillumocco elite/ceremonial sites continued to be occupied in the Tiwanaku period. Tumatumani is a classic Type 1 mound site. It represents one of the few major elite/ceremonial settlements in the region. Research at the Site of Tumatumani Historical Work Tumatumani was first reported in 1 976 by John Hyslop (1976: 246) in his survey of the Lupaqa area. He reported finding a variety of pottery styles on the surface of the mounds, including fragments related to Chiripa, Qaluyu, Tiwanaku I, and Pu- cara. These ceramic remains were concentrated on two partially artificial mounds that rose several meters above the natural ground level. The site sits in a low alluvial zone between the two large hills of San Bartolome and Sapacollo that flank the modem town of Juli (Fig. 4). A small river flows through the site area and actually runs directly between the two mounds. The relatively deep soils and proximity to the lake (lake shores have slightly higher ambient temperatures and therefore slightly longer growing periods) make the site area one of the richest agricultural zones in the Juli region. There is no surviving stone architecture on the surface of the mounds. However, the remains of several artificial platforms are clearly evident. Each of these platform faces is approximately 50 cm in height. They have been severely eroded by inten- sive agricultural use. Aboriginally, these platforms were considerably higher. We estimate that they were 2 m or more in antiquity. Despite its location near modem Juli, Tuma- tumani is relatively undisturbed and unlooted. The first 25 cm below the surface is badly disturbed from shallow agricultural plowing. One large mod- em house was built into the southern side of the west mound, destroying about one fifth of this mound. Also, the northern side of the west mound has been cut back for agricultural purposes. A stream cut the eastern side of the east mound, exposing a natural subsoil. Other than these dis- turbances, the site area is relatively intact. Research Design at Tumatumani The Tumatumani investigations included three principal field strategies: topographic mapping, in- tensive surface collections, and limited excava- tions. The mapping and surface collections were designed to define site limits, size, architectural and settlement complexity, architectural features, and intensity of occupation. The excavations were intended to separate major occupational periods from minor ones (given the great variety of surface ceramic types discovered by Hyslop and our team) and to help define architectural building episodes. The entire site was mapped with a T-2 Wild theodolite. This procedure provided a detailed, 1 -m contour map of the entire site area that was used in the systematic surface collection (Fig. 8) and in the location of other surface site features. The systematic surface collection utilized an in- tensive (100%), aligned, systematic (every 20 m) technique with 2 x 2-m collection units. The grid was placed over the entire potential site area and beyond and can be easily located by future re- searchers. The point N 2000/E 2000 served as the main datum and is the highest part of the west mound. The number designation is in meters, and the grid lines are oriented 4° east of north. To locate any part on the site, one can simply go to the highest point on the west mound, orient a da- tum line 4° off" north, and measure off" in meters to the unit designation. All units are numbered at the southwest comer. This grid system and surface collection method had been previously worked out at the large For- mative and Tiwanaku site of Lukurmata in Bolivia (Stanish 1989b: 45-50). This method is charac- terized by a systematic, aligned sampling tech- nique. The success of the Lukurmata research en- couraged us to use a similar method at Tumatumani. In an effort to characterize the Ti- wanaku and earlier occupations, all sherds iden- 10 Part One '^2200 2100 2000 1900 Fig. 7. Distribution of ceramics at Tumatumani. Background to the Tumatumani Investigations 1 1 Fig. 8. Contour map of Tumatumani showing location of platform faces. tifiable as post-Tiwanaku were removed from the unit collections and the remaining ones were counted. The density distribution (logarithmic) of this collected sample is presented in Figure 7. These results are interpreted below. The mapping and surface work also served to identify and measure the major architectural fea- tures preserved on the site. Figure 8 is a contour map of the site, and Figure 9 is a distant view and hypothetical reconstruction of the mounds. This involved the careful measurement of existing ter- race faces on the mounds and the subjective as- sessment of the size of the original mound areas. These reconstructions are useful for conveying the nature of the corporate architecture of the site. Of course, the aboriginal architecture is much more complex than is initially apparent from the sur- face. The final field procedure utilized at Tumatu- mani was the excavation of a single terrace face. A total of five adjacent excavation units (2 x 2 m and 2 X 1 m) were placed in one of the lower terraces on the more densely occupied sector of the west mound. This was an area where surface collections indicated a very dense scatter of fiber- tempered sherds and other pre-Tiwanaku and Ti- wanaku styles. The selection of the excavation units were therefore arbitrary, in an area of high-density surface materials. Results and Interpretations Surface Mapping— Figure 10 is a photograph of the site from a distance. The west mound is in the center left, with the house compound built into 12 Part One Fig. 9. Reconstruction of the site of Tumatumani. its southern side. The east mound is directly be- hind and to the right of the west mound. The site is approximately 8 m above ground surface at its highest point. It is significant that the lowest area of the site is only 7 m above the level of Lake Titicaca. If the site was continuously occupied from the Formative period to the Late Horizon, then the fact that the site is so low holds important implications for paleo-ecological reconstructions of Lake Titicaca levels. On the other hand, should good evidence indicate that the lake has indeed risen more than 7 m in some periods since the ^m^'^^iUmrf'^'^'^**" Fig. 10. Site of Tumatumani. Background to the Tumatumani Investigations 1 3 Formative, then our reconstructions of the pre- Hispanic settlement dynamics would have to ac- count for site abandonment and resettlement in areas as low as that surrounding Tumatumani. Certainly within the 20th century the lake has fluc- tuated at least 5 m in height. The extant surface architecture on the site was also recorded. We measured the width and height of terraces and platforms on the mound and in- terpolated for areas destroyed by modem con- structions. In the case of the destroyed terrace fac- es, we presumed that there was an architectural symmetry. This was a reasonable presumption, given other architectural reconstructions from contemporary sites in the Titicaca Basin (e.g., Chi- ripa as reconstructed by Conklin and Moseley [1988: 161]). These surface data permit us to reconstruct the architecture that characterized the last major con- struction of the site. On the west mound two large, low platforms were constructed, with the smaller one placed on top of the larger platform to create a two-tiered "temple" mound. Each of these plat- forms was faced with a terrace that was most likely made of uncut fieldstones. This suggestion of stone- faced terraces is based on the discovery of an uncut stone wall in the excavations that corresponded to an earlier terrace wall construction (see Excavation Results, below). Today the terrace faces are ap- proximately 50 cm in height, but they have been severely eroded by agricultural activities. The east mound is dramatically different from the west mound. It is built in an elongated horse- shoe or U shape with a small terraced platform between the two legs of the U. The terraced plat- form was probably faced with a stone wall. This we deduced from the presence of cut sandstone blocks near the site. Without excavations, how- ever, this inference remains unproven. We were able to locate areas on the mounds where natural subsoil was exposed on the sides. It was clear that each mound had been built on a natural hill. The aboriginal hills are characterized by a reddish yellow clay subsoil. By locating these natural surfaces or hills, we were able to determine the approximate contours of the pre-occupational natural hills. Because of the lack of disturbed areas around the mounds, these contours were too im- precise to present a profile but were sufficient for a verbal description of the quantity of artificial fill used in the site area. It is clear that both mounds were once low nat- ural hills that were later built up with artificial fill. The architectural features suggested in Figure 8 represent the last stage of the corporate building projects at the site. Prior to this final construction phase there was at least one other major recon- struction episode (see Excavation Results, below). Before these formal constructions, the population most likely lived on unaltered natural hills. The east mound is largely natural except for the west leg of the horseshoe, which was built with artificial fill. In contrast, the east leg of the U-shaped mound appears to have been excavated out of an aborig- inally larger hill area. This aboriginal excavation was necessary to create the desired architectural shape. The west mound was composed of artificial fill, although a substantial natural hill is evident from the distribution of subsoil sections. Several chulpas or aboveground slab-cist tombs are found on the east mound. These burial features are post-Tiwanaku phenomena. All are looted and do not appear to be contemporary with the cor- porate architecture on the site. This we deduced from the tombs' locations either on the top and off"-center on the mound or on the edge of the easternmost leg of the horseshoe. They are not found on the west mound. The chulpas are most likely the remains of a small post-Tiwanaku oc- cupation. Hundreds of similar aboveground tombs are found throughout the Juli-Pomata region. Systematic Surface Collections— The size and nature of settlement complexity of the For- mative and Tiwanaku periods in the Titicaca Ba- sin have rarely been addressed in a systematic manner. Complexity refers to differences in elite versus nonelite residences, the nature of corporate architecture, areas of economic specialization, and so on. Pucara is reported to have a habitation zone adjacent to the public architectural area on the mounds (Rowe 1963). The nature of the occupa- tion area around the Chiripa mound has never been published, but one of us (Stanish) has ob- served at least three large domestic terraces below the formal mound area toward the lake. Chiripa therefore has both a section with corporate archi- tecture in the mound area proper and a substantial area of domestic residence located directly below the mound. Little is known about the nature of residential or nonresidential occupation away from the architectural center of other sites in the region. The Tiwanaku site of Lukurmata in Bolivia has been systematically surface-collected using the systematic, aligned sampling technique (Stanish 1989b: 45). The surface density map indicates a huge site (1.5 km^) and a substantial variation in 14 Part One habitation intensity. A similar method was ap- plied to Tumatumani. Analysis of the systematic collections from Tumatumani provides a trend surface characterization of the settlement area and beyond. The fact that the mounds were built with cultural fill certainly prejudices interpretation of these surface data, since the existence of artifacts does not necessarily indicate that residential ac- tivities took place in that collection area. On the other hand, the absence of remains is a good in- dication that there was no habitation or site use other than possibly agricultural use. The total dis- tribution of the surface artifacts, therefore, serves to define the absolute limits of site activity. Our interpretations of the surface data are that Tumatumani was a substantial site in area and intensity of occupation. The two principal habi- tation areas total 5.3 ha, based on ceramic and other artifact distribution data. Tumatumani is therefore the largest Type 1 site in the Juli-Pomata area. Each mound occupies a roughly equal area, although the west mound has a substantially great- er concentration of artifacts. This probably reflects the fact that the west mound is built with a much greater quantity of artificial fill. The east mound, in contrast, more closely conforms to the natural contours of the original or natural hill. The distribution of Early Sillumocco pottery on the mounds of Tumatumani suggests that both were occupied early and continued to be used si- multaneously throughout the major occupations of the site. Since the fiber-tempered Sillumocco ceramics are associated with one of the oldest of the Formative cultures, we can conclude that the site was founded early in the Titicaca Basin se- quence. The most unexpected result of the surface in- vestigations was the complete absence of occu- pational debris away from the mound areas. All residential settlement appeared to have been con- centrated on the high areas; the lower sections be- low the mounds were virtually free of pre-Hispanic artifacts. This pattern was corroborated by the ex- amination of several disturbed areas in these low areas. A road cuts through the site area, for in- stance, and exposes a 50-cm-deep section through- out its entire length. There is no midden accu- mulation in this exposure. Likewise, adobe bricks in a small modem structure south of the site do not have any artifacts incorporated into their clay. The bricks were made from clay obtained im- mediately adjacent to the structure and therefore reflect the nature of the surface and immediate subsurface materials. In contrast to these "clean" bricks, those in other modem stmctures in the Juli area are full of pre-Hispanic artifacts. The systematic surface data indicate that all oc- cupation of the site was on the mounds proper. There is no indication whatsoever of a distinct residential or special-use occupation off the mounds. This is particularly interesting in light of the fact that the mounds were faced with stone walls. The evidence would therefore be consistent with the results from Chiripa, where residential structures were built directly on the mound both before and after the formal constmction of the mound (Kidder 1956; Chavez 1988). Excavation Results— Five excavation units were placed on the lower terrace face on the east side of the west mound (Fig. 1 1). The excavations did not reveal any undisturbed, stratified midden. Rather, we defined three major, pre-Inca occu- pational periods— an initial occupation plus two formal architectural reconstructions— on this ter- raced platform. Because of the unexpected com- plexity of the archaeological deposition in these units, we were able to complete only five relatively shallow units over a 2-month period. We were able to reach subsoil in one of the units. We think it likely that the depth of subsoil in the one unit is similar to that in the adjacent units. Analysis of diagnostic ceramics in the excava- tion levels permitted us to date the initial occu- pation of the site. The detailed results of this anal- ysis by Steadman are given in Part 2. The initial occupation is represented by Quluyu- and Chiri pa- related styles that date to approximately 800 B.C. Although the Early Sillumocco ceramics were re- covered from midden fill, which could have been brought from elsewhere for the construction of the platforms, the quantity of these ceramics and their extensive surface distribution at Tumatumani strongly suggest that the Early Sillumocco occu- pation was located at the site itself No formal architecture associated with this Early Sillumocco occupation was found in our excavations. The earliest formal architectural construction in this terrace area was represented by a stone-faced platform, a well-made clay floor, and several buri- als in or under the wall stones. The floor surface was clearly evident in the profile (denoted "intact surface" in Fig. 12) and was directly on top of the fill of the first construction. It was the same height as the top of the wall stones. The wall was used to support a platform, which in tum was con- structed with a massive quantity of redeposited Background to the Tumatumani Investigations 1 5 UNira 3CI^ B A L K 1 1 I'i'i'i'T !i!r]i!i!i '! UNTTl UNTr2 B A L • • : • i i-i vMt'tM,^ R '— hr^' .... B A L R ..•^7;^ :i^^^^ \ PLATFORM UNIT 5 WALL FACE UNIT4 Fig. 1 1 . Five excavation units on the west mound. midden. In other words, the inhabitants of Tu- matumani used existing midden as fill to construct a stone-faced platform surface. The platform as- sociated with this first construction was built over a natural subsoil. This platform may have been used as a residential area, a proposition supported by the absence of remains away from the mound area proper. This construction episode is repre- sented by the arbitrary excavation levels 9-19 in Unit 1 in Figure 1 2 and the natural levels below the floor in Figure 13. At least three human burials, two of which were excavated, were found underneath and inside the stone-faced wall (Figs. 12 and 13). The two ex- cavated burials were of an adult and child, in fetal positions, interred in unlined cists. The osteolog- ical remains were poorly preserved and could not be sexed. The bodies appeared to have been placed directly adjacent to or inside the wall. Part of the subsoil (and therefore a section below the wall itself) was excavated out to accommodate the bur- ials. Curiously, there were no grave goods asso- ciated with the burials, a pattern that Bennett ( 1 936: 431-436) found in many of his excavated burials at Chiripa. The burials, wall, and underlying fill were con- temporary and related to each other. The burials may have been sacrificial offerings of an adult and child at the base of a large terrace face, although this remains unproven. What is suggested from the stratigraphic relationships between the wall, subsoil, and the burials is that the bodies would have been interred before or during the construc- tion of the wall. They could not have been placed after the wall was constructed without removing the wall stones and replacing them, an unlikely burial practice. We can date the platform building episode by analysis of the ceramic diagnostics in the fill con- struction. The youngest nonintrusive sherds in the terrace provide us with a terminus post quern or later-than-or-equal-to date for the building period. The youngest sherds in the first constructions lev- els are Late Sillumocco in date. There are no un- equivocal Qeya diagnostics, although the total sample is small. This indicates that the construc- tion would have to be Late Sillumocco or post- Late Sillumocco in date. The dating of the second construction episode to the Tiwanaku IV period, however (see next paragraph), firmly places the 16 Part One UNIT1 UNIT 2 UNIT 4 ■^ — a ^^—r-^ V^__i -J — */ » ■ " UNIT 5 i ARBRITRARY LEVELS 1 M I« NO COMSTHUCnOM MTACT SURFACE Ti ■ ■ '_i tT coNsmucnoN -It. .1»_ UNEXCAVATED NATUML tUnON. UNIT 1 ENLARGEMENT Fig. 12. Arbitrary levels of excavation units. construction of the lower platform in Late Sillu- mocco times, between 200 B.C. and a.d. 400. The lack of Qeya diagnostics in this first construction episode may indicate that construction occurred before a.d. 200 (the oldest date for Qeya), al- though we refrain from making this conclusion with negative evidence. A second, later reconstruction of the terrace was identified in the excavations. This construction was represented by younger redeposited fill over the earlier terrace wall and intact floor surface (lev- els 1-8 in Figure 12 and noted as "2nd construc- tion"). Like the earlier fill episode, it was com- pletely homogeneous, with no indication of stratified levels. This fill contained artifacts similar to the earlier construction, with the addition of Tiwanaku IV ceramics (and some Inca intrusives). The second construction must therefore be Ti- wanaku IV or later in date. However, an Inca structure was uncovered in the upp>er levels of Unit 3 which is clearly intrusive into the second con- struction fill, indicating that the second construc- tion episode predated the Late Horizon. Since no securely identified Late Intermediate period ce- ramics were found at Tumatumani, either in the excavations or on the surface, the most likely date for the second construction is the Tiwanaku pe- riod, ca. A.D. 440-750. Two profiles from disturbed areas on the upper artificial terrace were cleaned in order to compare this pattern first seen in the formal excavation areas. These terraces stratigraphically correspond to the second construction period (levels 1-9) identified in the excavations. These profiles are similar to the excavation profiles that displayed an unconsolidated fill matrix with evidence of mottled subsoil mixed with midden. The subsoil appears to have been thrown in as a basketload of clean earth. In one of the profiles we noted a lens of clean yellow subsoil that slanted down and away from the center of the mound. At Chiripa, Bennett discovered yellow and red clay floors of about the same thickness as the lens in our profile. However, the floor surface that Bennett (1936: 426) discov- ered sloped toward the temple at the center of the mound. Although the slopes of the clay floors at Background to the Tumatumani Investigations 1 7 100 150 Fig. 13. Natural levels of Unit 1, west wall. the two sites differ, it is significant that each site shares this architectural feature. Each profile contained Sillumocco and Tiwa- naku period artifacts. No certain post-Tiwanaku fragments were discovered. The pattern first seen in the excavations is corroborated by the profile cleanings: the second major occupation was char- acterized by a rebuilding of the terrace with extant midden from the immediate site area. The youn- gest nonintrusive materials in the fill were Tiwa- naku, the period in which construction of the sec- ond terrace most likely took place. The Inca intrusives found in the upper levels of Units 1 and 3 were associated with an intrusive structure found principally in Unit 3. Only a few Late Horizon fragments were found in the rest of the excavations. There was a hearth associated with a concentration of wall stones in Unit 3. Based on the presence of a number of Late Horizon ce- ramic fragments in the upper levels, we interpreted the hearth and wall stones as a destroyed Inca period house foundation. One of the Late Horizon intrusives found in this excavation is virtually identical to a fragment illustrated by Ryden (1 947: 2 1 7, Fig. 89, fragment "b") that he recovered from Palli Marca in Bolivia. Late Horizon settlements are found throughout the Juli area (Stanish et al. 1 992). The house foun- dation location in the excavations most likely rep- resents a small Late Horizon occupation contem- porary with the large town site of Juli (Hyslop 1976) and smaller settlements in the immediate area near the lake. The Late Horizon intrusives are restricted in area and quantity throughout the excavations. In total, the post-Tiwanaku occupa- tions on the site are quite small compared to the Tiwanaku and Sillumocco occupations. Part One Part 2 Analyses of Artifacts from Tumatumani CERAMIC ARTIFACTS Lee Steadman The excavation data from Tumatumani presented a number of problems and limitations for the study of ceramics from the site. Because excavations re- vealed only construction fill episodes, a detailed level-by-level analysis of the Tumatumani ceram- ics was not warranted. Instead, after an initial clas- sification of the ceramics by level, the data were grouped into larger first construction episode and second construction episode samples for further analysis. Only the ceramics from Unit 1 were selected for study, for this unit was the deepest, and the only unit excavated into the first construction fill. Lev- els 9-1 9, plus Feature 2 (a burial) and the fill from the lower terrace wall, compose the first construc- tion levels of Unit 1 , while the second construction fill consists of levels 2-8. Level 1, located in the modem plow zone, was not included in the sam- ple. The total sample of body and rim sherds from the Unit 1 excavations was 4,970. Information from the Unit 1 excavations was supplemented by an analysis of selected sherds from surface collec- tions on the west mound and the material from excavation Units 2-5. All decorated Tiwanaku and pre-Tiwanaku sherds were chosen from this sam- ple for further analysis, as well as any undecorated sherds that supplemented the shape categories for the plainwares defined from the Unit 1 data. A total of 204 sherds was added to the excavation sample from these sources. Based as they are on data from fill excavations, conclusions about the Tumatumani ceramic se- quence rely both on internal evidence and on com- parisons with other previously defined ceramic chronologies elsewhere in the Lake Titicaca Basin. Internal evidence for the ordering of the ceramic material consists of the separation that could be made between the first and second construction episode fills. Based on the identifiable decorated sherds found in the first construction fill, the con- struction of the lower platform at Tumatumani was dated to the late Upper Formative period (Late Sillumocco). All ceramics found in this fill were therefore known to be late Upper Formative or earlier in date. Material found in the second con- struction fill, dated on the basis of identifiable sherds to the Tiwanaku period, was known to be a mix of Tiwanaku and pre-Tiwanaku ceramics. Any material found for the first time in the second construction fill and not found in the first con- struction sample could then be identified as per- taining specifically to the Tiwanaku period. The possibilities of error are obvious, for sampling er- ror might result in a particular attribute not being found in the first construction levels to which it actually corresponds. Difficulties also arose in dis- tinguishing possible cases of continuity, in which ceramics found in the first construction levels con- tinued to be made in much the same form and finish into Tiwanaku times. Refinement of the ar- chaeological evidence, incorporation of the sur- face-collected materials, and further subdivision of the larger pre- and post-Tiwanaku time frames were therefore based on visible differences within the assemblage itself and on comparisons of the Tumatumani material with other excavated Titi- caca Basin ceramic chronologies. Despite the limitations of the data, the Tuma- tumani sample provides a unique opportunity to study the ceramic traditions of the southwestern Titicaca Basin. Both local and imported examples of the major cultural styles of the Formative and Tiwanaku periods in the Titicaca Basin were found at the site, as were extensive samples of undecorat- ed plainware assemblages. Because chronological statements concerning the Tumatumani sequence Analyses of Ceramic Artifacts 1 9 ^ « E Os(NO r^ — >/^ (Sr- _ Tt — < ON >r> — < — (N r~- nvo^~00l?^ — (Nrn are still tentative, and because data cannot be pre- sented in a period by period format, organization of the ceramic data proceeds with an emphasis on ceramic wares. After an initial description of the pastes and surface finish categories, the four major plainware assemblages are described. Decorated wares, drawn mostly from surface collections, are then presented and discussed in roughly chrono- logical order. Ceramic Pastes Paste classification of the Tumatumani ceramics was undertaken using a lOx hand lens. Charac- teristics such as the density and texture of the clay body were noted, as well as the type of nonplastic inclusions present in the paste. Petrographic anal- yses have not been conducted on the Tumatumani material, so inclusions are described by their size, color, and luster only. Inclusions were measured with a reticule, and their shape and size are given with respect to the Wentworth scale. Inclusion density is represented by a count of the number of inclusions visible in a reticule grid of 5 x 5 mm. The inclusions of each paste are listed in descending order of abundance, and comparative information between pastes can be obtained through reference to the 5 x 5-mm figures. All the sherds from the Tumatumani Unit 1 excavations, plus the selected surface and Units 2-5 samples, were analyzed with respect to paste and finish categories. Table 1 lists the number of sherds of each paste by level in the Unit 1 fill excavations. Table 2 summarizes each paste's per- centage of the total sherd assemblage grouped into the larger first and second construction episode level divisions of this same fill. The temporal dis- tribution of each paste, at least as found at Tu- matumani, is given in Table 3. Pastes were dated based on their association with identifiable dec- orated ceramics and on evidence from finishes and plainware shapes. Dotted lines in Table 3 indicate uncertainties, which are explained in the text. Only Pastes 1,3,5, and 9 yielded a large enough sample to merit a complete discussion of plain- ware vessel shapes and the finish and firing tech- nologies particular to those shapes. The presen- tation of Pastes 1, 3, 5, and 9 in this section is therefore limited to a description of the paste itself and basic firing technologies. Further information on the firing and finish of vessels in these pastes can be found in the discussions of their respective 20 Part Two undecorated ceramics. The samples for Pastes 2, 4, 6, 7, 8, 10, 11, 12, and 13, in contrast, are smaller and consist mostly of painted or incised ceramics. There is no separate section on plain- wares for these pastes; a description of their firing and finish techniques has been incorporated into the paste and finish category sections that follow, and is also summarized in Table 4. Paste 1 The nonplastic inclusions in Paste 1 are (1) abundant opaque or matte white inclusions, sub- rounded and subangular form, medium size; (2) abundant lustrous and often translucent white in- clusions, subangular form, medium size; (3) rela- tively abundant matte black inclusions, subangular and angular form, fine size; (4) relatively abundant gold and silver mica flakes, fine and medium size; and (5) very scarce matte red inclusions, sub- rounded form, very fine and fine size. Density is 25-30 inclusions per 5x5 mm. Table 2 shows the overwhelmingly large per- centage that Paste 1 commands in the total Tu- matumani excavation assemblage. The manufac- ture of Paste 1 sherds also shows a remarkable persistence through time: sherds of this paste are found at Tumatumani dating from the Early Sillu- mocco through Tiwanaku periods. Paste 1 ceram- ics are believed to have been made locally in the Juli area. Many factors, of course, must still be examined to substantiate this conclusion. No di- rect evidence of pottery manufacture has been identified at Tumatumani itself A mineralogical analysis of Paste 1 , location of the clay and temper sources used in its manufacture, and more infor- mation on the geographic distribution of Paste 1 vessels would also contribute to identification of their production center. However, all information presently available points to the local manufacture of Paste 1 ceramics. This deduction is based on a number of characteristics exhibited by the Paste 1 assemblage. One is the sheer number of Paste 1 sherds found at Tumatumani, particularly the large number of utilitarian wares. Other factors being equal, bulky, low-prestige items, such as utilitarian pottery, do not travel far from their centers of production (Rice 1987: 200; Hodder 1974). Al- though any number of factors could result in the importation of some utilitarian vessels, the large quantity of Paste 1 utilitarian ceramics and the fact that they form the bulk of all utilitarian wares found at the site from the Upper Formative and Table 2. Percentage of paste types by construction episode. First Second Paste construction construction number (% of total) (% of total) 1 92.27 81.05 2 0.53 0.43 3 5.15 12.12 4 0.07 0.17 5 1.39 3.27 6 0.13 0.20 7 0.13 0.26 8 0.09 9 0.26 1.48 11 0.17 12 0.07 0.12 13 0.64 Total 100 100 Tiwanaku periods suggest that they were locally manufactured. The variability of Paste 1 ceramics provides fur- ther evidence for local production. Technological variability is visible in the lack of standardization in Paste 1 firing practices. Paste 1 vessels show evidence of being fired in a wide variety of at- mospheric conditions and temperatures, which vary within one vessel category and even within one vessel surface. Finish also takes many forms and again is not consistent even within one vessel shape category. Paste 1 itself shows more variation than any of the other pastes at Tumatumani, with more examples having a larger proportion or dif- ferent sizes of various nonplastic inclusions in the paste. This lack of standardization in the Paste 1 assemblage suggests production at the household level rather than manufacture by specialized pot- ters. Finally, the Paste 1 utilitarian wares show the widest range of shapes and vessel attributes of any of the assemblages at Tumatumani, which again suggests they were not a class of sp)ecialized im- ported wares but were manufactured on a local or regional basis. Paste 1 is characterized by a predominance of matte and lustrous white inclusions. It is a com- pact, or often semicompact, paste with a medium texture and a porous, somewhat laminated ap- pearance. The relatively abundant air holes are both rounded and lenticular in shape. It would appear to be a friable paste, since the break on a freshly snipped sherd is slightly jagged with a peaked, craggy look under the hand lens, but in fact it has relatively good cohesion and cannot be crumbled with a fingernail. In the large sample of Analyses of Ceramic Artifacts 2 1 Table 3. Tumatumani paste distribution by period. Paste number 1 2 3456789 10 11 12 13 Early Sillumocco ! Late Sillumocco 1 Tiwanaku 1 1 1 Late Intermediate period Late Horizon i — Y Paste 1 sherds, certain variations on the standard paste composition were found. These variations include specimens with a higher overall density of inclusions (30-35 per 5x5 mm), a higher density of mica with respect to other nonplastics in the paste, or sherds that have coarse rather than me- dium matte and lustrous white inclusions. The matte red inclusions in all Paste 1 specimens are very scarce and often are not found in the break of a small sherd. Firing technologies for the Paste 1 wares are varied. Vessels fired in a final oxidizing atmo- sphere have a red-brown paste color, in the Mun- sell color range of 2.5YR 3.5-5/6, or 4/4. The red color of oxidized Tiwanaku and post-Tiwanaku ceramics tends to be slightly brighter than the ear- lier wares. Oxidation of these red-brown vessels often was not complete, leaving a gray central core. Occasionally the fired paste color of Paste 1 sherds is a uniform light brown or light brown with a darker core. Since oxidized Paste 1 sherds do achieve a red-brown color, these light brown ex- amples are believed to be the result of incomplete oxidation rather than representative of a clay that naturally achieves a light brown color at full ox- idation. Rare in the Paste 1 ceramics, although common in the fiber-tempered wares, is the pro- duction of a red-brown or light brown color on the surface of the sherd through a brief period of final oxidation, achieved by pulling the vessel from the fire to cool in the outside air. This technique pro- duces a sharp line between the lighter interior and exterior edges and the dark core (Rye 1981: 117), rather than the diffuse margins of the incompletely oxidized wares. Approximately 60% of the sherds in the total Paste 1 sample have a black, dark gray, gray, or dark brown paste color. The gray and black colors were apparently achieved by firing in a final re- ducing atmosphere or by smudging, a procedure in which the fire is smothered with organic ma- terial at the end of the firing period, causing a deposition of carbon on the surface and in the pores of the vessel (Rice 1987: 335). The gray and dark gray wares appear to have been only super- ficially reduced or smudged, for the gray surface is uneven and rarely penetrates the entire vessel; more common is a red-brown core with darker interior and exterior edges. Frequently these edges are so thin as to appear as gray or dark gray "rinds" on an otherwise red-brown paste. The blackwares, with their deep black surfaces, were almost cer- tainly intentionally smudged, and the black color is usually found throughout the entire sherd cross- section. The uniformly dark brown examples could be the result of smudging or incomplete oxidation; refiring tests would need to be undertaken to de- termine this. Sherds suggesting the use of both techniques have been found in the sample. In gen- eral, there is a notable lack of standardization in the Paste 1 firing, with many different techniques used, even within one vessel category. Variations on a single sherd, where only the interior or ex- terior of a vessel is reduced or incompletely oxi- dized and the other face is oxidized, are also com- mon. 22 Part Two Paste 2 The nonplastic inclusions in Paste 2 consist of ( 1 ) abundant matte white inclusions, subrounded and subangular form, fine size; (2) abundant trans- lucent and lustrous white inclusions, subrounded and subangular form, fine size; (3) relatively abun- dant matte black inclusions, subangular form, very fine and fine size; (4) relatively abundant gold and silver mica flakes, very fine and fine size; and (5) very scarce matte red inclusions, subrounded form, very fine and fine size. Density is 20-25 inclusions per 5x5 mm. Paste 2 is a finer version of Paste 1 . It has the same type of inclusions occurring in the same pro- portional mix, but of considerably smaller size. The paste is compact and fine-textured. The break of a freshly snipped sherd is smooth, and the few air holes visible are round in shape. Most of the Paste 2 examples are oxidized to a red-brown, incompletely oxidized with a gray core, or incom- pletely oxidized to a light brown, in the same range of colors as the Paste 1 ceramics. Reduction or smudging is very rare in firing, and except for one example of a Tiwanaku black kero rim, is always superficial. Almost all Paste 2 sherds are from dec- orated or finely finished vessels and are burnished on both sides to a medium or high luster, mostly with a fine category 7 burnish. Monochrome- slipp>ed wares are predominantly red. Only Late Sillumocco (Figs. 127, 153-154) and Tiwanaku Paste 2 sherds have been found, with no examples known from the Early Sillumocco or post-Tiwa- naku periods. Because Paste 2 is so similar to Paste 1 , and because macroscopic examination suggests that the clay and temper used to make this paste were taken from the same sources as Paste 1 , Paste 2 ceramics are also believed to have been manu- factured locally in the Juli area. Paste 3 Nonplastic inclusions in Paste 3 are (1) abun- dant matte white inclusions, subangular and sub- rounded form, fine and medium size; (2) scarce matte black inclusions, subrounded form, fine size; (3) very scarce lustrous white inclusions, suban- gular and subrounded form, medium size; (4) very scarce matte red inclusions, subrounded form, fine size; and (5) very scarce silver and gold mica flakes, fine size. Density is 30-35 inclusions per 5x5 mm. Paste 3 is characterized by an abundance of matte white inclusions. These are visible to the naked eye as dense white grains in the sherd cross-sec- tion. It is a compact and medium-textured paste. The break is even, without the friable appearance of Paste 1 . Air holes tend to be round rather than laminated. Paste 3 is fairly standard in composi- tion, and only one variation, containing fine matte white inclusions rather than medium, was found to occur. Slightly more than half of all Paste 3 sherds were fired in an oxidizing atmosphere, resulting in a red-brown paste with a value of 2.5YR 4/6-7. Ti- wanaku period oxidized wares tend to be of a brighter red or orange color, closer to lOR 4/8 or 2.5YR 5/8. Incompletely oxidized Paste 3 sherds have a brown, orange-brown, or gray core with red-brown or light brown edges, or are a uniform light brown or orange-brown (5YR 5-6/5, 7.5YR 5.5/4 or 6/6). Only a very small number of Paste 3 sherds have cooled red-brown or light brown surfaces. Almost all diagnostic rim and decorated Paste 3 sherds are oxidized or incompletely oxi- dized. Incompletely oxidized jars, bowls, and Ti- wanaku period wares tend to have dark cores, while incompletely oxidized decorated Late Sillumocco ceramics are more likely to be a uniform light brown. Approximately 47% of Paste 3 sherds have pastes with dark brown or gray interior and ex- terior edges and red-brown, orange-brown, or brown cores. Rarely does a uniform gray paste occur. Deep black exterior surfaces and uniform black pastes, common in the Paste 1 wares, are not found in Paste 3. Paste 3 is the second most commonly occurring paste at Tumatumani, forming 5% of the sample in the first construction episode fill and 1 2% in the second construction episode fill. Paste 3 ceramics found at Tumatumani date from the Late Sillu- mocco and Tiwanaku periods; diagnostic sherds include a small selection of Late Sillumocco and Tiwanaku utilitarian shapes and Late Sillumocco polychrome painted and incised specimens. Given the small overall percentage of Paste 3 in the total Tumatumani sample, particularly as contrasted with the Paste 1 figures, one can hypothesize either that these vessels were made elsewhere and arrived at Tumatumani only in limited quantities, or that they were manufactured locally on a small scale and only for certain vessel shapes. However, the fundamental difference in the nonplastic inclu- sions of Paste 3 from the locally made Paste 1 wares, as well as the difference in rim shapes, points more to a nonlocal origin, as yet unidentified. It is unlikely that the manufacturing center for Paste Analyses of Ceramic Artifacts 23 3 ceramics was located a great distance from the Juli area, however, since undecorated utilitarian jars and bowls did make their way to the site, and Paste 3 is referred to in the remainder of this report as a semilocal paste. Paste 4 Nonplastic inclusions in Paste 4 consist of (1) abundant matte black inclusions, angular and sub- angular form, medium and some coarse size; (2) relatively abundant matte white inclusions, sub- rounded and subangular form, medium size; (3) relatively abundant lustrous white inclusions, sub- rounded and subangular form, medium size; and (4) very scarce mica flakes, silver only, very fine size. Density is 30-35 inclusions per 5x5 mm. Paste 4 is distinguished by a predominance of angular matte black inclusions. It is a compact, medium-textured paste. The break is slightly ir- regular with a peaked appearance but cannot be chipped or crumbled with a fingernail. Air holes are both round and laminated, although relatively few. In overall appearance the texture of this paste resembles that of Paste 1 . Firing was most com- monly in an oxidizing atmosphere, producing a red-brown color in the Munsell range 2. SYR 3-4/ 5-6, although superficial dark brown smudging also occurs. Body sherds are all unslipped. Diagnostic Paste 4 sherds are few and include only a painted Qaluyu-related rim (Fig. 1 04) and a light brown- slipped jar neck with an applique incised fillet lo- cated just below the rim (Fig. 14). This decorative technique is considerably different from the fillets on Pastes 1 and 3 jars, which occur at the joint between neck and body. The small number of Paste 4 sherds found at Tumatumani, combined with this difference in jar decoration, suggests a non- local origin for this paste. Little else can be said about Paste 4 at the present time. Its existence in the Early Sillumocco period is evidenced by the Qaluyu rim, while the three body sherds with Ti- wanaku and post-Tiwanaku finish category 1 1 sur- faces (see Table 4) seem to indicate a long period of production. Any conclusions based on three body sherds must be tentative, and a larger sample of this paste type is needed before further analysis can be made. Paste 5 Nonplastic inclusions in this paste are ( 1 ) abun- dant matte white inclusions, subrounded and sub- angular form, medium and occasionally coarse size; (2) abundant lustrous and translucent white inclu- sions, subangular form, medium and occasionally coarse size; (3) relatively abundant matte black inclusions, subangular and angular form, fine size; (4) relatively abundant gold and silver mica flakes, fine and medium size; (5) relatively abundant fiber fragments; and (6) very scarce matte red inclu- sions, subrounded form, fine and medium size. Density is 25-30 inclusions per 5x5 mm. Pastes 5 through 9 are the fiber-tempered pastes at Tumatumani. All of the fiber-tempered pastes, with the exception of Paste 9, have counterparts in the non-fiber-tempered group: i.e., the charac- teristics of paste texture and nonplastic inclusions are virtually identical, with the differences occur- ring only in the presence or absence of organic temper. This similarity suggests that these pastes were manufactured in the same general region as their non-fiber-tempered equivalents, although this cannot be proved with the data available. Paste 5, because of the marked similarity of its clay body and nonorganic temper to those features of Paste 1, is also believed to have been manufactured in the Juli area. It shows the same persistence through time, a numerical advantage over the other fiber- tempered pastes, particularly in utilitarian ceram- ics, and the same variability in vessel shapes and finishes that Paste 1 demonstrated in the non-fi- ber-tempered group. In addition, one category of Paste 5 undecorated ceramics was found to be virtually identical in shape and finish to contem- porary Paste 1 utilitarian wares. Paste 5 has the same characteristically abundant matte and translucent white inclusions as Paste 1 . It is a semicompact, medium-textured paste with a tendency toward laminated fractures. The break is slightly jagged and somewhat friable, and can be chipped with a fingernail. The fiber in Paste 5, as in all the fiber-tempered pastes described here, is represented by long narrow voids, which in many cases still bear the impressions of the grass seg- ments once present in the unfired paste. These fiber voids appear in cross-section in preferred ori- entation, parallel to the surface of the sherd. Paste 5 was used in the Early Sillumocco period for the production of Chiripa-style incised and painted ceramics as well as monochrome-slipped jars and bowls. Most Paste 5 sherds of this time period have a red-brown or light brown surface color, usually achieved through cooling of the vessel and leaving a gray core, although incompletely oxi- dized examples are also found. Reduced or smudged wares have light brown or brown pastes 24 Part Two Table 4 Finish categories in Unit 1 excavations. by construction episode. Con- struc- Total Paste num- tion epi- . sode Finish category Total number of ber 1 2 3 4 5 6 7 8 9 10 11 12 % sherds 1 1st 16.5 1.3 15.1 7.4 38.7 21.0 100.0 1,359 2nd 21.4 0.6 2.3 2.0 35.1 25.9 0.7 1.7 7.9 2.4 100.0 2,763 2 1st 25.0 37.5 37.5 100.0 8 2nd 6.7 6.7 13.3 6.7 40.0 6.7 20.0 100.1 15 3 1st 37.7 3.9 2.6 9.1 3.9 3.9 38.9 100.0 77 2nd 44.5 0.5 0.2 18.4 0.5 1.0 23.6 0.7 10.5 99.9 407 4 1st 2nd 100.0 16.7 33.3 50.0 100.0 100.0 1 6 5 1st 23.8 28.6 9.5 38.1 100.0 21 2nd 4.5 12.5 1.8 77.7 3.6 100.1 112 6 1st 2nd 14.3 100.0 85.7 100.0 100.0 2 7 7 1st 2nd 50.0 50.0 22.2 77.8 100.0 100.0 2 9 8 1st 2nd 100.0 100.0 0 3 9 1st 2nd 2.0 100.0 98.0 100.0 100.0 2 51 11 1st 2nd 40.0 40.0 20.0 100.0 0 5 12 1st 2nd 100.0 75.0 25.0 100.0 100.0 1 4 13 1st 0 2nd 4.8 9.5 4.8 19.0 57.1 4.8 100.0 21 with dark brown or gray inteiior and exterior edges and cores. A significant variation on the standard Paste 5 is a paste that, although the same in all other re- spects, contains a smaller quantity of fiber temper. No fiber voids are visible on the surface of the sherds, and without careful examination of the paste with a hand lens the pieces often appear to be manufactured from Paste 1 . This reduced-fiber version of Paste 5 forms 25% of the total Paste 5 sample. Firing technologies are the same as for the Paste 1 wares, with a roughly equal distribution between reduced and oxidized firing, the former producing red-brown pastes with black interior and exterior edges, the latter producing red-brown or light brown pastes, red-brown pastes with brown cores, or pastes with cooled red-brown edges. In contrast to the standard Paste 5 sherds, which are all finished with burnishes from the exclusively fiber- tempered finish category 8, the reduced-fiber sherds have category 1, 5, 6, and 1 1 finishes, fin- ishes found more commonly on the non-fiber- tempered pastes. Production of this reduced-fiber version of Paste 5 began in the Early Sillumocco period. Its use at this time was restricted to a small number of Qaluyu-style ceramics, and it was pro- duced alongside the more commonly found high- fiber version of Paste 5 in this period. Although production of the high-fiber version apparently ceased at the end of the Early Sillumocco period, the reduced-fiber version of this paste continued to be used through Late Sillumocco times in the production of domestic wares, and into the Ti- wanaku occupation of the site for the manufacture of polychrome Tiwanaku ceramics. Paste 6 The nonplastic inclusions of this paste are (1) abundant matte white inclusions, subrounded and subangular form, fine size; (2) abundant translu- cent white inclusions, subangular and subrounded form, fine size; (3) relatively abundant matte black inclusions, subrounded form, very fine and fine size; (4) relatively abundant gold and silver mica flakes, very fine and fine size; (5) scarce fiber frag- ments; and (6) very scarce matte red inclusions. Analyses of Ceramic Artifacts 25 subangular form, fine size. Density is 20-25 in- clusions per 5x5 mm. Paste 6 is best described as a Paste 2 with fiber temper. It is a compact, fine-textured paste. The break is smooth and the air holes tend to be round and few in number. The quantity of fiber itself is relatively small. Firing techniques for Paste 6 in- cluded both reduction or smudging, to produce black and dark brown pastes, and cooling, to pro- duce pastes with dark cores and sharp red-brown edges. Decorated wares were generally cooled. No diagnostic sherds that could be identified as Paste 6 domestic wares have been found; like Paste 2, Paste 6 appears to have been used more often for the production of fancy or decorated ceramics. All sherds are burnished, and Paste 6, of all the fiber- tempered pastes, has the highest percentage of sherds burnished on both sides. The majority of pieces are slipped dark red or brown (7.5R 3/4-7, lOR 3/6, 5YR 3/3), and the total number of sherds is predictably low. Paste 6 is the last of the pastes believed to have been manufactured in the Juli area, although nei- ther overall numbers nor variety and quantity of utilitarian wares can be used to support this con- clusion. As for Paste 2, the inclusion of Paste 6 in the 1/2/5/6 local group of pastes is based on the similarities of its clay body and nonplastic inclu- sions to other pastes in the group, which suggests that the raw materials were obtained from the same source and that the pottery was therefore manu- factured in the same production area. The man- ufacture of Paste 6 appears to have been limited to the Early Sillumocco period; of the three di- agnostic decorated sherds found, one is a local version of a Qaluyu painted type (Fig. 103) and the remaining two have Chiripa-related incised or yellow-on-red painted decoration (Figs. 113, 118). Paste 7 Nonplastic inclusions found in Paste 7 are (1) abundant matte white inclusions, subangular and subrounded form, fine and medium size; (2) rel- atively abundant lustrous white inclusions, sub- angular and subrounded form, medium size; (3) scarce matte black inclusions, subrounded form, fine size; (4) scarce fiber fragments; (5) very scarce matte red inclusions, subrounded form, fine size; and (6) very scarce gold and silver mica flakes, very fine and fine size. Density is 30-35 inclusions per 5x5 mm. Paste 7 is very similar to Paste 3, having the same abundant white inclusions visible to the na- ked eye. It differs only in the addition of fiber and in the higher quantity of lustrous white inclusions in relation to the other nonplastics. It is a compact and medium-textured paste. The break is even, with only a few small, round air holes. Firing is usually reduced or smudged, producing black or dark brown paste colors; only a few of the Paste 7 examples have cooled red-brown edges. Most pieces are burnished, but a higher percentage of sherds than in Paste 6 or 9 are burnished on one side only. About half the sample is slipped in red (lOR 4/6) or red-brown (2.5YR 4/6) tones; brown slips are rare. Undecorated or monochrome-slipped Paste 7 ceramics found at Tumatumani include only four rims representing two forms: a jar with a straight-sided, slightly everted neck (Figs. 15, 16), and a constricted-neck olla shape (Fig. 17). The only decorated sherd is a fragment of a ce- ramic tube, illustrated in Figure 1 28 and discussed in a later section. Both the undecorated rims and the ceramic tube fragment can be related to Chi- ripa ceramics. The production of Paste 7 therefore appears to be limited to the Early Sillumocco pe- riod exclusively. Paste 7 is probably not a local paste but, as for Paste 3, the small number of Paste 7 utilitarian shapes found at Tumatumani indi- cates that their center of production was probably not far from the Juli area. Paste 8 Characteristic inclusions of Paste 8 are ( 1 ) abun- dant matte black inclusions, subangular and an- gular form, medium and coarse size; (2) relatively abundant matte white inclusions, subrounded and subangular form, medium size; (3) scarce lustrous white inclusions, subrounded and subangular form, medium size; and (4) scarce fiber fragments. Den- sity is 25-30 inclusions per 5x5 mm. Paste 8, like Paste 4, has a noticeable predom- inance of angular matte black inclusions. It is semicompact and medium textured. The break is irregular, with a peaked appearance, and can be chipped with a fingernail. Air holes are fairly nu- merous and round in shape. The paste has no mica. Paste 8 is extremely rare at Tumatumani and is presumably nonlocal. Only three body sherds were found. These all have dark brown pastes and are burnished on one or both sides; two of the three specimens are slipped brown (7.5YR 5/3). Since no decorated or diagnostic sherds were found in the Paste 8 sample, the inclusion of this paste in 26 Part Two the Early Sillumocco ijeriod (Table 3) is based solely on the attribute of fiber temijer, which it shares with the other, more securely dated Early Sillumocco fiber-tempered pastes fi-om the site. ter of production lay not far from the local pro- duction area for the site. Paste 10 Paste 9 Paste 9 nonplastic inclusions are (1) abundant matte white inclusions, subrounded, subangular and angular form, fine, medium, and coarse size; (2) relatively abundant lustrous white inclusions, subangular form, medium size; (3) relatively abun- dant fiber fragments; (4) scarce matte black inclu- sions, subangular form, fine size; and (5) very scarce matte red inclusions, subangular and subrounded form, medium size. Density is 20-25 inclusions per 5x5 mm. Paste 9 is defined by its characteristic chunky white matte inclusions, plainly visible to the naked eye. The three size categories of these inclusions are found in equal proportions in the paste. Paste 9 is semicompact and of medium texture. The break is jagged, with a peaked appearance, and can be chipped with a fingernail. Air holes are fairly abundant and round in shape. Notably, the paste contains no mica. The fired paste color of Paste 9 vessels is divided about equally between reduced or smudged dark browns or blacks and dark cores with cooled red- brown or light brown edges. A small number of incompletely oxidized sherds with red-brown pastes and dark cores were found as well. Deco- rated ceramics all have red-brown surfaces, achieved either by cooling or by incomplete oxi- dation. The majority of sherds are burnished on one or both sides. Paste 9 sherds were found in sufficient quantities at Tumatumani to reflect at least part of the total Paste 9 ceramic assemblage, and both jars and open bowls, as well as Chiripa- style yellow-on-red and incised wares, were man- ufactured. No post-Early Sillumocco period ce- ramics appear to have been made. Paste 9 has no non-fiber-tempered counterpart in the Tumatu- mani pastes. Because of the fundamental differ- ence of its nonplastic inclusions from the Pastes 1/2/5/6 group, it is unlikely that Paste 9 was made locally. However, it does not appear to be an im- ported paste from the Chiripa area; Paste 9 Chi- ripa-style specimens actually bear less resem- blance to the type-site Chiripa ceramics than do the locally made Paste 5 and Paste 6 examples. As for Pastes 3 and 7, the number of Paste 9 utilitarian vessels found at Tumatumani suggests that its cen- Paste 10 inclusions are (1) relatively abundant matte white inclusions, subrounded and suban- gular form, very fine, fine, and occasionally me- dium size; and (2) scarce matte black inclusions, subangular form, very fine size. Density is 20 in- clusions per 5x5 mm. Paste 10 is a dense, very fine-textured paste. In cross-section, portions of the clay body have the appearance of being sintered or melted together. Paste 10 is the only paste with this characteristic found at Tumatumani. The break of a snipf)ed sherd is even, with a very few, small, laminate air holes visible. Paste 1 0 sherds are rarely completely oxidized; more common is an oxidized orange-red (2.5YR 5/7-8) with a light gray or brown core, or an oxidized exterior and reduced light gray inte- rior. This reduced edge can comprise up to half the cross-section of the sherd, in contrast to the thinner interior or exterior edges common to the other pastes at Tumatumani. No Paste 10 speci- mens were found in the Unit 1 excavations, so that the paste does not appear in Tables 1 or 2. Specimens are from the other excavation units (all in second construction levels) and from the sur- face. The Paste 1 0 specimens are all decorated and burnished and form a small homogeneous stylistic group within the total Tumatumani assemblage that can be related to the Qeya (Tiwanaku III or Early Tiahuanaco) painted ceramic tradition. Paste 1 0 is practically identical to the pastes of Tiwa- naku sherds from the site of Tiwanaku itself Dif- ferences lie in the fact that the Tiwanaku pastes have an even denser, more sintered look, tend to be fired to a more brick red color, and have a slightly higher quantity of temper, although com- posed of the same matte black and white inclu- sions as Paste 1 0. It is consequently believed that Paste 10 was imported into Tumatumani, prob- ably from somewhere near Tiwanaku, during a short time in the later Qeya period. Paste 11 Paste 1 1 inclusions are ( 1 ) relatively abundant lustrous white inclusions, subrounded and sub- angular form, very fine size, occasionally fine; (2) relatively abundant matte black inclusions, sub- Analyses of Ceramic Artifacts 27 rounded form, very fine size; (3) relatively abun- dant matte red inclusions, subrounded and sub- angular form, very fine size, occasionally fine; and (4) very scarce mica flakes, silver only, very fine size. Density is 20 inclusions per 5x5 mm. Paste 1 1 is a compact to dense paste with a fine texture. The red, white, and black inclusions are found in equal proportions and are fairly dense, but their fine size makes them difficult to see with the naked eye. The break is slightly irregular, somewhat similar to Paste 2, with a medium quan- tity of very small, round air holes. Paste 1 1 sherds are fired to a buff" or light brown color (7.5YR 6-7/ 4). Most are burnished on the exterior or both faces, although wiped and smoothed sherds also occur. Paste 1 1 dates to the Late Horizon, and all diagnostic Paste 1 1 sherds are decorated with stan- dard Inca motifs such as fields of black triangles and fern designs. Shapes include a shallow bowl with a thinned rim and decorated interior, a bowl with a direct rim and convex profile, and frag- ments of "aryballus" jar necks. The center of pro- duction of Paste 1 1 , as well as of Pastes 1 2 and 13, is presently unknown. The sample of these sherds at Tumatumani is small and the post-Ti- wanaku occupation of the site brief and too little studied to formulate conclusions at the present time. Paste 12 Nonplastic inclusions in Paste 1 2 are ( 1 ) scarce matte white inclusions, subangular form, very fine size; (2) scarce lustrous white inclusions, suban- gular form, very fine and occasionally fine size; (3) scarce matte black inclusions, subangular form, very fine size; (4) scarce matte red inclusions, sub- angular form, very fine and occasionally fine size; and (5) scarce mica flakes, silver only, very fine size. Density is 10 inclusions per 5x5 mm. Paste 1 2 is a dense paste with a very fine texture. In appearance it looks something like fired silt. Inclusions in this paste are very few and difficult to find, even with a hand lens. The break is smooth, and the air holes, which are both round and lam- inate, are small and very few in number. The fired paste color is buff'or soft brown (7.5YR 5-5.5/ 4). Burnishing is generally on the exterior only, with a wiped or smoothed interior. Slipping occurs only in a thin, very dark brown color that does not cover the surface of the sherd completely. None of the specimens are decorated. Of the five Paste 1 2 sherds found, two can be pieced together, and at least two of the others appear to come from the same vessel, a bowl form with a direct rim and convex profile similar to the examples known from Paste 1 1 . This form suggests that Paste 1 2 is prob- ably also Late Horizon in date. Paste 13 Inclusions in Paste 13 are (1) relatively abun- dant matte grayish white inclusions, subrounded form, fine and occasionally medium size; (2) rel- atively abundant matte pinkish white inclusions, subrounded form, fine size; (3) scarce matte black inclusions, subangular form, very fine size; and (4) very scarce mica flakes, silver only, fine and very fine size. Density is 20 inclusions per 5x5 mm. Paste 1 3 is a compact, medium- to fine-textured paste. The break is only slightly irregular with a somewhat peaked appearance. Air holes are round or laminate, small, and very few in number. The paste is characterized by its predominantly grayish and pinkish white inclusions which, because of their dull appearance, do not stand out to the na- ked eye as do the white inclusions of the other paste groups. Also distinguishing Paste 13 is its distinctive oxidized fired color, redder than the other pastes at Tumatumani and falling in the range of lOR 4/6-8 and 5/6-8. Although most specimens are completely oxidized, some have an incom- pletely oxidized light gray or grayish brown core. Very rarely a superficial smudging producing a brown interior and/or exterior edge is found. Paste 13 sherds occur in small quantities in most of the wiped and burnished finish categories at Tuma- tumani, but more than half of the sample is found in the late finish category 10. Most sherds are bur- nished on the exterior or on both faces; a minority are smoothed, wiped, or burnished on the interior only. Slipping is exclusively red, the colors 7.5R 3.5/6 and lOR 3.5/6 being the most popular. Diagnostic sherds in Paste 13 include two Ti- wanaku decorated specimens (Figs. 1 60, 161), and body sherds with Late Horizon Inca fern motif designs. The remainder of the sample appears to also contain specimens from the Late Intermediate period. Like the Paste 12 ceramics, these Paste 13 sherds are undecorated and cannot be easily coni- pared with other ceramics in the Titicaca Basin. Nor can they be compared locally, since they bear little relation to the Late Intermediate period Pu- karani wares defined from the site of Pucara Juli, only 4 km from Tumatumani (de la Vega 1990). However, both Late Horizon bowls with interiorly beveled rims and the characteristic Late Inter- 28 Part Two mediate p)eriod straight-walled bowls with a direct rim (Julien 1983: 153) were found in the assem- blage, suggesting that the production of Paste 1 3 wares was continuous from Tiwanaku through to the Late Horizon. Since this report focuses on the earlier period of occupation at Tumatumani, before and during the time that the platform terraces were built, the Late Intermediate period and Late Horizon Pastes 11, 12, and 1 3 sherds will not be discussed further. Of interest, however, no Late Intermediate period Pukarani domestic or decorated wares were found at the site in either the excavations or the surface collections. The only Late Intermediate period sherds that do occur at Tumatumani are those few Paste 1 3 sherds that may date to this time period. It appears that there was a hiatus in the occupation of Tumatumani during most of the Late Inter- mediate period, when settlements moved to higher hilltop or hillside locations. Reoccupation of the site occurred on a limited scale in the Late Hori- zon, as evidenced by the Inca domestic structure found in the upper levels of Unit 3. The presence of Pastes 11, 12, and 13 sherds in the second con- struction levels of Unit 1 (see Table 1 ) is trouble- some, given the hypothesized Tiwanaku date of this construction episode. However, these sherds occur in sufficiently small quantities that we are presently attributing their existence to intrusions from the Late Horizon occupation of the site (es- pecially because Unit 1 is located adjacent to Unit 3) or to errors that may have occurred during sift- ing, which was undertaken by a large untrained crew. Finish Categories Twelve finish categories were defined for the complete Tumatumani ceramic assemblage, in- cluding all pastes and time periods. These cate- gories were based on observation of surface fin- ishing techniques, i.e., wiping, scraping, and burnishing, including the width of the burnish stroke, its visibility, and whether burnishing was applied to a leather-hard or moist clay surface. Table 4 shows the distribution of finishes for each paste as found in the first and second construction levels of the Unit 1 excavations. Category 1 Category 1 includes the wiped or scraped fin- ishes at Tumatumani. Sherds classified directly into this category (see Table 4) have a wiped or scraped finish on both sides. Scraping, which is in fact more a shaping than a finishing technique, was performed with a hard-edged tool, such as a piece of wood or flaked stone, leaving even and parallel striations. The effect is similar to that il- lustrated by Shepard (1956: Fig. 13c right), al- though the striations are not quite as pronounced. There is little drag of the temper grains, and the surface contour of the sherds is even or slightly irregular. Wiped surfaces are more common in category 1 than surfaces that were left completely unelaborated after the initial scraping process. Wiping was undertaken on a plastic clay surface and produced fine, even and parallel clay ridges with a characteristic plastic flow, similar in ap- pearance to that illustrated by Shepard (1956: Fig. 13c left). Surface contours are generally even or slightly irregular. Wiping was performed with a soft, yielding tool such as a piece of cloth or leather, or the hand. Striations are particularly fine and even on rims, suggesting that this part of the vessel was finished with a cloth folded over the top of the rim. The even contours of wiped sherds in- dicate that they were first scraped to even out pits and bumps, and then remoistened and given a final finish with a wiping tool. Both wipe and scrape are invariably horizontal on the interiors of vessels in all pastes but can be either horizontal or vertical on the exterior, depending on vessel typ)e and where on the vessel this finish occurs. Category 1 finishes are the only ones to be cross- referenced with other finish categories. Thus, cat- egory 1 sherds are wiped and scraped on both sides, but category 1 finishes themselves also occur on the opposite faces of burnished sherds in cat- egories 3-11. Category 1 wipes and scrapes are found with the highest frequency on Paste 1 sherds, particularly on the interior and exterior of jar necks and on the interiors of almost all vessel forms, even bowls, which were only rarely burnished on the interior in the Late Sillumocco period. They are found with less frequency on the exteriors of bowls or on the exteriors of bases other than the utilitarian thickened edge base shape. Wiped and scraped finishes are rare on Paste 2 vessels, which are principally burnished, but they are found oc- casionally on the interiors of exteriorly burnished forms. Paste 3 has its own category of wipe (cat- egory 9), but category 1 wipes and scrapes are found with some frequency on the interior and exterior of jar necks and the interior of decorated and open forms. Wiping or scraping generally does not occur on the fiber-tempered wares; it is absent Analyses of Ceramic Artifacts 29 in Pastes 6, 8, and 9, extremely rare in Paste 7, and occurs on Paste 5 vessels only in the reduced- fiber examples, mostly on the interior. Category 1 finishes are also found only infrequently on the later Pastes 11, 12, and 13, which when unbur- nished are more likely to have smoothed or wiped surfaces from categories 10-12. Again, when cat- egory 1 finishes do occur on only one side of the vessel in these pastes, they are generally on the interior. Category 2 Sherds of this category have a rough grainy sur- face, with faint traces of a wipe executed while the clay was in a plastic state but without great atten- tion to detail. Many spots are missed, and temper grains show through onto the surface. Sherds in category 2 also occur with category 1 scraped in- teriors. Category 2 finishes are generally found on the bottom of vessel bases and occasionally on the exterior of Late Sillumocco period Paste 1 bowls. They occur only in Pastes 1 , 2, and 3, where in all cases they are a minor finish category. Category 3 Category 3 is the first of the burnished catego- ries. Burnishing strokes are wide, from 2.5 to 4 mm, and have distinctly visible clay ridges along their sides and at the end of the strokes. The effect is similar to that illustrated by Shepard (1956: Fig. 1 3e), but with even more pronounced burnishing troughs. The pushed clay ridges of category 3 bur- nishes suggest that burnishing in this case was per- formed on a somewhat moist clay, just before the leather-hard stage. The luster of category 3 bur- nishes is predictably low or matte, since a consid- erable amount of shrinkage would have occurred after the burnishing process, removing the luster from the surface. Burnishing in category 3 occurs only on the exterior of vessels; interiors have a scrape or wipe from category 1 . Variations include a burnish that achieves complete coverage over the surface of the vessel, or a more commonly found incomplete coverage, with narrow strips left between the burnish lines. These unfinished areas show the scraped surface below the burnish. An- other variation in which the strokes are extremely widely spaced and have a very haphazard ap- pearance occurs only on Paste 1 specimens. Cat- egory 3 burnishes are found principally on the exterior of Paste 1 utilitarian thickened edge bases and occasionally on large straight-sided bowls. Its presence mostly on bases is probably attributable to the fact that vessels were not dried completely before burnishing, so that the burnishing process was begun when the thinner rim area was dried to a leather-hard stage but the thicker base was still moist. The occurrence of category 3 burnishes on other pastes at Tumatumani is limited to Paste 3, where it is found only rarely. The entire category 3 group of finishes appears to have been used only during the pre-Tiwanaku period at Tumatumani; it is not found on Tiwanaku or post-Tiwanaku period vessels. Category 4 Category 4 burnishes are similar to those of cat- egory 3. Burnishing is executed just before the leather-hard stage, resulting in the same clay ridges along the edges of the burnish strokes, but the width of these strokes is slightly narrower, from 1 to 1.5 mm wide. The luster of the finished vessel, usually medium, is also slightly higher than that of category 3 specimens. Burnishing occurs only on the exterior; interiors are scraped or wiped. Complete and incomplete burnish coverage vari- ations occur, with the incomplete burnish in the majority. This finish category is restricted to Paste 1 sherds and is found on the same vessel shapes as category 3. Like category 3, it is believed to have been used only in pre-Tiwanaku times. Category 5 Category 5 is the most popular finish category at Tumatumani. It occurs on the largest percentage of the Paste 1 sherds and is used in some per- centage on practically all the pastes, with the ex- ception of two of the fiber-tempered examples. Burnishes in this category are executed on a leath- er-hard clay surface; there is no pushing of clay along the edges, and there are no burnishing troughs. The burnish strokes themselves are just barely visible in the form of parallel burnishing facets. Lusters range from low to high, but the majority fall in the medium range. Strokes are fairly wide, from 2 to 2.5 mm, and contours are even or slightly irregular. Both complete burnish coverage and incomplete coverage types are found. The vast majority of category 5 sherds are bur- nished on the exterior only, with a wiped or scrajjed interior, but variations include a complete and/or 30 Part Two incomplete category 5 burnish on both faces and, in small quantities, a wiped or scraped exterior and a category 5 burnish on the interior. When burnishing occurs on the interior, in this and all burnish categories at Tumatumani, it is invariably horizontal in orientation. Exterior burnishes can be either horizontal or vertical, depending on the vessel type and where on the vessel the burnish is placed. There is a certain lack of consistency in the orientation of exterior burnishes, particularly in the Paste 1 group, which has a larger sample to draw from. As a general rule, straight-sided vessels with either vertical or flaring walls have a vertical burnish and bowls have a horizontal burnish. Bur- nish directions, however, are often different even on the same vessel, with, for example, a horizontal burnish at the rim and a vertical burnish below. Category 5 burnishes are found with the highest frequency on the exteriors of Paste 1 bowls, bases, and burnished jar necks and somewhat less com- monly on the thickened edge bases. Category 5 burnishes also appear in small percentages on Pastes 3 and 5 plainwares. In the other paste groups, category 5 is most often represented in its complete coverage version on decorated wares. It occurs only rarely on the fiber-tempered pastes. wanaku period ceramics. Burnish marks are not visible at all and leave no grooves or facets. Con- tours are even and smooth, and burnish luster is high. Burnish occurs in complete and incomplete coverage versions, but the spaces left between the burnishing strokes of the incomplete version are narrower than in the incomplete coverage varieties of categories 3-6. The majority of category 7 sherds are burnished on both sides, with a smaller per- centage burnished on the exterior or interior only. Except for the fineware Paste 2 sample and some Tiwanaku and post-Tiwanaku specimens, cate- gory 7 burnishes are generally not found on plain or monochrome-slipped ceramics. However, they are the most popular for decorated or fancy wares in nearly all the pastes. They are rarely found on the fiber-tempered pastes; examples include only sherds from the reduced-fiber version of Paste 5 and one Paste 6 specimen. These sherds are from the surface collections and do not appear in Table 4. Also from the surface collections are decorated sherds finished with category 7 burnishes in Pastes 4 and 10. Category 7 finishes are not found on the late Pastes 1 1 and 1 2 but do occur on Paste 1 3 on both plainware and decorated specimens, includ- ing Late Horizon decorated sherds. Category 6 Category 6 is the narrow burnish stroke coun- terpart to category 5. Burnishing was again per- formed on a leather-hard surface, and the strokes are just barely visible, but in this case they are slightly narrower, from 1 to 1 .5 mm. As in category 4, the narrow burnish stroke category 6 tends to achieve a slightly higher luster than its wide-stroke equivalent. Lusters range from medium to high. While a majority of sherds have wiped or scraped interiors, minor variations in this category include sherds burnished on both sides and, in small per- centages, sherds with a burnished interior and a category 1 exterior. Category 6 is the second most popular finish for Paste 1 vessels, where it occurs most frequently on the exteriors of bowls and on jar necks. It is used less frequently on bases. The use of category 6 burnishes on ceramics of other paste types is again restricted mostly to the dec- orated wares. Category 7 Category 7 is one of the fine burnish categories at Tumatumani and the finest found on pre-Ti- Category 8 This finish category is restricted to the fiber- tempered pastes at Tumatumani, and virtually all the fiber-tempered sherds are burnished with cat- egory 8 burnishes. Burnish strokes are not visible and leave no grooves. Burnish luster is medium or high, and surface contours are even or only slightly irregular. Both complete and incomplete coverage examples exist. As is characteristic of fiber-tempered pastes, these sherds have fiber voids visible on the surface, ranging from sparse to con- siderably dense, depending on the paste and the individual variation in attention to surface ap- pearance. If the pieces are slipped, as they are in the majority, the slipped surface is sometimes crackled in hairline fractures. Sherds are usually burnished on both sides, although wiped and scraped interiors or exteriors do occur infrequently in all the fiber-tempered pastes. In these cases the wipe or scrape is similar to the category 1 versions. Occasionally a fiber-tempered equivalent to the category 2 finishes, a rough, grainy surface with a large number of fiber voids visible, is found on the exterior of a sherd, with a burnish on the in- terior. These sherds probably come from open Analyses of Ceramic Artifacts 3 1 forms, but no rim or base sherds of this finish type were found that would aid in identifying a specific shape. Category 9 Category 9 is a particular type of plastic wipe characteristic of Paste 3 sherds. Wiping in this category imparts a layer of fine clay particles to the surface of the sherd and gives it a distinctive muddy appearance. This layer of fine clay was probably achieved through repeated wiping with a moist cloth or hand, causing the fine clay par- ticles in the paste to float to the surface of the sherd. The floated clay layer does not succeed in covering the abundant temper grains of Paste 3, and surfaces consequently have a grainy appear- ance. Wiping ridges are visible but are somewhat less even and parallel than their category 1 coun- terparts. The contours of these sherds are even or slightly irregular. Interiors may have either the same fine clay wipe or a wipe or scrape from cat- egory 1 , in which case the temper grains still show through on the surface. A large number of Paste 3 sherds still have a category 1 finish on both sides; the fine clay particle category 9 appears to have been used only on part of the total assemblage, generally the interior and exterior of jar necks, and perhaps on the bodies as well, and on the exteriors of most of the bowl forms. Paste 3 sherds that have an incomplete burnish in categories 3-7 show, in the spaces between the strokes, the standard nonfloated plastic wipe or scrape below. Category 10 Finish categories 10-12 occur only in the second construction episode levels. They are found only on ceramics that date to the Tiwanaku and post- Tiwanaku occupation of Tumatumani and were used on both the early pastes that continued to be manufactured during Tiwanaku times (Pastes 1, 2, 3, 5, and perhaps 4) and on the exclusively Tiwanaku and post-Tiwanaku pastes (Pastes 1 1- 1 3). Category 1 0 is the finest of the three. It can best be described as a velvet-smooth finish. Un- bumished sherds have no traces of wiping or scraping of any kind visible on the surface of the sherd. Experiments show that they were probably compacted or rubbed with a soft cloth or piece of leather at the leather-hard stage, leaving a matte- smooth surface. Contours are even or very even and smooth to the touch. Burnishing was evidently performed with a wide tool on a leather-hard or almost dry clay surface and can have either a com- plete or slightly incomplete coverage. Burnish strokes are similar to, but finer than, those of cat- egory 7, leaving no grooves or facets of any kind. In the few spaces between the strokes of the in- completely burnished sherds, the velvet-smooth finish described above is visible. Contours are very even, and the luster of these sherds is medium or high. Category 10 finishes are found on Pastes 1, 2, 3, 11, 12, and 13. Burnished sherds are more frequent than sherds that are smoothed only on both faces. Burnishing is more likely to be on the exterior than on both faces of the sherd. Interior- only burnishes are less popular and in the sample of diagnostic sherds are found only on bowl shapes. Mixing with finishes from categories 1, 5, 6, and 1 1 is also found, with, for instance, a category 10 burnish on one side and a category 5 burnish on the other. Because the presence or absence of cat- egory 10-12 finishes was deemed to be a significant attribute in dating the Tiwanaku and post-Tiwa- naku sherds, examples with 1-6/10-12 mixes were coded as belonging to the later finish categories. It is important to note, however, that categories 1 0- 1 2 do not represent a complete change in finishing techniques but are in fact frequently combined with finishes from the earlier categories. Category 11 Category 1 1 is a less finely finished version of category 10. It occurs more commonly on Pastes 1, 2, 3, and 5, those pastes with a previous history of manufacture at Tumatumani, whereas category 10 was favored for Pastes 11, 12, and 13. Unbur- nished varieties of category 1 1 have a finish that only approximates velvet smooth. Over most of the surface no traces of finishing can be seen; how- ever, occasionally a faint wipe or scrape mark is found. Contours are even only, and rarely very even. The burnished variety is also less finely fin- ished. Burnish strokes leave no grooves or facets, but the luster is low or medium, and surface con- tours are even rather than very even. Spaces be- tween burnish strokes are slightly more frequent, and on the smoothed surface beneath an occa- sional faint finish mark can be seen. Sherds that are smoothed only on both sides form a majority in category 1 1 . Of the burnished sherds, an exte- rior-only burnish is the most common, while bur- nish on both sides or on the interior only occurs 32 Part Two less frequently, and in smaller percentages than in category 10. Like category 10 finishes, category 1 1 finishes are used along with earlier category 1, 5, and 6 wipes, scrapes, or burnishes on opposite sides of a single sherd. Category 12 Category 12 is principally a wiped finish cate- gory, similar to category 9. Wiped surfaces display even and parallel clay ridges, with the same layer of floated fine clay as the category 9 examples. Category 1 2 wipes are found only on Paste 1 sherds in the Tumatumani excavation sample, where they occur mostly on jar necks and on the interiors of exteriorly decorated Tiwanaku ceramics. Addi- tional category 12 wipes, also on the interiors of Tiwanaku decorated wares, are found on surface- collected sp>ecimens of Pastes 5 and 13. These wipes are not found on Pastes 1 1 and 1 2 ceramics, where unbumished surfaces are finished with an even, parallel plastic scrape or wipe from category 1 or have a smoothed finish from categories 10 or 11. Category 1 2 burnishes are less frequent and have been found only on the exteriors of Paste 1 body sherds. Burnishes have strokes similar to those of categories 5 and 6 on leather-hard clay, with bur- nishing facets just barely visible. Burnishing is al- ways incomplete, and in the spaces between the strokes the floated clay wipe can be seen. Interiors of these sherds can have the same category 1 2 wipe or a wipe or scrape from category 1 . Paste 1 Undecorated and Monochrome-Slipped Ceramics The locally manufactured Paste 1 specimens form the majority of ceramics found at Tumatu- mani, both decorated and utilitarian. What follows is a discussion ofone group of Paste 1 undecorated and monochrome-slipped ceramics drawn mostly from examples in the first construction episode fill of Unit 1 and pertinent sherds from the surface collections and second construction episode levels. In spite of their varied archaeological provenience, the undecorated wares described here are believed to be contemporary and to date to the Late Sillu- mocco period. The dating of this undecorated group is discussed at the end of the shape descriptions and is followed by an account of the Paste 1 Ti- wanaku period undecorated assemblage. Decorat- ed Paste 1 ceramics are addressed separately in a later section. Surface Color and Slip Table 5 shows the distribution of surface color categories for the total sample of body and rim sherds from the Unit 1 excavations. Categoriza- tion of unslipped sherds was based principally on exterior surface color. The category "blackware" in the table refers to those ceramics that were re- duced or smudged in firing, as described in the paste descriptions, or those with dark brown sur- face colors. These sherds characteristically have a medium to large amount of mica visible on their surfaces. Surface colors fall in the range of 2. SYR, SYR or 7. SYR 2-3/0-4. The category "redware" refers to sherds that were oxidized in firing, in- completely oxidized, or cooled to produce a red- brown, light red-brown, light brown, or light or- ange-brown surface color of 2. SYR S/4-8 or SYR S-6/4-8. They tend to have a slightly less mica- ceous aspect than the blackwares, principally be- cause they lack the contrast of the black ceramics, which causes the mica to stand out. The red slips on the Paste 1 monochrome-slipped ceramics include colors that range from red or dark red to red-brown. Munsell colors are 7.SR 3/3-7, 3.5/6; lOR 3/3-6, 3.S/6, 4/4-8; and 2.SYR 3/4-6, 4/4-6. Bright reds, such as 7.SR 3 or 3.S/8, do exist but are extremely rare. Red slips are generally thin, and application can vary from uniform and even to patchy. Applied to a micaceous surface, they do not cover the mica completely, so that these vessels are almost as micaceous as the un- slipped wares. Red slips are used more frequently on the oxidized wares but also occur on a fair percentage of black or gray reduced vessels, where their color range tends to be slightly darker. Bur- nishes applied to these latter vessels tend to result in a characteristic black streaked effect. Brown slips on Paste 1 wares are also thin but are applied more evenly than the red slips and tend to cover the surface mica more uniformly. They are used with just a slightly greater preference on the reduced wares. Colors are SYR S/3-6, 6/4; 7.SYR 4/3, S/3- 6, 6/4-S; and lOYR S-6/3. Both red- and brown- slipped vessels are frequently left unbumished, the slip being applied directly onto a wiped surface without further elaboration. As Table S indicates, sherds of the blackware group predominate over the redware, particularly in the material from the first construction levels. Of the slipped wares, the red exterior slip predominates, although the slipped Analyses of Ceramic Artifacts 33 •s e H s S o row teri red n X e _, © = .2 s ¥ ® .2 o a « + 4j M '-S .5 g E oj 9 * S.-C BQ '"S .5 I as + S ■5 .S S -i « .2 II « •- -S to « IT) O OS On 00 /^ — NO m •*" d r-' Tf rn so u-i r- vo r- >o vo OS ^ Tf r- cnj ov o o sd sd r-' vo (N 00 OS r^ OS od Tt' eN — ' d sqoo — rned red on both faces, the others are unslipped black- ware. One rim is decorated with a small incised 36 Part Two nubbin (Fig. 40), another has a horizontal strap handle (Fig. 41). Bowl Shape E (Figs. 42, 43)— The four bowls in the shape E sample have flat, direct rims. These vessels appear to come in two sizes: a small version in the 12-13 cm range and a larger 23-25 cm size. Three rims are unslipped with reduced or smudged gray surfaces; one of these is gray on the interior only, with a red-brown exterior. The fourth rim has a light brown slip on both faces. Interiors are wiped or scraped with an even surface contour, although one of the blackware examples has a cur- sory horizontal category 6 burnish on the interior. Exteriors vary and include a horizontal wipe, a horizontal wipe at the rim and a vertical scrape below this (recalling the jar neck finishes), and a horizontal burnish from categories 5 or 6, found on two of the blackware rims. Burnish coverage is not complete, and lusters range from low to high. Exterior surface contours are even or slightly ir- regular. Straight-Sided Bowls (Figs. 44, 45)— These shapes differ from the bowls discussed above in that their walls are not convex but rather straight in contour on both the interior and exterior faces and slightly flaring. The term straight-sided should not be confused with vertical-sided, which refers to a vessel with a wall at a 90° angle to the base. Straight-sided vessels at Tumatumani are difficult to identify, since a fairly large rim sherd is needed to adequately judge the vessel shape. The number of straight-sided vessels in the total assemblage is therefore probably larger than the two examples that can be confidently identified (Figs. 44, 45). (Bowl shape A appears to be, in addition, a cate- gory of basically straight-sided vessels, at least on the exterior face.) Both of the rims illustrated are unslipped, one red-brown and one gray in color, with an incomplete vertical low luster burnish from categories 3 or 6 reaching up to or within 3 cm of the rim. One also has a cursory category 5 burnish on the interior. The vessels are large and deep, with diameters of 22 and 28 cm. As the illustra- tions show, straight-sided bowls were made with both loop handles on the rim, or horizontal strap handles located just below the rim. The illustrated examples have flat or slightly rounded rims, while other possible examples of this shape are found with rims which are rounded and just slightly thickened. Miscellaneous Rims (Figs. 46-5 1)— A number of miscellaneous rims were found at Tumatumani, none with a large or consistent enough sample to be classified into any shap)e group. Two of these are decorated rims. The sherd shown in Figure 46 probably comes from a shape E bowl. The incised "shelf visible in the illustration was applied to a much smaller rim, which appears to have been flat. The slight bulge on the exterior, below the shelf, is also a by-product of the applique. The piece has an unslipped grayish brown exterior with a matte vertical burnish from category 5; the in- terior is scraped and then slipped light reddish brown. The "shelf is slipped as well. The sherd illustrated in Figure 47 also appears to come from a shape E bowl. The piece is too small to judge accurately whether the lower wall was straight or convex, but the interior and exterior finish, an even contour wipe, is more characteristic of bowl shape E than the straight-sided forms. The rim is unslipped and red-brown in color. The rim sherd illustrated in Figure 48 has a unique flattened lip shape. It would appear to be- long to a straight-sided bowl. The exterior is fin- ished with an incomplete category 6 horizontal burnish and has a patchy red-brown slip; the in- terior is a scraped unslipped gray surface. Figure 49 illustrates a distinctive rim of which only one example exists in the Tumatumani sample. The piece has an oxidized light red-brown surface with a light red slip applied only within the exterior groove of the rim itself. Vessel shape is in this case indeterminable, but judging from the medium lus- ter incomplete category 5 burnish on the interior as well as the exterior, the rim appears to belong to some type of open bowl form. Other single- example rims are illustrated in Figures 50 and 5 1 . Figure 50 is a small incurving bowl with an un- usual light red slip, an incomplete high luster hor- izontal burnish on the exterior, and an unslipped wiped interior. The example in Figure 51 has a particularly thin wall compared to other Paste 1 sherds and appears to come from a cup or small open bowl. It is of an unslipped red-brown color with a simple wipe on both faces. The possible existence of a shallow bowl form in the Tumatumani assemblage should be noted; however, the sample available is not adequate to distinguish this category with any confidence. All probable sherds are small, and some could be from everted necks. Two rims, each very different, can be more confidently identified. One has a flat hor- izontal direct rim, is scraped on both faces with a red slip on the exterior, and has a diameter of 8 cm and a height of approximately 2.5 cm; the other has a rounded direct rim, is finished with a hori- zontal incomplete category 5 burnish on both faces over an unslipped red-brown surface, and has a Analyses of Ceramic Artifacts 37 diameter of 20 cm and a height of approximately 5 cm. No plates were identified in the Paste 1 sample, a fact that is unusual insofar as plates do exist in both the roughly contemporary Pucara and Qeya (Tiwanaku III) ceramic assemblages (pers. observ. by the author at the Pucara site museum; Girault, 1990: 107-109). Bases— Thickened Edge Bases (Figs. 52-56)— Bases of this type are characterized by a thickened edge of clay around the very bottom of the base, producing the effect of a short, solid pedestal. This thickening can range from 5 to 9 mm high and usually projects outward slightly to form a pointed edge. Vessel walls themselves are straight and flar- ing. Thickened edge bases are the most common base type at Tumatumani, with 38 securely cate- gorized examples. Diameters are small and range from 5 to 12 cm, with an average of 8 cm. One particularly large example has a diameter of 17 cm. Virtually all examples have the same exterior finish. The thickened area and immediately above it is unfinished or has a horizontal wipe; the walls themselves are finished with a vertical burnish. Occasionally there are also one or two horizontal burnish strokes at the bottom of the base, above the thickening. Rarely are these vessels completely unbumished on the exterior. However, despite the emphasis on burnishing, the exteriors of the thick- ened edge bases are somewhat coarse in appear- ance. Surface contours are usually irregular or slightly irregular, and burnishing never achieves complete coverage. Burnishing frequently is of cat- egories 3 and 4, leaving burnishing troughs and pushed clay ridges, and of categories 5 and 6. Lus- ters range from matte to medium, with the ma- jority in the low range. Interiors are wiped or scraped, but with even contours. The majority of these pieces (58%) are unslipped blackware. Sur- face colors are a deep black or very dark brown, not the superficially reduced light grays of most other shapes. Pastes are black or brown with black edges. Only 1 9% of the sample consists of un- slipped oxidized redwares, and few bases of either red- or blackware are slipped; 1 3% have an exte- rior slip, 5% an interior slip, and 5% are slipped on both sides. Red and brown slips are found in equal numbers. Thickened edge bases are the only shape cate- gory at Tumatumani to display carbonized de- posits with any amount of frequency. These occur as distinct sooty accumulations on the surface of the sherds and are found in 23% of the thickened edge base sample. They are found most often on both sides of the vessel or on the interior in con- junction with a black burnt or fire-blackened ex- terior. These deposits appear to be the result of direct exposure to fire and/or the charring of food residues on the interior of the vessel and indicate that some of these thickened edge bases come from cooking vessels. Although the overall form of these vessels is unknown, several examples of horizontal and vertical strap handles with comparable soot- ing were found (Fig. 62), suggesting that at least some specimens had strap handles located on the body of the vessel. Other thickened edge bases in the sample, however, are carbonized only on the interior, with a slipped or oxidized red exterior. A small number of body sherds in the sample also show this same arrangement. These vessels were evidently used in a different type of fire-related activity than those with sooty or fire-blackened exteriors. Because of the coarse finish of many of these specimens and the large amount of carbon- ization on their interiors, a utilitarian function as a portable brazier is suggested. Some examples, however, may have been used as ceremonial burn- ers or incensarios. Interior blackening or burning is a trait often found on incensarios elsewhere in the Lake Titicaca Basin (Chavez 1985: 144, 147). Various lines of evidence indicate the presence of a ring-based decorated incensario form at Tu- matumani (see pp. 58-59). However, a less finely finished thickened edge base version may also have existed. The exterior finish on these thickened edge bases is commensurate with that found on the shape E and straight-sided bowl forms (Figs. 42-45), so that these braziers or ceremonial burners may have had an overall V-shaped or slightly convex profile. Pedestal Bases (Fig. 57)— A small sample of hol- low pedestal bases was found at Tumatumani. Un- fortunately, none are complete, so that the actual height of the pedestals themselves is unknown, but they are taller than the ring bases discussed below. Of a sample of three, diameters at the point of constriction between wall and base range from 1 0 to 1 1 cm. Pedestal bases are fairly well finished compared to the majority of Paste 1 shapes; all three examples are slipped light brown or red, ei- ther on both sides or on the interior or exterior only. One is also slipped on the interior of the pedestal itself The interiors of the vessels have a careful, even contour horizontal scrape or medium luster category 5 burnish. On the exterior, the ped- estal portion of the wall was wiped, while the vessel portion has a low or medium luster category 5 vertical burnish. One example (Fig. 57) has, in addition, a very small amount of localized carbon deposition on the interior of the vessel, indicating 38 Part Two its use in a presumably ceremonial fire-related ac- tivity. Ring Bases (Figs. 58, 59)— Five ring bases were found at Tumatumani. They are similar in size to the pedestal bases, ranging from 10 to 12 cm in diameter at the constriction between wall and base. The height of the rings themselves varies from a barely elevated 3 mm to a pronounced 1.3-cm ring. Rings flare out at an angle from the horizontal plane of the bottom of the vessel; vessel walls ap- pear to be straight or slightly convex. No prefer- ence exists for slipping the vessels: while two ex- amples are slipped red-brown on both faces, three are unslipped red-brown or gray-brown in color. Interiors are finished with a simple wipe or scrape, and three of the five are burnished horizontally or vertically on the exterior with a matte or low luster category 5 burnish. Burnishing is rarely complete in coverage, in only one case extending as far down as the ring itself Miscellaneous Bases— A fair number of mis- cellaneous bases were found in the Tumatumani excavations; unfortunately, most were too frag- mentary to reconstruct the vessel shape to which they belong. However, a number of generaliza- tions can be made about these specimens. All bases are flat; there are no round or convex specimens in the assemblage, nor are there any bases with vertical walls that join at a 90° angle. Wall contours indicate that these miscellaneous bases come from globular vessels or open forms, i.e., from the jar and bowl categories. Some show a very small amount of thickening at the joint between vessel wall and base, although not to a degree that ap- proximates the thickened edge base category. Sur- face finishes demonstrate a predominance of ver- tical burnishes or horizontal wipes on the exterior of all examples. Interiors are generally wiped, al- though the open forms may occasionally be bur- nished. Like the other Paste 1 base categories, any burnishing on the exterior almost invariably stops 5-10 mm from the bottom of the vessel. Dating the Paste 1 Undecorated Ceramics The Paste 1 ceramic assemblage just described is an associated group that shares similar char- acteristics of firing, finish, and vessel shape. Spec- imens in this group are consequently believed to be roughly contemporaneous and to date to the same period of the local Juli Sillumocco sequence. Although a definition of the basic assemblage can be made with some confidence, the dating of these ceramics poses more of a difficulty. The Paste 1 undecorated wares in question were found in the first construction episode fill levels. The construc- tion of the lower platform at Tumatumani has been dated, based on decorated ceramic sjieci- mens, to the Late Sillumocco period. The group of local plainwares found in the fill of this plat- form, i.e., the Paste 1 assemblage defined here, can therefore be assumed to be either Late Sillumocco or pre-Late Sillumocco in date. Based on a com- parison of the Paste 1 assemblage, including at- tributes of technology, decoration, and vessel shape, with ceramic sequences of known age elsewhere in the Lake Titicaca Basin, a Late Sillumocco date is suggested. The resemblance of the other major ceramic assemblage at Tumatumani, the fiber- tempered plain and decorated wares, to those from the site of Chiripa suggests that that assemblage dates to the Early Sillumocco period, the period in the Juli sequence that was contemporary with the Chiripa culture in Bolivia. As defined, the division between the Early Sillu- mocco fiber-tempered ceramics and the Late Sillu- mocco non-fiber-tempered Paste 1 assemblage is somewhat artificially abrupt. Obviously a period of transition existed at the site, with new Late Sillumocco traits appearing alongside a declining fiber-tempered tradition. However, owing to the nature of the archaeological evidence, this period of transition cannot be defined. There is, of course, no reason to assume that the production of fiber- tempered ceramics at Tumatumani ended with the decline of Chiripa influence at the site, or that a fiber-tempered and non-fiber-tempered tradition could not have coexisted. Bermann's (1990) ex- cavations at Lukurmata show that the majority of plainwares at that site were fiber-tempered through the end of the Qeya (Tiwanaku III) period, and a significant number had this trait even in early Ti- wanaku IV times. Examples of coexistence and continuity in the different temper traditions occur at Tumatumani as well, particularly in the deco- rated wares and in the plainwares of Paste 5. How- ever, the similarities of the Tumatumani ceramics to dated assemblages elsewhere in the Titicaca Ba- sin result in a basically successive tradition in the plainwares at the site. This is further supported by the fact that the Paste 1 assemblage described here shows considerably different characteristics of sur- face finish, firing techniques, shape, and decora- tion from the Early Sillumocco fiber-tempered wares, strongly suggesting that they are not con- Analyses of Ceramic Artifacts 39 temporary but belong to successive periods in the ceramic chronology at the site. A Late Sillumocco date for these Paste 1 ceram- ics makes them roughly contemporary with Clas- sic Pucara in the northern Basin and the early phases of the Tiwanaku culture (Qalasasaya and Qeya) in the south. If the Paste 1 ceramics date to the Late Sillumocco period, then similarities should be evident between the Tumatumani material and the ceramics of these cultures. Comparisons show that a number of specific similarities do exist be- tween the Tumatumani Paste 1 ceramics and par- ticularly Qeya and Pucara ceramics, similarities that are not found in other earlier or later Titicaca Basin ceramic traditions. The first set of these comparisons is related to ceramic technology. The predominantly black and brown micaceous wares from Tumatumani, for instance, are technologically similar to both Qeya and Pucara plainwares. Although Qeya painted ceramic vessels are generally fired to a light brown color, the majority of the plainwares are also an unslipped black or brown (Bennett 1934: 450). Pucara ceramics are likewise characterized by a combination of redwares and black, brown, and light brown ceramics (Franquemont 1 990) and ex- hibit a large amount of mica on their surfaces (Kid- der 1943: 5), a trait also found at Tumatumani. The closest resemblance of the Paste 1 plainwares, however, is to Qeya ceramics from Qeya QoUu Chico, which I had the opportunity to examine at the American Museum of Natural History in New York. These are manufactured from a paste very similar to Paste 1 and are fired to the same red- brown and gray tones as the Tumatumani ceram- ics. The Qeya Qollu Chico ceramics also exhibit one of the same characteristic firing techniques particular to the Paste 1 plainwares: superficially reduced or smudged interior and exterior "rinds" on an otherwise red-brown paste. In attributes of finish, the ceramics from Qeya Qollu Chico exhibit remarkably similar horizontal wiping and scraping techniques, which left the same parallel striations, and the same category 5 and 6 burnishes as are found on the majority of the Paste 1 Tumatumani ceramics. The vessel shapes represented in the Paste 1 plainware assemblage also exhibit similarities to contemporary Qeya and Pucara shapes. One par- ticular form characteristic of the Qeya Qollu Chico assemblage is a straight-sided or slightly convex- sided flaring bowl with two semicircular tabs at the rim, incised with four or five parallel lines (Wallace 1957: 46, Fig. 1: 10). Similar examples of these tabs exist at Tumatumani (Fig. 46), plus * variations on the idea of rim embellishments, such '. as vertical tabs and nubbins (Figs. 33, 34). These tabs are not found in other southern Titicaca Basin cultures and occur at Tumatumani only on the Paste 1 bowl shapes presumed to be Late Sillu- mocco in date. Horizontal lugs at the rim, similar in concept to the Qeya Qollu Chico tabs, are found ' on Pucara bowls (Rowe and Brandel 1971: Fig. 67), but they are never incised and are used only on finely slipped and burnished vessels. Another characteristic of the Paste 1 plainwares is the horizontal strap handle found on the rims of bowl shape D, the straight-sided bowl shape, and one of the miscellaneous shape examples (Figs. 41, 45, 48). These horizontal strap handles can be compared to similar examples occurring on the rims of Qalasasaya and Qeya vessels (Ponce San- gines, 1971: Figs. 1-13,4,12, Figs. 2-2,5,21,23; Bermann 1 990: Fig. 63 top; Albarracin-Jordan and Mathews 1990: Fig. 6b top; Bennett 1934: Figs. 1 3f,g), as well as the roughly contemporary plain- ware assemblage from Lukurmata (Bermann 1 990: Fig. 50b). Incurving bowls, such as in Figure 50, have not been identified in the Qalasasaya and Qeya assemblages but do occur at Pucara (Fran- quemont 1990: Figs. 71-75). Comparisons of the Tumatumani rim shapes are less conclusive. Many of the characteristic Paste 1 rim shapes, such as that of bowl shape A, are not found on either Pucara or Qeya ceramics, nor are the characteristic Pucara thickened rims (Franquemont 1990: Figs. 19-22, 37-40, 43-44) or interior beveled rims found at Tumatumani, except perhaps on bowl shape C. Paste 1 jar rims, however, do bear a general resemblance to Pucara jars (Franquemont 1990: Figs. 50-57). The flat or slightly rounded rims on Tumatumani bowl shapes B and E and on the straight-sided bowls are also comparable to bowl rims found at Qeya Qollu Chico and to some of the Pucara bowl rim shapes (Franque- mont 1990: Fig. 25). The thickening that occurs at the joint between vessel wall and base in the Paste 1 thickened edge base group, and to a lesser extent on the convex- walled bases, was one characteristic that set the Paste 1 assemblage apart from others at Tuma- tumani. This distinctive feature also occurs on Pu- cara jars (Franquemont 1990: 6-7, Fig. 49) and the bottles and tab bowls from Qeya Qollu Chico (Wallace 1957: 43, 46). Pedestal bases, found at Tumatumani (Fig. 57), also occur at Pucara and in the Qeya Qollu Chico collections. The Pucara examples are 5-6 cm high (Franquemont 1990: 4) 40 Part Two and the Qeya Qollu Chico example is 6 cm high; the height of the Tumatumani examples is unfor- tunately unknown. Perhaps the single most important diagnostic shape that places the Tumatumani Paste 1 group in the Late Sillumocco period is the incensario. Paste 1 incised and feline head incensarios are discussed in the section on decorated ceramics, and their similarities to Qeya incensarios are noted and described. Various plainware sherds have been found, however, that indicate the existence of an incensario form at Tumatumani. Qeya incensarios have flat bases or bases with an elevated ring no more than 3 cm in height, straight-sided flaring walls, and flat or slightly rounded rims elaborated with scallops or pierced loops (Wallace 1957; Ben- nett 1934). Ring-based bowl forms are also found in the Qalasasaya assemblage (Ponce Sangines 1971: Figs. 2-2,5,21,23) and at Lukurmata (Her- mann 1990: Fig. 50b). The height of the Tuma- tumani ring bases falls within the range of all these examples. However, the presence of rim scallops in the Tumatumani assemblage (Fig. 64), as well as the decoration on the incised examples, indi- cates the existence at Tumatumani of vessels more closely akin to Qeya incensarios, since rim scallops are not found in the Qalasasaya assemblage. Al- though Tiwanaku ceramics also have scalloped rims, they are slipped and decorated, and generally larger in proportion. Finally, pierced loops at the rim, found on Qeya incensarios and the ring-based bowls at Lukurmata (Wallace 1957: 36-37, PI. le; Hermann 1990: Fig. 52b), also occur at Tumatu- mani (Fig. 63). The one example that preserves a fairly large section of the rim (Fig. 44) suggests, from the amount of space to the right of the loop, that it functioned as a handle. However, the pos- sibility of multiple loops along the top of the rim of smaller vessels, approximating the Qeya ar- rangement, cannot be discarded. Tiwanaku Period Paste 1 Undecorated and Monochrome-Slipped Ceramics The Paste 1 ceramics described in this section are believed to date to the Tiwanaku period, con- temporary with the construction of the upper plat- form at Tumatumani. They are identified princi- pally by the presence of category 10-12 finishes on their surfaces, finishes that appear for the first time in the second construction fill. The definition of a Tiwanaku period plainware assemblage from Tumatumani at this time is tentative. The fact that finish categories 10-12 do not occur in the first construction fill gives them a post-Late Sillumocco date. Their presence in large quantities in the sec- ond construction fill, and the identifiable conti- nuity in vessel shapes between this sample and the Late Sillumocco Paste 1 wares, indicates that the Paste 1 category 10-12 specimens are not post- Tiwanaku intrusions. However, it is unlikely that the transition between the Late Sillumocco and Tiwanaku periods involved a total change in do- mestic ceramic production. At least some of the Late Sillumocco shapes described previously were probably made during Tiwanaku times in much the same form and with the same finishes as they were previously. With the archaeological evidence available, however, it has not been possible to distinguish in the second construction episode fill between Late Sillumocco specimens and similar or identical ceramics manufactured in the Tiwa- naku period. That ceramics with finish categories 10-12 fin- ishes do not form the entire Tiwanaku period plainware assemblage is suggested by the data pre- sented in Table 4. The percentage of the late finish category sherds in the second construction episode levels appears to be too low to represent the entire Tiwanaku period sample. If finish categories 1-6, the finishes found on the Late Sillumocco Paste 1 assemblage described earlier, were exclusively Late Sillumocco in date, then their percentages in the second construction episode levels would seem to be too high for a mixed fill that should have only representative samples of the earlier Sillumocco periods. This suggests that at least some of the sherds in finish categories 1-6 found in the second construction fill were manufactured in Tiwanaku times. This is supported by Girault's (1990) anal- ysis of excavated ceramics at Tiwanaku itself. The majority of his utilitarian wares are still finished with wipes, scrapes, or a burnish type that leaves visible burnish strokes. His finish categories "re- tocado con los dedos,'" "retocado con paiio," and a burnish that leaves no visible strokes, the prob- able equivalent of categories 10-12 here, occur as a minority on the utilitarian wares, although they are the most popular finishes for the decorated ceramics. Although Tiwanaku period ceramics that ex- hibit continuity with the Late Sillumocco period could not be identified on an individual basis, by comparing the sample of finish categories 1-6 Analyses of Ceramic Artifacts 4 1 sherds found in the first construction levels with those from the second construction levels (Tables 4 and 5) it is possible to make some deductions about this portion of the Paste 1 Tiwanaku assem- blage. The percentage of blackwares, for instance, the smudged or reduced sample, drops consider- ably in the second construction levels, with a cor- responding increase in oxidized or incompletely oxidized ceramics. Second construction period blackwares also tend to be more superficially re- duced and of a lighter color and are very rarely smudged. The Tumatumani sample, however, still differs from the ceramics excavated at Tiwanaku. Redwares remain a minority, even in the catego- ries 10-12 group, while at Tiwanaku a majority (77%) of the assemblage had a brick red fired paste color (Girault 1990: 123). Finish categories 3 and 4 show a marked reduction in percentage in the second construction episode fill, from 1 5% and 7% respectively to 2%. This suggests that they were no longer produced in Tiwanaku times at Tu- matumani. The persistently high percentages of finish categories 1,5, and 6, however, imply that these finishes were still used in the second con- struction period. Category 1, 5, and 6 finishes were, it will be remembered, often combined with cat- egory 10-12 finishes on opposite sides of the same sherd. Finish categories 3 and 4, in contrast, never appear in conjunction with category 10-12 finish- es, supporting the suggestion that they were no longer used in Tiwanaku times. The exclusive presence of category 7 in the second construction fill is probably an accident of sampling, since pre- Tiwanaku decorated Paste 1 sherds with category 7 finishes were found in the surface collections, but an increase in the use of this fine burnish finish, including its use for the first time on undecorated and monochrome-slipped wares, obviously oc- curred in Tiwanaku times. An increase in interior-only burnishes is another characteristic of Tiwanaku period ceramics, re- flecting the higher number of bowls, interiorly bur- nished everted jar necks, and burnished vessels in general in this time period. Even finish categories 5 and 6 show a modest increase in the percentage of interior-only burnishes in the second construc- tion sample, rising from 0.4% and 0% respectively in the first construction levels to 1% and 0.6% in the second construction levels. The finish cate- gories that appear only in the second construction levels have a markedly higher percentage of in- terior-only burnishes: 10% of the category 7 sam- ple, 1 1% of the category 10 burnished sample, and 1 0% of the category 1 1 burnished sample. Sherds in these categories are also more likely to be bur- nished on both sides: 58% of category 7, 22% of category 10, and 12% of category 11, in contrast to the relatively constant 2% figures for categories 5 and 6. Slipping techniques also show an increase in in- terior slip placement in the second construction levels. Slipping in general increases, with a higher percentage of slipped sherds in the category 7-12 finish group than in finish categories 1-6, while interior-only slips increase and brown interior-only slips appear for the first time. Red and brown slips on opposite sides of the same sherd are used for the first time in the second construction period. Of the category 10-12 finish groups, category 10, the finest burnished of the three, accounts for the highest percentage of slipped sherds. Of the earlier finishes, category 1 shows the most marked in- crease in the use of slips. Red slips continue to be more prevalent in the second construction levels, as they were in the first. However, while the per- centage of brown slips used on the earlier finish categories remains the same in the first and second construction levels, finish categories 10-12 appear on a higher percentage of brown-slipped sherds than do finish categories 1-7 in the second con- struction period. Although the exclusively Tiwa- naku period finishes show a greater use of brown slips, the ratio of brown slips to red slips on the Tumatumani utilitarian wares is still considerably less than that found by Girault ( 1 990) at Tiwanaku itself. Second construction period slips on utili- tarian wares at Tumatumani are again thin, and the range of colors remains generally the same as in the Late Sillumocco period. In the red group, there is a greater use of the color 7.5R 3.5/6 in the second construction period, while darker browns (5YR 3/2.5) and more orange-browns (5YR 5/7, 7.5YR 5/7, 6/6) are added to the palette of brown slips. Carbon deposits on the surfaces of identifiable second construction period ceramics are rare. In- teriorly carbonized thickened edge bases are found in the second construction fill and therefore may have been used during Tiwanaku times. Other se- curely identifiable second construction period sherds, all body sherds, have sooty deposits on both the interior and exterior faces. Shapes Shape categories for the identifiable Paste 1 Ti- wanaku period wares at Tumatumani, those with 42 Part Two category 10-12 finishes, include jars, small ollas, and two bowl forms. The number of diagnostic sherds is much smaller than for the first construc- tion period sample. Rims are in the majority rounded, with a few slightly flattened examples occurring on the jar forms. No thickened rims exist in the sample. The shapes themselves generally have smaller diameters than the vessels in the Late Sillumocco assemblage. Although the nature of category 10 and 1 1 burnishes means that the di- rection of burnish often cannot be determined, in the plainware examples, mostly from category 1 1 , where the strokes can be seen, burnish as well as wiping and scraping is always horizontal on both the interior and exterior. This contrasts with the varied vertical and horizontal arrangements found in the Late Sillumocco Paste 1 ceramics. Also dis- tinguishing the Tiwanaku period wares is the much smaller amount of mica visible on their surfaces. Almost all the Tiwanaku period diagnostic sherds are oxidized, usually to a slightly more intense red- brown than that found in the earlier ceramics. A few incompletely oxidized examples are also found, while superficial reduction is rare and occurs only on jar necks. Jars and Ollas (Figs. 65-68)— Tiwanaku pe- riod jars (Figs. 65, 66) have flaring, everted rims. Of a sample of eight specimens, all are slipped: two red on both sides, three red on the interior only, two brown on both sides, and one brown on the interior only. Three examples have a category 1 wipe on both sides, three a category 1 1 burnish on both sides, and two a category 1 1 burnish on the interior only. Diameters range from 10 to 15 cm and average 14 cm. Two examples of a short- necked vessel with a less constricted orifice were also found at Tumatumani. One (Fig. 67) appears to be from a utilitarian shape. It is an unslipped red-brown and has a category 1 1 burnish on both sides. The other (Fig. 68) is a much finer vessel with narrower walls. It is slipped red and is like- wise finished on both sides with a category 1 1 burnish. Bowls (Fig. 69)— Tiwanaku period bowls at Tu- matumani come in two forms. The first has a rounded rim and slightly convex sides, with di- ameters in the two examples known ranging from 12 to 17 cm. One is slipped red and exhibits a category 1 0 burnish on the interior, with a wiped unslipped red-brown exterior; the other is slipped red and has a category 5 burnish on the exterior, with an unslipped smoothed interior. A second bowl form has straight, slightly flaring walls with either a rounded or slightly thinned rim. These vessels are generally slipped red and finished with a category 7 or 1 1 burnish on both sides. Diam- eters range from 13 to 20 cm. One particularly interesting example, illustrated in Figure 69, shows a certain similarity to first construction Late Sillu- mocco Paste 1 bowls in its size, relatively thick walls, and the loop handle at the rim. However, this loop is considerably smaller than those on Late Sillumocco examples, and the hole within the loop does not pass through from one side to the other. The vessel is slipped red, with a category 1 2 wipe on the exterior and a category 10 burnish on the interior. Bases— Bases in the Tiwanaku period assem- blage are flat and appear to belong to bowl and convex-sided shapes. In addition, one example was found that is reminiscent of the earlier thickened edge bases. The vessel wall, although not thick- ened per se, jogs up almost vertically from the bottom of the base for approximately 5 mm before flaring out in an open vessel form. In contrast to the Late Sillumocco bases, the exterior finish on Tiwanaku period examples, whether a wipe or bur- nish, continues evenly down to the joint between vessel wall and base. Paste 3 Undecorated and Monochrome-Slipped Ceramics The Paste 3 specimens form the second largest group of sherds found at Tumatumani. Their char- acteristic grainy surfaces and distinctive thicker rim shapes make sherds of this paste easily dis- tinguishable from those of Paste 1 . The Paste 3 vessels described in this section are drawn from examples from both the first and second construc- tion episode fills and selected specimens from the surface collections. Like the Paste 1 assemblage discussed previously, the specimens described here are believed to be representative of a roughly con- temporary ceramic assemblage dating to the Late Sillumocco period. Further discussion of the dat- ing of these ceramics is addressed in the section following the shape descriptions. Surface Color and Slip Table 6 shows the distribution of surface color categories for the total sample of Paste 3 rim and body sherds in the Unit 1 excavations. The term "blackware" in this table is somewhat misleading. Analyses of Ceramic Artifacts 43 Table 6. Surface color and finish of Paste 3 specimens in Unit 1 excavations, by construction episode. Red- Brown- slipped slipped Red- Red- interior Brown- interior Total Finish Black- slipped slipped + slipped + number category ware Redware exterior interior exterior exterior exterior Total % of sherds 1 st construction 1 70.3 27.0 2.7 100.0 37 2 33.3 66.7 100.0 3 3 100.0 100.0 2 5 42.9 42.9 14.3 100.1 7 6 33.3 33.3 33.3 99.9 3 7 100.0 100.0 1 9 34.8 56.6 4.3 4.3 100.0 23 All finishes 51.3 40.8 5.3 1.3 1.3 100.0 76 2nd construction 1 27.7 57.9 11.3 0.6 2.5 100.0 159 2 100.0 100.0 2 3 100.0 100.0 1 5 14.9 43.2 32.4 2.7 1.4 5.4 100.0 74 6 66.7 33.3 100.0 3 7 33.3 33.3 33.3 99.9 3 9 40.6 37.5 2.1 15.6 1.0 3.1 99.9 96 10 33.3 33.3 33.3 99.9 3 11 28.8 53.3 2.2 11.1 2.2 2.2 99.9 45 All finishes 28.8 48.7 12.2 0.5 5.4 1.3 3.1 100.0 386 since Paste 3 reduced ceramics never have a pure black surface color. "Blackwares" of Paste 3 are gray, dark gray, brown, or a grayish dull brown in color. Redwares are an oxidized red-brown or light red-brown, or an incompletely oxidized light brown or orange-brown color, falling in the range of 2. SYR 4-5/6-7 or SYR S-6/4-6. Paste 3 sherds have no mica visible on their surfaces and in general are characterized by a dull, grayish appearance. Both the red and brown slips on the Paste 3 mono- chrome-slipped ceramics are thin and do not cover the nonplastic inclusions in the paste or marks left by surface finishing. Red slips fall in the range of 7.5R 3/4-8, lOR 3/6, 4/6-8, or 2. SYR 3/4-S, S/6, while brown slips can be SYR S-6/4, 7. SYR S/3- S, 6/4, or lOYR 5/3-4. Red and brown slips are applied with equal preference on both the reduced and oxidized sherds. Paste 3 blackwares form the majority of the first construction sample, while redwares are more common in the sample from the second construc- tion levels. The majority of sherds are unslipped, but slipping, when it does occur, is found princi- pally on the burnished sherds, particularly those in the fine finish category 7. Red slips predominate over brown, which are found in the first construc- tion episode sample only on wiped finish category 9 sherds. The presence of interior and exterior slipping on finish category 1 specimens reflects the use of this finish for jar necks, which when slipped are usually slipped on both sides. Other slipped jars are reflected in the finish category 9 figures, while exteriorly slipped bowls and open forms oc- cur in finish categories S, 6, 7, and 9. Interior-only slipping is not found in the Paste 3 first construc- tion episode sample, and only infrequently in the second construction sample. Carbonized Paste 3 sherds are extremely rare. Only two examples were found, and as both are body sherds, they provide no information on pos- sible vessel shape. Both have wiped finishes, one from category 1, the other from category 9. One of the sherds has a sooty deposit on the interior with a blackened exterior face, and the other is carbonized on the exterior only. In general, it ap- pears that Paste 3 vessels, at least as they were employed at Tumatumani, were not a class of do- mestic cooking ware but rather were restricted to jars, bowls, and fancier decorated ceramics. Shapes \ Only a small number of Paste 3 shapes can be j defined from the Tumatumani data, reflecting both I 44 Part Two the limited range of vessels found at the site and the smaller diagnostic sample available to define them. Almost all Paste 3 rims and bases are oxi- dized or incompletely oxidized; jar necks are the only vessel shape with superficially reduced ex- amples, although these form a minority. Jars (Figs. 70-74)— The largest sample of rim sherds for the Paste 3 shapes is a group of everted or slightly everted rims that have been classified as jars. Only one example (Fig. 71) has a flat lip shape; all others have rounded lips, often with an indeterminate or exterior medial thickening. Di- ameters of these vessels range from 11 to 18 cm, with an average of 1 4 cm. Diameters of the slightly everted rims tend to fall at the lower end of the range. Finishes on Paste 3 jar necks always consist of an even parallel horizontal wipe on both sides, in the majority of cases from category 9, although some category 1 examples are found. Occasionally the interior has a slightly more even finish. The characteristic switch to a vertical scrape or burnish below the neck joint, such as is found on Paste 1 examples, does not occur on Paste 3 rims, even on those sherds large enough to preserve a fair section of the wall. The majority of rims have a light brown, light reddish brown, light gray-brown, or orange-brown unslipp>ed surface color. Only two of the ten examples are slipped, one light brown and the other red-brown, on both faces. Like the Paste 1 examples, these necks occasionally had embellishments, consisting of diagonal or punctate incisions, at the joint between neck and body (Fig. 74). No fillet or applique examples were found, or any rim-to-body strap handles. All Paste 3 handles encountered in the excavations are horizontally or vertically oriented strap handles located on the body of the vessel. The majority of these probably belonged to the jars described here, although some may have come from other forms. Miscellaneous Rims (Figs. 75-77)— The Paste 3 sample includes a number of single example or unclassifiable shapes. The first is characterized by its distinctive rim, which is exteriorly thickened and somewhat rectangular in profile, forming a band of 2-3 cm around the rim of the vessel (Fig. 76). Diameters of the two examples known are 14 and 18 cm. Both examples are finished with a category 9 wipe on the interior and exterior; one has in addition a red-slipped exterior. These rims very likely belong to tall-necked jars, but the pos- sibility of a straight-sided bowl shape cannot be discarded. Another possible bowl is illustrated in Figure 77. Only two examples are known of this type. Of the two. Figure 77 is slightly more vertical in ori- entation. These rims have a category 9 wipe on the exterior and a category 1 wipe on the interior, which in both cases is slightly more even in con- tour. One has an exterior red slip. Diameters are 14 and 22 cm. The rim sherd illustrated in Figure 75 is perhaps the most important of the undecorat- ed Paste 3 pieces. It is a single scallop from the rim of a straight-sided, slightly flared vessel, prob- ably an incensario shape. The diameter of the ves- sel is approximately 18 cm. The sherd is well fin- ished, although an unslipped red-brown, with a category 3 complete coverage low luster vertical burnish on the exterior and a low luster horizontal burnish on the upper 3 cm of the interior. Bases (Fig. 78)— Paste 3 bases are flat and sim- ilar in profile to Paste 1 bases. Exterior finishing, whether wiped, scraped, or burnished, is always horizontal and, in contrast to the Late Sillumocco Paste 1 examples, is uniform down to the joint between wall and base. One ring base fragment (Fig. 78) was also found. It has a horizontal cat- egory 9 wipe on both faces, an unslipped light brown exterior, and a light red-slipped interior. The ring itself is almost vertical in orientation. No thickened edge bases exist in the Paste 3 sample. Dating the Paste 3 Undecorated and Monochrome-Slipped Ceramics A similar problem arises in trying to date the Paste 3 plainwares as was encountered in dating the Paste 1 wares. The presence of these sherds in the first construction fill gives them either an Early or Late Sillumocco date. Using the same process of reasoning as was applied to the Paste 1 assem- blage, the distinctiveness of the Paste 3 vessel shapes, finishes, and firing practices from the Tu- matumani Early Sillumocco Chiripa-related wares would indicate that they belong to a different pe- riod in the Tumatumani ceramic sequence. A comparison of the Paste 3 assemblage with other ceramic sequences in the Lake Titicaca Basin in- dicates a relationship to both Qeya and Qalasasaya ceramics. This would place the Paste 3 assemblage in the Late Sillumocco period and would make it roughly contemporary with the Late Sillumocco Paste 1 group. As with Paste 1 , scallops from un- decorated incensarios (Fig. 75) are found in the sample, indicating the existence of a scalloped in- censario shape in the Paste 3 assemblage and re- Analyses of Ceramic Artifacts 45 lating this assemblage to the Qeya period. Painted Qeya-related ceramics (Fig. 155) were also pro- duced in Paste 3. However, Paste 3 ceramics ex- hibit certain similarities to Qalasasaya (Tiwanaku I) ceramics not found in the Paste 1 sample. This suggests that at least some of the Paste 3 ceramics are slightly earlier than the Paste 1 examples or were manufactured in an area with closer ties to the Qalasasaya sequence. The ring base found in the Paste 3 sample, for instance, is vertical in ori- entation, in contrast to the flaring rings of Paste 1 specimens. Vertical rings are found only rarely on Qeya incensarios but occur on all examples of Qa- lasasaya ring-based shapes (Ponce Sangines 1971: Figs. 2-2,5,21,23). The thickened rim illustrated in Figure 76 also can be compared to Qalasasaya ceramics. This rim is distinctive at Tumatumani but bears a resemblance to a similar rim on a tall- necked jar in the Qalasasaya assemblage (Ponce Sangines 1971: Fig. 3-1; Boero Rojo 1980: 279), suggesting that the Tumatumani example is indeed from a tall jar rather than a straight-sided bowl. In characteristics of surface color and slip, the Paste 3 ceramics are also similar to Qalasasaya period wares. Unincised Qalasasaya ceramics are characterized by simple red bands on a light brown slip (lOYR 6/4), or are an unslipped dark yellow- brown, reddish brown, or dark gray (Ponce San- gines 1971: 11-12; Bermann 1990: 117). Un- slipped Paste 3 plainwares occur in the same color range, and of all the Tumatumani pastes, the slipped Paste 3 sherds are the most likely to have light brown slips in the SYR and 7.5YR range and are the only group to have a significant number of sherds slipped yellow-brown in the lOYR range, like the Qalasasaya ceramics. The Qalasasaya wares are said to have a light orange-brown paste or a paste similar to color to the light brown slips (Ber- mann 1990: 462; Ponce Sangines 1971: 18). Paste 3 wares are also in the majority fired light brown, as opposed to the more commonly found red- browns and blacks of Paste 1 . Other aspects of the plainwares are not comparable: the wiped, grainy appearance of Paste 3 plainwares is unlike Qala- sasaya finishes, nor is it likely that there is a great similarity between Paste 3 itself, with its dense white inclusions and scarce mica, and Qalasasaya pastes, which are characteristically mica-tempered (Ponce Sangines 1971; Bermann 1990; Albarra- cin-Jordan and Mathews 1990: 58). However, among the most common decorated ceramics in Paste 3 are incised specimens that share certain similarities with Qalasasaya decorated wares. The dates of the Qalasasaya phase at Tiwanaku are still unclear. Since it is possible that Qalasasaya is in part contemporary with Chiripa Mamani in the earlier phases of its development, any Qala- sasaya-related Paste 3 plainwares might be con- temporary with the early phase of the Qalasasaya sequence rather than the later, and thus date to the Early Sillumocco period at Tumatumani. Be- cause Qalasasaya and Chiripa ceramics have not been found in association (Bermann 1990: 85; Al- barracin-Jordan and Mathews 1 990: 1 87), it is dif- ficult to clarify their relationship. However, the radiocarbon dates associated with the Qalasasaya ceramics illustrated by Ponce Sangines (1971), to which the shapes of the Paste 3 ceramics are com- pared, indicate that these vessels date to the later part of the Qalasasaya period. This, plus the in- ternal evidence at Tumatumani, which implies a successive tradition of fiber-tempered and non- fiber-tempered ceramics, suggests that the bulk of the Paste 3 ceramics can be placed in the Late Sillumocco period. Tiwanaku Period Paste 3 Undecorated and Monochrome-Slipped Ceramics The Paste 3 ceramics described in this section are believed to date to the Tiwanaku period. Like the Paste 1 assemblage, they are identified prin- cipally by the presence of category 1 0 and 1 1 fin- ishes on their surfaces. However, as was the case for the Paste 1 ceramics, at least some of the earlier vessel shapes and finishes of Paste 3 probably con- tinued to be manufactured in Tiwanaku times. Although these sherds cannot be identified on an individual basis, the rim and body sherds from the earlier finish categories 1-9 found in the second construction fill exhibit a sufficient number of dif- ferences from those of the first construction fill (see Tables 4 and 6) to imply that at least some of these sherds were made in Tiwanaku times and that the differences are due in part to changes in ceramic technology that occurred in the Tiwanaku period. The percentage of blackwares, for instance, drops considerably in the second construction fill, with redwares becoming more prevalent than blackwares. The number of sherds from finish cat- egories 2, 3, 6, and 7 also drops, implying that these finishes were used less frequently or perhaps not at all in second construction times. The Ti- wanaku period finishes probably consisted mostly of categories 1, 5, and 9, as well as the new ad- 46 Part Two ditions of 10 and 1 1. Burnishing patterns, how- ever, remain the same within the first and second construction fill samples: burnishing occurs pre- dominantly on the exterior of specimens from both fills, with a few examples burnished on both sides. Unlike the Paste 1 sample, interior-only burnish- ing is very rare in Paste 3 sherds in the second construction fill, occurring on only one specimen, a finish category 5 body sherd. The use of slip in general increases in the second construction episode sample. Finish categories 5, 6, 7, and particularly 10 are represented by the greatest numbers of slipped sherds, followed by finish categories 1 , 9, and 1 1 . There are no slipped sherds in categories 2 or 3, nor are there any in the first construction episode sample. Finish cat- egory 5 shows the most marked increase in slip- ping, while the percentage of slipped sherds in fin- ish category 7 actually decreases. The total sample of sherds in category 7, however, is particularly small and does not form a sufficient base for anal- ysis. Red slips continue to be more popular than brown slips in the second construction episode sample, but the percentage of brown slips does increase. In contrast to the Paste 1 assemblage, the earlier finish categories 1-6 also show an increase in the use of brown slips, although categories 10 and 1 1 still show a larger percentage of brown- slipped sherds than the earlier finish categories. Interior-only slipping is again rare, occurring on only two category 5 sherds. In their lower per- centage of blackwares and an increase in the use of brown slips, the Paste 3 plainwares are more similar to the Tiwanaku utilitarian wares de- scribed by Girault (1990) than are the Paste 1 ceramics. Shapes (Figs. 79, 80) All Tiwanaku period Paste 3 rim sherds, both the two examples illustrated and smaller sherds whose shape could not be identified, have rounded rims and burnishes that, when burnish direction could be determined, were applied horizontally on both the interior and exterior faces. All pastes were fired to an oxidized red-brown color or were in- completely oxidized to a red-brown with a dark core. Diagnostic shapes for the Tiwanaku period Paste 3 undecorated and monochrome-slipped wares include only two vessel forms. One of these (Fig. 79) is apparently a relatively straight-sided shap)e with a slightly flared rim, perhaps a cup, bowl, or tall-necked jar. It is slipped red on the exterior and on the interior of the rim and bur- nished to a medium luster with a category 1 1 fin- ish. The other specimen (Fig. 80) is probably from the neck of a small jar or bottle. It is slipped light orange-brown and finished with a category 1 1 bur- nish on both sides. Paste 5 Undecorated and Monochrome-Slipped Ceramics The Paste 5 group is the most common of the fiber-tempered pastes at Tumatumani. However, vessel shape, surface color, and slip are distinct for the high-fiber and reduced-fiber versions of this paste and must be addressed separately. Ceramics of the high-fiber version of Paste 5, as well as those of Pastes 6, 7, 8, and 9, are believed to be contemporary with ceramics of the Chiripa culture in Bolivia and thus to date to the Early Sillumocco period in the Tumatumani sequence. The relation of the Tumatumani fiber-tempered ceramics to those from Chiripa is based not only on the shared characteristic of fiber temper, typical of Chiripa ceramics, but also on the marked similarity in ves- sel shape, finish, and decorative style between Tu- matumani fiber-tempered decorated ceramics and specimens from the site of Chiripa itself (see pp. 53-54). The fiber-tempered plainwares manufac- tured at Tumatumani in these same pastes are assumed to be contemporary with the decorated ceramics and also to date to the Early Sillumocco F>eriod. This assignment is reinforced by the nu- merous similarities between the Tumatumani fi- ber-temp)ered plainwares and the Chiripa plain- wares. There are fewer published examples of Chiripa plainwares than of the decorated ceramics, making comparisons more difficult, but a number of shared characteristics can be identified. Overall similarities between the Tumatumani fiber-tempered plainwares and those from Chiripa are evident in several aspects of vessel technology. The majority of Chiripa plainwares, for instance, are brown (Mohr 1966: 139; Bennett 1936: 439; Browman 1980: 1 10). Unslipped plainware spec- imens in all the fiber-tempered pastes at Tuma- tumani are also principally brown. Like Chiripa ceramics (Mohr 1966: 92; Albarracin-Jordan and Mathews 1990: 58), Tumatumani fiber-tempered plainwares are almost uniformly burnished, in contrast to the many wiped specimens in the non- fiber-tempered pastes. Similarities in plainware vessel shap>es can also be found. Straight-sided Analyses of Ceramic Artifacts 47 bowls with direct or thickened rims and direct- rimmed jars and ollas dominate the vessel shape assemblage at Chiripa (Mohr 1966: Table 7) as they do at Tumatumani. Chiripa ollas are said to be globular, with a slightly constricted orifice and a rim with differing degrees of flare (Bennett 1936: 439). Rim shapes for both jars and ollas at Chiripa can be either direct, exteriorly thickened, or thick- ened with a triangular profile (Mohr 1966: 37). Tumatumani jars and ollas are found with the first two of these rim shapes and are directly compa- rable to the examples illustrated from Chiripa (Mohr 1966: Figs. 9, 10a, 12, 27a). The charac- teristic Chiripa vertical straight-sided bowl, which occurs in both plain and decorated versions at Chiripa (Bennett 1936: 439, Figs. 27m,o, 28i; Wil- ley 1971: Fig. 3-54; Layman and Chavez 1966: 224), is also found at Tumatumani in plain (Fig. 85) and decorated (Figs. 113-115) examples. These straight-sided bowls can also be flared in profile (Layman and Chavez 1966: 225; Ponce Sangines 1970: Fig. 27) (Figs. 95, 1 16-118). As at Chiripa (Mohr 1966: 37, Figs. 5-8, 17, 18, 23), straight- sided bowls at Tumatumani occur with both direct (Figs. 105, 116, 117) and thickened (Figs. 113- 1 1 5) rims. A further shared form is the open bowl with slightly convex sides (Bennett 1936: 443, 449) (Fig. 106). Details of vessel shape offer further opportu- nities for comparison. Rectangular or oval body lugs, for example, which occur at Chiripa only on the plainwares (Bennett 1936: 439; Mohr 1966: 102, Figs. 391, m; Albarracin-Jordan and Mathews 1990: Fig. 6a lower left) are also found at Tu- matumani, where they occur on both plain and decorated wares in Paste 5, while circular lugs (Mohr 1966: Fig. 39k) occur in Pastes 5 and 9 and on one Paste 5 decorated specimen (Fig. 121). Rim- to-body handles on ollas (Bennett 1936: 449; Mohr 1966: 101) are also shared by the two assemblages. Finally, the typical Chiripa exteriorly thickened rim (Bennett 1936: Figs. 27a,h,k,m; Mohr 1966: Figs. 5-8) can be found at Tumatumani on both plain and decorated wares. Surface Color and Slip— Early Sillumocco Paste 5 All rim and base sherds as well as all body sherds of the high-fiber Paste 5 are burnished on one or both faces and are always finished from category 8. Forty-two percent of the body sherds are slipped, a much higher percentage than in the reduced-fiber version. Red slips, the most common, are lOR 3-4/6, 4/8 or 2.5YR 3/4-6; brown slips are usually a dark brown 5YR 3/2-3 color. In the unslipped group, a dark or very dark brown surface color is the most commonly found, forming 74% of the unslipped sample. Other unslipped sherds have a red-brown surface color, usually achieved by cool- ing the vessel after firing. Fiber voids are often visible on the surface of the sherds but are not as abundant as on Paste 7 and 9 specimens. Shapes Diagnostic shapes for the high-fiber Paste 5 as- semblage include jars, ollas, straight-sided bowls, and incurving bowls. Technologically, the as- semblage is fairly uniform. Rim and base spec- imens have a red-brown surface paste color, most- ly cooled, although some oxidized and incom- pletely oxidized examples are found. All specimens are burnished on both faces, and almost all have an overall red or red-brown slip. Only two diag- nostic sherds have brown slips, and only one is unslipped. Jars and Ollas (Figs. 8 1-84)— The most com- mon diagnostic rim sherds found in the higher- fiber version of Paste 5 belong to jar and ollas forms. Eight specimens exist in this group. The jars (Figs. 81, 82) have tall, slightly everted necks that curve to a convex body form. Diameters range from 9 to 1 8 cm. All examples are burnished hor- izontally to a high or medium luster on both the interior and exterior. Burnishes are of both the complete and incomplete coverage varieties of cat- egory 8. These jars show the most variation in the use of slips: one example is slipped brown, one red-brown on the exterior and on the interior neck portion, and one is an unslipped mottled black/ dark brown. Paste 5 ollas (Figs. 83, 84) have short necks, either everted or slightly everted, and often with a slight interior thickening at the lip. The joint with the body of the vessel is generally at a more acute angle than in the jar group. In more than half the sample, these vessels have an oval or squared oval rim-to-body strap handle. These handles are 2-3 cm wide. All examples are slipped red or red-brown and burnished horizontally to a medium or high luster on the exterior and the interior wall of the neck. The interior of the vessel itself has a wiped surface. Diameters range from 48 Part Two 9 to 17 cm, with the majority in the 9-10-cm range. Straight-Sided Bowls (Fig. 85)— Two speci- mens were found of a bowl with straight, almost vertical sides and an exteriorly thickened rim. They are slipped red or red-brown on both sides and have a high luster, somewhat incomplete cov- erage horizontal burnish on the exterior, and a similar low or medium luster burnish on the in- terior. Diameters are 18 and 26 cm. The example illustrated represents something close to the com- plete vessel height for this type. Although almost imperceptible in the drawing, it curves in slightly at the bottom, as if broken at the joint with the base. Miscellaneous Rims (Fig. 86)— Although they are too small for use in determining vessel shape, rounded exteriorly thickened rims typical of Chi- ripa wares (cf Bennett 1936: Figs. 27a,h,i; Mohr 1966: Fig. 5) have been found in the Paste 5 as- semblage, slipped red and burnished to a high lus- ter. Published examples of these rims all occur on straight-sided bowls, and it is likely that the Tu- matumani specimens also belong to this shape. Finally, one rim specimen was found of an in- curving bowl (Fig. 86). It is slipped red and bur- nished horizontally to a high luster on both faces. Bases (Figs. 87, 88)— All bases in the Paste 5 assemblage are flat, but two forms were found to occur. The first (Fig. 87) has straight, nearly ver- tical-sided walls and is probably the base for the straight-sided bowl shape. Diameters range from 1 2 to 18 cm. These bases are slipped red and have a medium luster horizontal burnish on both sides. Both finish and slip reach to the bottom of the vessel wall on the exterior. The other base type (Fig. 88) appears to belong to a more convex-sided vessel. Diameters are 16 and 18 cm. The bases are burnished on both sides to a low or high luster and are slipp>ed on the interior and exterior or on the exterior only. Lugs and Spouts (Fig. 89)— Two examples were found of lugs located on the exteriors of Paste 5 body sherds. Judging from their small size, they appear to be decorative rather than functional. One is a narrow horizontal ridge at least 4.5 cm long (the piece is broken), the other is a circular button 1 cm in diameter. The most unusual ex- ample in the Paste 5 assemblage, and indeed the only one of its kind at Tumatumani, is illustrated in Figure 89. It is a trough-shaped spout with a protruding lip from a surface sherd at the site, unfortunately now somewhat eroded. The perfo- ration for the spout opening is made at a diagonal angle, and the lower lip is cupped and projects outward for directing and pouring liquids from the vessel. The spout had an interior opening of ap- proximately 1.5 cm and is located just below the rim of a large (21 -cm) straight-sided bowl. Both the interior and exterior of the vessel are slipped red and burnished horizontally to a medium lus- ter. To my knowledge, this is the only published example of an open projecting lip spout on a Chi- ripa period vessel, although a similar spout, found on a Tiwanaku shape, has been illustrated by Gi- rault(1990: 218). Surface Color, Slip, and Vessel Shapes— Late Sillumocco Paste 5 Ceramics of the reduced-fiber version of Paste 5 generally date to the Late Sillumocco and Ti- wanaku periods at Tumatumani, although a small number of Early Sillumocco Qaluyu-related dec- orated ceramics were manufactured in this paste. Undecorated and monochrome-slipped ceramics in the reduced-fiber version of Paste 5 believed to date to the Late Sillumocco period show marked similarities in vessel shape and finish to the Late Sillumocco Paste 1 assemblage. Sherds of the re- duced-fiber Paste 5, particularly the plainwares, were more likely to be found in the first construction episode levels, where they form 39% of the total Paste 5 sample. Their numbers fall to 1 2% in the second construction levels, while the percentage of the high-fiber version rises, as is the pattern in the early/late mix of the second construction fill, a portion of which was evidently mined from a Early Sillumocco occupation in the area. Because of the similarities in paste composition and surface finish and the lack of fiber voids on their surfaces, sherds of the reduced-fiber version of Paste 5 are often indistinguishable from Paste 1 specimens. The percentages of blackwares, redwares, and slipped sherds in this portion of the Paste 5 sample are comparable to the Paste 1 data. The finishes used on these sherds are those more commonly found on Paste 1 and the other non-fiber-tempered pastes at Tumatumani; most of the sherds in finish categories 5 and 6, and all of those in finish cat- egories 1 and 1 1 , appearing in Table 4 are of the reduced- fiber version of Paste 5. The sample of Late Sillumocco plainware shapes for the reduced- fiber Paste 5 is small but is virtually identical to Analyses of Ceramic Artifacts 49 the Paste 1 shape vocabulary. Jar necks with ver- tical strap handles at the rim and a grooved flat lip were found, as well as straight-sided open forms and a large bowl shape similar to Paste 1 bowl A but with a less pronounced thickening at the rim. Tiwanaku Period Paste 5 Undecorated Ceramics Vessels of the reduced-fiber version of Paste 5 were also manufactured during the Tiwanaku pe- riod at Tumatumani, as demonstrated by the dis- covery of a number of Tiwanaku-style decorated sherds in this paste (Figs. 162-165). Unfortunate- ly, no diagnostic sherds were found in the sample to define possible contemporary Paste 5 utilitarian wares. Perhaps this version of Paste 5 was used in the Tiwanaku period only for decorated ceramics. From the small percentage of reduced-fiber Paste 5 sherds in the second construction episode fill, this was likely the case. However, if utilitarian wares were manufactured, they were, like their Late Sillumocco counterparts, probably similar to Tiwanaku period Paste 1 utilitarian wares. Infor- mation from the sample of body sherds shows that reduced-fiber Paste 5 sherds occur in the second construction fill in finish categories 1, 5, 6, and the late finish category 1 1 . In addition, the second construction level figures reflect changes compa- rable to similar trends that occurred in the Paste 1 plainwares in Tiwanaku times: a drop in the percentage of blackware sherds, a corresponding increase in the percentage of redware oxidized sherds, and an increase in the total number of slipped sherds. Paste 9 Undecorated and Monochrome-Slipped Ceramics Paste 9 is the only other paste group yielding enough examples of plain or monochrome-slipped wares to allow a discussion of shape categories. Like the high-fiber Paste 5 specimens. Paste 9 plainwares are believed to date to the Early Sil- lumocco period at Tumatumani. All Paste 9 body sherds and rims are burnished on one or both faces, and only one specimen, a body sherd, has a burnish from other than category 8. Burnishes lend to be of a medium or high luster, although some low luster examples are found. Fiber voids are often visible on the surface of the sherds. Paste 9 sherds are equally as likely to be slipped as not, but slipping, when it does occur, is usually brown (SYR 4/5, 7.5YR 2/2, 4/4, and lOYR 4/4), in contrast to the predominantly red-slipped Pastes 5 and 7. The occasional red slip can be either lOR 4/6 or a more red-brown 2.5YR 3/4 or 4/6. Of the unslipped wares, body sherds with a black or dark brown surface color are more common than those with a red-brown or light red-brown color. Shapes All rims and bases in the shape sample are slipped and burnished on both sides, and all have red- brown or light brown surface colors, mostly achieved by cooling. The range of shapes in the monochrome-slipped ceramic sample is similar to that of the Early Sillumocco version of Paste 5, including jars, ollas, and straight-sided bowls. Paste 9 ollas, however, with their unthickened rims, are more similar to Paste 7 ollas (Fig. 1 7) than those in the Paste 5 assemblage. Incurving bowls are not represented in the Paste 9 sample, but a new shape, a flared bowl, does occur. Jars and Ollas (Figs. 90-93)— Paste 9 jars and ollas are generally large, with diameters ranging from 13 to 25 cm. All rims are rounded and with- out thickening. Necks of the Paste 9 ollas are short and vertical in orientation, while those of the jars (Figs. 92, 93) are slightly taller and everted or slightly everted in profile. Specimens are slipped brown, dark brown, or reddish brown on both faces and are finished with a complete or slightly incomplete horizontal category 8 burnish of me- dium or high luster. The interiors of the vessels, even below the neck section, are also burnished, at least as far down the interior wall as is preserved in the sample. One specimen (Fig. 91) still pre- serves a complete vertical strap handle, 2.1 cm wide, extending from just below the rim to the upper body wall. Straight-Sided Bowls (Figs. 94, 95)— Two categories of Paste 9 bowls were found at Tuma- tumani. The first (Fig. 94) is straight-sided with a vertical or slightly flaring wall and a slight interior beveling of the rim. Diameters are large, 23 and 28 cm, in the two examples found. They are slipped brown on both faces and are finely burnished with horizontal strokes on either side. One example has a medium luster, the other has a medium luster exterior and a high luster interior. The other bowl form (Fig. 95) also has straight-sided walls but is 50 Part Two more flared in profile. The one example known is slipped brown and burnished horizontally to a high luster on the exterior and a medium luster on the interior. Miscellaneous— No Paste 9 bases were found at Tumatumani, but, judging from the other fiber- tempered examples, they were probably flat. One exteriorly thickened rim of a type illustrated by Bennett (1936: Fig. 27a) was found in the Paste 9 sample. Again, it is too small to determine vessel shape, but probably belongs to a straight-sided bowl. The Paste 9 sample also includes one ex- ample of a decorative circular lug on an unslipped dark gray body sherd. Incised and Polychrome-Painted Ceramics from Tumatumani In contrast to the previous discussion of undeco- rated and monochrome-slipped wares, grouped according to the paste in which they were manu- factured, the decorated ceramics from Tumatu- mani are discussed according to identifiable cul- tural styles or similar decorative techniques. The particular pastes from which the sherds are made are referenced under each subheading where ap- plicable. Most of the decorated sherds come from the surface collections at Tumatumani, with only a minority from the excavation fill units. Conse- quently, there is no stratigraphic archaeological evidence that can be used to link the sherds under each subheading together as contemporary. In fact, many of the groups probably contain sherds from different phases of the stylistic traditions to which they are presumed to belong, but the phases cannot be distinguished with available evidence. Thus, the sherds are grouped for discussion solely on the basis of shared cultural and stylistic features. Early Sillumocco— Qaluyu Incised and Painted Ceramics (Figs. 96-104) Incised and painted ceramics sharing stylistic features with ceramics from Qaluyu are found only in small quantities at Tumatumani and in a lim- ited range of shapes, decorative techniques, and motifs. No Qaluyu everted "shelf rims (Lum- breras and Amat 1968: Fig. Id), neckless ollas {ibid.: Fig. le) or exteriorly thickened rims with the point of maximum thickness at the lip {ibid.: Figs. lb,g,h) occur at the site, nor were any vessels found that could be identified as Qaluyu or Qalu- yu-related plainwares. Seventeen of the 1 8 iden- tifiable Qaluyu-related sherds found are of Paste 1, the reduced-fiber version of Paste 5, or Paste 6, indicating local manufacture. The only nonlocal sherd is of Paste 4. Although these Qaluyu-related vessels were made locally during the Early Sillu- mocco period at Tumatumani, there is otherwise little Qaluyu influence at the site; the Juli area during this period belonged more to the southern Titicaca Basin tradition, with stronger ties to Chi- ripa, and with the majority of domestic and dec- orated ceramics showing influences from the south. Qaluyu-related Early Sillumocco decorated ce- ramics show a remarkable degree of homogeneity in shape, decorative style, and finishing tech- niques. All examples are slipped and burnished on both sides. Burnish is from categories 6 and 7. It is always horizontal and usually of medium luster. The fiber-tempered sherds in Pastes 5 and 6 have few or no fiber voids on the surface and little fiber in the paste; indeed, superficially they do not ap- pear to be fiber-tempered. Vessels were always fired to achieve a final red-brown or light brown surface color, either through complete oxidation, incom- plete oxidation, or cooling. All rim sherds in the group come from the same vessel shape— a straight- sided, slightly flared, presumably always flat-bot- tomed bowl with a diameter range of 18-27 cm. Rim shapes are limited; the majority are thickened on the exterior with a medial point of maximum thickness, comparable to one of the rim shapes from Qaluyu (Lumbreras and Amat 1 968: Fig. 1 i). Significantly, this rim shape is not found on any other group of decorated or utilitarian ceramics at Tumatumani and is exclusive to the Qaluyu-re- lated specimens. A minor rim type in this group is a direct rim without thickening. Qaluyu-related incised sherds are illustrated in Figures 96-98. They compare very closely with the published examples of this type from the site of Qaluyu itself (Lumbreras and Amat 1968: Figs. lg,h). All examples are slipped on both sides, and colors range from red (lOR 3.5/6) to red-brown (2. SYR 3/4, 4/4-6) and dark brown (2.5YR 3/2). Incisions are U-shaped, 2-3 mm wide, and have a depth of approximately 0.8-1 mm. Figure 96 is somewhat aberrant in its slightly wider incision than the other examples in the group, but its cur- vilinear design, rim, and vessel shape fall within the parameters of the sample. All illustrated ex- amples are of Paste 1. The one, small Paste 5 specimen has a curvilinear design and exteriorly thickened rim and is notable mostly for the uneven width of its incision. Analyses of Ceramic Artifacts 5 1 Painted sherds are illustrated in Figures 99-104. Figures 99 and 100 are Paste 1 specimens. Figures 101 and 102 are Paste 5, Figure 103 is Paste 6, and Figure 104 is Paste 4. Sherds are decorated in black or dark brown (SYR 3/2) on a light back- ground, with a horizontal red border at the rim and a red-slipped interior. The red slips are most commonly lOR 3/6, although lOR 3/4 and 4/6 were also used. Background colors for the exteriors are most commonly light brown (SYR S/6, 6/5 or 7. SYR 6/S) but also include white (SYR 8/1), red- brown (2. SYR 4/6, 4/8), or brown (SYR 4/4-S, 4.S/4). Motifs consist of checkered elements, nar- row and wide crisscrossing lines or lattices, rect- angles, and concentric diamonds or chevrons. Ves- sels are again straight-sided flaring bowls with diameters ranging from 18 to 27 cm. Most rims are exteriorly thickened, but the few direct rims in the Qaluyu-related group are found in these painted versions. These ceramics from Tumatumani show affin- ities not only with the painted ceramics of Qaluyu, but also with specimens illustrated by Chavez (1969, 1981) from the site of Marcavalle, near Cuzco. Shared characteristics include the basic color scheme of black designs on a cream, light brown, or otherwise lighter background color and the decorative motifs of lattice, chevron, and checkered elements (Lumbreras and Amat 1968: Fig. lc,i; Chavez 1969: Figs, d, k, p; Chavez 1981: Figs. 41, 42, 49). One trait that distinguishes the Tumatumani ceramics from those of Qaluyu and Marcavalle, however, is the use of generally darker background colors for the black designs. Both Qa- luyu and Marcavalle ceramics have black deco- ration on what is consistently described as a cream- colored background. The variations of black on brown or black on red-brown found at Tumatu- mani therefore appear to be unique to the site. Nevertheless, the bulk of the Tumatumani back- ground colors, the light brown specimens, are di- rectly comparable to the Marcavalle and Qaluyu black-on-cream ceramics, since many of the colors defined at Tumatumani as "light brown" are the same as those listed by Chavez (1980: 30S) as "cream," the difference lying only in terminology and not in actual color. While for purposes of com- parison the Tumatumani ceramics can be referred to as black on cream or black on light brown, the larger range of colors in the sample should be taken into account. Further comparisons of the Tumatumani ma- terial disclose closer relationships to either the Qaluyu or Marcavalle ceramics. Black on cream or light brown is the only color scheme in the painted sample at Tumatumani. Black-on-cream decoration at Marcavalle, however, is basically a minor type, second to the more popular cream on brown, and forms only 20% of the painted sherds at its most popular in Phase D (Chavez 1981: 244). In contrast, black or dark brown on cream, as well as red on cream, are the major colors in the Qaluyu palette and are said to characterize this style (Cha- vez Ballon 1 9S0: 44; Kidder 1 9S6: 29; Rowe 1 9S6: 1 44). In attributes of vessel shape, the character- istic Tumatumani straight-sided flaring bowl is similar to bowls found in both the Qaluyu and Marcavalle assemblages. However, the Tumatu- mani bowls have flat bases. This attribute is rare at Marcavalle, where bases of the majority of bowls are convex (Chavez 198 1 : 1 37), but is found in all published examples of Qaluyu pottery (Lumbreras and Amat 1968: Figs. la,b,d,f; Lumbreras 1981: 8S). The characteristic Tumatumani thickened rim occurs in both the Qaluyu and Marcavalle assem- blages but does not predominate at either site the way it does at Tumatumani. The Tumatumani sample is more closely related to the Marcavalle assemblage, however, in its universal attribute of a red slip on the interior of the vessel and on the top and exterior rim. A fair portion, 2S% in Phase D, of the Marcavalle black-on-cream examples also have this same associated red slip, again found on the top of the rim and extending into the interior (Chavez 1981: 244). Interiors of Qaluyu black-on- cream ceramics are not slipped red but rather are covered in the same cream as the exterior or are unslipped. It is of some interest that the majority of black- on-cream decorated sherds found at Marcavalle, including those with an associated red slip, were not originally from the site but were manufactured in a nonlocal paste believed to come from near Pikicallepata somewhat to the south (Chavez 1981: 244, 334). Black-on-cream sherds in this same paste have also been found at Taraco {ibid.: 334). How- ever, to my knowledge, the Tumatumani examples represent the southernmost reported occurrence of black-on-cream painted sherds with associated red slip, made all the more significant by the fact that all but one of the specimens is locally made. Per- sonal work in the southern Department of Puno, including recent excavations at the site of Camata, near Chucuito, has also consistently turned up lo- cally made black-on-cream ceramics with associ- ated red slip in conjunction with the better known Qaluyu incised wares. Associated red slip itself is not new to the Titicaca Basin, since this same S2 Part Two pattern of red slip on the exterior rim and interior of the vessel is found in Cusipata ceramics from Pucara, believed to postdate and to be in part con- temporary with Qaluyu (Mujica 1 987). Designs on Cusipata ceramics, however, are executed exclu- sively in cream on a dark brown or black back- ground color. Although heretofore described in the literature only with reference to the Marcavalle and Pikicallepata assemblages, it is suggested that black-on-cream painting with associated red slip should also be included in the definition of the early ceramic styles of the Titicaca Basin, if not for Qaluyu itself, then for the Qaluyu-related local cultures on the west side of Lake Titicaca. Chavez's work at Marcavalle makes it possible to locate the Tumatumani sherds approximately in time. Black-on-cream decoration is found at that site in the Late Stylistic Division (Phases B- D) (Chavez 1981: 334). Specifically, lattice motifs in black on cream appear in phase C, and check- ered elements not until phase D (Chavez 1981: 339-340). Phases C and D are dated to approxi- mately 800-600 B.C. Tumatumani black-on-cream painted ceramics are likely to be roughly contem- porary and can therefore be placed toward the beginning of the Early Sillumocco period, coin- cident with the later Marcavalle and Qaluyu phases. However, the Tumatumani sequence may differ from the northern chronology, and an earlier or later date is also possible. Early Sillumocco— Chiripa Incised and Painted Ceramics (Figs. 105-122) The incised and polychrome-painted ceramics described here exhibit certain features that com- pare closely with the sherds illustrated by Bennett (1936) and Mohr (1966) from the site of Chiripa. Most of the Tumatumani examples, however, are manufactured in the local Pastes 5 and 6, while the Paste 9 specimens show sufficient differences in style and technique from the illustrated ex- amples to suggest that they were not made at Chi- ripa itself Rather, the Tumatumani decorated Chiripa-related ceramics are local or semilocal. Early Sillumocco versions of the Bolivian Chiripa ceramic style. Browman ( 1 980: 111) states that the examples published by Bennett and Mohr date to Chiripa Mamani Phases A and B, the last of his four-phase division of the Chiripa sequence. The similarity of most of the Tumatumani wares to these examples implies that the majority of the sample dates to later in the Chiripa sequence rath- er than earlier, placing it principally in the later part of the Early Sillumocco period. However, nar- row line incision, black-on-red designs, and care- fully executed rectilinear designs, which are said to occur with more frequency in the earlier Chiripa Llusco phase (Browman: 1980), are also present at Tumatumani, suggesting the possibility of a Llusco-related occupation at the site beginning early in the Early Sillumocco period. Certainly the Qaluyu ceramics described in the previous section are testimony to the occupation of the Tumatu- mani area during the earlier part of the Early Sil- lumocco period. Because the southern basin influ- ence, including that from Chiripa, is generally stronger in the Tumatumani ceramics than influ- ence from northern areas such as Qaluyu, one would reasonably expect to find contemporary— i.e., Llusco phase— Chiripa-related ceramics at the site. However, without more stratigraphic exca- vation in the southwestern basin and the definition of a local chronological sequence for the Chiripa- related ceramics in this area, it is impossible to determine the exact chronological position of the Tumatumani ceramics at this time, or to distin- guish between temporal differences and the pe- culiarities of local styles in the variation demon- strated by the Tumatumani sample. Incised Chiripa-related ceramics at Tumatu- mani exhibit considerable variation in slip color and incision technique. Finish and firing tech- niques on the other hand, are fairly standardized. Eleven of the 13 examples have a high luster hor- izontal category 8 burnish on both sides (the others have a medium luster burnish), and 10 of the 13 were cooled after firing to produce sharp red-brown interior and exterior edges. Single examples exist of oxidized, incompletely oxidized light brown, and gray pastes. The majority of the incised ex- amples are painted in two or more colors, most often with an overall red slip (lOR 3-4/6) applied to both faces of the vessel and with additional colors painted over the red. Incised specimens are found in Pastes 5, 6, and 9. Paste 5 examples form the largest percentage of the sample and show the greatest amount of vari- ation. Incisions on Paste 5 sherds can be a very narrow and shallow line, more like a tracing made with a stylus than an incision; a wide. U-shaped incision of 2.5-3 mm; or a medium-width incision of 1-2 mm. The stylus line-incised group (Fig. 107) exhibits the most standard Chiripa color scheme (cf Bennett 1936: Figs. 27ij,n; Mohr 1966: Figs. 26d,e): black and yellow (7.5YR 6/6-8) on a red base slip, although the use of such a nartow Analyses of Ceramic Artifacts 53 incision is probably a local Juli variation. The rim shape of Figure 107 is also very similar to one of Bennett's (1936: Fig. 27k) specimens. The wide- line incised examples are less like the ceramics from Chiripa. They are painted in two, not par- ticularly contrasting colors: one example is red on a reddish brown (2. SYR 4/6) interior and exterior base color, the other (Fig. 1 08) is painted red and light brown (7.5R 5/4) with no base red. This sherd has an unslipped red-brown interior and only a medium luster burnish. The medium-width in- cised Paste 5 sherds also exhibit an unusual range of colors: black and yellow on red occurs (Fig. 1 09), but also dark red ( 1 OR 3/2) and yellow on red (Fig. 105), grayish white (lOYR 7/2) on red, light brown (7.5YR 7/4) on dark red, red and light brown (lOYR 6/4) without a red base slip (Fig. 1 10), and monochrome-slipped brown (Fig. 106). The rim sherds in this medium-width incised group include specimens from a slightly convex-sided bowl (Fig. 106), mentioned by Bennett (1936: 443) as oc- curring at Chiripa, as well as a straight-sided form with a direct rim. Like most Chiripa bowls, these examples have large diameters. Monochrome-sUpped incised ceramics, such as the one illustrated in Figure 106, are a minor type at Tumatumani, as they are evidently also at Chiripa (Bennett 1 936: 44 1 ; Mohr 1966: Table 22). The Paste 9 incised ceramics are considerably different from the Paste 5 examples. Incision is of medium width but the edges are jagged, as if ex- ecuted on a dry clay surface. Burnishing reaches only a medium luster. The inward curve at the bottom of Figure 1 1 1 suggests that the sherd comes from near the vessel base. This sherd is slipped red on one side of the incision and dark red (lOR 3/3) on the other, without any base slip. The in- terior is slipped red. Figure 1 1 2 is decorated in black over a red slip, but is unslipped on the in- terior. The one Paste 6 incised sherd (Fig. 1 1 3) has a medium-width incision, slightly wider, deep- er, and more even than the Paste 5 examples. The interior and thickened rim portion of the exterior are slipped red; below the rim on the body of the vessel is a red slip streaked with black. Burnish luster is high on both sides. The sherd is from a vertical-sided bowl with an exteriorly thickened rim. The majority of the 2 1 Chiripa-related painted sherds are decorated in cream or yellow on a red interior and exterior base slip. As a whole they form a more homogeneous group than the incised ceramics and resemble fairly closely the published examples from Chiripa. Yellow-on-red painted sherds are found in Pastes 5 (Figs. 114-116), 6 (Fig. 1 1 8), and 9 (Fig. 1 1 7), with no major differ- ences in shape or decoration between the different pastes. The red slips are all in the Munsell range lOR 3-4/6-7; designs are painted in cream (lOYR 8/4, 7. SYR 8/4), yellow (7.5YR 6/7, 7/6), or yel- low-orange (5YR 5/8, 6/6). All examples but one are slipped red on the interior. Three of the Paste 5 body sherds are exceptions to this color range as they display darker, more brownish reds and more intense yellow slips than the rest of the group. Sherds are generally burnished horizontally to a high luster on both sides. Exterior designs are all geometric, although some have sharper lines and more angled comers than others. A few of the rim sherds have interior decoration as well, pendant triangles being the only recognizable motif (cf Bennett 1936: Figs. 27m,o, 28i). Vessel shapes for these yellow-on-red ceramics include straight ver- tical-sided bowls with exteriorly thickened rims as well as a straight-sided, more flaring bowl shape with a direct or only slightly thickened rim. Vessels are large, with diameters ranging from 18 to 30 cm. One example of a painted body lug was also found. Aside from the yellow-on-red specimens, two other groups of Chiripa-related painted ceramics were found at Tumatumani. These are less com- monly found in the published literature and were also a minority at the site. The first of these occurs in both Pastes 5 (Fig. 119) and 9 (Fig. 120) and consists of black designs on a red (lOR 3/5) or brown (7.5YR 4/4, 5YR 4/5) slipped background or an unslipped red-brown background. Interiors are an unslipped reddish brown or a slipped dark brown. Designs are rectilinear only and appear to consist mostly of stripes. The similarity of these designs to those on some of the Paste 9 jagged- edge incised wares is noticeable. Fewer of these sherds have high luster burnishes, but otherwise surface finish is the same as the yellow-on-red ex- amples. The second of these painted wares (Figs. 121,1 22) is found only in the locally made Paste 5 and perhaps is not properly called Chiripa-re- lated, for it appears to bear less relation to the Chiripa wares made in the southern Titicaca Basin than the other specimens from Tumatumani. It consists of red (lOR 3/4) designs on a red-brown- (2.5YR 3/6) or red-orange (2.5YR 4/8) slipped background. Motifs are lines and zig-zags. Surface finish is also slightly different from the other Chi- ripa-related painted wares: burnishing strokes in this case are slightly visible, in contrast to the more common undetectable burnished finish. 54 Part Two Late Sillumocco Polychrome Incised Ceramics (Figs. 123-127) The majority (six) of sherds in this group are of Paste 3; the rest of the sample consists of two Paste 1 specimens (Fig. 126), apparently less well-fin- ished local versions of the Paste 3 wares, and two, also local. Paste 2 examples (Fig. 127). Sherds are painted in red (7.5R 3/5-8, lOR 3/6), light red (7.5R 4/6) or pink (5R 5/6), white (5 YR 8/ 1 ), cream (7.5YR 8/4) or yellow (lOYR 7/6), and black on an unslipped or light brown (5YR 6/5, 7.5YR 5-6/ 5) slipped surface. Interiors are slipped light brown in the same color range or are an unslipped light brown or reddish brown. Sherds are burnished on the exterior to a high or medium luster with a category 7 burnish and have interior wipes or scrapes. Only Figure 124 is burnished on the in- terior. Fired paste color is almost universally light brown, with only one oxidized red-brown speci- men in the sample. The two rim sherds in this group both have rounded direct rims and belong to large (20 and 24 cm diameter), slightly convex- sided bowls. Incision techniques vary; the two rim sherds have a narrow, shallow (0.1 mm) stylus- type incision, while the other examples have wider, deeper (0.4-0.7 mm) incisions that can be either U- or V-shaped in profile. These polychrome sherds were found at Tu- matumani in both the first and second construc- tion episode fills as well as on the surface, so ar- chaeological evidence gives us only a pre-Tiwanaku date. However, the attribute of color areas delin- eated by incision and the decorative motifs found on these sherds are comparable to other non-fiber- tempered polychrome incised ceramics from the Lake Titicaca Basin, e.g., Classic Pucara from northern Puno and Qalasasaya (Tiwanaku I) from Tiwanaku. If similarity implies contemporaneity, then the Tumatumani specimens should be rough- ly contemporary with the ceramics of these two cultures. However, while the Tumatumani speci- mens can be related to both of these ceramic traditions, they are not strictly comparable. No examples of fancy, specifically Pucara iconogra- phy, such as felines, birds, running angels, or even trophy heads, have been found at Tumatumani, nor have any motifs, such as profile felines, been found that can be specifically related to Qalasa- saya. It is suggested that the Tumatumani exam- ples are contemporary in some way with both Pu- cara and Qalasasaya but are a local or semilocal production from near Juli, belonging to what was probably a pan-Titicaca tradition of non-fiber- tempered polychrome incised ceramics during the Upper Formative period, a tradition that had a long duration in the Pucara sequence but which was superseded in the Tiwanaku area by unincised painted decoration or simple incision on an un- painted surface at the beginning of the Qeya pe- riod. Specific comparisons of the Tumatumani ma- terial with Classic Pucara and Qalasasaya ceramics can be made with respect to vessel shape and de- sign motifs. The convex-sided bowl with a direct, rounded rim and a decorative band on the exterior border, such as is represented by the two Tuma- tumani incised rim sherds, occurs in both the Clas- sic Pucara and Qalasasaya assemblages (Ponce Sangines 1971: Figs. 3-24,25; Rowe and Brandel 1971: Figs. 54-56). The motifs represented at Tu- matumani are fairly simple, consisting of step blocks and rectilinear designs, and are also found in the design repertoires of both Classic Pucara and Qalasasaya (Ponce Sangines 1971: Figs. 3-24,25,27-29; Rowe and Brandel 1971: Figs. 49- 56). Other comparisons of the Tumatumani ma- terial with Pucara and Qalasasaya specimens re- veal a closer relationship to either one or the other. Qalasasaya incised ceramics are said to have a light brown fired paste color (Ponce Sangines 1971: 18), as do the Tumatumani examples; Pucara incised wares are fired to a red or red-brown (Franque- mont 1990). Also closer to Qalasasaya ceramics is the use of background colors at Tumatumani. On most Tumatumani specimens colors are ap- plied separately onto an unslipped surface, but in two cases (Figs. 123, 1 24) design colors are applied onto an overall light brown background slip. Qala- sasaya incised ceramics are also painted on an underlying light brown (lOYR 6/4) slip (Ponce Sangines 1971: 17), while Pucara ceramics, in con- trast, are almost always painted on a red back- ground slip color. In slip colors, the Tumatumani examples show more similarities to Pucara ceramics. Qalasasaya polychromes are reported to have only red, white, and dark gray decoration (ibid.). At Tumatumani, these colors as well as pink and yellow are found, additions that also occur, significantly, in the Pu- cara palette (cf Rowe and Brandel 1 97 1 : PI. 1 color key). The reds used in the Tumatumani wares, mostly in the 7.5R range, are also brighter than the lOR colors of the Qalasasaya ceramics and more similar to Pucara reds. The Tumatumani incised decoration, smaller and finer in scale and executed in narrower incision than the Qalasasaya Analyses of Ceramic Artifacts 55 specimens, is likewise more comparable to the Pu- cara examples. Both Pucara and possibly Qalasasaya have ear- lier phases that would be contemporary with the Early Sillumocco period at Tumatumani. The ce- ramics described here, however, are believed to date to the Late Sillumocco period, based on sev- eral lines of evidence. One is the apparent tem- poral position of the bulk of the non-fiber-tem- pered ceramics at Tumatumani after the Chiripa- related wares. As regards Qalasasaya, illustrated examples of incised ceramics of this style are known only from the later Qalasasaya period. These ves- sels were found in burials and offering caches in the lower levels of excavations in the Qalasasaya pyramid at Tiwanaku. Although several radiocar- bon samples were taken from the levels in which these burials and caches were encountered, only one is directly associated with the burials them- selves (Ponce Sangines 1971: 18, 1981b: Table 1). The date of this sample is the latest of all those associated with the Qalasasaya sequence and falls within the Late Sillumocco period. Examples of unincised Qalasasaya pottery excavated at Lukur- mata have likewise been ascribed a date pertaining to Late Sillumocco times (Hermann 1 990: 20). At Pucara, incised polychrome decoration is found in the pre-Classic phases of the site, phases that would be contemporary with Early Sillumocco. Pucara pre-Classic bowl shapes, however, all have beveled rims (Franquemont 1 990: Figs. 8-12; Mu- jica 1987: Fig. 7) and are very unlike the Tuma- tumani bowls. Bowls with direct rims do not occur in the Pucara sequence until Classic Pucara times. Step block designs are also found in the pre-Classic Pucara repertoire, but when they occur on bowls they are found almost exclusively on the interior of the beveled rim. This is very unlike the exterior decoration of the Tumatumani examples but is an attribute that also occurs on Classic Pucara bowls. Ceramic Tubes or Trumpets (Figs. 128-133) Nine fragments of ceramic tubes with grass im- pressions on their interiors, plus four fragments of tubes without grass impressions, were found at Tumatumani. These ceramic tubes, generally re- ferred to as trumpets (Kidder 1943: 3; Rowe and Brandel 1971: 15; Chavez 1988: 21) are familiar components of the Pucara (Rowe and Brandel 1971: Figs. 71-76; Franquemont 1990: Figs. 61- 65; Chavez 1988: Fig. 11) and Chiripa (Bennett 1936: Figs. 28g,h; Mohr 1966: Figs. 43-46; Cha- vez 1988: Fig. 7) ceramic assemblages, and have also been found at Taraco in the Department of Puno (Kidder 1943: Fig. 5) and in the Tiwanaku area (Ponce Sangines 1970: Figs. 30-1,2,4; Ber- mann 1990: Fig. 57c). Pucara specimens have thickened mouth pieces and a flaring distal end (Muelle and Bias 1938: PI. 71b; Bennett 1946: PI. 37c; Chavez 1988: Fig. 1 1), much like modem trumpets, while southern basin examples generally have less flare (Bennett 1936: Figs. 28g,h). The Tumatumani specimens are all body sherds, but their diameters of 3-6 cm are comparable to the diameters of complete clay tube specimens, and the presence of grass impressions on the interiors of the majority of the Tumatumani examples is a trait associated only with ceramic tubes. These grass impressions have been sug- gested to be evidence of the use of grass bundle molds to achieve a tube shape (Kidder 1943: 24; Franquemont 1990: 7; Chavez 1988: 23). Many of the published tubes from both the northern and southern basins have sooted or blackened interiors (Bennett 1936: 443; Kidder 1943: 24; Franque- mont 1990: 7; Chavez 1988: 21, 23), a trait that does not occur on the Tumatumani specimens. Figure 128, in Paste 7, and Figure 129, in the Paste 5 higher-fiber version, are the only two fiber- tempered examples in the group and appear to be slightly earlier, Chiripa-related Early Sillumocco in date, than the Paste 1, 2, and 3 sherds. Figure 128 has a gray paste and unslipped surface color. It is burnished to a low luster and decorated with red slip (pre-fire) in the incisions. It is the only grayware example in the collection and as such is comparable to the tubes from Chiripa, most of which are black or brown in color (Chavez 1988: 21). Figure 129, the Paste 5 example, does not have grass impressions on its interior, but its small diameter (3 cm) identifies it as a member of this same ceramic tube group. The modeled and in- cised design on this sherd appears to be zoomor- phic, representing the appendages of some type of animal, and bears a general resemblance to the Chiripa tubes (Bennett 1936: Fig. 28g; Chavez 1988: Fig. 7) and the examples from Taraco, which also display modeled zoomorphic figures on their exteriors. Figure 129 has a red-brown (2. SYR 4/6) slip on the modeled area and to its right, and a brown slip (7.5YR 4/4) to the left. It is finished with a category 7 burnish of high luster in the red- brown-slipped area and medium luster on the brown-slipped portion. The interior is a wiped un- slipped light brown, and the fired paste color is black with cooled light brown edges. The Paste 1, 2, and 3 ceramic tube specimens 56 Part Two form a fairly homogeneous group, with decoration characterized by rectiUnear incision and rows of dots. Eight specimens are manufactured in Paste 3. They are slipped light brown (7. SYR 4/4-6, lOYR 5/2.5), including within the incisions, and decorated with shallow (0.2-0.4 mm) U-shaped incisions and punctate dots. Figures 130 and 132 have in addition a pre-fire red slip (lOR 3/5, 7.5R 3/6) within the incisions. These two sherds, plus one example not illustrated, are finished with a category 5, 6, or 7 burnish of medium or high luster; the other examples have a faint plastic cat- egory 1 wipe. Fired paste color is generally light brown, with some red-brown specimens. The three locally made Paste 1 (Fig. 133) and Paste 2 ceramic tube fragments differ slightly from the Paste 3 group in the use of red or red-brown slips rather than brown and in their uniformly red-brown fired paste color. Two examples are slipped and burnished on the exterior with a category 6 medium luster bur- nish. Figure 133 is an unslipped red-brown and is finished with a faint plastic wipe. Again, incision is shallow and U-shaped. The Paste 3 specimen illustrated in Figure 1 3 1 stands out from the rest of the Tumatumani spec- imens as being the only polychrome example and clearly belongs to a different stylistic group. It is decorated with a fine-line incised cross and paint- ed in black, white (lOYR 8/3), and red. Incised lines delineate the central square of the cross only at the top and bottom; the sides are marked by the change of color between the black and white slips. Thus, while not a true "checkered cross," it bears more of a resemblance to the checkered crosses found on Pucara pottery (Rowe and Bran- del 1971: Figs. 63, 68, 69) than to the southern basin examples. Chiripa crosses are open with punctate dots at the end of each arm (Bennett 1936: Fig. 28g), while Qeya incised crosses are concentric or have + motifs in their centers (Ponce Sangines 1948: Fig. 38; Chavez 1985: Fig. 67). The ceramic tube fragments of Pastes 1, 2, and 3 are believed to date to the Late Sillumocco pe- riod. This conclusion is based on the similarity of the incision on these sherds, wide lines and punc- tate dots, to the Paste 1 Late Sillumocco incised ceramics described in the following section (e.g., Figs. 141, 143, 145). In addition, the faint plastic wipe found on some of the Pastes 1 and 3 tube fragments is identical to the finish on these Paste 1 incised sherds. The resemblance of the incised cross of Figure 1 3 1 to Pucara checkered crosses would also indicate a Late Sillumocco date for this unusual specimen in the sample. Their position in the Late Sillumocco period makes the Tumatu- mani trumpets roughly contemporary with Pu- cara, Qalasasaya, and Qeya. Although trumpets are known from Pucara, and are found in the Qala- sasaya assemblage as well (Ponce Sangines 1970: Figs. 30-1,2; Bermann 1990: Fig. 57c), no Qeya examples have as yet been identified. If the dating of the Tumatumani trumpets is correct, and given that the Paste 1 Late Sillumocco incised ceramics to which they show the greatest resemblance are related principally to Qeya ceramics, then the ex- istence of these incised ceramic tubes at Tuma- tumani suggests the possibility that trumpets may be found in the Qeya ceramic assemblage as well. Paste 1 Incised Ceramics (Figs. 134-146) The incised ceramics described in this section are all manufactured in Paste 1. Three groups, differing in firing techniques, surface finish, and use of slips, can be distinguished within the larger incised ceramic sample. Group A (Figs. 134-136) is composed of five sherds with dark brown, dark gray-brown, or black surface colors. Pastes are black, or have a red-brown core and thick dark interior and exterior edges. Sherds are incised with a medium-width U-shaped line. Incisions are made over a wiped surface, which generally has a more even contour and a smoother look than the sur- faces of the sherds in groups B and C. Unincised areas on group A sherds are clearly set off by in- cised borders (Figs. 135, 136) and are burnished to a medium or high luster with a complete cov- erage burnish from categories 5 or 6. Two exam- ples are burnished on the interior as well; the oth- ers have a category 1 scrape. Sherds of group A have a minimal amount of slipped decoration; on Figure 136, red post-fire paint (7.5R 3/6) was ap- plied in two of the incised dots, while Figure 135 has a red slip (pre-fire 1 OR 3/6) in the burnished area and within two of the incised circles. The six sherds in group B (Figs. 137-139) have black, gray-brown, or gray surface colors and are characterized by being only superficially reduced or smudged, producing a red-brown paste with a very thin interior and exterior dark "rind." Surface color is lighter than that of sherds in group A, and interiors are occasionally red-brown. Sherds in this group also have more mica visible on the surface than group A sherds. Surface contours are even or slightly irregular, and surface finish consists of a category 1 wipe ranging in appearance from a smoothed look with very faint wiping ridges to a Analyses of Ceramic Artifacts 57 surface with more prominent wiping marks. In- teriors are uniformly scraped. Incisions are of me- dium width and can be either U-shaped or rect- angular in profile. Motifs in this group are generally rectilinear and consist of nested rectangles, squares, and L-shaped elements, incised circles, and fields of punctate dots. Red post-fire paint (7.5R 3/6-7) is found in the incisions of all four examples in this group, while Figure 137 has both red and yellow (7.5R 6.5/8) post-fire decoration. A narrow band of red post-fire paint was also applied on the surface of the sherd shown in Figure 138. All post- fire paints are in a poor state of preservation, and it is likely that the vessels originally had paint in more of the incisions than the sample shows. The third group of incised ceramics, group C (Figs. 140-146), is distinguished by having slipped surfaces. Five specimens (Figs. 140-143) have red sHps (lOR 3.5-4/6, 7.5R 3/6) on the exterior and unslipped red-brown interiors, two specimens have a light brown (SYR 5/4, lOYR 6/3) slip on both faces, one (Fig. 1 44) has a red-slipped exterior and unslipped dark brown interior, and one (Fig. 1 46) has a cream (7. SYR 7.5/6) slip on both sides. All slips are thin and were applied after incision. The specimen in Figure 145, which has a plain red- brown surface, is the only unslipped example, but its incised decoration is sufficiently similar to the other specimens to justify its inclusion in this group. Motifs include parallel lines, double circles, and triangles or angles with fields of punctate dots or a single dot in their comers. Six of ten specimens have post-fire paint in the incisions; four have both red and yellow paint (Figs. 141, 142, 144, 146), and two have only red. Incisions have U-shaped contours and are generally wider and deeper than the incisions on sherds in groups A and C. One sherd (Fig. 141) exhibits the technique of jab-and- drag or stamped incision; the circle on this sherd was created by pushing the incising tool into the clay, then dragging it in a short arc. Surface finish on the sherds of this group is again a category 1 wijje; surfaces have only very faint wiping marks, and none show marked striations. Interiors are scraped. Most examples have oxidized red-brown pastes or red-brown pastes with an incompletely oxidized core, although very superficially reduced gray edges are occasionally found in the sample. Groups B and C of the Paste 1 incised ceramics show a noted similarity to a class of incised wares, generally scalloped-rim flat or pedestal base in- censarios, often with feline heads, found in both the northern and southern Lake Titicaca Basin. They form part of the Qeya assemblage as defined by Wallace (1957) and were included by Bennett in his corresponding Early Tiahuanaco period (Bennett 1 934: 409). Chavez ( 1 985), in an analysis of similar sherds from the Departments of Cuzco and Puno, has likewise assigned her sample a Qeya or Early Tiahuanaco date; incised incensarios have also been found in late Qeya contexts at Lukur- mata (Bermann 1990: 195, Fig. 77b). Aside from the illustrations found in the studies of Bennett (1934: Fig. 13a), Wallace (1957: PI. lc,e), and Cha- vez (1985: Figs. 1-58, 65-69), examples of com- plete incised incensarios can be found in various other publications (Eisleb and Strelow 1980: Figs. 3-5; Fellman Velarde 1961: PI. 6; Ponce Sangines 1948: Figs. 38, 39; Ibarra Grasso 1957: Fig. 74; Bandelier 1910: PI. 21 left; Means 1931: Fig. 56 lower left and center; Ruben 1952: Fig. 30 left). Similarities of the Tumatumani sherds to the Qeya incised wares are numerous and specific. Like the Tumatumani sherds, Qeya incised ceramics have unbumished matte luster smoothed surfaces, occasionally with visible striation marks (Wallace 1957: 48; Chavez 1985: 159). Both Wallace and Bennett mention only unslipped examples, but many of Chavez's specimens are slipped. Post-fire paint applied in the incisions of Qeya ceramics is also characteristic. Wallace (1957: 48) noted the colors red, yellow, and white for the Qeya Qollu Chico wares, and Bennett (1934: 409) described the same colors for his material, although he did not specifically state it was post-fire. Further ref- erences to the colors red, orange, and white can be found for the southern basin examples (Eisleb and Strelow 1980: 23) and red and white for the northern and southern basin (Chavez 1985: 141; Bermann 1990: Fig. 77b). The Tumatumani pal- ette appears to have been more limited, since nei- ther the white nor orange colors are found at the site. Although the Tumatumani sherds probably do not show complete preservation. Figures 139, 142, and 146 suggest that post-fire paint was ap- plied in alternating colors in the incisions, or per- haps in alternating painted and unpainted incised lines. Alternation of post-fire colors has previously been noted as a characteristic of Qeya period in- cised ceramics (Chavez 1985: 141). The design motifs on the Tumatumani ceramics also compare well with the published examples of Qeya incised ceramics. Nested rectangles and L-shaped motifs are found particularly in the examples from the southern basin— Tiwanaku, the Tiwanaku area, and Qeya Qollu Chico (Chavez 1985: Figs. 67, 68; Bandelier 1 9 1 0: PI. 2 1 left; Eisleb and Strelow 1 980: Fig. 3 [compare especially Fig. 137, this paper]). 58 Part Two Fields of dots within triangles are also found on specimens from this area (Eisleb and Strelow 1 980: Fig. 3; Ponce Sangines 1 948: Fig. 38; Chavez 1985: Figs. 65, 66), while parallel lines and circles are found on both southern and northern basin ex- amples (Chavez 1985: Figs. 21, 25, 29, 34). Some of the specimens from Tumatumani ap- pear to have belonged to the same scalloped-edge incensario shapes as those found in the Qeya as- semblage. Figure 137, for instance, with its squared rim and straight sides, may have come from one of these vessels, from a portion of the rim without scallops, or from an open, straight-sided incised vessel without scallops or feline head, such as that on view in the Tiwanaku Museum in La Paz. The diameter of this sherd, 12 cm, is comparable to the scalloped versions illustrated by Chavez (1 985). Unslipped brown or gray-brown rim scallops have also been found at Tumatumani (Fig. 64), as well as unslipped and red-brown-slipped ring bases (Figs. 58, 59), a base shape generally found on Qeya period scalloped incensarios. Figure 1 40 rep- resents an incised example of one these hollow bases, although the total height of the indented portion of the base cannot be judged. The base illustrated in Figure 138 also approaches the ring type. The existence of feline head incensarios at Tu- matumani is evidenced by the sherd illustrated in Figure 147. While somewhat difficult to recognize, this specimen is an abstract version of the feline heads found on scalloped incensarios from both the northern and southern Titicaca Basin. As Cha- vez points out ( 1 985: 149), the position of the head on the body of the vessel, with the rim connecting just below the ears, most closely resembles Qeya period examples; earlier feline heads are entirely on the vessel body, while Tiwanaku heads are en- tirely above the rim. The sherd has an unslipped dark brown micaceous surface, and while the up- per portion is scraped only, there are a few incom- plete burnish strokes beginning at about the level of the mouth. This, plus the lack of incisions around the mouth or chin as is common on other feline heads, suggests that the piece was not incised else- where on the vessel but had a general overall cat- egory 5 burnish. The rim shape, not visible in the illustration, is squared like that in Figure 137. Although the incised specimens of both groups B and C bear a relation to Qeya period incised ceramics and incensarios, some specific geographic comparisons differentiating the two can be made. Group B, the ceramics characterized by having a gray surface color with interior and exterior "rinds" in the paste, is more comparable to the southern basin incised wares. These too tend to have a tan or gray surface color (Chavez 1985: 151), in con- trast to the predominantly red-brown surface col- ors of the material from the Cuzco-Puno area. They are also more likely to be incised with nested rectangles and L-shaped motifs, as is Tumatumani group B. The most specific comparisons are with the incised incensarios, as well as other shapes, from Qeya Qollu Chico in the American Museum of Natural History (New York). Specimens in this collection not only have the same characteristic gray surface color but the same superficially re- duced "rinds" visible in the paste cross-section as well. Incisions on some of these pieces also have straight-sided rather than U-shaped contours, a trait found only in group B at Tumatumani. Group C ceramics, in contrast, share certain similarities to the material studied by Chavez, par- ticularly in the use of slips. Southern basin ex- amples are never slipped, while a fair number of Chavez's pieces are, including within incisions (Chavez 1985: 144), in the same red, red-brown, and light brown slips as the Tumatumani speci- mens. U-shaped incisions and a surface finish that shows no or only very faint finishing marks are predominant in the Department of Cuzco wares (Chavez 1985: 155-158) as they are in group C. Jab-and-drag or stamped incision, a particular trait of the Department of Cuzco ceramics (Chavez 1985: 140) and not found in the southern basin examples, is also found on one sp)ecimen in group C, where it was used to make a circular motif. The jab-and-drag technique is sufficiently diagnostic to link group C with the northern basin examples. However, it was employed in a slightly different manner at Tumatumani, since jab-and-drag inci- sion in the Department of Cuzco specimens was used only to make straight lines or dashes, while circles were stamped with a hollow stamping tool. While Tumatumani groups B and C can be pro- visionally identified as belonging to a Qeya tra- dition of incised ceramics made locally in the Late Sillumocco period at Tumatumani, group A stands apart. These sherds, it is suggested, are pre-Qeya and perhaps even Early Sillumocco in date. Incised ceramics with post-fire paint have been docu- mented from pre-Classic Pucara levels at Pucara by Kidder (Chavez 1985: 151) and Mujica (1978: 301; Mujica and Wheeler 1981: 51), and in pre- Pucara levels at Taraco (Chavez 1 985: 1 5 1 ). I have also found this type of incised ware in pre-Classic Pucara contexts at Camata. All of these ceramics have unslipped black or gray surface colors, which Analyses of Ceramic Artifacts 59 at least in the Camata examples are of a deeper color and include more of or all of the paste cross- section than the sherds in groups B and C, and closer approximate the group A firing technology. Refining the chronological order of the various local incised wares of the Titicaca Basin, however, is a task still to be completed. The Tumatumani excavations add more examples to the published corpus but cannot resolve the chronological am- biguities at this time. Late Sillumocco— Qeya Painted Ceramics (Figs. 148-155) A total of 1 1 sherds were found at Tumatumani, on the surface and in the second construction fill, that exhibit similarities to the painted ceramics of the Qeya period. Six sherds are manufactured in Paste 10 (Figs. 148-151), making this group the only one at Tumatumani to have a significant pro- portion of imported examples. The other speci- mens include two of the local Paste 1 (Fig. 152), two of Paste 2 (Figs. 153, 154), and one Paste 3 example (Fig. 155). Although these sherds share enough features to be discussed jointly, a number of differences separate the Paste 10 specimens from the Paste 1/2/3 sample. The similarities of the Tumatumani sherds as a whole to the published Qeya ceramics lie princi- pally in characteristics of design motif and slip. Like the Qeya examples (Bennett 1934: 399; Wal- lace 1957: 19a), the Tumatumani sherds tend to have thick, glossy slips, distinguishing them from the Tiwanaku period sample, which have some- what thinner slips. The color range of the slips on the Tumatumani specimens is also comparable to Qeya ceramics: black-and-red designs on a cream (lOYR 8/3-4), white (7.5YR 8/2, lOYR 8/2), light brown (7.5YR 6/5), orange (2.5YR 5/8), or yellow orange (5YR 6/6) background, or an unslipped light brown or light orange-brown (2.5YR 5/6, 5YR 5/6, 7.5YR 7/5, Fig. 149) background. Red on an unslipped buff, as described by Bennett ( 1 934: 399), occurs at Tumatumani (Fig. 1 55), as well as poly- chrome on buff. Qeya polychrome painting tends to occur in panels on the sides of vessels or on the interior rim, with the rest of the vessel remaining unslipped. Although most of the Tumatumani sherds are too small to ascertain if the same design arrangement predominates, the only large piece (Fig. 1 54) has this same pattern of an unslipped buff area below a design panel. Sherd interiors are always wiped and are generally slipped light brown or light orange-brown, although unslipped ex- amples occur in the Pastes 1/2/3 group. The paint- ed designs on the Tumatumani sherds, mostly di- agonal lines and triangles, generally compare well to the published Qeya examples, particularly to the triangle band motif (Bennett 1934: Fig. 13c; Eisleb and Strelow 1980: Figs. 18-23; Bermann 1990: Figs. 73, 75). The filler spots in Figures 148 and 154 and the squares with interior circles of Figure 152 are, however, somewhat outside the range of published Qeya examples. A limited number of shapes are represented in the Tumatumani sample. No "spittoons" (Wallace 1957: PI. 2d; Eisleb and Strelow 1980: Figs. 12- 16) or painted scallop-rimmed incensarios (Wal- lace 1957: PI. lb; Bennett 1934: Fig. 13b) could be identified. Most sherds appear to be from bot- tles, such as the body sherds with triangle band designs and perhaps the rim illustrated in Figure 1 53, although the diameter of this specimen seems unusually large. Most Qeya bottles do not have decoration on the exterior rim, as the Tumatumani examples do, but examples can be found in the literature (Ponce Sangines 1981a: PI. 2, 1981b: PI. 3). The base from what is perhaps a cup, illustrated in Figure 154, appears to be similar to an undeco- rated version encountered by Bennett in his ex- cavations (Bennett 1934: Fig. 13e), while Figure 148 may represent a slightly flared tall cup shape (Bermann 1990: Fig. 78). The Paste 1 0 specimens found at Tumatumani differ in several aspects from Qeya ceramics as defined by Bennett and Wallace. In fact, in some respects the locally made examples are more sim- ilar to Qeya ceramics than are the Paste 1 0 sherds. The Pastes 1, 2, and 3 specimens all have the characteristic thick glossy slips, while some of the Paste 10 specimens can approach Tiwanaku slips in their thinness. Qeya vessels generally are bur- nished in what would be, in the categories used here, a medium or high luster category 6 or a less fine category 7 burnish. The Pastes 1, 2, and 3 sherds have this type of finish, but the Paste 10 sherds have a fine high luster category 7 finish with a very even surface contour. Most significantly, Qeya ceramics are described as having a light buff or brown micaceous paste (Wallace 1957: 1 9a; Gi- rauh 1990: 109; Bermann 1990: 472, 474, 488). This description applies to the Paste 1 and Paste 2 examples, which are all fired to a light brown, and even to the light brown Paste 3 example as a variant. It does not, however, describe Paste 10, which is characteristically dense, hard fired, and red-orange in color, much more similar to the Ti- 60 Part Two wanaku pastes than those of the Sillumocco pe- riod. It is possible, therefore, that these Paste 10 sherds were made toward the end of the Qeya period, when firing and finishing practices had be- gun to approach those of Tiwanaku. Alternatively, they might themselves be Tiwanaku in date, and simply decorated in an earlier, more conservative style. The Pastes 1 , 2, and 3 painted sherds could theoretically also date to the Tiwanaku period, since all were found in the second construction fill or on the surface. That none of the sherds in these pastes, of which there are only five, were found in the first construction levels is considered at present to be due to factors of sampling. It does suggest the possibility, however, that the lower platform at Tumatumani was constructed at a time during the Late Sillumocco period when only Qeya-re- lated incised ceramics, rather than painted, were found at the site. This would imply that painted sherds in the Qeya style postdate Qeya-related in- cised decoration, at least at Tumatumani. These hypotheses, however, cannot be tested without more stratigraphic excavation in the Juli area. Tiwanaku Ceramics (Figs. 156-165) Polychrome and fancy monochrome-slipped ce- ramics dating to the Tiwanaku period are mostly of local manufacture. Of 20 specimens, eight are of Paste 1, eight of Paste 5, and one of Paste 2. The remainder of the sample is composed of one semilocal Paste 3 fragment and two Paste 1 3 spec- imens. Since the Qeya period ceramics were de- termined to be principally imported, the predom- inantly local production of decorated wares in the Tiwanaku period is notable. Because of the small size of both the sample and the sherds themselves, it has not been possible to make a meaningful temporal subdivision within the decorated Ti- wanaku sample at this time, but some specimens, such as the Paste 5 examples, can be suggested to be earlier than others. Exclusive of the Paste 5 sherds, which are dis- cussed separately, polychrome painted ceramics are found only in Pastes 1 and 13. The Paste 2 example is a rim from a smudged blackware hero, while the Paste 3 specimen is a red-slipped rim, probably also from a hero. Paste 1 specimens are slipped a monochrome red, or are decorated in black on light brown (7.5YR 5/5) or black, white, orange, and yellow on an overall red- or red-brown- slipped background (Figs. 156-159). Paste 13 ex- amples are painted in black and red on a yellow- orange (SYR 5/8) background, or black and yellow on a red background (Figs. 1 60- 1 6 1 ). All the poly- chrome examples have medium or high luster cat- egory 7 or 10 burnishes, while the monochrome- slipped examples have category 6 or 7 burnishes. When burnishing direction can be detected, it is generally horizontal on Paste 1 bodies and rims, especially the kero shapes, and vertical on the bas- es. The one Paste 3 specimen has a vertical burnish to the rim, while the Paste 2 kero is burnished horizontally at the rim, and then vertically below this. Surface contours are even or very even. Poly- chrome ceramics generally are slipped red or light brown on the interiors, which may be finished either with a burnish or with a category 1 or 12 wipe. Monochrome-slipped wares are more likely to have an unslipped, wiped interior surface. Paste 1 vessels are fired in an oxidizing atmosphere to produce a red-brown paste color or are an incom- pletely oxidized light brown or red-brown with a gray core. Except for the Paste 2 smudged speci- men, there are no black or gray surfaces in the sample. The Paste 3 and 1 3 ceramics are incom- pletely oxidized, with orange-brown or red-orange pastes and gray cores. Shapes in the Tumatumani assemblage are lim- ited; most sherds appear to come from kero forms, while some rims indicate a larger open bowl shape. The pronounced curve of Figure 156, with a di- ameter of approximately 10 cm, is also evidence of a small, convex-walled shape, and scalloped incensario vessels are represented by the isolated Paste 1 scallop illustrated in Figure 159, painted with yellow stripes on a red ground. All bases are flat with vertical or nearly vertical sides. The Paste 5 examples (Figs. 162-165) form an interesting subgroup of the Tiwanaku wares. They are all of the reduced-fiber version of this paste. Fiber voids are rare on the surface of the sherds but are visible with inspection on almost all of the examples. The vessels are painted in black, white, cream, and orange on a red background. Although in most examples the design motifs have a dis- tinctly local appearance, the lines thicker and less carefully applied than in the Pastes 1 and 1 3 ex- amples, Figure 163 is virtually indistinguishable from the non-fiber-tempered Tiwanaku wares. References to fiber-tempered decorated Tiwanaku sherds are rare, but do exist (Hermann 1 990: 203). Although the Tumatumani sherds are of local manufacture, made at the end of a long-standing fiber-tempered tradition in the Juli area, their ex- istence contributes more examples to this new fea- ture in the range of Tiwanaku ceramic technology in the southern basin. Analyses of Ceramic Artifacts 6 1 These Paste 5 specimens are somewhat less well finished than the other Tiwanaku polychrome sherds in the sample. All have burnished exteriors, but burnished finishes are from categories 6 and 8, with only two fine burnish category 7 examples. Burnishing reaches only a medium luster, and sur- face contours are less even. Interiors either are slipped red and burnished or have a category 1 or 1 2 scrape or wipe on an unslipped red-brown or light brown-slipped surface. Shapes present in the sample include a kero (Fig. 163) and a shallow slightly flared bowl (Fig. 162; the sherd is broken right at the joint with the base). All bases are flat. Fired paste color can be light brown, red-brown, red-brown with an incompletely oxidized gray core, or, with slightly higher frequency, can be cooled to produce a red-brown edge and dark core. These Paste 5 sherds are the only Tiwanaku period ex- amples to have cooled edges, a technique more characteristic of the Sillumocco period. Summary and Conclusions The presentation of the ceramic artifacts from the fill excavations and surface collections at Tu- matumani has been organized in this report ac- cording to the different plain and decorated ce- ramic wares. What remains is to present a unified chronology for the occupation of the site based on the ceramic and archaeological evidence available. The ceramics from Tumatumani are summarized in this section by chronological period, as much as current information allows, and some figures are presented to show the patterning of local and nonlocal ceramic production at the site. Early Sillumocco The ceramics of the Early Sillumocco period occupation at Tumatumani form a predominantly fiber-tempered assemblage. Undecorated and monochrome-slipped ceramics believed to date to this period are manufactured in Pastes 6, 7, 8, 9, and the high-fiber version of Paste 5, and thus are all fiber-tempered. Sixty-seven percent of the plainware sample is manufactured in Pastes 5 and 6 and is therefore believed to be locally produced (see paste descriptions). Pastes 7 and 9, semilocal pastes, form 31% of the sample, while nonlocal pastes account for only 1% of the sample. Early Sillumocco ceramics are characterized by a high percentage of burnished specimens (81% of the sherds in these pastes are burnished on both sides), and a high percentage of slipped examples, totaling 39% of the assemblage. Both red and brown slips were used, with red predominating over brown by a ratio of 2: 1 . Unslipped ceramics are in the ma- jority brown or dark brown and form 64% of the unslipped sample, with the remainder having a red-brown surface color. Firing that produced black or dark brown cores and cooled red-brown edges was preferred for the decorated and slipped ce- ramics. Shapes include globular ollas, jars, in- curving bowls, straight-sided bowls with thickened rims, and bowls with straight, slightly flaring walls. All Early Sillumocco period vessel bases are flat. As an assemblage, the fiber-tempered undeco- rated and monochrome-slipped ceramics from Tumatumani share many characteristics with Chi- ripa ceramics from Bolivia. This suggests that Ear- ly Sillumocco is roughly contemporary with the Chiripa period, including Chiripa Mamani and Chiripa Llusco. The question of whether any of the Pastes 1 and 3 plainwares found at Tumatu- mani also date to the Early Sillumocco period, or whether the fiber-tempered plainwares continue to be made during Late Sillumocco times, must at present be left open. As stated previously, there is no reason to assume that a fiber-tempered and a non-fiber-tempered tradition could not have co- existed at the site. A certain amount of overlap, principally in the decorated ceramics, can in fact be identified in both the Early and Late Sillumocco periods. At issue here, however, are the predom- inant characteristics of each assemblage. Although our conclusion is subject to revision as more strati- graphic evidence becomes available, we presently believe that the fiber-tempered and non-fiber-tem- pered plainware traditions at Tumatumani were successive, with a period of transition, presently not identifiable, existing between them. The non- fiber-tempered ceramics at Tumatumani are con- siderably diflferent from the fiber-tempered wares: they have a high percentage of wiped rather than burnished specimens, show different firing tech- nologies, and occur in different shapes or with dif- ferent vessel attributes than the fiber-tempered specimens. These differences suggest that the bulk of the non-fiber-tempered ceramics at Tumatu- mani belong to a different chronological period in the Tumatumani sequence which, through com- parison with other ceramic sequences in the Lake Titicaca Basin, can be shown to postdate Early Sillumocco. Excavation data in turn support the hypothesis that the fiber-tempered wares were not 62 Part Two produced in any quantity beyond the end of the Early Sillumocco period, for they were found in only very small percentages in the predominantly Late Sillumocco f)eriod first construction fill; most of the fiber-tempered specimens came from the mixed second construction episode levels or from the surface. Decorated ceramics of the Early Sillumocco F>e- riod are also in the majority fiber-tempered, but it is in this group that the few examples of non- fiber-temjjered specimens occur. The vast major- ity (83%) of the decorated sample are manu- factured in locally produced pastes, with 15% in semilocal and 2% in nonlocal pastes. Decorated sherds in the fiber-tempered Pastes 5, 6 and 9 were found that show affinities with Chiripa ceramics of the Mamani and possibly also Llusco periods, including the use of yellow/cream-on-red deco- ration and black and yellow on red combined with incision. Other specimens exhibit bichrome paint- ing in various color combinations, or bichrome incised techniques. Shapes are predominantly bowls with direct or thickened rims. Qaluyu-re- lated specimens in the Early Sillumocco assem- blage occur in Pastes 1 , 4, 5, and 6. The are incised and slipped red or brown, or are painted in black on a cream, light brown, red-brown, or brown background with a red-slipped interior and exte- rior rim. Notable in this case is that while the locally made ceramics in the Chiripa style are tem- pered with a large amount of fiber and have fiber voids visible on their surfaces, ceramics made at Tumatumani during the Early Sillumocco period which copy the styles of the more commonly non- fiber-tempered wares to the north were manufac- tured in pastes with no fiber, such as Paste 1, or with little fiber, such as Paste 6 or Paste 5 (the latter occurs in this group only in its reduced-fiber version). More semilocal and nonlocal ceramics, however, occur in the Chiripa-related decorated sample than the Qaluyu sample, with figures show- ing 2 1 % from these sources in the former as op- posed to only 6% in the latter. Late Sillumocco The Late Sillumocco period marks a change from a predominantly fiber-tempered to a predomi- nantly non-fiber-tempered assemblage. Plainware sherds of the Late Sillumocco period have char- acteristically wiped surfaces on at least one and sometimes both faces. Twenty percent of all sherds believed to be Late Sillumocco in date in the first construction fill are wiped on both sides, 78% are wiped on one side and burnished on the other, and only 2% are burnished on both sides. Dark brown slips are no longer used for the monochrome- slipped wares; red and light brown are now the prominent colors. Slipping, however, occurs in far smaller percentages than in the Early Sillumocco period; 89% of the total sample is unslipped, with only 10% slipped red and 1% slipped light brown. Firing practices move more toward oxidation and incomplete oxidation, as well as reduction, and less cooled ceramics were produced. New char- acteristics of vessel shape, such as flat-rimmed bowls, convex-sided bowls, incised tabs and rims, loops and scallops, horizontal handles, thickened edge bases, ring bases, and pedestal bases, appear in the assemblage. The plainware ceramics of this period are manufactured principally in Paste 1, with minor quantities of Paste 3 and 5 specimens. As such, they are more weighted toward locally produced wares, which form 89% of the sample, than the plainwares of the Early Sillumocco pe- riod. Paste 5 plainwares of the Late Sillumocco period differ from those of the Early Sillumocco period in that they are manufactured in the re- duced-fiber version of this paste, containing much less fiber and with few fiber voids visible on the surface of the sherds. They are apparently the only group of fiber-tempered ceramics made in the Late Sillumocco period, for none are found among the decorated wares. In this case the Late Sillumocco Paste 5 assemblage exhibits marked similarities in vessel shape, finish, and firing to the contemporary Late Sillumocco Paste 1 ceramics, even though differences in paste composition exist. In dating the Late Sillumocco period plainwares the issue of successive or contemporary fiber- tem- pered and non-fiber-tempered tradition is again raised. The possibility that some of these ceramics date in part to the Early Sillumocco period cer- tainly exists. Many of the Paste 1 vessels can be related to Qeya period ceramics from Qeya QoUu Chico, so that the bulk of this assemblage can be placed squarely in the Late Sillumocco period. The Paste 3 wares show some similarities to Qalasa- saya ceramics and therefore are more problematic. The illustrated Qalasasaya ceramics to which the Paste 3 examples are compared, however, appear to date to later in the Qalasasaya sequence. Infor- mation on the characteristics of a possible earlier Qalasasaya assemblage, dating to the Early Sillu- mocco period, is not available, so that compari- sons cannot effectively be made. The possibility still remains that they are made in a community Analyses of Ceramic Artifacts 63 that had closer ties to Qalasasaya or was producing early Qalasasaya-related ceramics during a time in the Early Sillumocco period when Chiripa-re- lated ceramics were being produced in the Juli area. Decorated ceramics of the Late Sillumocco pe- riod at Tumatumani are mostly incised. They are manufactured principally in Pastes 1 , 2 and 3, with a small number of Paste 1 0 specimens. An increase in semilocal and nonlocal ceramics occurs during this period; only 60% of the decorated sample was produced locally, with 29% produced semilocally and 12% nonlocally. Designs are almost exclu- sively geometric, and few or no iconographic or representational figures occur. The polychrome in- cised ceramics that can be related to Qalasasaya and Pucara specimens are considered to date to this period. They are painted in red, light red or pink, white, cream or yellow, and black on an unslipped or light brown-slipped background and are found in Pastes 1, 2, and 3. Other Late Sillu- mocco incised ceramics are unslipped or mono- chrome-slipped and are often found with post- or pre-fire paint rubbed into their incisions. Some of these are specimens in the form of ceramic tubes or trumpets. Examples were found in Pastes 1 , 2, and 3 and are decorated in a fairly homogeneous style of U-shaped incisions and punctate dots. Nei- ther the polychrome incised ceramics nor the trumpets appear to have been produced princi- pally in the Juli area; 60% and 72% respectively of these specimens are semilocal, from the Paste 3 production area, with the remainder manufac- tured locally. Other Late Sillumocco incised spec- imens are manufactured exclusively in the local Paste 1. These ceramics have rectangular or U-shaped incisions and a wide range of design motifs, including nested rectangles, squares, and L-shaped elements as well as parallel lines and punctate dots. Surfaces are unslipped or slipped red, red-brown, or light brown. All have a char- acteristic smoothed or wiped finish. The Paste 1 incised sherds were shown to be related to Qeya incised ceramics from the Tiwanaku area, Qeya Qollu Chico, and the departments of Cuzco and Puno. It is not until the end of the Late Sillumocco period that ceramics with exclusively painted dec- oration, specifically specimens that can be related to Qeya polychrome painted ceramics, first occur. The Qeya painted style appears to be a nonindi- genous development at Tumatumani, or one that was produced less characteristically at the site it- self, insofar as 55% of these sherds were manu- factured in Paste 10, which is believed to be non- local, and were imported from somewhere in the vicinity of Tiwanaku. Tiwanaku The Tiwanaku period at Tumatumani was also witness to changes in the ceramic assemblage at the site. These changes did not occur in paste com- position, however, but rather in aspects of vessel firing and finish. Although ceramics similar to those of the Late Sillumocco period plainwares un- doubtedly continued to be made into Tiwanaku times, new characteristics of ceramic production also appeared. Plainwares of the Tiwanaku period are found in Pastes 1 and 3. Some of these vessels were finished with a finer burnish that left no vis- ible marks of finishing, and some were rubbed or compacted to produce a matte, smooth surface. More vessels were fired in an oxidizing atmo- sphere, more vessels were burnished, particularly on the interior, and more were slipped. Mono- chrome slips are still red and light brown, but use of the light brown color becomes more popular. Bowls and jars found in the assemblage are gen- erally smaller in diameter and have thinner walls, and new shapes such as small ollas and cup forms appear. Decorated ceramics of the Tiwanaku period re- turn to a predominantly (85%) locally produced assemblage. Although made locally, they are fairly close copies of Tiwanaku decorative geometric styles. Specimens in Pastes 1 and 13 are painted in black, white, orange, and yellow on a red- or red-brown-slipped background, or are decorated in black on light brown. Shapes include keros, large and small open bowls with slightly flaring walls, and scalloped incensarios. It is in the decorated ceramics of the Tiwanaku period that we again see evidence of the persistence of the fiber-tempered tradition at Tumatumani. Eight Tiwanaku speci- mens were found in the reduced-fiber version of Paste 5, painted in black and white on a red back- ground. 64 Part Two LITHIC ARTIFACTS Matthew T. Seddon Although lithic debris can provide valuable in- sights into many aspects of past technological sys- tems, stone tools and debris have generally been given little coverage in reports on Tiwanaku pe- riod sites. This report describes the lithic materials recovered from surface collections and excava- tions at Tumatumani and investigates the tool as- semblage and raw material use during the two ma- jor occupations of the site. To carry out the descriptive analysis, a typology was developed to classify the tools. This typology was based primarily on formal characteristics, al- though some functional hypotheses are suggested for certain tool types. Clearly, though, thorough use-wear studies and experimental work need to be carried out to move analyses of Titicaca lithic materials beyond descriptions of their formal characteristics. The lithic materials at Tumatumani can, how- ever, provide a great deal of information regarding human behavior at the site. As was pointed out in Part 1 , the public architecture of Tumatumani was extensively altered during the Tiwanaku pe- riod. The lithic assemblage at Tumatumani can serve as a useful if limited starting point for an investigation of the nature and effects of Tiwanaku expansion into this area of the altiplano. It would be valuable information, for example, if we could identify changes in the organization of lithic tech- nology before and after the reconstruction of the site by the Tiwanaku state. The investigation is limited by the nature of the context from which the lithic materials were recovered. Because the mound structures at Tumatumani are composed of mixed secondary deposits, stratigraphic rela- tionships are not clear, and the main change in lithic technology that we might be able to observe at Tumatumani would be additions or deletions of tool categories. Drastic additions or deletions of tool types in the pre- and post-Tiwanaku fill at Tumatumani could be indicative of changes in technology that may have been related to the ar- rival of the expansive Tiwanaku state. With this in mind, the other goal of this report is to compare the materials recovered from the different con- struction fills at Tumatumani to investigate the nature of the lithic technology before and after the restructuring of the site. Lithic Typology of the Collection The overall analysis focused on finished and un- finished stone tools and tool fragments as well as debitage. The typology was established for all lithic materials, other than raw material or debitage, with evidence of secondary use or modification. The raw material analysis was applied to both second- arily modified materials and to debitage. All lithic material collected from the five ex- cavation units and a random sample of lithic de- bris from the systematically collected surface units was examined. In addition, nonsystematically col- lected surface materials that were chosen by Stan- ish were considered. Comparisons will be made between materials recovered from the first con- struction of the site (levels 9-19, Unit 1), the upper construction (levels 1-8, Units 1-5), the sampled surface collection, and the combined Tu- matumani sample. Other than research on stone tools from the Ar- chaic period, little systematic work on stone tools has been done for the Lake Titicaca region. Uhle (1889) reproduced figures of many types of stone tools from Ecuador, Peru, and Bolivia, including Analyses of Lithic Artifacts 65 lithic tools from Tiwanaku, but his report is ex- tremely brief and descriptive. Bennett (1934: 425- 427) mentioned the stone tools recovered from excavations at the site of Tiwanaku, and Ryden (1956: 66) also briefly discussed stone tools from the Nordenskiold collection, but neither discus- sion is particularly in depth. Hyslop (1976: 442) noted the lack of work on lithic technology for the Titicaca Basin, and whereas his report does sum- marize tool types from a range of sites in this area, systematic definitions are needed to facilitate fu- ture intersite and diachronic comparisons. One problem, however, is that my tool definitions are based on materials from just one site. Fortunately, the tool collection at Tumatumani is similar to the collections recovered from sites that Hyslop (1976: 440-449) examined in his broad survey, indicating that the Tumatumani collection is not drastically anomalous. With these caveats in mind, the Tu- matumani interpretations should be considered as preliminary or working definitions that can be modified as more research is done in this area. As mentioned above, types were defined pri- marily on the basis of form, although preliminary functional interpretations based on form are pos- ited. Tools that appeared to be different were sep- arated visually, and then metric measurements were used to refine definitions of tool types. The major types I studied were hoes, adzes, hoe frag- ments, adze fragments, hoe/adze fragments (uni- and bifacial), discoids, cores, grinding stones, and projectile points (Table 7, p. 68). Other types in- clude axes and scrapers, but these were too few in number for separate analysis and were examined as part of an "other" category (Table 8, p. 69). The lithic typology that follows is limited by the material present at Tumatumani. Clearly, addi- tional types need to be added to the list as lithic materials from other sites are analyzed. Hoes and Adzes Bennett (1934: 426) referred to these tools as axes. Ryden (1956: 66) quotes Nordenskiold's de- scription of similar tools as "probably toy axes, lost by boys playing here some hundreds of years ago." Hyslop (1976: 442-444) termed these tools "elongated andesite implements." I have further subdivided this category into two types, hoes and adzes, on the basis of metric characteristics. Both hoes and adzes are rectangular and thin and pos- sess a beveled bit (arbitrarily defined as the distal end); they frequently exhibit visible use-striations that run distal to proximal along the blade of the tool (one adze did, however, exhibit lateral stria- tions). The tools often have a smooth, glossy pol- ish, typically near the bit, which I attribute to use. The tools also display large retouch flake removals along the lateral edges to shape the piece and also along the bit to give a beginning to the bevel. The primary difference between the two types is that hoes are generally longer, wider, and thicker than what I have termed adzes (Figs. 166, 167). For the few recovered complete examples of each type, which were originally separated on subjec- tive visual criteria (which tool is larger, etc.), the following measurements apply. Hoes: length, 1 1- 25.5 cm; width, 1 1.5-14 cm; and thickness, 1.9- 2.15 cm. Adzes: length, 7-14 cm; width, 5-8 cm; and thickness, 0.9-1.9 cm. It is extremely likely that what I have termed hoes most probably functioned as the distal dig- ging and prying end or q'orana (Rivero Luque 1987: 26) of the traditional Andean chakitaqlla (Quechua term) or uysu (Aymara term) (Rivero Luque 1987; see also Tschopik 1946: 515, Fig. 44). Forms of these traditional field preparation implements are still very much in use today in many parts of Peru and Bolivia. A survey in 1972 by the Nacional Agropecuario determined that 716,000 chakitaqllas were in use (Rivero Luque 1987: 8), although the distal digging bit, or q'ora- na, is now more typically manufactured from met- al. Rivero Luque (1987: 82-83) reported the pres- ence of a single chakitaqlla from the province of Paucartambo in the Department of Cuzco, where the q'orana is manufactured from stone with di- mensions (28 X 12 cm) very similar to those of implements recovered from Tumatumani. The adzes may also be related to agricultural instru- ments or to the smaller makitaqlla (Rivero Luque 1987: 12-13), which was a hand-held implement used for digging of individual tubers, much like a gardener's trowel. The polish and striations seen on both of these tools from Tumatumani could easily be explained by wear from cutting through rocky soil in field preparation and harvest. In ad- dition, the chakitaqlla, when used for field prep- aration, is thrust into the ground and then used to pry up large sections of earth (Rivero Luque 1987: 33-44). This would put stress along the vertical (or proximal-distal) aspect of the q'orana, which could easily result in fractures from one lateral edge to the other, the most common type of frac- ture seen on the hoes and adzes at Tumatumani. While it is extremely likely that what I have termed hoes are in fact the q'orana aspect of the 66 Part Two chakitaqlla, this is not firmly established. The forms recovered from Tumatumani do not precisely match the forms of the q'orana, both ancient and modern, that Rivero Luque (1987: 17, 29) noted in his study. These forms were typically stemmed at the proximal hafting end and expanded distally to the bit or boca (Rivero Luque 1987: 29). The complete examples recovered from Tumatumani tend to be very symmetrical (Fig. 166). In addi- tion, whether or not the adzes represent forms of the hand-held makitaqlla is even more question- able, particularly since Rivero Luque (1987: 12) devoted little attention to this type and it may not be common. Because of these questions, I am re- luctant at this point to formally define these tools as q'orana and makitaqlla. I prefer to continue to use the terms hoe and adze as a preliminary mea- sure (Earle et al. [1980: 1 1] also referred to simi- larly described tools as hoes). Use of the English terms should be taken for now to convey the im- pression that these tools are not precisely defined and refers to their generic digging aspect. There are few complete examples of these types of tools in the Tumatumani collection. Only three complete hoes and four complete adzes were re- covered. The vast majority are fragments. Some nearly complete fragments clearly belong to one or the other category, but a large number are too small to be easily classified. The minimum infor- mation needed to determine the type of tool rep- resented by a fragment is the presence of both finished lateral edges and both dorsal and ventral faces. In other words, the piece should not display evidence of a proximal-distal snap fracture or a dorsal-ventral break. Fragments that did not meet these criteria were defined as hoe/ adze fragments (discussed below). Criteria are also needed to distinguish hoe frag- ments from adze fragments. Length was not a suf- ficient attribute because the lateral axis of most fragments was broken. Separation was therefore based on thickness and width. First, fragments were subjectively classified as hoes or adzes, then length and width were measured (Fig. 168). Based on these measurements, width proved to be the more relevant dimension for separating one type from the other. The respective confidence intervals overlap between the two categories for the thick- ness variable, but there is no overlap for the width variable. Fragments that looked like hoes had widths greater than 8 cm, whereas adzes had widths only up to 8 cm, although fragments with maxi- mum widths of 7-8 cm should probably be con- sidered unclassifiable. Therefore, this benchmark of 8 cm in width can usefully serve as a means of distinguishing hoes from adzes. Examples of hoe and adze fragments are shown in Figures 1 69 and 170. Hoe/ Adze Fragments The type hoe/ adze fragments was defined sep- arately from hoe fragments or adze fragments, each of which is considered a unique type. Hoe/adze fragments, which are generally quite small, possess polish, striations, and/or worn surfaces that in- dicate they are from hoes or adzes. However, hoe/ adze fragments do not have two lateral edges and often have merely one original dorsal or ventral face. These fragments include pieces broken lat- erally as well as pieces that were flaked from larger tools. I further subdivided this type into two subtypes based on the type of break represented by the piece. Unifacial hoe/ adze fragments have one worn, pol- ished, or striated surface, but appear to have been flaked off" the surface of a hoe or adze. Some of these pieces possess a bulb of percussion. Bifacial hoe/adze fragments have two surfaces that are worn, polished, or striated, but are missing one worked lateral edge. These pieces are probably the result of a snap break or bending fracture. All of the above types— hoes, hoe fragments, adzes, adze fragments, and uni- or bifacial hoe/adze frag- ments—were scored separately for each compo- nent at the site (first construction fill, second con- struction fill, surface collection). The majority of these tools are composed of gray andesite, with a small number manufactured from dark black ba- salt. Discoids The fourth major lithic type corresponds to Hys- lop's (1976: 442) "circular stones." I renamed these types discoids. The majority are composed of an- desite. These artifacts were formed by removing flakes from along all edges to make a roughly cir- cular stone typically 2-3 cm thick. They also have two flat, sometimes slightly worn (although not extensively ground or polished), opposite faces or plateaus in the center of each side of the discoid (Fig. 171). They range from 2.2 to 3.2 cm in thick- ness with an average maximum dimension of 9.5 cm (range, 8.5-11.5 cm). The function of these stones is unclear. Hyslop (1976: 442) suggested Analyses of Lithic Artifacts 67 Table 7. Tumatumani lithic type collection. Type First (lower) Second (upper) construction construction Surface-collected Total No. (%) No. (%) No. (%) No. (%) 0(0) 0(0) 3(2.1) 3 (0.4) 6 (4.7) 9(1.7) 10(7.1) 25(3.1) 0(0) 0(0) 4 (2.9) 4 (0.5) 3 (2.4) 5 (0.9) 20(14.3) 28 (3.5) 76 (59.8) 406 (76.3) 42 (30.0) 524 (65.6) 26 (20.5) 91 (17.1) 24(17.1) 141 (17.6) 5 (3.9) 0(0) 6 (4.3) 11(1.4) 5 (3.9) 2 (0.4) 4 (2.9) 11(1.4) 2(1.6) 6(1.1) 8 (5.7) 16 (2.0) 0(0) 4 (0.8) 9 (6.4) 13(1.6) 4(3.1) 9(1.7) 10(7.1) 23 (2.9) Hoe Hoe fragment Adze Adze fragment Unifacial hoe/adze fragment Bifacial hoe/adze fragment Discoids Random cores Grinding stones Projectile points Other (see Table 8) that they were grinders or dirt clod busters. They could also be cores. It should be noted that some items I defined as hoe/adze fragments could be fragments of these tools. The type is difficult to interpret. It is particularly intriguing that discoids are present in fairly abundant quantities. Uhle (1889: PI. 19) did not report any similar items from Tiwanaku. However, Ryden (1956: Fig. 36, Item W) reproduced what is clearly one of these discoids from the Erland Nordenskiold collection, although he did not comment on it. This collection was recovered from the Mizque Valley in Bolivia, suggesting that the distribution of these items is extensive. covered at Tumatumani, they were scored in the "other" category. O Grinding stones Grinding stones are typically sandstone. They are round or oblong, ground or pecked stones with one flat, worn working surface clearly visible, and were probably used as manos in combination with a pestle or metate-\\\ie stone. I want to avoid use of the term mano, as most of the stones do not have a classic mano shape (flat and oval), and there is a great deal of variation in shape for these tools. Random Cores Random cores are the fifth major type. They are typically no larger than a fist, with flake removals on all surfaces. Some have a cortex, and some appear to have also been used as hammerstones or pulverizers, as suggested by crushing along the edges of the flake scars. Nearly all of the random cores were manufactured from basalt. Axes The two axes found at Tumatumani were formed from hardened sandstone and are little more than rounded rocks with a pecked, shallow groove for hafting. Uhle (1889: PI. 19, Item 19) illustrated a similar type of ax from Tiwanaku. There was little effort to form a bit, and the axes seemed to have functioned primarily as crushing rather than as cutting devices. Because so few of these were re- Chert/Obsidian types These forms are primarily projectile points and scraping tools. The projectile points are bifacial, although several examples appear to have been made of flakes, as they possess large areas on one or both sides of the artifact that have not been scarred by flake removals. The collection of points is roughly half chert and half obsidian, with some quartzite examples. I noted three basic types: (1) triangular pieces (Fig. 1 72), (2) pieces with concave distal ends (Fig. 173), and (3) comer-notched of tanged types, which tend to be quite tiny (Fig. 1 74). Uhle (1889: PI. 20, Items 21-30) illustrated nine tanged points, very similar to the ones from Tu- matumani, which he recovered from the site of Tiwanaku itself, which suggests that the distri- bution of these points was extensive and may have been related to the spread of the Tiwanaku state. One concave base type was neatly denticulated along both lateral edges. Scraping tools exhibited 68 Part Two Table 8. "Other" category. Lower component Upper component Surface Stone bowl (1) Random core of jasper (1) Andesite hammers tone (1) Basah core used as hammerstone (1) Basah stone ball, bashed (1) Basah core used as hammerstone ( 1 ) Pohshed basaU tube (1) Retouched flakes, chert (3) Tested piece, chert ( 1 ) Possible point preforms on flakes, chert (2) Axes (2) Retouched flakes, chert (4) Endscraper, jasper ( 1 ) Sidescraper, chert ( 1 ) Random core, chert (1) retouching on one or more edges (Fig. 175). One clear endscraper was found (Fig. 1 75). The sample size of bifacial types is not sufficiently large to merit further subdivisions. Other In this category were placed two kinds of arti- facts: objects that clearly had been deliberately shaped but whose form defied definition, and types that were too few in number to merit analyzing as a separate type— i.e., objects that could best be analyzed as part of a group of assorted tools (Table 8). Type Representation by Site Component All artifacts were examined and scored sepa- rately according to the above typology for each of the site components— lower (first) construction fill, upper (second) construction fill, and surface col- lection. Counts and percentages of each type for each component as well as for the total Tumatu- mani sample are given in Table 7. Because the abundance of hoe/adze fragments could skew the representation of formal tool types, results were also generated after exclusion of the hoe/adze frag- ments (Table 9). What becomes clear is that there is a relatively even distribution of types in each sample, with hoes and adzes, as complete or fragmentary spec- imens, being slightly more frequent than other tool types. It is also evident that there are few or no sig- nificant differences in representation of tool types between samples within Tumatumani. Arrange- ment of the types into a series and presentation of the results in a cumulative percentage graph shows minor variations between the two constructions, but little or no difference between the surface and total collections (Fig. 176). Kolmogorov-Smimov tests failed to yield significant differences at the 0.05 level for any of the subsamples. There are very few clearly significant differences in tool type composition between any of the subsamples. The few notable differences are the complete lack of discoids in the upper component, the complete lack of projectile points in the lower component, and a drop in the number of random cores between the first and second construction episodes (from 20% of the first construction sample to 6% of the second construction sample). Although these dif- Table 9. Type collection with "hoe/adze fragment" category omitted. First (lower) Second (upper) construction construction Surface-collected Total Type No. (%) No. (%) No. (%) No. (%) Hoe 0(0) 0(0) 3(4.1) 3 (2.2) Hoe fragment 6(24.0) ' 9 (25.8) 10(13.5) 25(18.7) Adze 0 (0) 0(0) 4 (5.4) 4 (3.0) Adze fragment 3 (12.0) 5(14.3) 20 (27.0) 28 (20.9) Discoids 5 (20.0) 0(0) 6(8.1) 11(8.2) Random cores 5 (20.0) 2(5.7) 4 (5.4) 1 1 (8.2) Grinding stones 2 (8.0) 6(17.1) 8(10.8) 16(11.9) Projectile points 0 (0.0) 4(11.4) 9(12.1) 13(9.7) Other 4(16.0) 9 (25.7) 10(13.5) 23 (17.2) Analyses of Lithic Artifacts 69 Table 10. Type collection raw materials (including fragments) by component. Compo- Andesite Basalt Chert Obsidian Jasper Quartzite Sandstone nent No. (%) No. (%) No. (%) No. (%) No. (%) No. (%) No. (%) Lower Upper Surface Total 117(92.1) 510(95.9) 104(74.3) 731 (91.5) 6 (4.7) 6(1.1) 7 (5.0) 19 (2.4) 0(0) 7(1.3) 1 1 (7.9) 18(2.3) 0(0) 2(*) 4 (2.9) 6(*) 0(0) 2(1.4) 3(*) 0(0) 2(1.4) 3(*) 3 (2.4) 6(1.1) 10(7.1) 19 (2.4) * Less than 1.0%. ferences are intriguing, the small sample size pre- cludes making substantial conclusions at this point. Further testing is necessary to determine if this pattern is repeated elsewhere in the Titicaca re- gion. The entire sample is dominated by implements that would have been very useful for agricultural activities, although adzes may also have been used for butchering or woodworking. Again, use- wear studies would be invaluable in determining the specific functions of these tools. Manufacture and Distribution Raw material frequencies of the tool types for each component were also compared (Table 10). Although projectile points are primarily composed of chert and obsidian, with three quartzite ex- amples (Figs. 172-174), the most frequently used raw material at the site was the gray andesite com- mon to the region. Indeed, in all the subsamples, only one hoe, two hoe fragments, and two hoe/ adze fragments were manufactured from basalt, the other major raw material present at the site. Interestingly, nearly 100% of the random cores are basalt. This information diverges greatly from the results of a similar analysis reported by Hyslop (1976: 440-449). Hyslop briefly analyzed the stone tools from all sites in his survey (52 sites occupied from Tiwanaku to Colonial periods). He reported that 60% of the stone tools and debris were basalt and only 1 9% were andesite (the remainder were quartzite, obsidian, and jasper). The Tumatumani sample is dominated by andesite. Andesite tools range from 74% to 96% of each subsample, while basalt tools range from 1% to 5% (Table 10). A major factor that probably accounts for the pre- dominance of andesite in the Tumatumani col- lection is that there is a large outcropping of an- desite on the nearby mountain, San Bartolome, making andesite more common than basalt in the immediate area around Tumatumani, while basalt outcrops are located elsewhere throughout the re- gion (KJinck and Palacios M. 1984). Since Hyslop grouped all lithic materials that he analyzed from a wide geographic area, the difference in local raw material outcropping throughout the region may account for the difference between his sample and the Tumatumani sample. The difference in raw material use between the Tumatumani sample and j Hyslop's sample may also be due to the fact that Hyslop grouped tools and debitage together in his analysis. The debitage results from Tumatumani were very different from the tool results, as will be discussed below. However, Hyslop also grouped together 52 sites occupied at different time periods. If changes in raw material use had occurred over time in this region, the temporally constricted Tu- matumani materials may then be expected to vary from Hyslop's grouped results. Hyslop (1976: 442) reported that he saw few differences in raw ma- terial use between the periods he defined for his survey. Clearly, though, the Tumatumani mate- rials differ from Hyslop's results, and this differ- ence may be due to geographic or temporal vari- ables. Raw material composition of the debitage yield- ed very different results, and comparison of the raw material frequencies for the debitage and type collections raises a number of interesting ques- tions. The debitage collection consists primarily of andesite, basalt, and chert flakes (Table 11). Again, subsamples differ little in their percentages of raw material. However, basalt ffakes form a substantial portion of each sample (12%-20%), whereas basalt is present in much smaller pro- portion in the type collection (l%-5%). Further- more, all the random cores are made from basalt. One wonders what type of artifact was manufac- tured from the basalt cores. Whatever this artifact was, it clearly was not deposited at the site. It is possible that the flakes themselves were produced as flake tools. It is also possible that basalt tools 70 Part Two Table 1 1 . Debitage raw material composition by component. Compo- nent Andesite Basalt Chert Obsidian Jasper Quartzite No. (%) No. (%) No. (%) No. (%) No. (%) No. (%) Lower Upper Surface Total 646 (78.5) 6,506 (83.8) 255 (72.6) 7,407 (82.9) 159(19.3) 958(12.3) 61 (17.4) 1,178(13.2) 16(1.9) 239 (3.0) 30 (8.5) 285(3.1) 0(0) 14 (0.2) 0(0) 14(*) !(*) 25(*) 5(1.4) 31(*) 0(0) 17(*) 0(0) 17(*) Less than 1 .0%. manufactured at Tumatumani were used else- where or were moved out of the area via trade or some other mechanism. Use- wear studies of the flakes and regional analyses of stone tools need to be carried out to address these questions. Furthermore, while obsidian projectile points compose 3% of the type collection, the percentage of obsidian debitage does not exceed 1% in any of the subsamples. In fact, only 33 obsidian flakes in total were recovered, indicating that manufacture of the obsidian artifacts probably occurred else- where. The different percentages of raw material types between the tool collection and the debitage col- lection suggest that a complex system of lithic manufacture and use occurred within the Juli re- gion. The results from Tumatumani indicate dif- ferential spatial distribution of raw material pro- curement, manufacture, use, and distribution. On the one hand, it appears that basalt artifacts man- ufactured at the site may have been transported elsewhere, either for trade or for use, while the obsidian artifacts were manufactured elsewhere and then traded or transported for use at Tumatumani. In addition, there is an intriguing suggestion from the decrease in the number of random cores be- tween the first and second construction fills that the manufacturing system may have changed over time in the area. However, the fact that all the samples are small and from mixed construction fills dictates that this must remain merely a sug- gestion until further testing can clarify the pattern. At any rate, a different set of manufacturing and final deposition locales is indicated by the Tu- matumani materials. Analysis designed to locate and clarify this pattern at other sites in this period in the Titicaca Basin could yield further infor- mation about trade, lithic procurement, elite use of goods, and other aspects of the pre-Tiwanaku lithic technological systems in this area. Conclusions The results of a systematic definition and anal- ysis of the Tumatumani lithic debris in terms of assemblage composition and raw material use sug- gest a number of interesting patterns. As expected, since the two constructions at the site were com- posed of mixed fill, results from the analysis show little clear change in lithic technology after the Tiwanaku restructuring of the site, although there are some interesting suggestions from the data that merit further exploration. However, raw material use, particularly of basalt and obsidian, differs sub- stantially between the debitage collection and type collection and could well indicate a system of dif- ferential use, manufacture, and final deposition of stone tools. Also, the degree to which the raw ma- terial frequencies from Tumatumani, essentially a Sillumocco and Tiwanaku period site, differ from Hyslop's multi-period results indicates possible temporal changes in raw material usage. Further work, particularly in the analysis of raw material use at different sites within this period and be- tween temporal periods, is clearly needed. Preliminary results from a systematic survey of this area (Stanish et al. 1 992) suggest further av- enues for work along these lines. Limited and sub- jective viewing of the artifacts uncovered at a number of other Tiwanaku period sites in the area suggests that these sites have lithic collections (hoes, discoids, adzes, flakes, etc.) similar to the Tuma- tumani collection. These different sites appear to have been part of a broad system of lithic man- ufacture, use, and distribution throughout the Si- llumocco and Tiwanaku periods in this region. The remaining question regards the nature of this system and how it compares with earlier patterns of tool manufacture and use in the area. Further analysis of lithic debris from these sites could clar- ify this pattern. Analyses of Lithic Artifacts 7 1 Part 3 Summary and Conclusions Charles Stanish The results of our recently completed survey pro- vide the regional context for the Tumatumani data. The results of collecting and intensive mapping at Tumatumani provide a good characterization of surface data from the entire site. Finally, the ex- cavations provide reliable data on the lower ter- race found on the eastern side of the west mound. We think that the conclusions about the architec- tural constructions drawn from the excavation data can be interpolated to the entire west mound as a whole. With this in mind, the following summary of Tumatumani is offered. Tumatumani is a major site with artificial mounds located in prime agricultural land next to Lake Titicaca. In our typology of sites developed from the survey data, Tumatumani can be labeled a Type 1, Large Artificial Mound (Stanish et al. 1992). Type 1 sites are elite/ceremonial settle- ments, and there are at least three Type 1 sites, including Tumatumani, in the Juli-Pomata survey area. The other elite/ceremonial site type. Type 3, is characterized by low, nondefensible, terraced hills with corporate architecture on the top. There are at least three Type 3 sites that date to the Sillumocco and/or Tiwanaku periods in the Juli- Pomata region as well. Tumatumani is therefore one of at least six elite/ ceremonial sites in the Juli-Pomata region that date to the Late Sillumocco and/or Tiwanaku pe- riods. It represents one of two principal elite/cer- emonial settlement types (Types 1 and 3) for the Late Formative and Tiwanaku periods in the Lake Titicaca region. Tumatumani is 5.3 ha in size and is composed of two principal mounds. The formal architecture of the west mound is characterized by a low plat- form with a second, smaller platform built on top of the first. The east mound, in contrast, is shaped like an elongated U or horseshoe, 150 m at its maximum length. This U shape is a classic Initial period and Early Horizon construction technique; in the central Andes as a whole, but it is rare in the Titicaca region. It is the only U-shaped struc- ture in the Juli-Pomata survey, with the possible exception of one small mound in the El Molino area. The major occupations at the site of Tumatu- mani include an Early Sillumocco one, a Late Si- llumocco one, and a reconstruction of the site by the Tiwanaku state. Minor occupations at the site include Inca, Early Colonial, Republican, and Modem ones. The existence of Early Sillumocco pottery with similarities to Qaluyu (ca. 800 B.C.) in construction fill indicates an earlier occupation prior to the construction of the formal terraces. The nature of the earliest occupation has not been precisely defined: the fill from the first terrace oc- cupation seen in the excavation profiles was col- lected on the site and reused for the construction of the first, formal terrace base. This fill was then faced with a large wall. Contemporary with the construction of this wall was the placement of several burials (two of which were excavated this season) at the base of the wall into the natural subsoil. The latest ceramic frag- ments in this fill (Late Sillumocco) give a terminal date for this construction. The first major terrace at the site was therefore constructed in Late Si- llumocco times, ca. 200 b.c.-a.d. 400. The earlier Early Sillumocco village that existed on the low natural hills at the site area was disturbed to obtain the fill to build the formal terrace during Late Si- llumocco times. The precise nature of this earlier village site remains to be defined. 72 Part Three The second major construction activity at the site was characterized by a rebuilding of the eariier terraces into larger ones with the same basic ar- chitectural pattern. This second occupation would correspond to the platforms now seen on the sur- face of the mounds. Although no wall stones are now evident, the quantity of stones in the vicinity of the site strongly suggests that these were used to face the platforms in a manner similar to the earlier one. A construction date for this rebuilding episode would be Tiwanaku IV (a.d. 400-750). This date was corroborated by two profile cuts into disturbed surface platform walls that also con- tained Formative and Tiwanaku materials. The final occupation discovered in these inves- tigations was a small Inca period house built over the Tiwanaku period terrace. The stones used in the construction of the Inca house were probably those originally used in the construction of the earlier terrace. Reuse of stones and the post-Inca agricultural activities would help explain the lack of any terrace wall stones associated with the sec- ond formal construction on the site. The initial Early Sillumocco occupants appar- ently chose two naturally high areas near the lake for their settlement. The agriculturally rich alluvial zone was left for cultivation, and this pattern con- tinued throughout the occupation at the site. In fact, Tumatumani is considerably less complex than expected, with all residential occupation and nonresidential architecture located on the mound areas proper. There is no evidence of nonelite res- idence away from the mound proper with related high-status habitation or elite/ceremonial areas near the core. Likewise, the surface survey did not detect any appreciable differences within the res- idential area itself Rather, the terraces constructed on the mound are interpreted as principally resi- dential in function, and terraces that were used by the entire population. One of the most important results of this re- search is the ceramic analysis of Steadman, re- ported in Part 2. This analysis constitutes one of the primary justifications for proposing the Si- llumocco culture. I will now summarize some of the important conclusions of her analysis as they relate to the nature of the Sillumocco ceramic as- semblage, culture, and time period. The Tumatumani ceramic assemblage is a local one, but it displays many similarities to the south- em Titicaca Basin traditions (Chiripa, Qeya, and Qalasasaya). These differences are sufficiently marked to warrant its separate cultural designa- tion. The Pre-Tiwanaku IV ceramic assemblage also is characterized by a less intensive but none- theless significant borrowing from the northern Titicaca Basin tradition, the ancestral home of the Qaluyu and Pucara styles. A substantial number of the diagnostic plain- wares and decorated pieces in the pre-Tiwanaku IV periods were locally produced. The only sig- nificant exception is a group of Qeya imports made in nonlocal pastes. The earliest decorated pieces are local imitations of Qaluyu. Other Early Si- llumocco incised and painted pieces are very sim- ilar to Chiripa, but most were made in local or semilocal pastes. The narrow incisions and un- usual color combinations on some of the these pieces appear to be purely local innovations in a "Juli" style (pp. 53-54). In the Late Sillumocco period, the most com- mon undecorated paste type. Paste 1, is used in vessels that are technologically and morphologi- cally similar to Qeya (pp. 40—41), yet most of the rim shapes of these vessels in the Tumatumani materials are not found in Qeya assemblages (p. 40). The characteristic and highly diagnostic Pucara thickened rims also are not found in the Tumatumani assemblage (p. 40). In other ceramic characteristics we see borrow- ing from both the north and the south, as well as independent stylistic innovation at the site. In- curving bowl forms are found in Pucara but not in the south (p. 40). The Tumatumani ring-based, scalloped-edge incensario shape is very similar to Qeya specimens, while the incised sherds associ- ated with this shape are similar to examples from both the northern and southern Titicaca Basin (pp. 58-59). Pedestal bases are found in both Pucara and Qeya assemblages (p. 40). The Late Sillu- mocco polychrome incised ceramics borrow from Pucara and Qalasasaya but are distinctive enough to represent a new, local type produced in Juli (p. 55). The contrast with Tiwanaku ceramics is quite marked. The Tiwanaku IV period ceramic mate- rials represent a significant change in the source of stylistic borrowing: all influence was from the Tiwanaku state, and there is far less evidence of local stylistic innovation in the decorated ceramic vessels vis-a-vis contemporary styles from the Ti- wanaku area itself By Tiwanaku IV times, both the ceramic and settlement data strongly suggest the incorporation of the Sillumocco populations into a larger political entity controlled by Tiwa- naku. These data suggest that Tumatumani represents a major site that was part of an autonomous po- Summary and Conclusions 73 litical entity that had borrowed styles from both the southern and northern Titicaca Basin during the pre-Tiwanaku IV periods. By the Tiwanaku IV period, Tumatumani had been incorporated into the Tiwanaku state. The research at Tumatumani, combined with our survey data, will force us to revise our under- standing of the Upper Formative period in the southwestern Titicaca region. The existence of a new Upi>er Formative period culture— Sillumoc- co— along with Pucara to the north and early Ti- wanaku and/or Chiripa (Tiwanaku I-III and Chi- ripa Mamani) to the south, points to a much more complex political and economic landscape in the region prior to the emergence of the Tiwanaku state as a regional power. A traditional conception of the Upper Forma- tive period Titicaca Basin is that it was divided into two great spheres of cultural innovation and political control as represented by Late Pucara and Early Tiwanaku (Tiwanaku III or Qeya). This as- sumption, implicit or explicit, will have to change. Sillumocco represents an as yet poorly understood political and economic entity of substantial com- plexity. Almost certainly, economic and political relationships between the Sillumocco sites and their northern and southern neighbors were constantly in flux, as would be expected in such a political landscape. In a similar vein, the Tumatumani data, as well as the survey data, indicate that the absorption of the Juli-Pomata area by the Tiwanaku state was not accompanied by major disruptions in settle- ment patterns. The Late Sillumocco elite/cere- monial centers continued to exist as part of the Tiwanaku system. This suggests that the absorp- tion was gradual, perhaps involving a long history of fluctuating elite alliances that ultimately crys- tallized into more direct, formal control by the Tiwanaku state. Research on the Sillumocco and Tiwanaku pe- riods in the Juli-Pomata area is ongoing. Future research will clarify many of the political and eco- nomic relationships between the southwestern Ti- ^ ticaca region and its neighbors to the north and south. 74 Part Three Literature Cited Albarracin-Jordan, Juan, and James E. Mathews. 1 990. Asentamientos Prehispanicos del Valle de Ti- wanaku, vol. 1. CIMA, La Paz. Bandelier, Adolph. 1910. The Islands of Titicaca and Kaoti. Hispanic Society of America, New York. Bennett, Wendell C. 1934. Excavations at Tiahua- naco. Anthropological Papers of the Museum of Nat- ural History, 34(3). . 1 936. Excavations in Bolivia. Anthropological Papers of the American Museum of Natural History, 35(4): 329-507. . 1946. The Archaeology of the Central Andes, pp. 61-147. In Handbook of South American Indians. Bureau of American Ethnology Bulletin 143, vol. 2. Washington, D.C. Berenguer, Jose. 1978. La problematica Tiwanaku en Chile: Vision retrospective. Revista Chilena de An- thropologia, 1: 17-40. Bermann, Marc. 1990. Prehispanic Household and Empire at Lukurmata, Bolivia. Ph.D. diss.. University of Michigan, Ann Arbor. Boero Rojo, Hugo. 1 980. Discovering Tiwanaku. Los Amigos del Libro, La Paz. Browman, David. 1978a. Toward the development of the Tiahuanaco (Tiwanaku) state, pp. 327-350. In Browman, D. L., ed.. Advances in Andean Archae- ology. Mouton, The Hague-Paris. . 1978b. The temple of Chiripa. El Hombre y la Cultura Andina, 2: 807-813. . 1 980. Tiwanaku expansion and altiplano eco- nomic patterns. Estudios Arqueologicos (Antofagas- ta), 5: 107-120. Chavez, Karen Mohr. 1 969. Excavations in the Cuz- co-Puno area of southern highland Peru. Expedition, 11(2): 48-51. . 1 980. The archaeology of Marcavelle, an Early Horizon site in the valley of Cuzco, Peru. Part I. Baes- sler-Archiv, N.F., 28: 203-329. . 1981. The archaeology of Marcavelle, an Early Horizon site in the valley of Cuzco, Peru. Parts II and III. Baessler-Archiv, N.F., 29: 107-205, 241-386. . 1985. Early Tiahuanaco-related ceremonial burners from Cuzco, Peru. Dialogo Andino (Arica), 4: 137-178. . 1988. The significance of Chiripa in Lake Ti- ticaca Basin developments. Expedition, 30(3): 1 7-26. Chavez Ballon, Manuel. 1950. Arqueologia del sur andino. Tradicion (Cuzco), 1(2): 41-50. CiEZA DE Leon, Pedro, n.d. La Cronica del Peru. In Chonicas de las Conquisla del Peru. Editorial Nueva Espaiia, Mexico. Conklin, Willl\m J., and Michel E. Moseley. 1988. The patterns of art and power in the Early Interme- diate period, pp. 145-163. In Keatinge, R. W., ed., Peruvian Prehistory. Cambridge University Press, Cambridge. DE LA Vega, Abel Edmundo. 1 990. Estudio Arqueo- logico de Pucaras o Pobladas Amurallados de Cumbre en Territorio Lupaqa: El Caso de Pucara Juli. B.A. thesis, Universidad Catolica "Santa Maria," Arequi- pa, Peru. DiEZ DE San Miguel, Garci. 1 964 [1567]. Visita He- cha a La Provincia de Chucuito. Ediciones de la Casa de la Cultura del Peru, Lima. Earle, Timothy K., Terence N. D'Altroy, Catherina J. LeBlanc, Christine A. Hastorf, and Terry Y. LeVine. 1 980. Changing settlement patterns in the upE>er Mantaro Valley, Peru: Prelimnary report for the 1977, 1978, and 1979 seasons of the Upper Mantaro Archaeological Research Project. Journal of New World Archaeology, 4: 1-49. EiSLEB, Deiter, and Renate Strelow. 1 980. Altepe- ruanische Kulturen: III. Tiahuanaco. Veroffentlichun- gen des Museums fur Volkerkunde Berlin, N.F., vol. 38. Erickson, Clark. 1988. An Archaeological Investi- gation of Raised Field Agriculture in the Lake Titicaca Basin of Peru. Ph.D. diss.. Department of Anthro- pology, University of Illinois, Chicago. Fellman Velarde, Jose. 1961. LosImperiosAndinos. Editorial Don Bosco, La Paz. Fish, Suzanne K., and Stephen A. Kowalewski. 1 990. Introduction, pp. 1-5. //iThe Archaeology of Regions. Smithsonian Institution Press, Washington, D.C. Flores Ochoa, Jorge A., and Magno Percy Paz Flores. 1983. La agricultura en lagunas del altipla- no. Nawpa Pacha, 21: 127-152. Franquemont, Edward. 1990. The ancient pottery from Pucara, Peru. Nawpa Pacha, 24(1986): 1-30. Literature Cited 75 GiRAULT, Louis. 1990. LaCeramicadelTempleteSemi- subterraneo de Tiwanaku. Centre de Estudios de la Realidad Economica y Social, La Paz. GoLr)STEiN, Paul. 1989. Omo, a Tiwanaku Provincial Center in Moquegua, Peru. Ph.D. diss.. Department of Anthropology, University of Chicago, Chicago. HoDDER, Ian. 1 974. Regression analysis of some trade and marketing patterns. World Archaeology, 6(2): 1 72- 189. Hyslop, John. 1976. An Archaeological Investigation of the Lupaqan Kingdom and Its Origins. Ph.D. diss., Department of Anthropology, Columbia University, New York. Ibarra Grasso, Dick. 1957. Un neuvo panorama de la arqueologia Boliviana, pp. 233-285. In Ponce San- guines, C, ed., Arqueologia Boliviana (Primera Mesa Redonda). Biblioteca Pacena, Las Paz. JuLiEN, Catherine. 1983. Hatuncolla: A View of Inca Rule from the Lake Titicaca Region. University of California Publications in Anthroplogy, vol. 15. Kidder, Alfred II. 1943. Some early sites in the north- em Lake Titicaca Basin. Peabody Museum of Amer- ican Archaeology and Ethnology, Papers, vol. 27(1). . 1956. Diggingin the Titicaca Basin. University Museum Bulletin, 20(3): 16-29. Klinck, B. A., AND O. Palacios M. 1984. Mapa Geo- logico de los Cudrangulos de Juli e Isla Anapia. Minis- terio de Energia y Minas, Institute Geologico Minero y Metalurgico, Lima. KoLATA, Alan. 1983. The South Andes, pp. 241-285. In Jennings, J., ed., Ancient South Americans. Free- man, San Francisco. . 1986. The agricultural foundations of the Ti- wanaku state: A view from the heartland. American Antiquity, 51: 748-762. KoLATA, Alan, Charles Stanish, and Oswaldo Ri- vera S. 1987. The Technology and Organization of Agricultural Production in the Tianaku State: First Preliminary Report of the Proyecto Wila Jawira. Uni- versity of Illinois, Chicago. Lanning, Edward. 1967. Peru Before the Incas. Pren- tice-Hall, Englewood Cliffs. Layman, Frederic, and Karen Mohr Chavez. 1 966. Petrographic analysis of pottery from Chiripa, Bolivia. Proceedings of the Pennsylvania Academy of Sciences, 39: 220-225. Lumbreras, Luis Guillermo. 1974. Los reinos post- Tiwanaku en el area altiplanica. Revista del Museo Nacional (Lima), 40: 55-85. . 1981. Arqueologia de la America Andina. Ed- itorial Milla Batres, Lima. Lumbreras, Luis G., AND Hernan Amat. 1968. Se- cuencia arqueologica del altiplano occidental del Ti- ticaca. Actas y Memorias del XXXVII Congreso In- temacional de Americanistas, 1 966 (Buenos Aires), 2: 75-106. Means, Phillip. 1931. Ancient Civilizations of the An- des. Charles Scribner's Sons, New York. Mohr, Karen. 1966. An Analysis of Pottery from Chi- ripa, Bolivia: A Problem in Archaeological Classifi- cation and Inference. M.A. thesis. University of Penn- sylvania. Muelle, Jorge Clemente, and Camilo Blas. 1938. Muestrario de arte peruano precolombino: I. Ce- ramica. Revista del Museo Nacional (Lima), 7(2, II semestre): 163-280. MujicA, Elias. 1978. Nueva hipotesis sobre el desa- rrollo temprano del altiplano del Titicaca y sus areas de interaccion. Arte y Arqueologia, 5-6: 285-308. . 1985. Altiplano-Coast relationships in South- Central Andes: From indirect to direct complemen- tarity, pp. 103-140. In Masuda, S., I. Shimada, and C. Morris, eds., Andean Ecology and Civilization. University of Tokyo Press, Tokyo. 1987. Cusipata: Una fase pre-Pukara en el cuenca norte del Titicaca. Gaceta Arqueologica An- dina, 4(13): 22-28. MujiCA, Elias, and Jane Wheeler. 1981. Produccion y recursos gandaderos prehispanicos en la cuenca del Titicaca, Perii. Trabajo del Campo, 1979-1980. Final report submitted to the Institute Nacional de Cultura, Lima, and Plan COPESCO (Cusco). PoNCE Sangines, Carlos. 1948. Ceramica Tiwana- cota: Vasos con Decoracion Prosopomorfa. Emece Editores, Buenos Aires. . 1970. Las Culturas Wankarani y Chiripa y su Relacion con Tiwanaku. Publicacion 25. Academia Nacional de Ciencias de Bolivia, La Paz. . 1971. La Ceramica de la Epoca I de Tiwanaku. Publicacion 28. Academia Nacional de Ciencias de Bolivia, La Paz. . 1981a. Descripcion Sumaria del Templete Semisubterraneo de Tiwanaku. Editorial Juventud, La Paz. 1981b. Tiwanaku: Espacio, Tiempo y Cultura. Los Amigos del Libro, La Paz. PuLGAR Vidal, Javier, n.d. Historia y Geografia del Peru: Los Ochas Regiones Naturales del Peru. Uni- versidad Nacional Mayor de San Marcos, Lima. Rice, Prudence. 1987. Pottery Analysis: A Source- book. University of Chicago Press, Chicago. RivERO LuQUE, Victor. 1987. La Chakitaqlla en el Mundo Andino. Proyecto de Herramientas e Imple- mentos Agricolas Andinos, Cusco. RowE, John. 1956. Archaeological explorations in southern Peru, 1954—1955. American Antiquity, 22(2): 135-151. . 1 963. Urban settlements in ancient Peru. Naw- pa Pacha, 1: 1-27. RowE, John H., and Catherine Brandel. 1971. Pu- cara style pottery designs. Nawpa Pacha, 7-8(1969- 70): 1-16. Ruben, Walter. 1952. Tiahuanaco, Atacama und Ar- aukaner. Otto Harrassowitz, Leipzig. Ryden, Stig. 1947. Archaeological Researches in the Highlands of Bolivia. Elanders Boktryckeri Aktiebol- ag, Goteborg. . 1956. The Erland Nordenskiold archaeological collection from the Mizque Valley, Bolivia. Ethnolo- giska Studier, vol. 22. Rye, Owen. 1981. Pottery Technology. Manuals on Archaeology 4. Taraxacum Inc., Washington, D.C. Schreiber, Katherina J. 1992. Wari Imperialism in 76 Literature Cited Middle Horizon Peru. University of Michigan Mu- seum of Anthropology, Ann Arbor. Shepard, Anna O. 1956. Ceramics for the Archaeol- ogist. Publication 609. Carnegie Institute of Washing- ton, Washington, D.C. Smith, C, W. Denevan, and P. Hamilton. 1 968. An- cient ridged fields in the region of Lake Titicaca. Geo- graphical Journal, 134(3): 353-366. Stanish, Charles. 1989a. Household archaeology: Testing models of zonal complementarity in the south central Andes. American Anthropologist, 91(1): 7-24. . 1989b. Tamano y complejidad de los asenta- mientos nucleares de Tiwanaku, pp. 41-57. In Kolata, A., ed., Arqueologia de Lukurmata, vol. 2. Instituto Nacional de Arqueologia, La Paz. 1992. Ancient Andean Political Economy. University of Texas Press, Austin. Stanish, Charles, Edmundo de la Vega, Lee Steaek MAN, LuPERio Onofre, Cecilia Chavez J., Percy Calisaya, Kirk L. Frye, and Matthew Seddon. 1992. Archaeological research in the Juli-Pomata re- gion of the Titicaca Basin, Peru. Final report submit- ted to the National Science Foundation. Tosi, Joseph. 1960. Zonas de Vida Natural en el Peru. Technical Bulletin #5. Organization of American States. TscHOPiK, Harry, Jr. 1946. The Aymara, pp. 501- 573. In Steward, J. H., ed.. Handbook of the South American Indians. Vol. 2. The Andean Civilizations. Smithsonian Institution Press, Washington, D.C. Uhle, Max. 1889. Kultur und Industrie Siidameri- kanischer Volker. Verlag Von A. Asher, Berlin. Wallace, DwiGHT T. 1957. The Tiahuanaco Horizon Styles in the Peruvian and Bolivian Highlands. Ph.D. diss.. Department of Anthropology, University of Cal- ifornia, Berkeley. Willey, Gordon R. 1971. An Introduction to Amer- ican Archaeology. Vol. II, South America. Prentice- Hall, Englewood Cliffs. Literature Cited 77 Key to Ceramics, Figures 14-165 ^ Red Dark red Light red Pink Red-brown Brown Dark brown Light brown Black White ^ N N Cream iilii; Yellow '!■!'; Yellow-orange _~_~_ Orange • • • • Post-fire red Post-fire yellow Munsell soil colors corresponding to key: Red: 7.5R 3/6-8, 3.5/6; lOR 3-4/6, 3.5/6. Dark red: lOR 3/2-4. Light red: 7.5R 4/6, 3.5/8; lOR 4.5/ 6-8. Pink: 5R 5/6. Red-brown: 2.5YR 3/4-6, 3.5/6, 4/4-8. Brown: 5YR 4/3-5, 4.5/4; 7.5YR 4/4; lOYR 4/4. Dark brown: 2.5YR 3/2; lOYR 3/ 2. Light brown: 5YR 5/5-6, 6/4-5; 7.5YR 5/4-6, 5.5/4-6, 6/4-5; lOYR 5/3, 5.5/4, 6/4. Black: lOR 2.5/1; 2.5YR 2.5/0; 5YR 2.5/1-2; 7.5YR 2- 2.5/0; lOYR 3/1. White: 5YR 8/1; 7.5YR 8/2-3; lOYR 8/2-3. Cream: 7.5YR 7.5/4-6, 8/4; lOYR 8/4. Yellow: 7.5YR 6/7-8, 7/6-7, 8/5-6; lOYR 7/6. Yellow-orange: 5YR 5/8, 6-7/6. Orange: 2.5YR 5/8. Post-fire red: 7.5R 3/6-7; lOR 4/8. Post-fire yellow: 5YR 6/8; 7.5YR 7/7-8. 78 Ceramics J D - 11 cm D -9cm m 18 7 10 r D-9cm ^ 20 15 ? 16 \ 17 D -24 cm 21 D - 16cm 7 22 1^ ^^ ^ D Scm Figs. 14-22. 14, Paste 4 jar; 15, 16, Paste 7 jars; 17, Paste 7 o//a; 18-22, Paste 1 jars. Ceramics 79 D • 20c(n D - 18cm D- IScm 7 0 -21 cm 25 D - 11 cm ( I 26 D - 17 cm D - 17 cm D-13cm 7 27 7 28 29 30 31 32 i-^ ^-^ M 5cm Figs. 23-32. Paste 1 . 23-30, jars; 31, incised fillet at neck joint; 32, bowl shape A. 80 Ceramics 34 35 36 M M H 0 Son Figs. 33-36. Paste 1. 33-35, bowl shape A; 36, bowl shape B. Ceramics 8 1 37 38 I 39 ^r 40 41 hH h-H f-H 5cm Figs. 37^1. Paste 1. 37, bowl shape B; 38, bowl shape C; 39-41, bowl shape D. 82 Ceramics 7 7 45 0 Figs. 42-45. Paste 1. 42, 43, bowl shape E. 44, 45, straight-sided bowls. Ceramics 83 \ f ^^^ mi 47 ^ [ \. I 0 5cm Figs. 46-5 1 . Paste 1 , miscellaneous rims. 84 Ceramics 53 55 1-^ ^ ^ 56 5cm Figs. 52-56. Paste 1 , thickened edge bases. Ceramics 85 [-(. ^9 59 « "" 60 f. HH hH hH 0 Scm B2 Figs. 57-62. Paste 1. 57, pedestal base; 58, 59, ring bases; 60, 61, rim-to-body strap handles; 62, strap handle. 86 Ceramics 63 \ 65 7 68 67 ? 68 f I— I >— I hH 0 Scm Figs. 63-69. Paste 1 . 63, rim loop; 64, scallop; 65, 66, Tiwanaku period jars; 67, 68, Tiwanaku period small ollas\ 69, Tiwanaku period bowl. Ceramics 87 D-14 cm I ^W 70 D-10 cm D-18 cm 7 t 74 D-16 cm 72 73 75 76 77 4 78 I— I t— I I— I 5cm Figs. 70-78. Paste 3. 70-73, jars; 74, punctate incision at neck joint; 75, scallop; 76, tall-necked jar; 77, bowl- 7o, nng base. j < ■> 88 Ceramics 7 79 ^ 80 81 82 83 HH hH hH 0 5cm 84 Figs. 79-84. 79, 80, Paste 3 Tiwanaku period shapes; 81, 82, Paste 5 jars; 83, 84, Paste 5 ollas. Ceramics 89 85 86 87 ^ 1— ^ ^ 0 5cm 88 Figs. 85-89. Paste 5. 85, straight-sided bowl; 86, incurving bowl; 87, 88, bases; 89, spout. 90 Ceramics V 91 92 l-H hH hH Figs. 90-92. Paste 9. 90, 91, ollas; 92, jar. 93 7 / 7 ^-1 I— I I— I 0 5cm Figs. 93-96. Paste 9. 93, jar; 94, 95, bowls. 96, Paste 1 Early Sillumocco-Qaluyu incised bowl. Ceramics 9 1 97 I— I I— I I— I 5cin I 98 I 99 100 I 101 7 102 Figs. 97-102. Early Sillumocco-Qaluya incised and painted ceramics. 97-100, Paste 1; 101, 102, Paste 5. 92 Ceramics 1 103 1 gi^M^ i-i I— I ^-l I 106 ^^ Figs. 103-104. Early Sillumocco-Qaluyu painted ceramics. 103, Paste 6; 104, Paste 4. Figs. 105-106. Early Sillumocco-Chiripa incised ceramics in Paste 5. I I T r I 110 I 112 T M I— rT=i 0 Scffl Figs. 107-1 13. Early Sillumocco-Chiripa incised ceramics. 107-110, Paste 5; 111, 112, Paste 9; 113, Paste 6. Ceramics 93 r 7 116 M M Figs. 1 14-117. Early Sillumocco-Chiripa painted ceramics. 114-116, Paste 5; 117, Paste 9. 94 Ceramics o I 119 118 I 120 121 ? 122 I— I I— I ^ 0 5cm Figs. 118-122. Early Sillumocco-Chiripa painted ceramics. 118, Paste 6; 119, Paste 5; 120, Paste 9; 121, 122, Paste 5. Ceramics 95 I 123 1 124 I 125 I 126 I 1 27 b ^ ^ ^ 0 5cm Figs. 123-127. Late Sillumocco polychrome incised ceramics. 123-125, Paste 3; 126, Paste 1; 127, Paste 2. 96 Ceramics 128 130 129 I 131 I 132 I 133 I— I I— I ^—^ 0 5cm Figs. 128-133. Ceramic tubes or trumpets. 128, Paste 7; 129, Paste 5; 130-132, Paste 3; 133, Paste 1. Ceramics 97 I I 134 136 135 138 137 139 I— I l-H hH 0 5cm Figs. 134-139. Paste 1 incised ceramics. 134-136, group A; 137-139, group B. 98 Ceramics I 141 1 143 I 145 140 I 142 144 I 146 l-H KH hH 0 5cm Figs. 140-146. Paste 1 incised ceramics, group C (140-144 are slipped red, 145 is unslipped, 146 is slipped cream). r' Ceramics 99 147 ^ I 149 148 I 150 I— I ^ I— I 0 5cm Fig. 147. Paste 1 incised and modeled feline head. Figs. 148-150. Late Sillumocco-Qeya painted ceramics in Paste 10. 100 Ceramics 'A ? 151 7 152 I 153 154 I 155 1-^ ^— I ^ 0 5cm Figs. 151-155. Late Sillumocco-Qeya painted ceramics. 151, Paste 10; 152, Paste 1; 153, 154, Paste 2; 155, Paste 3. Ceramics 1 0 1 I 156 157 158 159 I 160 \ 161 BUBUB 0 5cm Figs. 156-161. Tiwanaku painted ceramics. 156-159, Paste 1; 160, 161, Paste 13. 1 02 Ceramics 162 7 163 164 I 165 0 Scm Figs. 162-165. Tiwanaku painted ceramics in Paste 5. Ceramics 1 03 Fig. 166. Complete hoe. Fig. 167. Complete adze. 104 Lithics Cm Measurements - Fragments T r Hoe Length Celt Length Hoe Width Celt Width Fig. 168. Hoe and adze (celt) dimensions. 1 r Hoe Thick Celt Thick Fig. 169. Hoe fragment. Lithics 105 Fig. 1 70. Adze fragment. 106 Lithics Fig. 171. Discoids. Lithics 107 5 CM 0126 -Obsidian 0652 - Chert 0096 - Chert 0095 - Obsidian 0500 - Obsidian Fig. 172. Triangular projectile points. 108 Lithics 5 CM 0652 - Chert 0471 - Chert 0102 - Chert 0471 - Chert 0448 - Quartz Fig. 173. Concave-base projectile points. 5 CM 0309 - Obsidian 0302 - Quartz 0373 - Quartz Fig. 1 74. Comer-notched projectile points. Lithics 109 Fig. 175. Scrapers. 110 Lithics LOWER COMPONENT* UPPER COMPONENT** SURFACE TOTAL 1 2 3 4 HOE HOE ADZE ADZE FRAG. FRAG 5 6 7 DISCOID RANDOM GRNDNG CORE STONE TYPE * FIRST CONSTRUCTION * * SECOND CONSTRUCTION Fig. 1 76. Cumulative percentage of tools by component. 8 9 PROJ. OTHER PNT 10 Lithics 1 1 1 X G- A Selected Listing of Other Fieldiana: Anthropology Titles Available } An Fth.noj^" Northern Sakhalin island, i' idian i ratio urnamcnts in liic Coilecnons oi Mcld Museum of • Fieldiana: Anthropology, n.s., no. 13, 1989. 40 pages, 32 illu PnhJiraiion L361. SI 0.00 Publication 1404, $11.00 K.iUke-Related Pottery from Cuzco, Peru, in the F ield Museum of Natural Histoi an S. Haucr and Charles Stanish. F/V' 7 ,. ; '; - ro. : ^ ! 9f''0. 17 pages, 17 illus. Publication 14119, v'f ) ■^ Late Pre-Hispanic Ceramic ( h"MiMlogy for the Upper Moquegu; Fieldiana: Anthropology, n.s , 1991. 68 pages, 308 illus. ;;ioi^l, Material Culture of the Blackfoot (Blood) Indian no. 19 !00T 90 r>;>7.:>s, *= 3 '!1"!' Publication 1-}Z^, bl '.ivj .Lihcrn Alb Publication 1439, $19.00 tin Athapasknns of West Central British Columbia: Collectirms in the Field Museum of Natu! lO'M ^" n-rcs r'S illus. W. VanStone. Fieldian Publication 1446, $12.00 iiress and Warfare Among the Kayenta Anasazi of the Thirteci ' : ' '■> luiiathan Haas and W'm'iWxd Crcduicr. I- ichli ana: Anthropolo:,^ ;• i 1 pages, 74 illus. Publication 1450, $32.00 ider by publication number and/or ask for a li' 11 orders r; ilmois residents add current destination tax. All ibreign orders are payauie in U.S. dolia^-cln.■L^^ uiawn >'i any U.S. bank or the U.S. subsidiary' of any foreign bank. Prices and terms subject to change without ' rlress all requests to: FIELD MUSEUM Oi Library— Publications ui vision Roosevelt Road at Lake Shore Dvivt Chicago. Illinois 606' ^ - Field Museum of Natural History Roosevelt Road at Lake Shore Drive Chicago, Illinois 60605-2496 Telenhonc: HI 2^ 922-9410 CENTRAL GRCULATION BOOKSTACKS The person charging this material is re- sponsible for its renewal or its return to the library from which it was borrowed on or before the Latest Dote stamped below. You may be charged a minimum fee of $75.00 for eacli lost boolc. Th*ft/ nwtilcitloa, and UMiMllRing •! books oro roosons for discipliiMiry ocHon and may ruidl In dismissal from tho Univorsity. TO RENEW CALL TELEPHONE CENTER, 3M-8400 UNIVERSITY Of ILLINOIS LIBRARY AT URBANA-CHAMPAION FEB 1 9 1997 When renewing by phooe, write new due date below previous due date. Ll^ UNIVERSITY OF ILLINOIS-URBANA 3 0112 028183785