THE INDIAN CREEK EXCAVATIONS Irving Rouse In the spring of 1969, Fred Olsen invited me to Antigua to examine the newlydiscovered Indian Creek site. It surpassed my expectations in size, depth, and because it appeared to offer the best possibility for reconstructing the local ceramic sequence-- a sequence I was interested in obtaining so as to fill a gap in our knowledge of Caribbean chronology between the Greater Antilles, where I had previously worked, and the southern half of the Lesser Antilles, where the most recent chronological research has taken place. Accordingly, I welcomed Olsen' s proposal that I dig the site, but stipulated that I would first have to complete two projects already under way. While neither of these projects has yet been completed, Olsen persuaded me to proceed with the Indian Creek excavations during late May, June, and early July, 1973, in order to make the preliminary results available at the Fifth International Congress for the Study of the Pre-Columbian Cultures of the Lesser Antilles, which was to be held in Antigua in late July. It was understood that preparation of the final report on the excavations will have to wait until I have fulfilled my previous commitments. My interest in the local chronology led me to formulate a two-part program: (1) excavations under my direction at Indian Creek, in order to work out the ceramic part of the chronology; and (2) research by Dave Davis, a graduate student, in the more recently discovered non-ceramic sites, in an effort to reconstruct the preceramic part of the chronology (this volume, pp. 65-71). I am indebted to the National Science Foundation of the U.S.A. for awarding me a grant in support of this program (No. GS-37970), and to the Antigua Archaeological Society for generously matching that grant. Halfway through the Indian Creek excavations, I became ill and was confined to the hospital for the rest of my stay in Antigua. Nevertheless, the excavations continued according to my plan under the direction of Olsen and of Dr. E. J. H. Boerstra, government archeologist in Aruba, N. A., who had come to observe the research. I am indebted to Olsen and his daughter, Mrs. Elizabeth Kyburg, for processing the specimens as they came from the excavations, and to Boestra for supervising the excavations after I became sick and for making the trench profiles. The map of the site (Fig. 1) has been prepared by Davis with assistance from Boestra. Finally, I wish to thank Desmond Nicholson for handling the logistics of the program so competently, and the government of Antigua for allowing us to use the former N. A. S.A. tracking station as a laboratory. The Site Indian Creek is situated in the southeastern corner of Antigua, between English Harbour and Marmora Bay. When Charles A. Hoffman, Jr. (1963) surveyed the island in I96I, he located traces of Indian occupation on the rocky shore at the mouth of the creek. The site under discussion here lies about a half mile up the creek, separated from the shore by an encircling ring of hills. Thus it differs from a common or late settlement pattern among the ceramic Indians in the Lesser Antilles, which is for the sites to be located on the shore, just back of a sandy beach. This was one of the site' s attractions; its settlement pattern resembles that of the earliest ceramic sites in the 166
ROUSE 167 Greater Antilles, such as Hacienda Grande in Puerto Rico (Alegría 1965:247-8), and as a result it seemed likely to yield the beginning of the ceramic sequence on Antigua, which had hitherto been unknown. The site lies on a gentle slope just west of Indian Creek. From the air, it has the appearance of an oval ring, colored grey against a dark background. On the ground, it can be seen that the grey color is due to a concentration of shell refuse in a series of middens extending around the periphery of an oval area measuring approximately 283 by 165 m. (Fig. 1). The refuse in the remainder of the oval area appears to be sparse and shallow, and one wonders whether part or all of it may have been carried there by the plow, to which the site has been subjected in recent years. At present, the middens are simply areas of greater concentration and depth of refuse. Originally, I counted six of them, but our excavations revealed that in one case I had been misled by a natural depression in the site. This depression (indicated in Fig. 1 by a ticked line) falsely seemed to form the uphill side of a midden, with its other side formed by the downward slope of the hill. The three valid middens on the uphill side of the site (Fig. 1, 1-2, 4) are considerably smaller and shallower than the two on the downhill side (Fig. 1, 5-6). It was evident from two gullies eroded in the southernmost of the uphill middens (on either side of Excavation 1 in Fig. 1) that its height was only about 50 cm. A gully in the larger of the two downhill middens (just north of Excavation 6 in Fig. 1), revealed that midden to be about 2 m. high. Various people have collected artifacts from the gullies, and their finds indicate that the downhill middens are considerably richer than the uphill middens. The gentle slope on which the site is situated narrows to the south of the refuse deposit to a point where Indian Creek passes through a gap in the hills on its way to the sea. The slope broadens to the north, and it is here that Olsen found the possible ball court (not shown in Fig. 1) he discusses elsewhere in this volume (pp. 11-12). Plan of Excavation To achieve maximum results from an excavation, one must design it in terms of one' s objectives. For example, if I had intended to investigate the structures built by the Indian Creek inhabitants, I would have planned the kind of research Boestra carried out in Aruba (this volume, pp. 13-20). I would have made two large excavations, one in the refuse area in order to recover traces of the Indian houses, hearths, storage pits, etc., and the other in the area where Olsen has identified a ball court to search for further traces of such a structure. In each case, I would first have stripped off the topsoil that has been disturbed by plowing, and in the habitation area I would also have removed the underlying refuse, in order to lay bare the subsoil and search it for discolorations indicative of human activity. If, instead, my aim had been to obtain the best possible examples of the local artifacts, I would have concentrated on the two downhill middens, because they are the richest and because their depth offers greater possibilities of preservation than in the shallower middens on the uphill side of the site. My aims, however, were chronological rather than descriptive. I intended to distinguish ceramic complexes, establish their sequence, and study the variations within and among them. In order best to achieve these objectives, I chose to dig six rela-
168 THE INDIAN CREEK EXCAVATIONS tively small trenches, one in each midden (including the falsely identified midden on the west side of the site (Fig. 1, 3). Each trench was to be divided into four sections, 2 m. square, and each section was to be divided vertically into arbitrary 25-cm. levels. My excavation units therefore measured 2 m. on a side and 25 cm. deep, a size which experience has shown to be the most efficient for the purpose of distinguishing ceramic complexes and studying their variation. In any archeological or paleontological fieldwork, no matter what the objectives, one should be careful to design the excavations so that they test one' s assumptions about the deposition of the remains under study. My use of 25-cm. levels was based upon the assumption that the Indian Creek refuse had been deposited gradually and continuously, so that the depth of the remains beneath the surface is a function of the passage of time. I chose to arrange the excavation units in linear trenches in order to test this assumption. I anticipated that the profiles of the trenches would reveal any irregularities of deposition, whether these have resulted from erosion by natural agencies, from gaps in the occupation of a particular part of the site, or from disturbance of the deposit by burial or some other kind of cultural activity. The aim of chronological research is to work with undisturbed refuse, where the process of deposition has been continuous and regular. Any evidence of disturbance or irregularity of deposition in the walls of a trench indicate that its contents are not suitable for use in such research. This, of course, is one reason for digging a number of trenches in a site when studying its chronology. One can then select the trenches which show the most regular depsoition of refuse and which are therefore most suitable for chronological study. Another reason for digging multiple trenches is to test the possibility that the refuse in one part of a site may have been deposited at a different time than the refuse in another part of the site. In planning the excavations at Indian Creek, I was faced with a choice between two alternative hypotheses: 1. All five of the Indian Creek middens might have been deposited at the same time; for example, they might all have been occupied by contemporaneous households. This seems to have been true at the sites of Ostiones in Puerto Rico (Rouse 1952:393-7) and Carrier in Haiti (Rainey 1941:29-33), each of which contains a series of middens arranged in an oval as at Indian Creek. The excavations in each of these sites revealed that their middens were inhabited more or less simultaneously by a single group of people. 2. The Indian Creek middens might have been occupied successively by different groups of people, each distinguished by a different ceramic complex. This would have produced what is technically known as a multi-component site, with each component (part of the site) being characterized by a different ceramic complex, indicative of a separate group of people (Rouse 1972; Fig. 34). The preliminary survey of the Indian Creek site which Olsen and I made in 1969 revealed that it was a multi-component site, as in (2) above, though with some overlapping of the middens. Not only did the site appear to have been too large to have been inhabited all at one time by only a single group of people, but also the potsherds we collected from the surface and in the gullies resembled sherds from other islands that are known to have been laid down at different times and by different peoples. This is what made the site so attractive for chronological research.
SCALE before reduction I: 1600 lcm.= 16 m. I In. : 133ft. L i 'J3*t CONTOUR INTERVAL^ I m Fig. 1. Map of Indian Creek site locating tests. A BENCHMARK U EXCAVATIONS (2 x8m APPROXIMATE LIMITS VISIBLE SURFACE APPROXIMATE LIMIT MIDDENS
170 THE INDIAN CREEK EXCAVATIONS By distributing the trenches evenly among all the middens, I hoped to sample the entire sequence of components, and hence of ceramic complexes indicating different groups of people. I designated the trenches Excavations 1 through 6, numbering them in the order in which they were dug (Fig. 1). I began with the midden closest to the shore, in the southwestern part of the ring, and then proceeded clockwise around the site, digging first the middens on the higher (western) side of the site and then the two larger, deeper, and richer middens on the lower (eastern side) adjacent to the creek. I chose this order because my examination of the previous collections indicated that the earliest inhabitants of the site had lived in the southwestern corner and that subsequent occupants had moved upstream along the top side of the slope. I preferred to dig the top side of the site first because its deposits were shallower, poorer, and therefore less likely to have been subject to disturbance and to mixture of the pottery laid down by different peoples than the deposits on the downhill side of the site. I was attempting to define the local sequence of ceramic complexes in terms of the relative simplicity of the situation in the uphill middens before confronting the possibility of greater complexity in the downhill middens. When Olsen and Boerstra carried out Excavation 6, they found that the deposit sloped sharply downward beyond the area in which I had laid out that excavation' s four sections. As a result, they had to add two more sections to the downhill end of Excavation 6 in order to reach the heart of the midden. These two sections yielded such a wealth of objects that Olsen and Boerstra subsequently expanded Excavation 6 sideways, adding a seventh section to the south (downstream) and two more to the north (upstream), so that Excavation 6 became T-shaped (Fig. 1). The additional sections were not needed for chronological purposes; rather they were intended to provide specimens for the museum which it is hoped to establish on the former N. A. S. A. tracking station. Ceramic Sequence In discussing the results of this chronological research, I am obliged to limit myself to the conclusions I had reached before entering the hospital, since I have not yet seen the specimens obtained subsequently. Excavations 1-4, which were completed before I became sick, yielded evidence of three successive ceramic complexes, the first in Excavation 1, the second in Excavation 2, and third in Excavations 3 and 4. I shall briefly define these complexes and shall correlate them with the complexes already known from the other Leeward Islands (including Guadeloupe) and the Greater Antilles, farther north. I shall not, however, also attempt to relate the Antiguan complexes to those farther south in the Windward Islands (including Martinique). I had expected to elicit comments about the southern relationships from the members of the congress who are more familiar with the archeology of the Windward Islands than I, and prefer to defer discussion of the subject until I have had the benefit of their comments. Indian Creek complex. Pottery from Excavation 1 was unlike any I had seen from sites elsewhere in Antigua. I therefore chose to apply the name "Indian Creek" to the ceramic complex of which it is representative. Sherds (Fig. 2) are thin, fine, and always have smooth surfaces. They typically come from keeled bowls with gracefully out-flaring rims, although a number of other shapes are also represented. Flanges are not uncommonly attached to the rims. "Zic" decoration (zoned incised crosshatching) is limited to this complex. White-on-red painting is also characteristic but is not of high quality; the designs tend to be zonal. There
ROUSE 171 Fig. 2. Pottery of the Indian Creek Complex.
172 THE INDIAN CREEK EXCAVATION Fig. 3. Pottery of the Mill Reef Complex. Scales in centimeters. I 6 I 7 l' 1 r> I- i f;'!lil:i'iillllim". MI'll'lil l lhi l
ROUSE 173 Fig. 4. Pottery of the Marmora Bay Complex. Scales above in inches, below in centimeters.
174 THE INDIAN CREEK EXCAVATION are modeled head lugs, D-shaped strap handles, circular button lugs, and an occasional incised design, sometimes enclosing red or black painted areas. The griddle sherds are plain. The Indian Creek complex, (Fig. 2) is typically Saladoid. It appears to belong to the earliest ceramic horizon, marking the migration of the Saladoid Indians out into the Antilles from the mainland. It is thus to be correlated with Morel 1 in Guadeloupe (Clerc I968), Golden Rock on St. Eustatius (Josselin de Jong 1947:19ff. ), and Hacienda Grande in Puerto Rico (Alegría 1965). If these correlations are valid, the charcoal samples from Excavation 1 that have been submitted for radiocarbon analysis should date from the first centuries A. D. Mill Reef complex. Since the pottery of Excavation 2 is similar to that obtained by Olsen, Hoffman (1963), and others at the Mill Reef site, I shall apply the name "Mill Reef" to the complex of which it is representative. This does not mean, however, that all the pottery found at the Mill Reef site pertains to the Mill Reef complex; occupation of that site apparently continued into the subsequent Marmora Bay period, as at the Indian Creek site. Pottery of Indian Creek Excavation 2 continues to be relatively thin and fine and to include the bell-shaped profile characteristic of the previous Indian Creek complex. Some sherds now have their surfaces roughened by scratching. "Zic" decoration disappears and there is greater emphasis upon painting. Polychrome (white, orange, and red) painting is limited to this complex. White-on-red painting reaches its highest quality; curvilinear negative designs are diagnostic, but rectilinear positive designs also come into existence. Beveled, red-painted rims are another diagnostic. Lugs and incised designs continue very much as before, with the addition of broader- and shallower-line incision. Griddles now begin to have legs. The Mill Reef complex (Fig. 3) continues to be Saladoid and, in fact, appears to be the climax of the Saladoid development on Antigua. It may be correlated with Morel 2 on Guadeloupe, the latter part of the occupation of the Golden Rock site on St. Eustatius, Coral Bay-Longford in the Virgin Islands (Hatt 1924), Cuevas and Early Ostiones in Puerto Rico (Rouse 1952:336-44), and La Romana in the Dominican Republic (Rouse and Allaire, MS). These correlations indicate that the complex should date between A. D. 300 and 900; and Fred Olsen has already obtained dates of A.D. 445 ± 85, 445 ±85, 510 ±85, 720 ±85, 850 ±85, and 870 ±85 for charcoal from the Mill Reef-complex levels in Excavation 5, which he submitted in order to date the fragment of a stone "belt" he found in that excavation (this volume, pp. 11-12). Marmora Bay complex. The name of this complex is derived from the site of Marmora Bay, just east of Indian Creek. Hoffman (1963) noted that Marmora Bay pottery resembles certain atypical specimens he obtained at Mill Reef, but was unable to isolate and define a separate complex. We have succeeded in doing so in Excavations 3 and 4 at Indian Creek. Pottery of the Marmora Bay complex (Fig. 4) tends to be thicker, heavier, and cruder than pottery of the earlier complexes. Sides of the vessels characteristically curve upwards or inwards, forming the so-called cazuela shape. This shape is sometimes accentuated by adding a thick ring or fold of clay to the rim. The rings and folds take the place of the flanges that had been present in the previous complexes. Scratching
ROUSE 175 to roughen the sherd surfaces becomes more common. Painting is now limited to positive, rectilinear white-on-red designs, typically in the form of chevrons, and to a red or black slip, applied areally. Lugs are simpler, cruder, and rarer; and handles are virtually non-existent. Shallow-and broad-line curvilinear incision is now diagnostic. Griddles continue to be provided with legs. The Marmora Bay complex appears to be a composite one. In material and shapes, it is strongly reminiscent of the Santa Elena pottery in Puerto Rico (Rouse 1952:344-7) and for that reason, I would tentatively assign it to a new "Elenoid" series, along with Morel 3-4 in Guadeloupe (Bullen 1973), The Bottom on Saba (Josselin de Jong 1947), and most of the Magens Bay-Salt River pottery in the Virgin Islands (Hatt 1924). Such an Elenoid series has recently been independently proposed by Gary Vescelius on the basis of his Virgin Islands research (Alfredo Figueredo, personal communication). On the other hand, decoration of the Marmora Bay complex appears to be a purely local development. It resembles that of the other complexes I have assigned to the Elenoid series only in its relative simplicity, crudity, and scarcity. I was particularly surprised to find rectilinear white-on-red designs surviving so late--until A.D. 1200, if we may judge by the radiocarbon dates for the Santa Elena complex in Puerto Rico and by the latest radiocarbon dates for the Mill Reef site in Antigua, which are presumably to be associated with the Marmora Bay complex (Rouse and Allaire MS). We have submitted for analysis a series of charcoal samples from the Marmora Baycomplex deposits at Indian Creek, and they should provide a better idea of the chronological range of the complex. Pending completion of these analyses I would tentatively date the complex between A.D. 900 and 1200. General comments. The foregoing discussion is subject to revision when the Indian Creek collections reach the United States and I am able to check my preliminary notes against the material from all-parts of the site. Additional conclusions would be even more speculative, but I think it worth noting that the three complexes appear to form a single, continuous development. To be sure, there is evidence of a marked change in the diet between the periods of the Indian Creek and Mill Reef complexes: crab claws predominate in Indian Creek refuse whereas shellfish are characteristic of Mill Reef and Marmora Bay deposits. This change, however, is not likely to have resulted from an influx of new people from the direction of South America, as Rainey (1940) and Alegrfa (1965) have postulated on the basis of their finds in Puerto Rico. In Antigua, the change occurs before the climax of the Saladoid development, not at the end of that development, as in Puerto Rico. Hence, the change is more likely to be due to an expansion of the collecting activities of the previous inhabitants of the Indian Creek site into new micro-environments, including the sandy beaches some distance away from the site. Our non-ceramic finds from Indian Creek reinforce the impression of continuity. For example, we obtained small three-pointed stones and shell carvings like those of the Arawakan-speaking Indians of the Greater Antilles in association with all three ceramic complexes. To my mind, this indicates that we may attribute the entire Antiguan ceramic sequence to the ancestors of the Tainos or Island-Arawaks. I therefore doubt that the Carib Indians arrived in the vicinity of Antigua until after the Marmora Bay complex had passed its prime, that is, until after A.D. 1200 if the dating suggested above is correct.
176 THE INDIAN CREEK EXCAVATION Bibliography Alegría, Ricardo E. 1965 On Puerto Rican Archaeology. American Antiquity, vol. 31, no. 2, pp. 246-9. Salt Lake City. Bullen, Ripley P. and Adelaide K. 1973 Stratigraphie Tests at Two Sites on Guadeloupe. Proceedings of the Fourth International Congress for the Study of Pre-Columbian Cultures of the Lesser Antilles, pp. 192-6. Castries. Clerc, Edgar 1968 Sites Précolombiens de la Grande-Terre de Guadeloupe. Proceedings of the Second International Congress for the Study of the Pre-Columbian Cultures of the Lesser Antilles, pp. 47-60. Barbados. Hatt, Gudmund 1924 Archaeology of the Virgin Islands. Proceedings of the Twenty-first International Congress of Americanists, vol. 1, pp. 29-42. The Hague. Hoffman, Charles A., Jr. 1963 Archeological Investigations on Antigua, West Indies. Master' s thesis at the University of Florida, Gainesville. Josselin de Jong, J. P. B. de 1947 Archeological Material from Saba and St. Eustatius, Lesser Antilles. Mededelingen van het Rijksmuseum voor Volkenkunde, Leiden, no. 1. Leiden. Rainey, Froelich G. 1940 Porto Rican Archaeology. The New York Academy of Sciences, Scientific Survey of Porto Rico and the Virgin Islands, vol. 18, pt. 1. New York. 1941 Excavations in the Ft. Liberté - Region, Haiti. Yale University Publications in Anthropology, no. 23. New Haven. Rouse, Irving 1952 Porto Rican Prehistory. New York Academy of Sciences, Scientific Survey of Porto Rico and the Virgin Islands, vol. 18, pts. 3-4. New York. 1972 Introduction to Prehistory: a Systematic Approach. McGraw-Hill Book Company, New York. 1973 Analytic, Synthetic, and Comparative Archeology. In Research and Theory in Current Archeology, pp. 21-31,. edited by Charles L. Redman, John Wiley and Sons. New York. Rouse, Irving, and Louis Allaire MS Caribbean Chronology. To be published in Chronologies in New World Archaeology, edited by Clement W. Meighan and R. E. Taylor.