Saladoid Draft

Based on the archaeological evidence in the Lesser Antilles and Puerto Rico, the migration of a new group of people into the Caribbean in the last few centuries B.C. was impossible to miss. Their beautiful and unmistakable ceramics colored in brilliant white and red designs were completely different from the artifacts of the preceramic people who were living in the islands when they arrived. Also, the new colonists lived in large, permanent villages, grew crops, and so had a whole range of new artifacts. Given the dramatic appearance of this new archaeological evidence, the arrival of Saladoid people was viewed as a major migration of South American people up the Lesser Antillean chain, overwhelming, displacing, or absorbing the previous occupants. Saladoid sites with remarkably similar archaeological characteristics have been found on most of the islands. Caribbean archaeologists have viewed this as a wave of migration creating a Saladoid horizon, which swept through the Antilles as far as eastern Dominican Republic and had a major impact on all subsequent Caribbean history.

In recent years this picture has become somewhat more complicated. It now seems likely that the migrating group was more diverse than previously thought, possibly involving multiple groups. The process of migration probably involved voyages of reconnaissance, short-lived settlements, and retreats. Moreover, the interactions with Archaic people who were already living in the islands probably played a more important role than was thought in shaping the Saladoid advance, their adaptation to the Caribbean environment, and Saladoid culture itself.

This chapter explores the long trajectory of population movement, from its complex beginnings in South America to the point at which colonization stalled along the eastern shores of the Dominican Republic. Both the points of origin and the processes of movement -- the complexity of which is only now being recognized -- are considered. One of the most striking example of this complexity is the enigmatic presence of archaeological remains called La Hueca, or Huecoid, on Puerto Rico, Vieques, and a few other places. As will be discussed, La Hueca shows that groups in the early Saladoid period were not as homogeneous as they once seemed, and it invites us to consider the significance of artifactual differences in terms of their cultural meanings and importance.

Mainland Origins

Where did the Saladoid people come from? This has been a source of heated debate in Caribbean archaeology since the 1950s. Some, including Donald Lathrap and Irving Rouse, saw the ancestors of the Saladoid people coming from the Amazon basin, crossing into the Orinoco drainage, and eventually moving into the Caribbean. Others, including Betty Meggers, Clifford Evans, Mario Sanoja and Iraida Vargas, saw their origins to the west, in the large Colombian river valleys at the feet of the Andes. These differences in interpretation have not been the result of animosity or personality conflicts among the participants, as has sometimes been suggested, but rather they illustrate the difficulties of approaching a very complex set of archaeological questions with a still very fragmentary and partial archaeological record. The variables include multiple possible points of origin, the possibility of systematic problems with the contamination of radiocarbon dates from the Middle Orinoco, and perennial difficulty in archaeology of comparing stylistic and technological modes of poorly dated ceramic assemblages from far-flung locations and trying to decide which is related to which. Long geographic extrapolations are necessary because in lowland South America there are so many large areas lacking in basic archaeological research.

Nevertheless, with the benefit of ongoing research, the origins of the indigenous people of the Caribbean are clearer than they were in the past (Boomert 2000, Oliver 1989, Vargas 1990). In light of this, we can reexamine the evidence for the mainland origins of the Saladoid people.

From a long career of work in lowland South America, particularly in the upper reaches of the Amazon basin, Lathrap (1970) constructed a scenario in which the Saladoid and Barrancoid originated in the Amazon. He argued that they crossed into the Orinoco drainage at the Casiquiare Canal, an area of marsh and small river channels that connects the upper reaches of the Río Negro and the upper Orinoco (Figure FF). Across this canal, Lathrap and others reasoned, people could paddle or pull their canoes from one river system to the other. Boomert (2000: 115, and section 2.1) sees other possibilities for crossing this divide, arguing that ethnohistoric evidence supports use of the Pirara Portage between the Río Branco and the Rupununi/Essequibo systems in southern Guyana. As will be discussed, the archaeological evidence for Saladoid people using this route is also stronger than via the Casiquiare Canal. There are other close connections between northern tributaries of the Amazon and the rivers flowing north from the Guiana Highlands, and several routes of movement across the divide was possible.

Lathrap saw the strong connections between the first Saladoid people in the Orinoco and the pottery makers from high in the Amazon basin in eastern Peru. His reconstruction was that, "[t]he distribution of these complexes around the whole periphery of the Upper Amazon Basin suggests that this homeland is to be sought in the Central Amazon near the confluence of the Upper Amazon, the Negro and the Madeira. The ceramic evidence points to a pattern of population movement similar to that indicated by the distribution of the more divergent branches of Arawakan. It is of greatest interest that the Antillean Saladoid ceramics can be attributed without reasonable doubt to the speakers of Proto-Taino, the northernmost of the non-Maipuran Arawakan families, while there is every reason to believe that the Nazaretequi tradition was the work of the speakers of Proto-Amuesha" (Lathrap 1970:112).

Betty Meggers and Clifford Evans (1978) and Sanoja and Vargas (1978, 1983) saw Saladoid and Barrancoid origins in the northwestern part of South America, in particular in the major river systems draining the Andean highlands of Columbia, Ecuador and Peru. Meggers, who with Evans worked in many parts of lowland South America, from the mouth of the Amazon to the shadows of the Andes in Ecuador, proposed an extensive "Amazonian Zoned Hachure" tradition that includes ceramic complexes from Puerto Hormiga in northern Colombia to the mouth of the Amazon. In this tradition she includes both La Gruta, the earliest Saladoid pottery in the Orinoco, and Tutishcainyo ceramics from the Upper Amazon as well (Meggers 1987, 1991).

One thing is clear: the Saladoid ceramic series, and what we have come to understand as Saladoid culture, did not develop in situ in the Orinoco. The people making these ceramics certainly changed and developed them once they were there, but Saladoid-related ceramics came in the Orinoco drainage as highly sophisticated ceramic complexes. Although the Alaka people of the Guianas, who were living around the lower Orinoco, did make pottery in the their later period, it was comparatively crude and unrelated to the Saladoid ceramics either stylistically or technologically.

The picture that has been emerging in the last 20 years is that the Orinoco basin was a zone of intensive cultural contact and interaction for thousands of years, and particularly so in the last 2500 years B.C. Given the substantial influences coming in from outside the region, from both the river systems and llanos to the west and the Amazon basin to the south, it now seems much less productive to debate whether the origins of the first ceramics-using people of the Caribbean were strictly from one region or another. The Middle and Lower Orinoco was the crucible in which Saladoid, Barrancoid, and many other archaeological cultures developed, and there was usually a high degree of interaction, even over large distances, between the many groups living there.

What this means is that the archaeological situation in the Orinoco drainage and the Guyanas is very complex, much more so than in the Caribbean. The multiple overlapping and intermixed pottery traditions clearly reflect a cultural geography with multiple interacting groups. An example of this is the widespread and long-lived distribution of the sort of Cedeñoid pottery excavated by Alberta Zucchi and Kay Tarble at the sites of Agüerito and Cedeño in the Middle Orinoco (Zucchi and Tarble 1984; Zucchi, Tarble, and Vaz 1984). These two sites are near one another about 600 km by river from the sites of Saladero and Barrancas, and another 600 km from the divide between the Orinoco and Amazon drainages. Cedeñoid pottery is present in the Middle Orinoco at about the same time as Saladoid pottery, and in fact Cedeñoid occurs in small quantities in many of the sites that are predominantly Saladoid. Also, both Cedeñoid and Saladoid pottery occur as minority or "trade" pottery at important Middle Orinoco sites such as Corozal, Ronquín, and La Gruta (Boomert 2000: 109-110; Roosevelt 1997: 155-162; Zucchi 1991). As Roosevelt notes, "Thus, early Corozal seems a period of rapid, extensive, and parallel change in several aspects of ancient lifeways. . . . The great distance between the new Corozal pottery wares, shapes, and decorations and the older Saladoid-Barrancoid pottery on the one hand, and the evidence for interaction between pottery of the different traditions on the other, suggest that people of different geographic origins may have come together in the same community. . ." (Roosevelt (1997:161).

This suggests that the period from 1000 B.C. until the first centuries A.D. was a time of flux and change in the Orinoco, involving several groups who were trading long distances and sometimes living together in the same places. Roosevelt suggests an analogy to modern times in the lowlands, in which individuals or groups might migrate long distances in search of better economic possibilities (1997: 161). Many others have documented the complex cultural landscape of the lowlands, with patterns of occupation, trade, and population movement that would be extremely difficult to reconstruct on the basis of archaeological data (Basso 1995; DeBoer 1981; Forte 1996; Whitehead 1988; Wilbert 1996).

Another way of trying to unravel the complexities of the cultural interactions of the Orinoco in the last 2500 years B.C. is by using the techniques of historical linguistics. For our purposes the most important of these techniques is linguistic glottochronology, which attempts estimate the point in time at which two modern languages diverged from a common ancestor. In northern South America, given the cultural complexity, movement of groups, and difficulties in interpreting the archaeological record, it is very difficult to employ these methods with any precision. Oliver (1989) has done the most careful and thorough job with this in reconstructing the Arawakan expansion into northwestern Venezuela and eastern Colombia. He divides the expansion of Arawakan speakers into northern South America into three major stages (pp. 173-182). In the first stage, one group of Proto-Maipuran speakers (the presumed ancestors of the people who went into the Caribbean) diverged from proto-Arawakan in the central Amazon before 1500 to 2600 B.C., and moved into the upper Orinoco and upper Río Branco. From there one group of them continued westward into the Llanos of western Venezuela and Lake Maracaibo, while another remained in the upper and middle Orinoco. In Oliver's second stage, the group that had remained in the Orinoco moved to the Orinoco delta, some moving out into the Antilles, and others moving southeastwards into the coastal Guyanas. In his stage three, further movement and linguistic differentiation takes place in western Venezuela and the lower Amazon, but by that time the groups that became the historically known people of the Caribbean - the Taíno and Island Caribs - had diverged from their mainland relatives.

In considering the mainland origins of the Saladoid people, it is also important to recognize the important relationship between the Saladoid people and another group living in the Lower Orinoco at the same time, who are generally known by their Barrancoid pottery style. The relationship between the Barrancoid and Saladoid groups has important implications for the prehistory of the Antilles, because interactions with Barrancoid people continued from 200 to 600 A.D. and after in the Lesser Antilles. It influenced the initial movement of people into Antilles, and had a ongoing influence through interaction with the mainland for centuries afterward. Barrancoid influence can be seen in Caribbean pottery after A.D. 600, even in the Greater Antilles, as potters moved away from painting flat surfaces to suing clay more sculpturally, with heavy modeling and deep, broad-line incision.

Although the Saladoid ceramic tradition had deeper roots in the Orinoco basin, extending back to about 2500 B.C., the period of Barrancoid interaction was a powerful historical force motivating change. This interaction led to one of the fruitful periods of creativity and cultural expression that I believe are a part of a pattern - that periods of cultural interaction are also periods of great creativity and innovation in the Caribbean (and probably human societies generally). As Arie Boomert notes, "Undeniably, Barrancoid ceramics form the artistic climax of pre-Columbian culture in the Caribbean, being both highly aesthetic and imaginative" (2000: 118). Sanoja and Vargas (1974:97) describe Barrancoid art of the period as containing a wealth of decorative motifs, "representing mythical beings that were part human and part animal (bats, felines, fish, etc.), which indicate a rich ceremonial life and the existence of an animist religion in which the deities or natural spirits had a great importance."

Around 800 B.C. people making Barrancoid pottery show up in the Lower Orinoco, in the region around the present-day city of Barrancas, where the Orinoco begins to break up into the dozens of rivers of the 200-km-wide Orinoco Delta. Their occupation seems to be intrusive, rather than a local development, and their origins have been the subject of considerable debate (Lathrap 1970, Rouse and Cruxent 1963, Sanoja and Vargas 1983). Boomert's recent review (2000: 93-125) considers the problems of poor dates and uneven archaeological information and concludes that the Central Llanos of Venezuela and Colombia are the most likely immediate source, but that the northern Andes and the river systems of northern Colombia are still possible sources of influence. Thus in the period of cultural interaction in the Middle and Lower Orinoco in the last millennium B.C. both Andean and Amazonian ancestries may have had some expression and influence on the cultures of the Caribbean (Rouse 1991). The Barrancoid influence extended into the southern Lesser Antilles until 600-800 A.D., and extended as far as the Leeward Islands.

As noted at the beginning of this chapter, Saladoid archaeological sites are distinctive because of their white-on-red painted ceramics. Many writers in the early 20^th century commented on the finds, and it is clear from these early writings that there was the general sense that several different cultures had existed in the Caribbean, that these different cultures had flourished at different times, and that the sequence of fluorescences was related to their order of migration to the Caribbean. In 1924, Swedish archaeologist Sven Loven published his Über die Wurzeln der tainischen Kultur in German, and in English translation in 1935. His widely shared view was that there were, "four distinct races in the West Indies properly speaking, from historical sources. The names are given in the order which they must have immigrated" (1935:2). He then listed (1) Guanahatabeyes (Siboneyes); (2) Island-Arawaks; (3) Maçoriges (Ciguayos); and (4) Island-Caribs.

Jesse Walter Fewkes, in his analysis of the artifacts in the Heye Collection (Fewkes 1922, see also his 1907 work on Puerto Rico and neighboring island), had proposed "three cultural epochs, grading into each other, which may indicate a sequence in time or distinct cultural stages. These epochs were of the cave dwellers, the agriculturalists, and the Carib" (Fewkes 1922:56). Fewkes noted that all three groups were still living in the Caribbean at the time of European conquest. Fewkes's objective with the Heye collection were not so much to establish a chronology, but to look for geographical areas of cultural similarity, of which he identified several.

An important step came in the 1920s, when Gudmund Hatt, a professor of archaeology at Copenhagen University in Denmark established a more explicit provisional archaeological sequence for the region. He carried out archaeological research on the island of St. Croix in the early 1920s, excavating about 30 sites in the Virgin Islands (Morse 1995:473-4). The Virgin Islands had been a Danish possession in the Caribbean until the United States bought them in 1917 for $25 million. They remained an area of Danish interest thereafter. Although others had excavated in the Caribbean - indeed Theodoor de Booy of the American Museum of Natural History had worked on some of the same sites as Professor Hatt (de Booy 1919) - Hatt carefully analyzed the stratigraphy of the sites he excavated, and based on it he proposed a three-phase sequence for the prehistory of the Virgin Islands (Hatt 1923, 1924). He called the first phase Krum Bay, after the preceramic site he explored on the southeast coast of St. Thomas. As discussed in the previous chapter, it was characterized by ground and flaked stone tools, lots of shellfish, and a lack of pottery. Hatt's second phase was Coral Bay-Longford, referring to sites with white-on-red pottery and other Saladoid characteristics. The third and most recent phase he called Magens Bay - Salt River. These sites contained somewhat less carefully made pottery vessels with modeling and incision, and without white-on-red painting. The site of Salt River, on the northern coast of St. Croix, also had a ball court comparable to those at Taíno sites in Puerto Rico and the rest of the Greater Antilles.

Hatt's chronological sequence for the Caribbean is still essentially intact today, and it provided a much better conceptual framework for understanding human history in the Caribbean than had existed before. There is considerable variability in how people have interpreted this sequence, and in the names they have given to the phases, as they have variously constructed them. Because this has been a relatively important issue in Caribbean archaeology, we must briefly explore the terminology issue here.

In the years following Hatt's work, the relative chronology he proposed has been refined and elaborated upon considerably. The work of Irving Rouse in Haiti (1939, 1941) and Cuba (1941), and the work supported by the New York Academy of Sciences in Puerto Rico (Mason 1941, Rainey 1940, Rouse 1941, 1952), provided a great deal of new and systematically collected archaeological data. Based on this and his ongoing research in the 1950s, Rouse proposed a relative time scale with periods, identified by Roman numerals I through IV, extending from the earliest human occupation to the time after the arrival of Europeans (Rouse 1964). These periods were subsequently subdivided (for example into periods III-a and III-b) and refined. Based on the characteristics of the artifacts, Rouse and others employing his methodology defined "series" grouping complexes and styles. Series were designated by the suffix "-oid" and in a refinement suggested by Gary Vescelius (1980), were further subdivided into subseries, which ended in "-an". Thus, for example, the earliest preceramic assemblages in the Greater Antilles were called Casimiroid, after the site of Casimira, whose artifacts were deemed representative of the series. Within this series, a further differentiation was proposed between Cuban sites, of which Cayo Redondo was typical, and Hatian sites, of which Couri was typical. Under the system of series and subseries, the two subseries came to be called Redondan Casimiroid and Courian Casimiroid.

Other systems for describing the chronological and archaeological variability of the Caribbean are also in use in Caribbean archaeology. Cuban archaeologists have developed a system of classification based on the economic systems of the prehistoric people. This system was influenced by Tabío and Rey's focus on the relations of production in their 1966 book Prehistoria de Cuba, and developed by subsequent generations of Cuban archaeologists. In 1995 a great deal of information on Cuban archaeology was assembled on the CD-ROM "Taino: Arqueología de Cuba" by the Centro de Antropología. The work was produced by nearly forty collaborators, and was edited by Jorge Febles (1995). The organizational framework for classifying archaeological remains is widely used in Cuba and elsewhere. It consists of two stages of economic development -- the economy of appropriation and the economy of food production - representing the delineation between hunter-gatherers and horticulturalists. Within the economy of appropriation are three phases -- hunting, fishing and collecting, and the protoagricultural phase. The economy of food production is dominated by the agricultural phase. Within this overall economic framework, the archaeologists discuss all aspects of life, including social and political organization, language, religion, art, and all of the technologies represented in the archaeological record.

Another system of classification that in a sense combines the two previous schemes. Luis Chanlatte Baik (1995) proposes one major division between an early period called Archaic and a later period of farmer-potters (arcaicos and agroalfareros). Within the Archaic, Chanlatte defines a preceramic period in which there was little or no interaction with pottery-making groups outside the islands, and an aceramic period of development that took place with some interaction between Archaic and people who made ceramics and practiced horticulture. Within the Agroalfarero or farmer-potter period Chanlatte describes two migrations - Agroalferero I and II - representing separate Huecoid and Saladoid migrations into the Caribbean. From the interaction of these groups and Archaic people emerges the Antillean Formative, which includes all subsequent prehistoric people. Agroalfarero III includes people between about 460 and 1025 A.D., and Agroalfarero IV including the late Taíno fluorescence seen by Spanish chroniclers.

It has been difficult for many people interested in Caribbean archaeology to make sense of what seems like a daunting profusion of series, phases, styles, periods, subseries, complexes, and other descriptive or taxonomic units. On one hand it seems like a problem of translation, because Caribbean archaeology is such a polyglot enterprise. It is actually, however, a more complex issue, because there is no simple way of classifying human cultures in time and space, and one's classificatory scheme depends a great deal on what you want to know. If a scholar is interested in the changing economic strategies of prehistoric people through time his or her system of classification will likely be based on economic data. If one is interested in the history of cultural groups as they move through a region and change through time, changes in the characteristics of durable artifacts such as pottery or other non-perishable artifacts may take precedence. In fact, however, most scholars are interested in a wide range of phenomena about human societies in the past, and so we have a confusing situation with multiple classificatory schemes, aims, and research directions. José Oliver (1989: 212-222) makes a compelling argument that multiple levels of taxonomic resolution are needed: "Each of these classificatory units has been used in response to specific problems, ranging in scope from a single site to an entire country. . . . Undoubtedly, each approach has its merits. However, classification is only useful to the degree that it can help to address or better understand the problems at hand" (1989:314). Oliver proposed a hierarchical classificatory model going from macro-tradition to tradition and then sub-tradition, with complex or style and phase being the lowest taxonomic level (1989:218-19). This seems very reasonable to me. However, in this volume my attempt has been to place emphasis on describing the scholarly conclusions about Caribbean prehistory, and less on the methods used in reaching those conclusions. Nevertheless, very quickly in the following discussion we encounter the question of how to interpret the differences between two different kinds of pottery, found in early Saladoid sites, and this returns us to the question of classification. One type of pottery has been mentioned briefly above. It is characterized by white-on-red (which is often abbreviated WOR) and polychrome painting, and has modeled and incised ornamentation on the rims or handles called adornos. Incision or some sort of painting is found on nearly half of this pottery. The pottery vessels included bowls, dishes, and jars, and they also made pottery griddles and other specialized pieces, for example potstands and incense burners. The other kind of pottery has been called Huecoid or Huecan Saladoid, and tends to be thinner and unpainted. It is often decorated with incision and modeling, and the classic decoration is called zone-incised-crosshatched (abbreviated ZIC), in which areas of the vessel surface are delineated with marked incision and filled with a fine crosshatching of lighter incision.

The problem, as we shall see, to that it is difficult to understand the cultural relationship between the people who made these two kinds of pottery, which are sometimes found mixed together, and sometimes quite separate from each other.

Before radiocarbon dating began to be used in the Caribbean in the 1950s, there was no way to anchor a relative archaeological chronology to absolute calendar years. There was also little way of determining whether a phenomenon like the region-wide appearance of Saladoid sites occurred as a sudden wave or a protracted trickle of people. As more radiocarbon dates have accumulated, even through the 1990s, it appears that it might have been both a relatively coherent movement, perhaps with some earlier explorations.

Figure FF shows the distribution of the Saladoid sites with the earliest radiocarbon dates in the Lesser Antilles and Puerto Rico (the figure shows those sites dated earlier than 100 B.C. based on the midpoints of each date's calibrated 2-sigma range). These dates carry with them varying degrees of confidence and as with all radiocarbon assays, are subject to many potential problems. New sites are being studied, and others, including some of those mentioned, have research ongoing, so this is a very provisional set of data from which to draw some tentative impressions. Nevertheless, some possible patterns may be observed. First, there is no suggestion of a gradual migratory wave of settlement moving from south to north in the Lesser Antilles, with, for example, beachheads on Grenada, the Grenadines, St. Vincent, and St. Lucia. On the basis of current research and existing dates, the early Saladoid migrants bypassed the southernmost islands. Instead, the earliest dates are farther north, on St. Martin, Montserrat, Martinique, and Puerto Rico. Second, these early dates do not appear to be just the earliest of a large number of dated sites and deposits. In other words, they do not seem to represent the arrival of the founders of lasting and expanding Saladoid occupations. Instead, they seem to predate the sites that came after by several centuries in some cases. This suggests that the earliest Saladoid strategy in the Caribbean was one of exploration and isolated, small-scale, and possibly intermittently-occupied settlements.

A third potential pattern is one of variability in the earliest Saladoid sites. While they have similarities, there are also significant differences to be seen among the eleven early sites shown on Figure FF. Three of the earliest are the Hope Estate site on St. Martin, Trants on the island of Montserrat, and Cedros, on Trinidad's southwestern peninsula. Some further discussion of these sites may help illuminate some of the characteristics of this period.

The site of Hope Estate is in the hills of northern St. Martin, about 2 km from the sea both to the east and the west. It sits at about 100 meters above sea level. It has been the subject of research by many archaeologists, including Haviser, Henocq, Hofman, Hoogland, Bonnissent, Richier, and Weydert, among others. The site was occupied several times in prehistory, but it is the first occupation that we are most concerned with here. Hoogland and Hofman carried out excavations at the site in 1993 and reconstructed three main phases of occupation. The earliest period occupation is thought to date from about 400-300 cal BC (Hoogland 1999: 129-147).

The ceramics from this early component are Huecan Saladoid - unpainted pottery decorated with incision, including zone-incised or ZIC, and some modeling (Hoffman 1999, Bonnissent 1995). This pottery also contained a rare decorative mode called zone-incised-punctate, in which instead of filling incised decorative areas with fine crosshatching, the potter filled it with punctations. The incised areas are bounded by curvilinear lines and filled with either crosshatching or punctation (see Figure FF). Simple, unrestricted bowls and jars are most common, and the assemblage also contains restricted contours. In Hofman's study (1999: 159-165) the average vessel diameter is 25.3 cm, and the average wall thickness was 6.8 mm. Rims were simple, typically rounded and sometimes flattened. Some vessels were ornamented with modeled adornos, usually zoomorphs with round eyes. A very small fraction of the pottery from this occupation phase, about 0.2%, have decoration associated with Cedrosan Saladoid ceramics - white-on-red painting and red slips - but these artifacts may be intrusive from higher in the site's complex stratigraphy (Hoogland 1999: 143-147).

Hofman and Hoogland interpret the second stratigraphic unit of the site to be a predominantly Cedrosan Saladoid occupation. The decorative modes on the ceramics include white-on-red and polychrome painting (about 20% of the decorated sherds), and some zone-incised crosshatching (about 10% of the decorated sherds). Over 50% of decorated sherds are incised, and there is also some modeling. Within this stratigraphic unit seven zone-incised punctate sherds were found (out of 12,928), but they were deemed to be a product of mixing during site formation processes. For the most part, the artifacts from this second component, which Hoogland estimated to date to about cal AD 450-600, were typical for a Saladoid site in the northern Lesser Antilles of similar age. As noted, it had both WOR and ZIC ceramics, griddles, highly variable chert artifacts, and beads and other artifacts of exotic stone (Haviser 1999). The site probably had a succession of houses dating from the earliest occupation onward (Hoogland 1999, figure 11.13).

In summary, Hope Estate is a site with a very early occupation that contains an unusual component that is exclusively Huecoid in character. It contains zone-incised punctate ceramics that are found in only a few other sites. After a long period of abandonment, Hope Estate is reoccupied with a more typical Saladoid settlement, containing both ZIC and WOR ceramics.

The site of Trants, on the east coast of Montserrat, is another very early Saladoid site, but with some characteristics that separate it from Hope Estate. It does have pottery that could be considered part of a Huecoid or Huecan Saladoid ceramic assemblage - i.e. unpainted vessels sometime with zone-incised-crosshatched decoration -- but unlike Hope Estate, it does not have a discreet deposit of this pottery without white-on-red pottery present, nor did it have zone-incised-punctate ceramics. Instead, the ZIC and WOR ceramics are consistently found together, and Petersen and Watters (1991: 292) note, "some of the WOR and ZIC vessels show common techniques and even nearly identical motifs between different site areas on the basis of surface and subsurface finds. For example, the earliest deposits at the site thus far identified, dated ca. 480-440 B.C. in fields 8 and 9, share a common unslipped, bowl form with a flanged rim that was decorated using a common form of broad line ZIC and a similar punctate adorno, along with WOR decorated vessels. The former occurrence, among others suggests that these areas were occupied more or less contemporaneously by people sharing very similar, if not precisely identical ceramic styles." From their descriptions and profile drawings, it seems as if the largest part of the midden at Trants, stratum B, dates to the last century B.C. to about A.D. 200 (see also Watters 1994, Watters and Petersen 1995). However, the investigators believe that the site's basic configuration, a 200 by 300 meter oval midden ringing a central plaza, was established before 400 B.C., and early dates come from several parts of the site (Petersen, Bartone, and Watters 1995: 44).

Of all the early Saladoid sites, the closest one to the South American mainland is the site of Cedros, on the southwestern peninsula of the island of Trinidad. It looks south onto the channel called the Serpent's Mouth, and to the mainland 15 km away. Like many early Saladoid sites, it is located on a rise with a commanding view about 200 m back from the coast. It is close to freshwater swamps and mangrove areas, and also close to the geothermal features called mud volcanoes that are common on the southwest coast. It is hard to know the site's original extent, but the relatively modest size of the remaining midden suggests that it was not large, perhaps consisting of a few houses, with the trash midden on the slopes of the hill downwind from the houses (Boomert 2000: 264). The early component probably dates to around 100 to 300 B.C. (Boomert 2000: 127-145), so it probably postdates the sites just mentioned by a century or two.

Cedros is another classic example of the early Saladoid sites, and its name was given to Rouse's Cedrosan Saladoid subseries (1992: 77-85). Boomert analyzed the pottery excavated at Cedros and the nearby site of Palo Seco, and provides a detailed description (Boomert 2000: 127-145; see also Harris 1978). Of the decorated sherds, 51% were painted in some fashion, and 14% had white-on-red painting. Others had simple red or orange paint, white or black paint alone, or polychrome painting. Sixteen percent of the decorated sherds had ZIC decoration and another 19% had other kinds of incising. Boomert identified seven common vessel forms and seven rare forms. The common forms are mostly unrestricted bowls with rim forms varying from thickening and flanges to elaborate compound rims. The functions performed by the vessels included storage, cooking, food preparation that did not include heating, serving, and transport of the vessels' contents (Boomert 2000:136-139).

As may be seen from these three sites, there is considerable diversity among the early Saladoid sites. Of critical importance is the relationship between the two dominant decorative styles of the pottery - white-on-red painting, and zoned incision. What does the presence of these two modes mean in terms of the human processes of exploration, colonization, and population movement, and how should sites like La Hueca and those similar to it be seen in Caribbean prehistory?

In the late 1970s, Luís Chanlatte Baik and his colleagues at the University of Puerto Rico made one of the most dramatic and important archaeological discoveries of the late 20^th century at the site on La Hueca, on the island of Vieques, just off the coast of southeastern Puerto Rico (Chanlatte 1981, Chanlatte and Narganas 1980). Until that time the zone-incised-crosshatched (ZIC) pottery (which above I have called Huecan) that was typically found in Saladoid archaeological sites was thought by nearly all Caribbean archaeologists to be a minority constituent of the Saladoid assemblages, diagnostic of the early Saladoid period. At the site of La Hueca, however, only ceramics affiliated with this ZIC tradition were found, without the white-on-red pottery characteristic of the abundant Saladoid sites of the Lesser Antilles and Puerto Rico. Moreover, La Hueca produced spectacular examples of small carvings in exotic stone and shell - frogs, bats, amulets, beads, disks, and raptorial birds holding human heads (Figure FF).

Some sites, including Punta Candelero, La Hueca, and the earliest levels of Hope Estate, have only early Huecoid ceramics. There are other early sites that have comparable assemblages, including decorative modes with curvilinear incised zones filled with punctations, or zoned hachure. As noted, similar early Huecan ceramics have been found at Trants (Peterson 1996, Watters 1994), as well as at Morel (Durand et Petitjean Roget in 1991: 53-8) and Anse Patat (Hofman 1999: 311), and perhaps Folle Anse on Guadeloupe, and perhaps El Convento and Maisabel on Puerto Rico (Siegel 1992). For some of these sites it is likely that the early Huecan material occurred alone, without white-on-red pottery or other Saladoid diagnostics.

When all of these Huecan-related sites are considered together, three important things must be acknowledged about the sites in which Huecan material is found in isolation from WOR ceramics. First, the people who lived at La Hueca and similar sites cannot possibly be viewed as a group which preceded later Saladoid people and had disappeared or were on the way to disappearing during the early Saladoid (in Puerto Rico, during Hacienda Grande times). The charcoal dates from Punta Candelero span more than 1000 years, from around cal BC 250 until after cal AD 1000 (Rodriguez 1989: table 1). A similarly long sequence has been documented at La Hueca, and it is also quite well dated through the first millennium and into second millennium AD. Indeed in the "New Extension" excavations of Block Z, a region of the site that has only La Hueca style ceramics, dates from about cal AD 1050 until after the time of European conquest or after (Narganes 1991). Following Oliver (1999: 271) I agree that we cannot ignore this persuasive evidence for a very long-term occupation at La Hueca and Punta Candelero, and must fundamentally reassess what it means for Caribbean prehistory.

Second, it is now clear that Huecan and Saladoid sites existed side by side for long periods. The site of Sorcé is immediately adjacent to La Hueca and has a more conventional Saladoid date range of cal BC 100 to cal AD 700 (Narganes 1991, Oliver 1999), as well as a general style of WOR pottery decoration characteristic of the bulk of contemporary Saladoid sites. Based on the radiocarbon chronology for the sites, they were both occupied for all that time. Punta Candelero also has adjacent non-Huecan components. Third, it must be acknowledged that there are significant differences between the Huecan and the Saladoid artifacts, differences that may have deeper cultural and historical ramifications than ceramic decorative modes might suggest. One important area of research on this is the lithic analysis done by Reniel Rodríguez Ramos (2001) on the sites of La Hueca and Sorcé, and on both the Huecan and Saladoid (Cuevas period) components of Punta Candelero. He concluded that there were identifiable differences between the lithic assemblages of the Huecan and Saladoid sites (Rodríguez Ramos 2001: 178-1888). He saw differences in the raw material that was used, and differences in the way that raw material, in the form of cores of nodules of chert, was turned into usable tools. Chert was far more prevalent in the Huecan components; establishing comparability in units was complicated but in general the Huecan components had 3 to 8 times as much chert by weight per excavated meter than the Saladoid components. The Huecan reduction strategy, which aimed to produce flakes from discoidal cores, was different from the rather haphazard Saladoid approach, and it left different kinds of cores. Within the ground stone assemblages, the most common tool - the biconvex celt - is found in both components, but the Saladoid deposits had ground stone tool types absent in the Huecan deposits and vice versa (Rodríguez Ramos 2001: 144-174, 183-187). In sum, the lithics from these two kinds of sites were related in some ways, yet clearly distinct.

It is increasingly clear from recent research that La Hueca and Punta Candelero, and probably others, are distinct and special communities, different in material and cultural senses from the broader Saladoid community that surrounded them. As noted, that distinctiveness was very long lived. Given that, one might be tempted to see the Huecan community as isolationist, rejecting the Saladoid mainstream, in a manner analogous to the Amish. The Amish, Mennonites and other groups of "Plain People" have advocated a separation from the modern world since the inception of the Anabaptist movement in the early 1500s, following the biblical mandate, "Be not conformed to this world" (Romans 12:2). In places like Lancaster County, Pennsylvania, in the U.S., the technology and material culture used by Plain People is quite distinct from their secular neighbors, and would appear so archaeologically if their settlements were excavated. Similarly, some groups of Hasidic Jews maintain group identity and observe scriptural law in ways that mark them as culturally distinct. To some degree this distinction would also register archaeologically.

In Caribbean prehistory, however, the analogy of a cultural or religious minority within a substantially different mainstream culture is not entirely applicable. In Saladoid times, the artifacts related to the Huecan phenomenon -- ZIC ceramics -- are found as a minority constituent of nearly every contemporaneous ceramic assemblage, and they continue to be present for nearly 1000 years. Clearly there was some relationship between the people at Huecan sites and the others. Perhaps a better analogy can be seen in monastic communities such as Buddhist monasteries, or the monastic orders with the Catholic or Greek Orthodox churches. Religious communities such as these share some characteristics with sites such as La Hueca and Punta Candelero: they are distinct from, but embedded in a cultural matrix to which they are related; their distinctive artifacts, symbolically rich and non-utilitarian, are found in the dominant culture as minority constituents. And while they share many material traits with surrounding people they are distinct and identifiable. Among the distinctions is a markedly greater prevalence in the Huecan sites of artifacts related to shamanistic hallucination or trance rituals. These special purpose artifacts are used for inhaling consciousness-altering powders or smoke, or drinking special drinks, and are clearly related to the cohoba rituals of the historic Taíno (Rodríguez 1997: 85-86). These practices also related to divination and communication with supernatural beings.

Envisioning La Hueca and Punta Candelero as a cultural enclave, perhaps with religious functions and certainly very long lasting, runs counter to a view of Caribbean prehistory that sees large periods of cultural homogeneity as the norm. It fits better with a view of the Caribbean, during most of its prehistory and in historic times, as a relatively stable mosaic of cultural difference. Whether we imagine them as enclaves with special social functions or persistent cultural isolates involves great speculation on our part. Clearly the Huecan phenomenon is one of the most interesting topics in Caribbean prehistory and an exciting area for continued research.

Apart from the relationship between Huecan and Saladoid people just discussed, and other internal complexities of Saladoid society, there were several other important groups they interacted with.

The most important of these were the Archaic people whom the Saladoid people joined in the Caribbean. As noted in Chapter 2, the Archaic people in the Antilles had two principal areas of origin, one in Central America and one from South America via the Lesser Antilles. These two groups had met and interacted in the territory between western Puerto Rico and Antigua for more than a thousand years before the Saladoid people arrived, and no doubt in that time had diverged into many regional groups, specializing in different areas of the Caribbean.

Saladoid migrants from the Orinoco moved up into regions that the Archaic people had long occupied, and apparently early Saladoid settlement patterns reflect this. Higher islands seem to have been selected over lower ones and places such as the island of Antigua, strongly favored for Archaic settlement, were avoided at first. Early Saladoid settlements were often situated in places with good views, on the windward sides of islands (although this generalization may mask a great deal of variability in locations). Some early sites, such as Hope Estate on St. Martin, seem almost hidden from view from the coast or sea. It would not be hard to imagine that some tensions developed between these two populations, but it does not show up in the archaeological record as it often does in more violent confrontations - as defensive architecture, armaments, or characteristic patterns of injuries in the burial population (Carmine 1997, Lambert 2000).

There is some evidence for cultural interaction or "transculturation" between Archaic and later people. At the Huecan sites of Punta Candelero and La Hueca, Reniel Rodríguez found continuities between Archaic and Huecan use of lithics. The characteristic "edge grinders" or edge-ground cobbles that were probably important in processing plant materials continued in the Huecan sites. These artifacts, and round, pitted stones, are also found in some early Saladoid (Hacienda Grande period, in the Puerto Rican ceramic sequence) sites. The degree of technological interaction seems to vary among the early sites. As Rodríguez Ramos notes, ". . . the [La Huecan] context of Punta Candelero seems more "Archaic-like" in its technological emphasis than that from La Hueca site. The greater emphasis on freehand flaking and other overall unmeasurable qualities in that artifactual repertoire, make us estimate that the [Huecan] occupation of Punta Candelero is earlier than that of La Hueca site, as is also reflected by the radiocarbon dates collected from these sites. Also, the lithic remains seem to reinforce Haviser's (1991) argument that the [Huecan] materials [from the] La Hueca site presents a higher degree of Saladoid influence when compared to other Early Ceramic contexts such as that from Hope Estate." Rodríguez cautions that some of these ideas are speculative and must be evaluated with future research, but they are certainly plausible if one envisions the early Saladoid period as one of multiple interacting cultural groups.

At the site of Hacienda Grande, Rouse and Alegría (1990: 66) found similar "hammer-grinders" and "edge-grinders" that might be the result of mixed deposits, but they could be evidence of Saladoid people adopting the tools (1990: 20-25). Walker (1985: 186, 190-91) and others are convinced of the edge-grinding tool's continuity from Archaic to Saladoid times. If these tools were involved in processing vegetable foods, as can be inferred from their use-wear, they may indicate the transfer of local knowledge and ways of adapting to the Caribbean environment from one group to the other, if not the interaction and fusion of the two in a longer-term cultural sense.

A very important example of the possible interaction of Saladoid and Archaic people occurs in eastern Dominican Republic. Marcio Veloz Maggiolo, Elpidio Ortega, Plinio Pina and others from the Museo del Hombre Dominicano excavated the site during a period of widespread and intensive archaeological research in the Dominican Republic in the 1960s and 1970s. In their 1976 publication Ortega and his colleagues noted, "The site of El Caimito was worked for the first time in 1972, and it was considered to be preceramic since it produced a radiocarbon date of 15 B.C. . ." (Ortega et al. 1976: 276). When they discovered ceramics they returned to the site for a more extensive program of field work and excavated most of a small mound. Four radiocarbon dates yielded a range from cal 390 BC to cal 410 AD. Apart from the ceramics, they found a range of artifacts that would typical of Archaic period sites - flake tools, edge grinders or martillos, manos, metates, and mortars, fragments of petaloid axes. The excavators did not find cassava griddles among the ceramics, and the faunal and botanical analyses conducted suggested a dependence on collecting wild foods, and the possible absence of cultivation, at least cultivation or domestication on a large scale (Ortega et al. 1976: 280).

In the 1970s Caribbean archaeologists in general were paying a great deal of attention to the sequence of prehistoric migrations and their impact on the archipelago (Keegan 2000). Initially, El Caimito was interpreted from the widely-held perspective that different kinds of assemblages usually meant different migrations from the mainland. Ortega et al. (1976:280) concluded that "El Caimito showed that non-‘Saladoid' groups arrived in the Greater Antilles area at the same time or before the Saladoid groups, and made or acquired ceramics of good quality." Alberta Zucci also related the El Caimito ceramics to some material at the site of Agüerito in the Middle Orinoco (Zucci 1984). Indeed, the site may still prove to be evidence for an exploratory foray into the Caribbean by that or some unknown group from the mainland. In recent years, however, other explanations apart from migration have received wider consideration in the Caribbean. Luis Chanlatte, Marcio Veloz Maggiolo, Irving Rouse, and others have assigned greater importance to processes of cultural interaction going on in place within the Caribbean. From that perspective, El Caimito could be interpreted as a manifestation of the interaction of Archaic people with Saladoid and Huecan people, which was what was going on in the eastern Greater Antilles and particularly in areas such as eastern Dominican Republic. The appearance of typically Archaic implements in Huecan and Hacienda Grande sites in Puerto Rico and Vieques suggests a transfer of both technology and of adaptive strategies. El Caimito probably shows a similar transfer of a Saladoid or Huecan technology - pottery making - to the people who had lived in the Greater Antilles so successfully for so long. In that interaction it was the Saladoid people who had the most to gain, since over millennia the Archaic people had developed sophisticated knowledge about the land and resources, climate, possibilities, and dangers in the islands.

In some ways the pottery of El Caimito is strongly reminiscent of later pottery on the island of Hispaniola, especially that known as Meillacan, the most widespread pottery on the island (Rouse 1992: 92-101, Veloz Maggiolo, Ortega, and Caba Fuentes 1981). Its decorations include parallel line incision and, like some of the pottery associated with the historic Taíno, incisions ending in punctations. In some ways both the El Caimito pottery and the Meillacan, with their areas of parallel incised lines patterned at perpendicularly to each other, and occasional basketweave applique, seem intentionally to evoke the decoration of baskets made of organic materials (Figure FF, also see Veloz Maggiolo et al. 1981). It is as if the material technology of ceramics was adopted based on Saladoid or other models, but to a large extent the potters' decisions about decorative treatment were based on existing basketry, rather than Saladoid decorative modes. This preference was not exclusive, however, because some early Hispaniolan pottery does follow Saladoid pottery in stylistic motifs and execution (Veloz Maggiolo 1972: 308-311). As will be discussed in the next chapter, Meillacan pottery clearly has strong continuities with Archaic material culture, and El Caimito is very likely the earliest known example of the process of cultural interaction that gave rise to it.

By the time the Saladoid people moved into the Caribbean, and for several thousand years before, they had been living in permanent villages and subsisting in part on domesticated plants. Their economy was adapted through hundreds of human generations of refinements to the lowland riverine environments, and they made use of hundreds of species of plants, both wild and domesticated. They also hunted, fished, and collected a wide range of animals, from large mammals such as deer and tapirs, to shellfish and crabs. Apart from using plants and animals for food, the mainland ancestors of the Saladoid people who came to the Caribbean also used the forest resources for medicines, adornment, and nearly every part of their material culture. As with the Archaic people who preceded them into the Caribbean, more than ninety-five percent of their material culture was made of perishable materials.

The main plant in their horticultural system was manioc (Manihot esculenta, also called cassava and yuca), a root crop that was a good source of carbohydrates and calories (Piperno and Pearsall 1998). Manioc is a staple throughout much of the lowland tropics, and there are hundreds of different kinds. They can be roughly divided into "sweet" and "bitter" varieties, depending on their levels of toxicity. Without careful processing, compounds in raw manioc break down into hydrocyanic acid, which is poisonous. Sweet manioc can be processed fairly easily so that the toxins are eliminated through cooking. Bitter manioc receives more elaborate treatment, involving peeling, grating, washing, straining of the mash, drying, and ultimately cooking into large, thin cakes. Part of the rationale for this process is to eliminate the hydrocyanic acid, but it also turns the manioc, which cannot be stored for long once dug out of the ground, into a flour which can be stored. The rhythms and demands of manioc run through many parts of the lives of people who depend on it. Well drained fields must be cleared and planted, and the roots take between six and eighteen months to mature, depending on the variety. Soil fertility declines under manioc cultivation and when it is possible, people will move their gardens after a few years, clearing another patch of river terrace and letting the old ones be taken over again by the forest.

Manioc has a special place in the world of the Saladoid people and their descendants in the Caribbean, just as the primary staple plant does in most horticultural societies. But it was far from the only crop they grew. As in much of the Americas, maize, beans, and squashes were probably significant domesticates, and chili peppers, in their vast range of forms from small to large and mild tasting to explosive, were also a part of the South American cuisine taken to the Antilles (Newsom 1993). Some of the many fruits that were in use by the historic Indians of the Caribbean may have been native to the islands, and others were brought from the mainland. Most of these are still common plants in the West Indies, and demonstrate the considerable extent to which indigenous knowledge and lifeways have survived into the present. For most of these there are many and varied common names and spellings, but some of the most common are mamey apple (Mammea americana), custard apple (Annona reticulata), soursop (Annona muricata), guava (Psidium guajava), papaya (Carica papaya), pineapple (Ananas comosus), avocado (Persea americana), and passion fruit (Passiflora edulis).

The archaeological evidence from the Saladoid period is not complete enough to claim with confidence that all of these plants were taken to the Caribbean Saladoid times. There is, however, enough evidence to conclude that the lowland South American economy transported to the Antilles by the Saladoid people (and probably to some extent by their Archaic predecessors as well) was rich and varied. Even among the root crops they had a dozen other alternatives to the many kinds of manioc they grew (Boomert 2000:97). With the addition of all of the kinds of wild plant foods to be had in the islands, and the wealth of marine and terrestrial animals, they had an extremely productive economy.

To summarize, the food production system of the Saladoid people had been developed over millennia on the lowland rivers of the Orinoco and Guyanas, and brought into the Caribbean. Over time they adapted this economic system to the island environments, adding endemic plants and animals to their list of useful species and learning the different climatic rhythms of the islands. After an initial period of exploration and probably more frequent movement, Saladoid settlements became very stable, with sites being occupied continuously for centuries with little change in their size. Many other features of Saladoid cultural life also show a similar conservatism. The consistent use of some iconographic motifs in Saladoid pottery and other media over time reflects a relatively unchanging cosmology and world view, and many elements of this cultural system of symbols and shared meanings persists into historic times as part of Taíno culture. Much of our understanding of Saladoid cosmology comes from their pottery and other media such as carved precious stone and shell. Recently a great deal of new insight has been gained through the archaeological exploration of Saladoid houses and village layouts. It is there we will begin, before turning to pottery and other forms of cultural expression.

A remarkable change in archaeologists' vision of Saladoid life began in the 1980s with the large-area excavation of substantial parts of Saladoid villages. This was done at the site of Golden Rock on St. Eustatius, Maisabel in northern Puerto Rico, Punta Candelero in southeastern Puerto Rico, Tibes in southern Puerto Rico, Tutu village on St. Thomas, Trants on Montserrat, and on smaller scales at several other sites. Aad Versteeg's research at Golden Rock from 1984 to 1988 was a pioneering example of this research strategy. Before this project the majority of archaeological explorations of Saladoid sites focused on stratified midden deposits, where over time trash and food remains had accumulated. In these middens pottery, lithic tools, and nearly all classes of artifacts could be recovered in the greatest numbers with the least amount of excavation. That kind of concentrated block excavation was essential in reconstructing a chronological sequence of occupations, and in discerning changes in artifact type and frequency through time. Indeed the overall chronology of Caribbean prehistory was largely based on this sort of concentrated excavation strategy. Apart from occasional postmolds, however, archaeologists encountered very few remnants of the Saladoid people's built environment - the houses they lived in and the other structures they built and used in their settlements. Many archaeologists prior to these research projects (and I speak from personal experience) tended to see trash middens as the most significant indicators of the presence of archaeological sites, without a clear concept of how these middens related to the houses, plazas, and activity areas that were so revealing about the lives of the people who lived there.

At Golden Rock, Versteeg excavated a great deal of the midden deposits, but also used mechanized equipment to remove the overburden from 2,800 m sq of the village. The area had been in cultivation for centuries and the material that was removed had probably been plowed dozens if not hundreds of times. What was revealed by this procedure was a stratum of light colored soil that would have been below the village's ground surface when it was occupied. Careful scraping revealed hundreds of soil stains where posts or pits or other features had penetrated the original surface and cut through the light soil below. These features, large and small, were strewn across the site's surface, and could be connected in connect-the-dots fashion into any number of patterns of houses, large and small. Each feature was excavated by slicing it down the middle from top to bottom, revealing in profile its depth and shape. When postmolds of similar depth and profile were compared, the layout of individual houses became apparent (Figure FF) (Versteeg 1989, Versteeg and Schinkel 1992).

Versteeg and his colleagues found houses dating to two periods at Golden Rock. Three smaller houses, of about 7 m in diameter, date to the fifth century AD. The posts of these structures are shown on Figure FF as unfilled circles connected by dotted lines. The earlier houses are spaced about 20 m apart, and may have been occupied at the same time. Two of the houses from the later period, dating into the ninth century AD, are built in the same spots as two of the houses from the earlier period, and there is other archaeological evidence for continuity of occupation between the two building phases. The later houses, shown on Figure FF as filled circles connected with solid lines, are about 20 m in diameter and are more robustly built than the earlier structures. The pattern of large posts, sunk nearly two meters into the ground, shows strong similarities to historically known communal houses used in lowland South America. These houses are conical in shape and nearly 20 m tall. The major timbers were tied together with a framework of smaller poles and the whole exterior was thatched. The patterns of post molds also suggest that there were attached or free-standing walls away from the structures. These are interpreted as windbreaks or screens, creating areas on the leeward side sheltered from the prevailing east wind, and to some degree segregated from the rest of the settlement.

The northern part of the excavated area seems to be an area of ceremonial importance. There are several caches of artifacts and burials in this area, and a rectangular structure, also with an associated windbreak. Because of its association with the caches and burials, and again through comparison with ethnographic information from South America, the rectangular structure is seen as having a ceremonial function, perhaps as a men's house (Versteeg 1989, Versteeg and Schinkel 1992).

The large area excavation carried out at Golden Rock by Versteeg and the University of Leiden team revealed a settlement of large houses that was probably occupied continuously for five-hundred years. Even so it is only part of the site. St. Eustatius' airport either destroyed or lies atop other parts of the site. Josselin deJong's map from 1947 shows a cluster of five middens occupying an area that is about 500 m in diameter. Several other houses are almost certainly associated with these middens. As has been noted, Golden Rock was occupied relatively late in Saladoid times. Other sites in the Leeward Islands preceded it by many centuries. It was also not unusual for its size or architecture, but is important precisely because it is probably unexceptional and representative of a great number of late Saladoid sites in the Lesser Antilles and Puerto Rico.

The spatial organization of settlements, including the spatial relationship of central spaces, special buildings, and burial areas, yields information on many other aspects of Saladoid life. The relationships of buildings and other features helps inform us about indigenous social organization, conventions about residence and inheritance, and patterns of social inequality.

At Golden Rock, on St. Eustatius, most of the burials were located in the central plaza. This is the case at most Saladoid sites. On Puerto Rico, three well-excavated sites show the same pattern and yield evidence about its possible meaning, and about important changes that occurred at the end of the Saladoid period, about A.D. 600.

The site of Maisabel is on the north central coast of Puerto Rico, about 30 miles west of San Juan. It is on the island's rich coastal plain, in an area with many lagoons and wetlands. It is right on the coast on a low rise of less than 10 m, about one km from the mouth of the Río Cibuco. The site dates from early Saladoid times and continues into the Ostionoid period, or from a few centuries B.C. until around A.D. 1000. Like most sites containing the impressive white-on-red Saladoid ceramics and lapidary work, the site was known to local artifact collectors, and a few archaeologists, including Ovidio Dávila and Peter Roe, had carried out excavations in the late 1970s and 80s. Peter Siegel and the Centro de Investigaciones Indígenas de Puerto Rico led extensive excavations in the mid and late 1980s (Siegel 1989, 1992).

In the Maisabel site's original, early Saladoid configuration, it was a circular village with houses and mounded middens surrounding a ceremonial complex and cemetery (Siegel 1992: 188-191). Throughout the site's long occupation, the cemetery remained a constant feature at the center of the site. Siegel concluded that, "the overall village layout of Maisabel can be interpreted from the perspective of South Amerindian lowland cosmology. The central sacred space (cemetery ringed by the mounded middens) provided a focal point, around which daily and ritual life revolved, spatially and temporally. Perhaps this area also served as a plaza" (Siegel 1992:190). Burial practices changed though time at Maisabel, as later people chose to bury most of their dead inside their large communal houses. A few individuals were still buried in the central cemetery, however.

A similar shift in burial patterns -- from a centrally placed cemetery to a practice of burying the dead in surrounding houses - may be seen at the site of Tibes, along the Río Portugués near the modern city of Ponce on Puerto Rico's southern coastal plain. Tibes is remarkable for its early examples of ceremonial precincts lined with stone borders or pavements. These are the predecessors of the stone or slab-lined "ball courts" or bateyes of later prehistory and the period of European contact (Alegría 1983). The site is also very important because it was occupied continuously from at least A.D. 400-1200, and shows a transition from a typical Saladoid site to, in Santa Elena times (A.D. 900-1200), what is arguably the Caribbean's first "chiefdom" or ranked society (Curet and Oliver 1998:224, Veloz Maggiolo 1991). Antonio Curet, Lee Newsom and a team of researchers from the Field Museum of Natural History and Penn State University have been working at Tibes since the mid-1990s, building on the earlier work of Alvarado Zayas and González Colón. Their research has revealed a sequence of the development of the site that is similar to that found at Maisabel. Again there is a ring of middens with a central cleared space in the center. There were two clusters of Saladoid burials within the central space, one in what was to become the largest ceremonial precinct of the site, and another under another stone-demarcated feature or ball court (Curet and Rodríguez Gracia 1995). Once the plazas were constructed, the people at Tibes tended to bury their dead in association with houses or midden areas, as they had at Maisabel.

At Tibes and similar sites the reuse of the cemetery area as a ceremonial space was clearly not a coincidence. These areas continued to be important to the community, but in a transformed way. Curet and Oliver (1998:230-31) argue that this marks an important change in Caribbean social organization. The earlier practice of using central cemeteries, they argue, "were instituted, in part, to legitimize both the extended, kin-based corporate groups that controlled rights over the resources within their territory and the ideology to perpetuate them." They further argue that, "in post-Saladoid times the corporate group had been replaced by an elite group or household that was to assert control over and manage the critical, ideological, and symbolic resources that were once in the hands of communal institutions." This may well be the case, particularly at a site such as Tibes, with early evidence for the development of social complexity. At smaller sites, the replacement of Saladoid cemeteries with central plazas and the transition to burying people in domestic contexts may reflect a shift in the way people understood and represented their identities - a move from identifying with a single cohesive community to seeing oneself as a member of one or another distinct lineage or kinship groups.

These changes in burial patterns and community organization are related to a broad set of transformations beginning to take place by around A.D. 600. Beginning at this time, and continuing until the time of the European conquest of the Greater Antilles, new and more complex forms of social and political organization emerged in the Caribbean. These developments will be explored in the following chapters.

The Saladoid Phenomenon