EXCAVATION REPORT ON THE CERAMIC SITE OF GOLDEN GROVE, TOBAGO

EXCAVATION REPORT ON THE CERAMIC SITE OF GOLDEN GROVE, TOBAGO Peter O' B. Harris Golden Grove is a coastal site in southwest Tobago. It lies on a low coral plateau, some 10 m above sea level and 5 km from the sheltered west coast waters of Buccoo Reef and the Bon Accord Lagoon. A low swampy area, presumably an old arm of the lagoon, forms its immediate northern edge. At present the area is a grass paddock immediately behind and southeast of the Golden Grove estate house. Some years ago, the owner had to dig a pit on the slope above the swamp, and found great quantities of sherds and shells. Subsequently the perimeter of a large roughly finished canari or bowl was found among the debris; also a small Maritime Saladoid head at the eastern edge of the paddock. (Maritime Saladoid is used throughout this report rather than Insular Saladoid because the style includes mainland sites along the north coast of Venezuela. ) We dug five small test holes (25 by 25 by 25 cms) around the paddock: three on very slight eminences to the northwest, north and east; and two on flat ground to the south and in the centre. The holes to the north and south showed plenty of sherds and shells under 2 and 7 cms of topsoil respectively. The east hole produced a smaller quantity below 5 cms. The northwest one produced only shells below 15 cms. The one in the centre produced no shells and only two sherds down to 28 cms. There is some indication therefore that occupation may have been in a semicircle facing west towards the coast with a clear area in the centre. Sherds at the north test hole included two pieces of crudely modeled rim extension, and our sample excavation of two adjacent 1- by 1-m units (referred to below as units A and B) was made in this northern eminence, more or less along the line of the presumed site perimeter. However the south hole included two sherds with rim incision, which we now know to be characteristic of the upper level of the site. When time permits, I feel a further excavation should be made near the south test hole, at right angles to the line of the presumed perimeter. Stratigraphy After removing 8 cms of topsoil (level 1), four 25 cm levels were excavated down to 109 cms. Notes and wall sketches show the presence of four distinct layers, all more or less horizontal as follows: Topsoil (0-8 cms): occasional sherds; coincides with level 1. Shell II (8-21 cm): abundant shells and sherds; this coincides with level 2 of unit B, and is the majority material in level 2 of unit A (8-21 cms). Earth II (21-56 cms): dark grey soil, with intermittent shells and sherds; possibly some shallow {4 cm) lenses of pure earth occur in the top 10 cms; levels 3 (31-56 cms), and 2-2 (21-31 cms) of unit B contain this material; some, fortunately not much, is mixed in with level 2 of unit A. Shell I (56-85 cms): abundant large oysters with intermittent sherds; some 22 whole conchs (Strombus gigas) were found in excavation unit A; largely coincides with level 4 (56-81 cms). Earth I (85-109 cms): yellowish soil, with intermittent shells and sherds; largely coincides with level 5 (81-108 cms). These alternating layers suggest alternating periods of temporary and permanent settlement, or of small and large populations, or even shifts in the area of deposition. 145

146 TOBAGO Food Remains The shells and crab claws fall into four habitat areas, all occurring close to the site: Land (4 shell species); Lagoon (5 shell, 1 crab, 1 sea egg); Sandy Seashore (12 shell); and Reef (12 shell, 1 crab, 1 chiton). Lagoon is the main area exploited, however the emphasis changes slightly from layer..to layer. During Earth I the main collecting area is Lagoon/ Sandy Shore (oyster, venus/lucina). Reef and Land shells are also present in reasonable quantity. Six occasional species in this layer and in Shell I do not reoccur subsequently. During Shell I the main area is Lagoon (oyster, many of which are extremely large; and there is a tremendous increase in blue crab). Reef is now a definite secondary collecting area (tree oyster, mussel). Sandy Shore and Land continue to be reasonably represented. During Earth II the collecting area seems more restricted. The main area is Lagoon (oyster, venus); Reef is reasonably represented; but Sandy Shore and Land are hardly utilized. New occasionals include juvenile gastropods (Lagoon), and rocky shore species (Reef). During Shell II the emphasis returns to the combination Lagoon/Sandy Shore (venus, macoma, melongena and Strombus gigas juvenile s/lucina). A new lagoon species, melongena, appears here for the first time. Again there is a sharp increase in the consumption of blue crab. Reef and Land are reasonably represented. Occasionals still include rocky shore species. The number of species present in the shell levels 2 and 4 (26 and 21) is slightly more than in the soil levels 3 and 5 (19 and 17). This, together with the sharp increase in level 2 and 4 consumption of crab, and the increased number of juveniles in level 2, favours the interpretation that the shell layers reflect permanent settlement or increase in population. The prominence of melongena, after a total absence in earlier levels, is also significent. It may indicate an environmental change, or some new element associated with the second settlement or population. Two interesting details may be noted. Twenty-five specimens of Strombus gigas apparently form the edge of a pile, whose centre probably lies outside the excavation. They slope down from a depth of 62 cms (measured from the bottom of the shell) at the southwest corner of the excavation, to 91 cms at the perimeter, some 80 cms radially distant from the corner. All are juveniles or young adults, 10-24 cms long. None shows a hole in the spire to extract the animal, as is the practice today. Two Strombus raninus (levels 4 and 2-2) show cracking of the glaze and a hole on the inner shoulder, apparently from being placed at the edge of a fire. Fish and animal bones are present throughout occupation, fish consistently exceeding animals in volume by about 2.5 to 1. The fish jaws indicate some 13 species, including parrotfish, ray, gar, barracuda, and a reptile -iguana. The animal vertebra indicate a probable 6 species, possibly also including reptiles. Other bones confirm the presence of manicou, tatou, turtle, deer and probably agouti and rat. Large specimens are relatively few. These include several fish vertebra (25-29 cms) and jaws, 5 animal limb bones, 2 vertebra, and soem turtle bones. They are not confined to any one level. The two shell levels 2 and 4 have twice as much bone volume as the two earth levels 3 and 5, with level 2 slightly in the lead. Based on fish jaws and animal vertebra, the number of species present increases from level to level: 4 (level 5), 10 (level 4), 12 (level 3), 14 (level 2). This is similar to the increase in shells, and is apparently as much a result of length of settlement as of increased population demand in the shell layers. In level 2 there are three changes in the fish bones: the number of both parrot fish and ray jaws increases sharply; and shark vertebra, of which only one small example has occurred previously (level 4), are suddenly present in reasonable quantity. As with the shell melongena, this hints at a new element in the second settlement or population: either a slight

HARRIS 147 difference of food repertoire, or a greater proficiency in the marine environment. There is little evidence bearing on food preparation. The vertebra usually occur in series of 2 to 6 joints, suggesting breaking of the backbone into chunks of a reasonable size. A pair of broken knee joints from a medium sized animal (level 2) suggests the same approach. There are so few burnt bones as to give the feeling that these came about by accident, and that meat was cooked in a container. Finally there are 5 curious bundles of tiny bones, probably fins (levels 2, 3,4), one of which is inserted into the mouth of a Strombus raninus shell. The disposition of bones into a bundle shape suggests that the flesh was removed by sucking, which in turn suggests a liquid form of cooking such as stewing. In summary, food debris suggests a people who mainly exploited the coastal environment. The consistency of the shell and bone remains when comparing earth and shell levels, favours the interpretation of the shell layers as periods of increased population. An increase in crab consumption may well be an indicator of population increase. Some new elements of lifestyle, hinting at greater proficiency, seem to be associated with the Shell II population influx, Other Debris Throughout occupational levels small quantities of broken stone occur in large and small chunks and flakes. A minority of the stone maybe local to the site area; but the nearest likely source for the grey volcanic rock is on the coast 5 kms to the north, and for the diorites and igneous greenstones also on the coast 8 kms to the north. Artifacts include celts and workstones, to be described later. Also present, (particularly in levels 1, 2, 3) are small pebbles and chunks of black and white diorite, and small flakes of white quartzite. Artifacts from the former include beads (mainly present in level 4). Use for the latter is unknown, although it is a normal minority debris in all local preceramic and ceramic sites; possibly it is for grating manioc. Occasional items include reddish-brown ochre; 4 small fragments of hematite (levels 1, 2); 4 pieces of fine sandstone apparently used as a workstone (levels 1,2,3); 2 fragments of rock crystal, one of which shows signs of work (level 2); and 1 soft mica flake. Interestingly chert flakes and pebbles appear to be totally absent. Rods of staghorn coral occur throughout; only 4 out of 115 are at all smooth, on the other hand only 2 retain their pristine roughness. I think it is safe to conclude this is wear from use, possibly from scr«aping fish. Some finger coral is also present; and 6 flat pieces of starlet coral, 3 of which show definite wear or shaping. In addition pieces of dead conch shell (mostly gigas) are present throughout. The same parts occur as with the live shells: points, outer wall, and inner whorl. It is true that 2 conch artifacts were found, a celt (level 4) and a partially shaped bead (level 5); also 5 areas of outer wall (4 dead shell, 1 live) may have been roughed out into blanks, preparatory to making a scoop (?). However I have no satisfactory explanation for the presence of so much dead shell. Enough charcoal for dating was found in levels 2, 3, 4, 5. Pieces were scattered unevenly throughout each level, and were rather small (2-17 mms). In some instances they included a wood with long fibrous grain, presumably from the palm family. Pitch was found in levels 3,4, 5 in small pellets (4-15 mms) and flat nodules (14-37 mms). From the waterrolled appearance of the former, they may have been collected on a beach. Non-Ceramic Artifacts Celts number 8-10, one of which is made out of shell. The stone specimens are: 1 small celt, 2 butts, 1 blade, 1 part blade, 2 exterior flakes, 1 doubtful exterior flake, 1 partly unfinished replica celt or pendant. All are green except the doubtful flake, which is grey.

148 TOBAGO All are made from stone which is available in Tobago: 6 in volcanic stone outcropping along the nearby coasts, as mentioned above; and 3 in metamorphic rock available on the main inland ridge (20 kms away). Excluding the replica, 5 come from a medium-sized implement (8-13 cms) and 1 from a small tool (6 cms). Four have the thickness proportional to a celt, and 1 is thin enough to be defined a hatchet. Both blades show the wavy line associated with resharpening. Three butts are tapered and truncated, 1 shows signs of secondary use at its broken end as a hammer stone. The replica celt or pendant (level 1) appears to be made from a flake off an old celt. It is 7 cms long and is so thin (5 mms) as to appear non-functional. Although it has a partially ground blade, it is of similar size and shape to a pendant found in level 2 (6 1/2 cms x 3 mms), and could equally be considered in that category. The shell celt (level 4) seems to be made from the lip of a dead Strombus gigas. The blade is ground unifacially, and the butt is missing. The above implements were found throughout the excavation, with the greatest numbers coming from levels 2 and 4, the periods of probably increased population. Only 10 stones show any signs of use or work. A flat triangular piece of smooth sandstone, say 10 cms long including its missing apex, and 3 fragments (levels 1,2, 3) have edges ground through use or intention. The material would achieve a good finish as beads or celts. Of 4 elliptical stones, 4 1/2 - approx 8 1/2 cms long, the longest is broken (levels 3,4, 5). All show varying signs of use as a hammerstone. Because of their size and weight, they would be limited to small work such as beads; the probably very small size of the flakes removed may account for the apparent absence of these in the excavation. A possible result of another use occurs on one of these hammerstones and another elliptical stone (both level 2). Both stones are roughened all over as though from intense weathering, except for a small clean patch in the centre of one side. This may be associated with cooking. Finally a small stone fragment (level 5) shows signs of work on 3 surfaces; it could be a piece from the 22 mm thick rim of a shallow stone mortar, but there is no other evidence in support of this. It may be worth noting that every celt or workstone over a certain length (around 7 cms) was found broken. In the case of the triangular implement in level 1, one piece was found in each excavation unit; the third piece is still missing. The next category includes 11 beads, mainly of black and white diorite, and 2 pendants. Most are made from stone: 2 drilled beads, 4 undrilled, 1 partially shaped; 1 lamellar pendant, 1 unfinished probable pendant. Two are of shell: 1 partially shaped bead of conch, 1 Olivella shell with a hole in the apex. Two are of bone: 1 burnt bone bead, 1 shark vertebra with central hole which appears to be chipped rather than' drilled. The beads are normally cylindrical (10-25 mms long); however 1 is a small disc. Six of the 7 stone beads are of diorite, which as mentioned above is available on the nearby coasts; 1 is of pale green felspar probably from the Guyana shield. The lamellar pendant is of green volcanic stone available from the same area as the diorite; the unfinished pendant is of muscovite mica and possibly beryl, probably also from the Guyana shield; both are from level 2. As usual, these items are present in greatest number in levels 2 and 4, especially if we add the 2 pieces of rock crystal (level 2), and 5 probably worked pebbles of diorite (levels 1, 2). The two examples of overseas stone come from levels 2 and 3. Like the conch celt, the conch bead is early (level 5). A small waterworn piece of rose-pink shell (level 4) should probably also be mentioned here. Two further items are clearly worked, but their purpose is unknown. The first (level 3) is a narrow implement of bone, 4 mms thick and a present broken length of 5 cms. The complete implement would probably be 8-9 cms long. The unbroken end is rounded, and thickness tapers gently over the last. 5 cm as though from wear. It may be connected with small-scale weaving, as the wear pattern on the point is reminiscent of a weaving sword. The second implement (also level 3) is a 2. 5 cm length of strong marine worm shell, with 11 mm diameter and both ends ground smooth, but otherwise unfinished. It would make a

HARRIS 1.49 satisfactory inhaling tube for either nostril or mouth. Its lack of finish, let alone decoration, suggests a very ordinary level of activity. Possibly worked coral and shell forms a rather indefinite category. Two pieces of flat coral show wear on the flat surface, and 1 has been shaped to form a tapered edge, as though from a medium-sized platter. A raninus and cone have had their spires chipped off; the interior of the cone has also been removed. Coincidentally perhaps, the raninus contained one of the bundles of tiny parallel bones mentioned earlier. Two oysters have a hole near the hinge, 1 drilled (this could also be by a marine predator) and 1 chipped. The mother- of-pearl interior would make a pretty but fairly short-lived pendant. Several Lucina clam have part of their edge apparently worn, say from use as a scraper. In the absence of chert flakes, this may have been the tool for peeling manios. A piece of whelk has been neatly cut into an oval. The 5 pieces of dead and live conch shell mentioned before have been roughly trimmed as though preparatory to manufacture, and 1 lip from an adult dead shell has been separated from the main body. The common denominator of this group is an extreme lack of finish; suggesting casual selection, use for the job at hand only such as scraping fish scales, peeling manioc, etc., and more or less immediate discard. Three historic artifacts should also be put on record: a piece of glass; and another of cemented slate (both from level 1-topsoil); and an iron water pipe running diagonally a- cross excavation A at a depth of 9 cms. Fortunately this 18 mm 'pipe seems to have been installed with minimum soil disturbance. In summary, the well finished artifacts are made from stone, with a minority use of conch in levels 4 and 5. Beads and celts were probably made within the settlement. Coral and shell were also used, but with minimum work and finish, and probably therefore for utilitarian purposes. The greater part of the stone used is available in Tobago, mostly from the coasts 5-8 kms to the north. However 3 specimens would only be available inland 20 kms away, and 2 probably come from the Guyanas. The impression is given of a coastal people, with some trading contact inland and overseas, but otherwise rather local. Pottery The sherd contents of excavation units A and B are very similar in type and quantity. There is an appreciable difference only in levels 1 and 3, where A exceeds B in both cases. For the purpose of pottery analysis this was ignored and both units were combined. Sherds were initially divided into those with good finish and those with poor to average finish. The ratio remains more or less constant throughout occupation, with good finish sherds comprizing an average 30% of the total for each level. In the good finish group there is a noticeable decrease of decorated sherds from 50% (level 5) to 33% (level 2). In the poor/average group the percentage of decorated sherds remains more or less constant at 3%. Shell levels 2 and 4 differ from the earth levels 3 and 5 by containing generally larger sherds, and a greater quantity of probable vessels. In quantity of sherds the order is surprisingly: 535 (level 2), 485 level 5), 397 (level 3), 335 (level 4), 139 (level 1). In level 2B, where sherds were collected separately for the shell and earth layers, there is a tremendous difference in sherd density. In the shell 249 sherds were obtained from 14 cms of deposit or 17 sherds per cm of depth; in the earth 59 sherds were obtained from 11 cms, or 5 sherds per cm. The paste is characterized by a fine to very fine white temper, with polished finish and surface sparkle. Some examples do occur of a medium grit temper with the occasional large particle. Finishing lines are usually visible. Thickness varies 3-12 mms, and most sherds fall within 6-9 mms. Although vessels appear to have an even thickness, normal

150 TOBAGO variation on a single vessel is quite high, 5-9 mms. The repertoire includes 8-9 vessel shapes: Shallow Bowl, Effigy Bowl, Lobed Bowl, Bowl, Coui, Biomorphic Urn, Small Urn, Vase. Oval Bowl is the probable ninth shape; the sherds are covered for the time being under Shallow and Effigy bowls. Enough data is present with the Shallow Bowl to trace its development during occupation. With the others, only enough sherds were recovered to attempt a description.. Shallow Bowl - counting unlike rims only, 143 of these vessels can be enumerated. These are distributed among levels more or less according to shell density, and appear to be the constant major vessel in the good sherd group throughout occupancy. Sizes seem to average 3 small and 6 medium for every 1 large. Diagnostic is the shallow 25-35 degree slope of the vessel wall. The rim is usually plain, however some triangular thickening and downsloping flanges also occur. Bases include inset, ring and flaring ring. Decoration mainly takes the form of broadline incision round the interior rim. However a few examples of red paint, and exterior, flange and rim decoration are also present. Several changes in technology, and decorating technique, area and motif seem to occur during settlement. The few flanges occur decreasingly with occupation. The triangular thickening of the rim receives a sudden emphasis in level 2. The emphasis on bases moves from inset and low ring in levels 4 and 5, to ring in level 3, and flaring ring in levels 1 and 2. Exterior decoration is limited to levels 3, 4, 5; as is all but 1 example of red painting. Rim decoration occurs mainly in levels 1 and 2. The percentage of decorated sherds decreases noticeably during occupation, from 91 to 67%. Principal motif in levels 3, 4, 5 is an interior single incised line running parallel with the rim. This, together with an occasional exterior line, appears to be a minor theme on a similar vessel shape at Los Barrancos (ca 450-1000 AD). The pottery finish, temper and thickness are also typical of Los Barrancos. The two main motifs of level 2 are different as 2 3 loops are added to this single line, probably at each quarter of the vessel; in one instance there is a finger indent in the rim above the loops; in another, the loops form a continuous pattern round the inner rim. These loops are not reported from Venezuela; however they are similar to Aruka Incised Motif 4 on a similar vessel shape from the middle to late Mabaruma phase in northwest Guyana ^ (ca 850-1600 AD). The second main motif in level 2 is a series of short lines at right angles to each other incised on the flat top of the thickened triangular rim. In two instances, both level 2 motifs are combined. There is some evidence also to suggest that the vessels with short line rim incision are oval. One example of this motif occurs on a large fragment of oval bowl with flaring rim base found just below the surface at Guayaguayare, Trinidad, a generally Barrancoid site with a carbon-14 date for 0-25 cms of 690 AD ± loo 1 *. A shallow bowl with thickened triangular rim and parallel line incision on the rim seems to occur commonly at Los Barrancos, and boat shaped bowls are reported in the Orinoco delta from the time of Las Barrancas, but the only Venezuelan reference to short line rim incision is at Los Caros in the llanos near the Middle Orinoco. This is probably a mid-period III (750 AD) site with a Ronquin/Arauquin pottery style. This is, however, similar in concept to other examples of the interior rim incision of Aruka Incised Motif 4 from Guyana, mentioned above. Minor motifs include interior multilinear Barrancoid designs throughout occupation, 2 exterior spirals (level 4), 2 triangular rims with single line rim incision (level 4, 5), and 2 rims with an interior incised zoned red band (level 5). The normal flange decoration is 1-4 incised lines running parallel with the rim and/or red paint; 1 red example has incised oval (level 5). The above data seems to show a history of local development in the Barrancoid tradition, plus new cultural influences accompanying the suggested population influx of layer Shell II, which brought with it certain elements probably related to northwest Guyana.

HARRIS 151 Effigy Bowl - diagnostic is the presence of recognizably biomorphic lugs and features. The category includes small and inturned bowls with rim or wall lugs, shallow vases with exterior red incision and lugs, and in level 2 oval and round shallow bowls. Included within this definition are 21 vessels plus 7 other possible sherds. To me turtles appear to be the dominant theme, although others may think differently. Frogs and birds should be mentioned as less likely alternatives. Lugs can be tentatively classified as : 1 turtle, 3 bird/turtle, 1 frog/turtle, 1 bird, 1 gopher, 1 alligator, 2 bird/turtle tails, 2 probable turtle flippers, 1 "praying man", and 4 missing. There is also a sherd which appears to be part of a later effigy shallow vase. The lugs occur mostly in layers Shell II and upper Earth II. Two pieces of a distinctive inturned bowl, probably a turtle effigy, should be mentioned (levels 4, 5), the earlier one with zoned red paint. Also two oval bowls (level 2) belong, for me, with the series I attempted to define last year^ as starting with the out-flaring bowl with interior crosshatch and pedestal base of early Maritime Saladoid and continuing as the hammock-shaped vessel of the Taino in Santo Domingo. An interesting piece of data is supplied by the pieces of this bowl. The orientation of the sherds in the excavation shows clearly that the vessel was broken before discard, and by the proximity of the sherds, just before. It is also notable that all lugs, except those on a very small vessel (level 4), were broken off their vessel prior to discard. Two lugs show clear stylistic relationships: the "alligator" head is Barrancoid (level 4), the "praying man" is pure Los Barrancos. Lobed Bowl - includes 9 shallow bowls, all from lower layers (levels 2-2, 3, 4),with inturned rim, and a possible vertical vase, the rim of which extends or rises to form a lobe. Decoration is mainly exterior, consisting of incision, model/incision, and in one case red paint. One, the vertical lobe (level 4), depicts a Barrancoid face. One with exterior red paint and model/incised flipper has also been included under Effigy Bowl. Bowl - in all levels rim sherds occur with steep to slightly inturned walls. Nineteen can be enumerated, small to medium in size and mostly'undecorated. Coui - this steep-sided, shallow, round or oval bowl reminds one of a half calabash cut vertically; and for this reason I have tentatively called it Coui. It would normally appear to have a thickened rim and an all over exterior incised pattern of diagonals and spirals, apparently also covering the base area. Only 2 definite examples can be identified (levels 4, 5), but 17 rims and 15 exterior incised sherds (mainly from levels 3, 4, 5) probably also belong. Two rims have single-line rim incision. It is interesting that in levels 4 and 5 two Shallow Bowl rims have this decoration, and two others have exterior incised spirals, suggesting that Coui may have exercised a cultural influence outside its vessel shape in earlier times. Four curious exterior-incised inset bases from levels 3, 4, 5 may belong here despite the absence of a recognizable base postulated in the definition. Equally these may belong to a separate vessel category, which would then probably include some of the exterior incised sherds. One base has a red interior, and one appears to be oblong. Biomorphic Urn. - 8 very thick (17 mms) sherds occur with, in most cases, an exterior white slip and heavy exterior modeling /incision (including arms, hands/paws, and a spiral). These are spread over levels 2, 3, 4 and apparently belong to a large heavy urn-like vessel, whose body is either an applique model/incised figure or face. One sherd is of Arauquinoid type paste (level 3). The decoration style is similar to Akawabi Model/Incised Motif I from the Mabaruma phase in northwest Guyana (ca 500-1600 AD), however an urn shaped vessel is not reported there. Small Urn - 8 sherds occur in levels 3 and 5 which suggest a small thick-walled (9-12 mms) globular vessel with perhaps a tall flaring base. Six have an unfinished interior, with exterior red paint or incision; 1 is part of a 4 cm high flaring ring base, and the last has an urn shaped rim, with however a finished interior.

152 TOBAGO Vase - 2 sherds only (level 2) probably belong to this shape, a carinated vertical rim and an angled sherd both with exterior incision. Four other plain angled sherds (level 5) may or may not belong in this group. Eleven small sherds with exterior red incision and polished interior cannot be provisionally allocated. Ten possibly belong to Effigy Bowl (7 curved, 3 slightly curved); 1 may indicate a new vessel shape. This last has a curved profile meeting an incised inset base, and the quality of the red paint is particularly good, almost glazed. Besides the single example of Arauquinoid paste (Biomorphic Urn, level 3), 2 other sherds of a dark red, very fine sandy paste occur in the same level while some 74 rather small sherds of a very fine sandy paste occur (2 in level 4, 72 in level 5). Most of this last group have a poor apparently eroded--finish, surface sparkel, and a normal thickness of 4-8 mms. Vessel size appears to be mostly small. Shapes include 3 outsloping rims, 1 inturned rim, 4 possible flanges, a short spout, a wide thin D strap handle, an inset base, and a flat base. Other sherds are flat to slightly curved. Decoration includes interior and exterior fine line incision or scratching (4 examples), and rim pellet, wall brake, punctate pellet, broken off rim lug, and interior red paint (all 1 each). Three items were measured during excavation at depths of 98, 105, and 107 cms--the very start of occupation. This appears to be a separate culture, perhaps a degenerate form of Saladoid, which occurs as a minority ware very early at Golden Grove. In summary, the sherds with good finish suggest that three reasonably distinct cultures were present during the history of Golden Grove. The first, characterized by thin sandy sherds, apparently co-exists briefly as a minority element with the people of the second culture and also may predate them. The better class pottery of the second culture is characterized by slightly thicker sherds, fine white grit temper, and polished surface. The characteristic vessel is a Shallow Bowl, normally decorated with interior single line incision. Insufficient data suggest that the next most frequent vessel is the Coui, with exterior incised spirals and diagonals. This culture lasts through the Earth I, Shell I and Earth II layers. Other characteristic vessels are probably the Lobed Bowl, Turtle Effigy Inturned Bowl, Animal Effigy Shallow Vase, Biomorphic Urn and Small Urn. Minority decorations are exterior single line incision; flange red incision; rim incision, and pellets; interior multilinear Barrancoid incision; and zoned red paint. Again insufficient data suggest that rim incision and pellets may cease at the end of Earth I; while Lobed Bowl, Shallow Effigy Vase, and Biomorphic Urn may come in with the Shell I population influx. In general the stylistic affiliations are with a modified form of Los Barrancos; certain Maritime Saladoid elements are present in level 5; one Arauquin sherd is present in level 3; and in the Biomorphic Urn there may be an indirect relationship with Mabaruma. Shell II appears to mark the arrival of a new group of people. They are apparently from the same cultural area as the above, since three major elements remain the same; paste, dominant decorating technique of broad line incision, and dominant Shallow Bowl vessel shape. On the other hand, this same vessel undergoes a change of decoration motif involving loops, while a probably oval form appears with a new decoration motif of short line incision on a new part of the vessel, the thickened triangular rim. With regard to minor elements, the data is again insufficient. However the following are suggested: disappearance of Coui, Lobed Bowl, Effigy Inturned Bowl, Effigy Shallow Vase, red paint, and exterior incision; continuity of Biomorphic Urn, and the dominant lug theme of "Turtle"; and appearance of a possible new vessel, exterior incised Vase. General stylistic affiliations remain Barrancoid; and the new designs strongly suggest some form of relationship with Mabaruma, or less likely the Middle Orinoco. Some tentative general conclusions may also be drawn. Firstly, the steady decrease in decoration throughout occupation suggests a waning of the cultural influence which this

HARRIS 153 decoration may represent. Secondly, there is considerable difference in the numbers of each vessel shape recovered; i.e. 143 Shallow Bowls (average per level 36); 27 Effigy Bowls (av. 7), 19 Coui (av. 6), 19 Bowl (av. 5); 9 Lobed Bowl (av. 3), 6 Biomorphic Urn (av. 2), 6 Small Urn (av. 2), 1 Vase (av. 1). If this excavation is a representative sample, and if frequency of midden deposit reflects frequency of use, there seems to be, even allowing for errors, an interesting variation in the cycle of socio/cultural activity associated with each vessel shape. For example if Shallow Bowl were associated with a monthly acitivty; then Effigy, Coui and Bowl might be used annually; and the others once in a lifetime. Thirdly, the good condition of the sherds suggests that vessels were discarded on completion, not of their functional life, but of their cultural purpose. There is some evidence that they were broken just prior to discard, presumably also for r eligió/cultural reasons. Sherds with Poor to Average Finish The paste has a fine to medium evenly ground white grit temper with occasional large pieces of grit. Normal finishing shows marks suggesting a wet smoothing instrument; however a few sherds in Shell II show exterior scratching or brush marks. Interiors are normally better finished than exteriors. Thickness varies considerably, 2-18 mms; however normal thickness is 5-12 mms. A large amount of variation occurs on the same vessel. The distinction between different vessel shapes seems less definite than with the well finished sherds, and only a few relatively large sherds can be assigned with any degree of certainty. The following appear to be normally present: Neckless Olla, Bowl, Sloping sided Bowl, Necked Vessels, and Griddles. Unusuals include 2 probable Potrests, and 2 Angular Inturned Bowls. Neckless Olla - seems to be a large/medium vessel occurring throughout occupation. Rims are squared, rounded or tapered in cross section. One large vessel fragment (level 2) has a crude D handle decorated with a knob. The probably associated base has a very eroded interior. Two sherds (level 2) have a scratched or brushed exterior, another (level 5) has an exterior incised single line. Bowl - is a large/medium/small vessel which also occurring throughout occupation. In level 2 the rim is sufficiently inturned to make a normal sized rim sherd indistinguishable from Neckless Olla. In the lower levels it cannot be distinguished from the next category, Sloping Bowl. One very large fragment (level 3) has a crude D handle attached to a rectangular lobe extension to the rim, the associated base has a blackened interior. Another large fragment (level 4) is unusual for its good finish and out-curling rim. It also has a knobbed D handle, but without the lobe. Two inturned sherds (level 3) are decorated, 1 with an exterior tapered thickening to the rim, and 1 with an exterior incised line. Sloping Bowl - is a large/medium vessel with an open out-sloping wall, apparently limited to the lower levels. Six varying size fragments (levels 3, 4) have D handles: 3 with rectangular lobe and knob, 1 with lobe only, 1 with knob only, and 1 plain. A sherd in level 5 has an exterior incised line. This rectangular lobe, without a handle however, is a distinctive feature of Koriabo Plain vessel type 1, a medium to large open bowl from northwest Guyana^. The Koriabo phase is felt to be contemporary with late Mabaruma (1200-1600 AD). Necked Vessels - is a rather loose category, probably comprizing 3 different vessel shapes. The first has a broad outsloping mouth above a body of unknown shape. It is large/ medium and occurs throughout occupation. One sherd (level 2) has a brushed exterior. Another (level 4) has a broken off area, suggesting there was a lobe extension to the rim. The second shape includes 3 vessels (levels 4,5) with a suggested globular body and a broad vertical neck. One sherd has a mouth-shaped lug, apparently projecting horizontally from the neck; and 1 a thickened triangular rim with an exterior incised line. The third shape probably

154 TOBAGO represents bowls or ollas with a strengthened or everted rim. Vessels are large/medium and occur throughout occupation. Potrest - 2 fragments occur in level 4. They have massively thickened rims and an exterior incised line. One narrow ring and 1 square thickened rim or base (both level 5) may also belong to this group. Potrests are reported from Los Barrancos, and crude potrests from Mabaruma and Koriabo^» '. Angular Inturned Bowl - only 1 rim sherd from a small vessel (level 2) definitely belongs to this category. One or 2 angular sherds occur in each level, but they may not be related. Another sherd (level 3) with simple model/incision on a very sharp angle belongs to a different, clearly Barrancoid, shape of vessel. Griddles - occur decreasingly throughout occupation. Thickness varies (8-26 mms), and in many cases the paste is extremely friable. Only 4 edges were found: 2 show the imprint of a probable grass binding round the perimeter (level 5); the other 2 have a tapered edge (levels 2, 3), however their unusually well finished underneath suggests they may belong to another category such as a platter. Underneaths are normally unfinished, showing not an imprint of leaves, but probably of a well brushed piece of ground. In levels 3, 4, 5 the ground seems to have been smoothed or flattened, and there are faint brush marks of the dust. In level 2 a less prepared piece of ground seems to have been used, perhaps further from the living area, as the underneaths are very uneven and somewhat crinkled. Handles and Bases - in most cases cannot be related to a particular vessel shape. Large, medium and small crude D handles occur throughout occupation. From a total of 40, 9 have variously shaped simple knobs, 2 have a sort of heel or thumbrest, 1 with a better than usually finished paste (level 2) has a small modeled Barrancoid adorno, and 5 (levels 3, 4) have a lobe extension to the rim. Nine canbe allocated to avessel shape: 1 Neckless Olla (level 2), 2 Bowl and 6 Sloping Bowl (levels 3, 4). It is interesting to note that D handles are absent from the well finished vessels. Throughout occupation the majority of bases are round (37); however flat (9), built-up (9) and inset (6 - limited to levels 3, 4, 5) also occur. One broken off circular flat base (level 4) possibly belongs to an urn shaped vessel. Many of these bases show sighs of use. Thirty-five have a blackened interior, often with the black penetrating deep into the clay, and in some cases with a thin black deposit. Sixteen, most from level 2, have an eroded interior. If we exclude griddles from the utilitarian vessel totals, the major categories are undoubtedly Bowl/Sloping Bowl (av. 48% per level) and Neckless Olla (av. 38%, except in level 5). The bases with blackened and/or eroded interiors presumably belong to these two vessels, and indicate the possibility of a usage difference between them; like for instance the cooking of solids and liquids. Only 2 bases can be linked with a vessel shape; one blackened interior belongs to a large Bowl (level 3), and one eroded interior probably belongs to a Neckless Olla (level 2). Mention should also be made of a fragment of possible figurine base (level 5). Lugs - is a rather loose category. It includes 7 adornos (2 anthropomorphic, 4 zoomorphic and 1 unknown), 1 "mouth", 1 foot, 2 small circular knobs (1 plain, 1 punctate), 1 parallel-sided sherd with 2 plain knobs, 1 punctate pellet on sharply angled wall, 1 pellet on curved wall, 2 "ears", and 2 pieces of applique (1 straight, 1 S-shaped). To these should be added the 9 handles with knobs. The 2 anthropomorphic heads (level 2) are on stems, one of which is attached vertically to the side of a small bowl. Both have bulbous eyes, similar on the one hand to late Akawabi Model /Incised Motif 4 from the Mabaruma phase (ca 1000-1600 AD), and on the other to many presumably late Lesser Antilles adornos and the face on the handles of Greater Antilles stone pestles (ca 1200 AD). The foot (level 2) has 3 incised lines on its sole. A zoomorphic head (level 2) is on the handle mentioned in a preceding para

HARRIS 155 graph, and another (level 2-2) is attached to the rim and wall of a small bowl. An S-shaped applique (level 3) would seem to belong to a globular bodied vessel, possibly the second type of Necked Vessel. Surprisingly the distribution is not in accordance with shell density; the greatest numbers come from Shell II and late Earth II. Other pastes include: 3 sherds (level 5) of a soft paste, blue-grey one side and pink the other and 3 others (level 2) also of a soft paste, 2 white and pink, and 1 with white interior/ginger exterior. Several minor differences of lifeway, technology, and style are indicated during occupation. Most occur between Earth II and Shell II, paralleling the change of cultural influence suggested by the well finished sherds. Some elements--sloping Bowl, lobed handles; and possibly, potrests, the globular bodied Necked Vessel, occasional exterior single line incision, and rim thickening--disappear. Others change: eroded bases increase from 11% to 45%, suggesting possibly an increase in liquid cooking; griddles decrease from 26% to 16%, suggesting possibly alternative food to cassava breads; place of griddle manufacture seems to be different; and handles possibly change from Bowl to Olla. New elements are: occasional exterior brushing or scratching; and possibly, the Angular Inturned Bowl. It is interesting to note a report of 1 exterior brushed sherd from Koriabo, and this is of course a diagnostic of the proposed Suazey style of the Lesser Antilles (1020-1400 AD). These scratch marks appear to have been made by a stiff brush of dried grass. There is some evidence to suggest that they were part of the manufacturing process in the earlier levels, but that at that time they were subsequently removed by a final wet smearing finish. In any event very few exterior scratched sherds are present in the Shell II layer. Insufficient data suggest that minor differences may also occur between Earth I and Shell I. Griddles decrease from 87% to 26%; the Neckless Olla increases from 10% to 38%, apparently replacing Wide-mouthed Necked Vessel (28%); lobed handles and potrests may come in. These possible changes should be checked during future excavations into this culture and period. Two unexpected data are supplied by the lugs. Unlike the decrease in decoration during settlement indicated by the well finished sherds, lugs increase towards end of occupation. In addition, as many occur on the utilitarian vessels as on the well finished. In a culture so poor in lugs compared with its Maritime Saladoid and even Los Barrancos predecessors, this seems to be significant. The knobs on the crude D handles, for instance, seem a degenerate memory of the small but sophisticated adornos on the large Barrancoid vessels apparently used for burial - Is this evidence of a tremendous wane in religio-cultural influence? Or of conquered women potters clinging to a former tradition? In summary, the poor/average sherds confirm the population phases indicated by the well finished sherds. Relationships are suggested with Guyana, degenerate Barrancoid, and, towards the end, the Antilles. It is interesting to note that certain trends in the well finished pottery are reflected in the poor/average sherds. In both, the minor occurrence of exterior incision is restricted to levels 3, 4, 5; the globular bodied minority vessels of Small Urn (well finished) and of the second type of Necked Vessel (poor/average) are restricted to the same levels; and lugs occur mostly in level 2 and immediately below. Conclusions Apart from the minority presence of sandy sherds in the Earth I layer, Golden Grove seems to have been inhabited throughout by people from a single culture which may be considered to form part of the Barrancoid tradition. The people mainly exploited the coastal environment, as is witnessed by their choice of location, major food debris, and sources of stone. The last suggest common contact with the Tobago coasts 5-8 kms to the north, and occasional contact, perhaps trade, with inland Tobago and the Guyanas. There is some sug-

156 TOBAGO gestion that the settlement formed a semicircular pattern facing west, or towards the coast. The steady increase in number of shell, fish and animal species in the settlements food debris, together with a lack of difference between the pottery repertoires of the earth and shell layers, support the feeling of permanent settlement. I am therefore assuming that the alternating layers of earth and shell represent alternating periods of small and large population. Both the population increases, especially the second, bring with them new elements; but the overall framework of culture and lifestyle remains the same. The pottery suggests population overspills from settlements in the same cultural influence area; and in the case of the second, from a settlement considerably nearer the cultural centre. These population increases appear to be accompanied by: a sharp increase in the consumption of crab; in the earlier, an extension of the major shell collecting area from Lagoon to include Reef; and in the later, an increased consumption of juvenile shells, and an apparently superior exploitation of the coastal environment. The last is suggested by a sharp increase in the number of parrot fish and ray jaws. In all, pottery and stratigraphy suggest 4 phases of population. The first, represented by thin sandy sherds and probably the use of conch for celts and beads, is very distinct from the people described in the above paragraphs. There is some suggestion of degenerate Saladoid, but far more data are required to attempt a definition. The 2 deepest recorded sherds belong to this culture, and it would seem to exist as a minority element throughout the Earth I layer. The second population phase has a pottery which belongs to the Barrancoid tradition described above, with Maritime Saladoid elements of flanges, pellets, incised spirals, and red paint. This is the majority component in the Earth I layer. The third phase covers the Shell I layer and probably Earth II. The population influx apparently brings with it a new utilitarian vessel shape and associated decoration - the sloping sided bowl with rectangular lobe extension for the handle. The fourth phase covers layer Shell II. A large influx of population is characterized by an increased ability to exploit the coastal environment and an apparent decrease in griddles. It is tempting to ask if this could be related to climatic change. There is evidence of a possibly world wide "Little Optimum" lasting a few centuries around 1000-1200 AD, which involved wind circulation displacement, higher temperatures and different rainfall patterns. Design motifs on the major cultural vessel change to include loops at the quarters and a probable oval bowl with short line rim incision. While an example of the latter is known from southeast Trinidad, both appear to represent further influence from northwest Guyana. The very minor presence of exterior brushing, rim indentation, and anthropomorphic adornos with bulbous eyes seems to look toward the late populations of the Lesser Antilles, where the first two are major elements. Throughout there is a decrease in decoration; however the few adornos present appear to increase towards end of occupation. Also they occur as frequently on utilitarian as on ceremonial vessels. Some are clearly Barrancoid, others apparently Maritime Saladoid, and 2 from vessels with poor/average finish from the most recent population phase appear to represent influence from northwest Guyana. There is also throughout occupation a steady development in the bases of cultural vessels. These move from inset and low ring, to ring, and finally flaring ring. In summary Golden Grove appears to reflect the slow changes of strength and influence of a much wider cultural continuum. For our purposes, this is exemplified initially by unreported material from the Tobago site of Lovers Retreat' [Lower level. Ed. Note], This is mainly Barrancoid with strong elements of Maritime Saladoid, and appears to belong to a period of cultural high. A date of 650 AD ± 120 has been obtained for this site, but the depth of the sample is unreported. Around this time there is some evidence for a Caribbean-wide weakening of religio-

HARRIS 157 cultural influence. If cultural highs and lows tend to be more or less contemporary in the Caribbean area, then the cultural incline evident in phase 4 should be in the general time period of Boca Chica in the Greater Antilles. These rather theoretical considerations, together with other data mentioned in this report, lead me to feel that the approximate dates for Golden Grove settlement are 700-1200 AD. It will be interesting to see how the Carbon- 14 results compare. Footnotes (1) Stones discussed with Dr. John Tomblin, University of the West Indies, Trinidad. (2) Cruxent, J. and I. Rouse - Archeological Chronology of Venezuela (1961). (3) Meggers, B^ and C. Evans - Archeological Investigations in British Guiana (I960). (4) Harris, P. - Summary of Trinidad Archeology (1973). (5) Harris, P. - Origin of Lesser Antilles Crosshatch (1974). (6) Bullen, A. K. and R. P v and E. Branford - The Giraudy site, St. Lucia (1971). (7) Olsen, F. - On the Trail of the Arawaks (1974). (8) Lamb, H. H. - The early medieval warm epoch and its sequel (1965). (9) Cambridge, T. - unreported collection, Port of Spain, Trinidad. (10) Rouse, I. - Dating of Caribbean Cultures (1963).