THE MANZANILLA 1 (SAN-1) SITE, TRINIDAD.

a. at first a measurement grid was set u p that covered the total extensions of the hill where the site islocated in order to map it and be able to take height measurements for a contour map.b. secondly, this grid with a 20 m interval was used for the augertest program. By using an edelmanauger with a diameter of 10 cm, every 20 m an auger test was done. A total number of 98 auger testswere made. This in order to examine the subsurface; locating the exact extensions of the site, the locations and thickness of the middens, and location and extensions of vacant spaces. Of every auger testthe soil type texture and colour was recorded per stratigraphie layer. Furthermore the thickness (incm) and density (shell weight) of midden deposits that where encountered were also recorded.Artefacts that were found during auger testing were collected per stratigraphie layer or, when discovered in a midden context per midden deposit.c. Based on the results of this auger test program three test pits where made. Two test pits measuredl x l m and the third measured 1x2m. Their main purpose was to acquire artefacts and ecofacts fromtheir stratigraphie context, so all test pits where located in a midden deposit.The location of these test pits have been selected by the following criteria:1. the midden deposit had to exceed the maximal depth of the auger test (110 cm),2. a clear and interesting stratigraphy of superimposed layers,3. a high percentage of all the archaeological material categories,4.the location of the midden.All test pits were excavated in arbitrary, stratigraphie levels of 10 cm and each level was dry sieved ona 0.25-inch screen in order to collect ceramics, stone fragments, chert and large faunal remains.Because smaller faunal remains like small fish and reptile bones could not be collected using this meshwidth an additional 50x50 cm sample was taken from each level of the 1x2 m test unit. These sampleswere also used for research on shells, which were all counted using the M.N.I, method. The collectedmaterial from these samples was water sieved using a 1 / 6-inch screen. When a feature was encountered, a plan of this particular stratum was drawn accompanied with a profile-plan of this feature.Field ResultsThe total extension of the Manzanilla 1 site measured by the total area where midden material is visible at the surface is some 350 x 250m, which is approximately the complete extension of the small hillwhere the site is located (Figure 3). However, the area suitable for habitation is much smaller andirregular in shape.Three different morphological layers can be distinguished at the Manzanilla 1 site. The firstlayer is composed of humic clayish sand or sandish clay, predominantly loose in texture and has acolour ranging from black to very dark greyish brown. The thickness of this layer ranges from some 7cm up to 50 cm where it did not yield cultural materials, but reached often a 100 cm depth were it didcontain artefacts. The second is a layer of sandish clay or clay, sticky and compact in texture with acolour range from lOyr 4 / 4 dark yellowish brown, 2,5 yr 4 / 3 olive brown to 10 yr 7/6 yellow meaning that this layer contains not much humic material (using Munsell color chart). This layer continues176

u p to a 110 cm in depth in the greater part of the auger tests and transforms at these depths into a thirdlayer of very compact brownish red clay.In must of the augers tests midden material was found in the form of Chip Chip shells(Donax). The average thickness of this deposit was some 80 cm, but in some 13 tests the thicknessexceeded the 1.10 m. On the plateau four "empty" zones were localised which yielded no middenmaterial at all (Figure 4). Three of these areas were of somewhat the same size, each have an extensionof some 20 x 40 m and probably cover an area of 800? m. The fourth empty space covers almost thecomplete right part of the hill measuring approximately 40 x 60 m with an surface of some 1600? m.The thickest midden deposits are situated at the top of the slopes just over the edge of theplateau and closely encircle the empty areas on the plateau. Probably, those locations represent thevacant spaces where houses were located. By measuring the distances between the auger tests yielding thicker midden deposits and their closest living areas, it is possible to assign the midden locationsto one of the suggested vacant spaces, dividing the site into four habitation areas, vacant spaces andsurrounding middens. In an ideal situation each of those habitation areas could represent a singlecomponent phase of habitation of an Amerindian group that successively shifted their habitationspace four times on this small hill.Test pitsThe first test pit (Unit 1) measured 1x2 m (squares 1A and IB and was located somewhat in the centre of the larger plateau near auger test 47. A larger test pit on this location was chosen because it wasmore likely less disturbed by colluvial processes and the stratigraphy of this auger test showed that asterile layer of sandish clay intersected the midden deposit.A second test pit of one square meter (Unit 2C) was situated near auger test 28 on the slope ofthe western smaller plateau. This location was also chosen to obtain stratigraphie information aboutthe middens on the slopes. The location of the last test pit (Unit 3D) was determined by the situationof a midden deposit just on the edge of the western side of the larger plateau.In Unit 1 (Figure 5, Table 2) cultural material was found in a deposit of at least 160 cm. A pit or gullyintersected this midden deposit (XI) and it was filled with several thin lances of very different textureand composition, like sterile lances, some filled with ashy material and burned shell while others onlycontained small or fragmented chip chip shell. The lances mostly consisted of Donax striatus, withsome small quantities of other molluscs. One lance in unit 1 contains only small Donax striatus shells(Chip-Chip) of approximately two centimetres in length against a normal average length of 3.4 centimetres. This could mean that the Cedrosan people selected Donax on the site, after collecting them inthe surf.At a depth of 60 cm a burial (1F1) was found. This burial pit was dug from stratum II. Theexcavated part of this burial contained only the bones of a left and right foot. Both were not articulated and very well preserved while both the left and right tibia and fibula were missing. The right footwas found in a somewhat crossed position over the left foot.Three charcoal samples from Unit 1 were collected for radiocarbon dating.177

One sample was obtained from one of the lower levels, level 14 corresponding with the thick, homogeneous shell midden and revealed a date from 406 AD -556 AD. The last sample was obtained fromthe lens (stratum IV) in the depression/pit from the final habitation phase that contained only burntmaterial, mainly shell and came u p with a date of 688 AD- 892 AD. The sample from the middle levels proofed to be contaminated.These dates indicate a habitation span roughly between 406 AD and 892 AD correspondingwith the start of the Late Palo Seco up to the beginning of the Arauquinoid series.Unit 2 yielded cultural material (such as an Late Palo Seco anthropomorphic modelling in level 4(Figure 6)), in a midden deposit u p to a depth of 130cm represented by five strata of midden materialand a lower stratum of sterile soil. It consisted only of midden deposition, no features were found inthis unit. The midden existed of a more or less overall homogeneous deposition of mainly shells andfaunal remains. Numbers of artefact finds are significantly less than in the other Units. This can bepartly ascribed to colluvial processes.Unit 3 had a midden deposition with a depth of 160 cm. In Unit 3 two features were encounteredwhich were both dug from a depth of somewhere between 12 cm and 40 cm corresponding with stratum II. Both represent postholes. Unit 3 only yielded Late Palo Seco material in all levels, what couldbe an indication that the most south-western habitation area only represents a Palo Seco phase occupation.All Units yielded mainly ceramics from the Late Palo Seco complex dating 350-650 AD. Barrancoidinfluence is clear, flanged rims are dominant, the major part of the ceramics is covered with red slib.WOR is present is low numbers, while ZIC is still very well presented. Dominant temper types forthese Palo Seco ceramics is mixed quartz sand. In the past is suggested that Palo Seco ceramics areoften tempered with crushed pottery or grog. However, microscopic analysis showed that these redparticles are to soft and rounded to represent crushed pottery. Comparison with natural clay samplesshowed that similar reddish siltstone particles are also present in the natural clay. This proofed to bedeteriorated material from the soft sandstone formations in the surrounding Brigand Hill area (Dorst,2000).The upper levels in the midden deposit of both Unit 1 and 3 show ceramics with different characteristics. The numbers of decorated sherds are very low, and red slib and flanged rims are almost absent.Decorations that are found consist of simple incisions (Figure 7). Furthermore, other temper types arefound in ceramics in these levels, like crushed quartz, organic material and cauxi. All these characteristics indicate that these represent ceramics from the Arauquinoid Series. This series is found inTrinidad from 650 till 1200 where it is called Guayabitoid and originates from the Orinoco region andGuyana's, where it replaces the Barrancoid series. The recovered habitation area in the Palo Seco midden deposit represented by the upper levels in Units 1 and 3 most likely date from this period, whichis supported by the radiocarbon date, somewhere between 688-892 AD.178

Stones and LithicsWithin the stone and lithic assemblage of the Manzanilla 1 site four types were determined: lithicmaterial in the form of chert, sandstone debitage and artefacts, metamorphic rock artefacts and semiprecious stone. Chert material formed the largest group. Water worn chert pebbles were used whichcould be collected in the near surroundings. Pebbles with the same structure and colour were foundas washed-out material in the riverbed of the nearby (within 1 km range) L'Embrange River.Clear flakes with flake characteristics make up only 3.2% of the chert assemblage and nointentional retouch was found at all. Probably no clear technique was used to obtain chert flakes andsharp pieces were just selected after striking a chert nodule. Sandstone (quartzite) artefacts consist ofhammer stones and percussors. The sandstone that was used at the Manzanilla 1 settlement, both inthe form of water worn pebbles and unweathered rock is of a different type then the sandstone thatnaturally occurs in the near surroundings of the site. The more quartzite-like character and large quantity of schist suggest a more northern provenance (the Northern Range) of this sandstone. Water wornpebbles of both sandstone and chert are more abundant here and more accessible due to the existenceof riverbeds consisting of water worn rocks. One axe fragment was found made of a metamorphicgreenstone. The nearest metamorphic rock formations are found in the

This site had been classified in the past as an Early Ceramic site and is located on Trinidad's East Coast. This initial survey of the SAN-1 site had the following objectives: 1 To provide an accurate map of the site's extension, height and spatial division, which include the determination of the site's middens and vacant spaces.