The whole test unit consisted of very compact shell midden with limited sediment matrix. There was very little if any topsoil, and below this some distinctions can be made within the midden deposit. The top 20cm (Layer 1) below the current surface level contained a disturbed and highly fragmented shell deposit with a light grey brown silty matrix. Charcoal, coral and limestone inclusions were very common. The presence of many small rootlets combined with a lack of protective topsoil contributed to the disturbed nature of this uppermost deposit. Below this was a region of more compact shells to a depth of around 35cmbd (Layer 2) that were also highly fragmented within a coarse yellow brown silty clay matrix. Higher concentrations of charcoal were noted throughout the deposit and possible volcanic ash lenses. The transition between Layer 2 and 3 is not level, indicating that the surface of Layer 3 was undulating.
Layer 3 was a thick (35-60cmbd) deposit of loose large shell fragments with a friable yellow red clay matrix and large amounts of charcoal throughout. More pottery was located between 50-60cmbd, and large fragments of Tridacna were found at the base of this deposit. A small feature (Feature 1) was cut into Layer 3 from above and had more compact shell. It is possible that this was a disturbance from a tree root observed within this deposit. Layer 4 was another undulating layer below this with very compact large shell fragments, charcoal fragments, and a red clay matrix that extends to 80cmbd. At the base of this layer was another
120 possible feature (Feature 2) which was rounded with a diameter of 8cm, and cut into Layer 5. A higher density of shell was noted in the north-west corner of the test unit. The final and bottom layer (Layer 5) of midden was a reddish clay sediment with small crushed shell and charcoal at the base of the midden. The base of the layer slopes towards the north-west corner, and so this was recorded as Feature 3, however this may just reflect the undulating nature of the original ground surface. Importantly, there were also small fragments of dentate-stamped and plainware pottery found inter-mixed throughout the lower half of this assemblage, and near the base of the test unit at 85-90cmbd in the south-west corner. Sterile orange brown clay was encountered at 95cmbd.
Figure 6.2 Stratigraphic diagram of cultural deposits within Talasiu TP2 Leka (J17)
All three test pits at Leka revealed an interesting construction sequence for the langi or tomb, with features that were associated with the ditch for slotting the stone facing of the first tier of the tomb into place, and limestone cobbles providing support for these. Below these construction features were cultural deposits associated with occupation at the site prior to the development of monumental architecture at Lapaha. This occupation seemed to be centred near the location of TP4, where the deposit was deepest and contained the highest concentration of archaeological material.
The stratigraphy observed within TP2 represents a range of deposits associated with the construction of the langi, and the older cultural surface upon which the langi was built. The uppermost layer below the surface, Layer 1 was a coral gravel deposit with red brown clay matrix that was more concentrated near the surface and more diffuse towards the base. This
121 layer is most likely construction debris from the langi. Below this was deposit of reddish brown clay (Layer 2) with dispersed charcoal that capped an older cultural deposit (Layer 3). Layer 3 was a medium brown silty clay cultural deposit with many charcoal inclusions. Some red clay from the base of this layer was inter-mixed with the cultural material from 105-125cmbd. This layer sat upon the surface of the sterile red brown clay at 125cmbd.
In TP3 a shallow topsoil deposit (Layer 1) was visible below this surface from 5-15cm below the datum. Layers 2-5 were various coral gravel fill deposits within the ditch cut to allow the placement of the limestone tomb stones. At the base of this ditch was a surface of large limestone cobbles up to 20cm in diameter that were used to hold the beachrock slabs in place. Layer 2 also appeared to extend out to be a coral gravel path that surrounded the base of the langi. Layer 6 was a sterile red brown mixed clay in which was a tephra ash lens (Layer 7). It is believed that the large amount of sediment that created this layer was the result of ongoing tephra ash falls after 1000 BP. These deposits capped the mid-brown silty clay cultural layer (Layer 8) with small dispersed shell fragments and charcoal. This deposit was the only layer sampled for flotation and microbotanical analysis.
The stratigraphy of TP4 resembled that seen within TP2 and TP3, although the cultural deposit below 95cmbd contained a more dense concentration of shell, bone and charcoal. Similar to TP2, this cultural deposit was located below various mid-reddish brown clay deposits (Layers 1-3). The top of the cultural deposit was a mid-brown silty clay layer (Layer 4) from 95cmbd, with some dispersed shell, charcoal, fishbone, fire-cracked rocks and small limestone cobbles. Shell midden composed the lower half of the cultural deposit (Layers 5-6) from 125cmbd, and had internal stratigraphic differences. A mid-brown silty matrix was observed at the top (Layer 5), above a more red clay matrix at the base (Layer 6). The frequency of charcoal within the deposit increased between 135-140cmbd. By 140-145cmbd the cultural deposit began gradually transitioning into red clay with fewer shells and charcoal. The base of the excavation was sterile orange brown clay (Layer 7).
Figure 6.3 Stratigraphic diagram of cultural deposits within Leka TP2
Figure 6.4 Stratigraphic diagram of cultural deposits at Leka TP4
The initial test pit at TO-Nt-2 (TP2) revealed the edge of two deposits, one possibly related to the construction of the ‘Esi, and the other deriving from a pre-stonework phase of occupation at the site. Below the topsoil within this unit was a layer of red brown volcanic clay to a depth of
123 around 20-25cm below datum (bd) (Layer 1). Below this was a layer of beach-rock sand debris inter-mixed with some of this red brown clay to 25-30cmbd (Layer 2). A dark brown clay deposit with some humic material was beneath this to 30-35cmbd (Layer 3), and capped the cultural material. Layer 4 was a deposit of cultural material with dark brown clay matrix and charcoal fragments which was observed below these surfaces. The density of cultural material in this deposit was sparse. The base of this feature was sloping towards the south, and so the deepest point of the cut was at 70cmbd in the western-facing profile at the south-east corner. The shallowest point of the base of the deposit was at 53cmbd. A small pocket of shells followed the contour of the base of this deposit, about 3cm above the sterile clay. This deposit was possibly cut into another older occupation layer that had higher concentrations of shell, coral and charcoal and thus appeared darker from 70-93cmbd within the profile (Layer 5). The base of this deposit grades to less shell and more charcoal at the base. Both cultural deposits sat on top of a surface of sterile orange brown clay (Layer 6). This test pit was dug to approximately 99cmbd when the sterile clay was encountered.
The profile of stratigraphy within TP3 mostly resembled that of TP2. Red brown volcanic clay below the topsoil extended to a depth of 25cmbd (Layer 1). Below this was the same beach-rock sand inter-mixed with this clay (Layer 2), most likely related to the shaping of the limestone slab for the ‘esi. This layer of debris capped a cultural deposit (Layer 3) at 30- 35cmbd. The cultural deposit consisted of dense large shell and charcoal fragments with a dark silty loose clay matrix. There was some tree root disturbance within the deposit in the south- west corner, from 45-60cmbd. Near the base at 85cmbd was a white ashy lens (Layer 5) within a deposit of concentrated burnt crushed shell, charcoal, large whole shells, and many diodont or puffer fish spines (Layer 4) from 80cmbd to the base of the feature. Very high densities of mussel shell and a whole Lambis was located directly underneath Layer 5, grading to higher concentrations of Diodontidae spines in with medium orange brown clay from the base of the layer. The base of the deposit was encountered at 95cmbd, where the feature was cut into the orange brown sterile clay seen in TP2 (Layer 6).
Figure 6.5 Stratigraphic diagram of cultural deposits within Heketa TP3
AMS dating of cultural contexts
In order to identify samples for dating, the wet-sieved, flotation, and in-situ charcoal samples were sorted into those that could be identified as coconut, other endocarp or nuts, and other charcoal (wood and parenchyma). Coconut endocarp was identified based on the surface texture, which consists of a ‘cross-hatched’ type of pattern where clusters of three or four elongated parenchymatous cells are orientated at 45 degree angles to each other, as well as the thickness and density of the cross-section. Coconut is also very difficult to fracture, due to the density of the endocarp, but when it does fracture it does so relatively evenly and cleanly. The methods employed to extract macrobotanical remains from these sites were successful in producing small amounts of coconut from the wet-sieved material and in situ samples at all three sites. In order to develop a chronology for these sites, three samples of charred coconut endocarp were selected from spits (levels) located near the top, middle and bottom of one test- pit from each site for dating at the Waikato Radiocarbon Dating Laboratory (WRDL).
The results of AMS dating indicate that all three sites are from separate and relatively discrete time periods within Tongan prehistory. The midden at Talasiu (TO-Mu-2) can be dated to around 2750-2650 cal BP, and appears to have accumulated over a period of around 100 or less years. All three radiocarbon samples had narrow age ranges that overlapped very closely with one another, indicating a relatively short depositional period. This site therefore represents refuse from a settlement that is on the cusp of late-Lapita and Ancestral Polynesian Society (APS). These dates are supported by the ceramic assemblage from Talasiu, which is composed primarily of plain ware with a few dentate-stamped sherds in lower levels of the midden (Clark et al. In press; Golson 1957; Spenneman 1986).
Figure 6.6 Calibration of radiocarbon dates from Talasiu (TO-Mu-2), Leka (J17) and Heketa (TO-Nt-2)
Occupation underneath Langi Leka (J17) can be dated to around 1300-1000 cal BP, and so falls within the Formative Period (1550-750 BP) in Tongan prehistory (see Chapter 1), of which little is known (Burley 1998:365-8; Davidson 1979). The abandonment of pottery traditions around this time leaves an impoverished archaeological record, with a total of only 16 sites on Tongatapu representing these 800 years of occupation. The cultural deposit sampled for radiocarbon dating represents another relatively short period of deposition, with all three date ranges overlapping within a 100-200 year period.
Finally, the sampled Heketa (TO-Nt-2) material was from two deposits. One is around 800-600 cal BP and at the end of the Formative Period, while the other is a later deposit that is related to the development of early monumental architecture in Tonga during the Classic Tu’i Tonga chiefdom after 600BP (Burley 1998:368-79; Clark and Reepmeyer 2014). Two radiocarbon samples were from this lower and earlier deposit, and the remaining sample originates from the more recent feature and fill that caps this older pre-stonework architecture surface.
A combination of micro- and macrobotanical techniques was implemented to research and identify plant use at three stages in Tongan prehistory. Sites were selected based on the results of previous archaeological investigations on Tongatapu and sampled for botanical remains using current archaeobotanical methods, modified for tropical environments. A combination of flotation, wet-sieving and bulk stratigraphic sampling was used to extract all micro- and macrobotanical material from each test-pit. Laboratory processing of these botanical remains built on current techniques after experimentation revealed issues with contamination in the laboratory environment and also methods for the dispersal of microcharcoal concentrations within archaeological sediments, and the isolation of starch during heavy liquid separation. A revised protocol for starch extraction is published here for future use in microbotanical analysis. The extracted starch and parenchyma was analysed using both light microscopy and SEM, but was then classified to taxa using a combination of multivariate statistical analysis in the form of Discriminant Function Analysis, identification flowchart keys and visual checking of matches that is outlined in Chapter 8