CHAPTER 7WO THE ROCKSHELTER EXCAVATIONS by Patrick V. Kirch A LTHOUGH TE PARTICuLAR STRATEGIES of sampling design varied for the three rockshelters, excavation procedures were the same in all cases. All sites were first mapped at a scale of 1:50 with plane table and telescopic alidade. Surface artifacts were plotted and collected at this time. A metric grid was then established over the site surface, with the grid oriented to the long axis of the shelter. Trowels and brushes were used exclusively in excavation. Natural stratigraphic boundaries served as the basis for excavation control, although within homogenous strata finer control was exercized through the use of 5 cm arbitrary levels. In this manner we were able to combine the advantages of both natural and arbitrary excavation methods. Features (such as hearths and ovens) were excavated as individual units, and their contents bagged separately. Depth control was maintained both by measurement below surface and by taking below-datum instrument elevations. All excavated sediment was passed through nested sieves with mesh sizes of 1/4 and 1/8 inch. All artifacts, macroscopic floral remains (including charcoal fragments), faunal materials (both vertebrate and invertebrate), and any evident non-local lithic materials (including waterworn pebbles) were collected from these sieves for laboratory analysis. In addition, the contents of some features were bagged whole for later flotation of plant materials in the laboratory (see chapter 6). Following the completion of excavation, several continuous column samples were taken from cleaned profiles in Sites D6-36 and -60 (because of stratigraphic disturbances, no continuous columns were removed from Site D6-58). These column samples provided materials for detailed laboratory studies of sediment, archaeobotanical remains including charcoal, and of non-marine molluscs, as reported in chapters 3, 5, 6, and 7. During the excavation of each unit, standard grid record sheets were maintained showing horizontal plans (1:10 scale) for each 5-cm increment or natural stratigraphic boundary; all artifacts recovered in situ were plotted on these sheets. Features were recorded both on the grid sheets, and with a separate feature recording form. After excavation, cleaned profiles were drawn at a scale of 1:10 and described in terms of color, texture, structure, boundary, and other physical characteristics. Photographic records maintained during excavation include 35 mm color slides and black-and- white negatives, and selected 120 black-and-white roll film negatives. KUOLULO ROCKSHELTER (Site OA-D6-60) Kuolulo rockshelter is situated on the south, or Wai'anae, side of the Anahulu Valley, at an elevation of 97 m above sea level. The shelter lies within the indigenous Hawaiian land unit, or 'ili, of Ke'eke'e and, more specifically, within the kuleana or ancestral estate of Kuolulo (Lamd Commission Award 4308-1). We have given the name of- this mid-nineteenth century Hawaiian cultivator and principal resident of 'Ili Ke'eke'e to Site D6-60. Anah1du Figure 2.1. View of Site D6-60, Kuolulo Shelter, from the east. The shelter consists of a large "notch" in the precipitous cliff at the southern end of one of the valley's alluvial benches (fig. 2.1). The shelter was formed at a time when the Anahulu Stream was higher than at present (probably during the late Pleistocene), and hydraulically eroded a less resistent stratum of aa lava clinker (about 2-3 m thick) sandwiched between denser strata of flow basalt. As seen in figure 2.1, the dense basalt forms a jagged overhang rising 20-30 m above the shelter floor. Above the shelter, the valley wall is very steep, ascending rapidly in a series of cliffs and scree slopes to the edge of the tablelands at 209 m above sea level. Kuolulo shelter sits about 5 m higher than the level of the adjacent alluvial bench, and is separated from the latter by a gentle talus slope. The alluvial bench would have provided excellent terrain for indigenous gardening, although unlike other alluvial terraces in the valley, it was not modified for pondfield irrigation of taro. The present vegetation in the vicinity of Kuolulo rockshelter is almost wholly exotic. Several banyan trees (Ficus sp.) have implanted themselves in the overhanging cliff, and their aerial roots cascade down the dripline (fig. 2.1). The talus slope in front of the shelter is shaded by Java plum trees (Eugenia jambos), with a ground cover of honohono grass. The steep valley wall above the shelter is dominated by Christmas Berry (Schinus terebinthifolius), koa haole (Leucaena leucocephala), and lantana (Lantana camara). A feral remnant of indigenous Hawaiian cultivation existing on the alluvial bench near the shelter is clumps of the ti plant (Cordylinefruticosa), the large roots of which provided a source of sugar for the prehistoric Hawaiians. Kuolulo shelter (fig. 2.2) has a main floor area under the drip line of about 25 m long and 12 m deep, with a usable dry floor area of approximately 300 m2. The overhanging ceiling of dense basalt is restrictive only at the very rear of the shelter, where it declines in height to about 2 m; elsewhere it is 4 or more meters high. Although protected from the elements, the shelter is airy and well lighted. When the valley was more open, shelter occupants would have had a clear view out over the adjacent alluvial bench to the Anahulu Stream, and across to the north side of the valley. In 1982 when we commenced work at the site, there were no obvious signs of recent disturbance, and various surface features and artifacts appeared to be in situ. One feature, a small crypt (fig. 2.3), had been constructed against the shelter wall east of the main occupation floor. This crypt consists of a semi-circular wall of basalt cobbles (.3-.8 m high) held in place with a reddish clay "mortar." The crypt had been opened at some time in the past, and was empty. Similar features elsewhere in the valley are known to contain human 10 The Rockshelter Excavations burials and we presume that such was the original function of this structure as well. The eastern boundary of the main floor area of the shelter is defined by a low wall of basalt cobbles and boulders (fig. 2.4), two to three courses high (.4-.7 m), which runs from the back of the shelter, across the floor, and down the talus slope to the alluvial bench. This wall is not substantial enough to have kept pigs or other animals out of the shelter. It may have served as a symbolic boundary of the secular activity space of the shelter, as distinct from the burial crypt area to the east (for further discussion of the semiotic value of east-west spatial relationships on Hawaiian sites, see Weisler and Kirch 1985). As seen in the plan map (fig. 2.2) and in figure 2.5, the main shelter floor displayed a number of surface structures and features. A large roof-fall block of dense basalt at the front of the shelter (behind A. Saxe in the photo, fig. 2.5) had been modified with a grinding surface on its flat top. The circular grinding area was probably used for sharpening the bevels of stone adzes or other artifacts. Immediately south of the grinding- stone boulder is a small, earth-filled terrace, defined by a cobble retaining wall on the west, and by the base of a second retaining wall on the south. Referred to henceforth as the "outer terrace," this area measures about 4 by 2 m. (The Nikon survey level in fig. 2.5 is set up on this outer terrace.) The innermost part of the shelter was leveled by the construction of another terrace (the "inner terrace"), defined on the north by a 6.5 m long retaining wall that incorporates a large roof-fall boulder. Much of our excavation effort was directed toward these two terraces and their sub-floor deposits. The eastern portion of the main occupation area is not marked by terraces or other structures. While the terrace retaining walls were constructed largely of angular basalt cobbles derived from the shelter overhang, there were also seven large waterworn, stream cobbles on the shelter floor (see fig. 2.2). These had been purposely carried to the site from the Anahulu Stream, possibly for use as seats or as anvils. Lying on the dirt floor in the eastern part of the shelter were four massive wooden timbers, sawn and notched (fig. 2.6), of unknown function. These timbers certainly date to the historic period, and may have been associated with the construction by Waialua Plantation, in the early twentieth century, of a large irrigation syphon immedi- ately west of the rockshelter. A variety of smaller artifacts were also found scattered over the site surface in 1982, and these were plotted and collected prior to exca- vation (fig. 2.2). Indigenous artifacts included basalt and volcanic glass flakes, a grooved pebble sinker, a drilled dog tooth pendant, a basalt adze preform, and a polished adz flake. Historic period artifacts included bottle glass fragments, ceramic fragments, and a gun flint. Excavation Strategy Following mapping and establishment of the metric grid, excavation commenced in the P-trench (fig. 2.7), which bissected the retaining wall of the inner terrace, thus providing a stratigraphic section through the terrace fill, and north toward the drip line. We then expanded the excavation of the inner terrace area (fig. 2.8), which proved to have a relatively shallow depth of cultural deposit (ranging from 5-25 cm). Some 8 m2 of the inner terrace were excavated, revealing several small hearth features. Attention was then directed to the outer terrace, which was sampled by excavating the M-trench through the retaining wall of the inner terrace, and out across the outer terrace to abut the large rock-fall boulder. The M-trench thus provided a second stratigraphic profile, through both the inner and outer terraces. While most of our efforts were directed at the westem half of the main occupation area, with its two structural terraces, we also tested the eastem half. Unit 125 was excavated in the approximate mid-point of the shelter, toward the rear wall. This unit revealed a very shallow, and nearly sterile, deposit overlying bedrock. A second unit, D20, was then excavated in the center of the flat occupation floor in the eastern half of the shelter. This unit revealed a deep cultural deposit, extending to 80 cm below surface. While it would have been desirable to expand our sample of these deep deposits, we were restricted by project time constraints. Future archaeological work at Kuolulo Shelter should be directed at the eastem half of the site, to expose what is probably a sizeable area of well-stratifled cultural material dating to as old as the fifteenth centry A.D. A total of 18 m2 was excavated at Kuolulo Shelter, with an approximate total volume of 4.5 m3 of cultural deposit. Physical Stratigraphy The physical stratigraphy of the shelter deposits differed substantially between the westem and eastern portions of the main occupation area. The western area, with the two structural terraces, was characterized by relatively shallow, homogeneous cultural deposits overlying bedrock or pre-occupation clays. The stratigraphy in this area was best exposed along the M- and P-trench profiles, described in detail below. The eastern area of the shelter contains a deep deposit, exposed only in the D20 test unit, also described below. Stratigraphic sections were drawn and described in the field, at which time column samples of sediment were also taken for laboratory analysis. The results of the laboratory studies are presented in chapter 3. 11 Anahulu C 0LJ H- ~cn 7 0 QO 12 0 (D (0 0 0 I- E 0 0 U? co w I, 11 0 - 0 0* -C', o-i I 0 I H wr (D * Z (9 \ Co 0% Cu C.) 4)A UB co o co 0 (0 0 Cu Cl Al U ..- I I I I -I I I I I J. II f I 1. 1. I Il I . 1. 11 1, -1 :~ t I t I 1'. . Gc) The Rockshelter Excavations E N' 0 l 0 1.(9 I (0 0 0 13 QA U 'V 0 d. co Co ._4 4. m cs 0 I" co Co 0 4- 0 EZi U 4. Anahulu Figure 2.3. Stone crypt located east of the main occupation area at Site D6-60; view from the west. Figure 2.4. Low cobble wall defming the eastern boundary of the main occupation area at Site D6-60. 14 The Rockshelter Excavations Figure 2.5. View from the east of the main occupation floor of Site D6-60; the Nikon level is situated on the outer terrace. Figure 2.6. Sawn and notched wooden timbers on the floor of Site D6-60. 15 Anahulu Figure 2.7. Commencement of excavation in the inner terrace area of Site D6-60. The cobble retaining wall of the inner terrace is clearly visible on the left. Figure 2.8. The inner terrace of Site D6-60 after completion of excavations. 16 The Rockshelter Excavations SITE D6-60 M TRENCH MAUKA FACE O lm !- I * * - a SITE D6-60 M TRENCH MAKAI FACE M23 I M24 M25 I E E Figure 2.9. Stratigraphic sections of the M-Trench in Site D6-60. 17 M22 Anahulu The M-Trench Profile. The east and west faces of the M-trench exposed parallel sections bissecting the low cobble retaining wall of the inner terrace, as well as the cultural deposits of both the inner and outer terraces. These sections are shown in figure 2.9, and a view of the west face of the M-trench is provided in figure 2.10. The deposits were comprised of three strata: Layer I, a dark reddish brown (5 YR 3/2-3) deposit 4-6 cm-thick, representing a post-occupational buildup of sediment (derived from roof decomposition and aeolian materials) and organic litter. The boundary between Layer I and II was abrupt, grading over a 1 cm zone. Layer II, a dark brown to very dark grayish brown (7.5 YR 3/2 to 10 YR 3/2) cultural deposit, largely homogeneous from top to bottom. The deposit had a very fine granular structure, largely unaggregated, with a soft consistence when dry. Fine gray ash appeared to be a significant constituent of the deposit, as well as larger charcoal fragments, with individual pieces of charcoal ranging up to 1-1.5 cm in diameter. The presence of ash and charcoal was doubtless related to the large earth oven (Feature 10) occupying the center of the outer terrace (see below). The basal boundary of Layer II was abrupt, grading into Layer III over no more than 1 cm. Layer III, a dark red (2.5 YR 3/6), culturally-sterile deposit, with very fine granular structure. At the Layer 11/I1 interface we noted a high frequency of large (woody) charcoal fragments, many of which appeared to have been pressed down into the top of Layer III. This phenomenon was suggestive of an initial clearance of woody vegetation at the site by burning, immediately prior to human occupation. The P-Trench Profile. The east and west faces of the P-trench are shown in figure 2.11. Throughout the length of this trench, the cultural deposit was identical to Layer II in the M-trench, described above. In the upper terrace (south of the retaining wall), the cultural deposit rested directly on the bedrock of the shelter floor. North of the retaining wall, the bedrock itself is capped by a deposit of sterile, dark red gravelly mud, identical to Layer Ill described in the M-trench. At the interface of the cultural deposit and this sterile gravelly mud, we noted the presence of numerous large charcoal flecks (as in the M-trench), and also several examples of the large, endemic arboreal snail Achatinella (A.) cf. decora. As described fiuther in chapter 5, the presence of this species strongly suggests that the shelter was covered in an endemic forest association immediately prior to human occupation, and that the flirt occupants of the site cleared this vegetation by burning. Figure 2.10. View of the west face of the M-Trench in Site D6-60, after the completion of excavations. The trench bissected the cobble retaining wall of the inner terrace. 18 The RockshelterEzaatos19 a- a- U- wuI z 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~a D CL~~~~~~~~ C) 0~~~~~~~~~~~~~~~~~~~~~ (0<< F- o rA~~~~~~~~~~~~~~~~~~~~~ 0 %~~~~~~~~~~~0 i~~~~~~~~~~~~~~~~~ C14~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~C IL~~~~~~~~~~~~~~~ 0-j %~~~~~~~~~~~~~~~~~~~~~~~~~( 0~~~~~~~~~~~~ 0 E 0 0 C) Anahadu I BEDROCK 0 50cm Figure 2.12. Stratigraphic section of the west face of Unit D20, Site D6-60. The Unit D20 Profile. The west face of Unit D20 is shown in figure 2.12, which also indicates the position of Column C, utilized for geoarchaeological analysis, and for extraction of landsnail, charcoal, and seed samples (see chapters 3, 5, 6, and 7). Three depositional units are represented in the profile: Layer I, a dark-reddish brown (5 YR 3/24) deposit which accumulated after abandonment of the site; Layer II, a thick dark brown to very dark grayish brown (7.5 YR 3/ 2 to 10 YR 3/2) cultural deposit with substantial quantities of charcoal, relatively homogeneous throughout the section; and Layer III, a basal deposit of dark reddish brown (5 YR 3/34) sedint with some angular gravel, presumably derived from roof fall. Layer II[ rested directly upon the, bedrock floor of the shelter. A detailed sedimentological analysis of this stratigraphic section is provided in chapter 3 (see table 3.3). Cultural Content and Features The excavations at D6-60 yielded a total of 1,636 indigenous Hawaiian artifacts, of which 1,407 (or 86%) are small flakes and cores of volcanic glass. Of partcular interest in this assemblage are five small fishhooks and tree line sinkers, 10 hammerstones, and 52 small cone-shell beads. The upper cultural deposits also yielded artifacts of Euro-American manufacture. These portable artifacts are described in detail in chapter 8. The site also produced an array of vertebrate and invertebrate faunal materials, analyzed in chapter 4. Eleven discrete feames were recognized during the excavation. These are listed and described in table 2.1. Ten of the feames consisted of small hearths, characterized by shallow, basin-shaped concentrations of white ash and charcoal. Significantly, none of these features was stone-lined, or showed other signs of permanent use. Rather, each hearth feature appeared to represent a single burning event The most interesting of these hearths was Fe 3, situated in Unit Q26, which contained the partial cranium, and a mandible, of a Polynesian dog (Canisfamiliaris), as well as burned endocarp fragments of candlenut (Aleurites moluccana) and marine shell (fig. 2.13). Of particular note is the large earth-oven feaure (Fe 10) situated in the middle of the outer terrace. The plan and stratigraphic section of this feature are shown in figure 2.14, while figures 2.15 and 2.16 show the oven during excavation, before and after sectioning. This feature is a typical indigenous Hawaiian earth oven, and was filled with angular, fire-altered volcanic oven stones. A rather unusual feature (Fe 12) was exposed in the M-trench after the removal of the Fe 10 oven (fig. 2.17). TIis was an embament or ridge of hard-packed |- s ~ ~ ~~ __ 20 I1 The Rockshelter Excavations TABLE 2.1 FEATURES IN KUOLULO ROCKSHELTER FEATURE UNIT DEPTH'BELOW TYPE DESCRIPION NO. SURFACE (cm) 1 P26-27 2-8 Hearth Small hearth, measuring 19 by 14 cm in plan view, maximum depth 6 cm. Roughly oval in outline. Consisted of white ash with some included charcoal, and contained burned bone and marine shell. 2 P25 10-17 Hearth Small hearth, measuring 15 by 30 cm, maximum depth 7 cm. Consisted of white ash with some charcoal flecking. 3 Q26 7-10 Hearth Small, oval hearth, measuring 20 by 30 cm, maximum depth 3 cm. The hearth lay just above the contact between the cultural deposit and bedrock. Consisted of a lens-like concentration of "greasy" charcoal and ash, containing burned Aleurites endocarp, a dog cranium and mandible, burned shell, volcanic glass flake, and coral abrader. 4 D20 23-29 Hearth Small hearth only partially exposed in D20 excavation. Consisted of charcoal and ash overlying oxidized soil. 5 D20 21-29 Hearth Small circular hearth, approximately 20 cm in diameter, maximum depth 8 cm. Consisted of charcoal and ash with burned bone, burned Aleurites endocarp, and burned marine shell. 6 (no. not assigned) 7 D20 26-35 Hearth Small concentration of charcoal and ash, measuring 20 by 25 cm, with maximum depth of 9 cm. 8 D20 0-30 Hearth Only a portion of this feature was exposed in D20; most apparently lies in adjacent Unit D19. 9 N25-24 8-25 Hearth Small, circular hearth about 20 cm in diameter, maximum depth 17 cm. Consisted of white ash and charcoal. 10 M22-23 4-39 Earth Oven Large, circular earth oven, approximately 1 m in diameter, maximum depth 35 cm. The oven was filled with vesicular, angular, fire-altered volcanic stones. The base of the oven was marked by a 3-4 cm thick deposit of white ash, and charcoal in pockets. 1 1 M22 9-17 Hearth Basin-shaped hearth about 40 cm in diameter, maximum depth 8 cm. Consisted of white ash with some charcoal flecks. About 8 fire-altered volcanic stones were also included in the feature. 12 M22 32-44 Clay Ridge Hard-packed ridge or embankment of clay running pependicularly across trench near base of cultural deposit 21 Anahulu Figure 2.13. Feature 3 hearth in Unit Q26, on the inner terrace of Site D6-60, containing a dog mandible. clay, running perpendicularly across the trench at a depth of 32-44 cm. The ridge appeared to have been an artificial construction, but its full extent could not be determined, and its function remains unknown. 'he horizontal exposure of 16 m2 in the western par ,f she rockshelter, behind the inner terrace and across the outer terrace, provides a picture of the distribution of major classes of cultural material. Figure 2.18 depicts the frequencies of vertebrate faunal remains, marine molluscan fauna, charcoal, and the endocarp remains of candlenut (Aleurites moluccana) by excavation unit. Shell and bone midden are most densely concentrated in the fill of the inner and outer terraces. Charcoal fragments are heavily concentrated in Units M22-M23, reflecting the presence of the large earth oven in the outer terrace (Fe 10). Endocarp frag- ments of candlenut (Aleurites) are heavily concentrated on the inner terrace, suggesting that candlenut lamps were most frequently used in that part of the site. The Depositional and Occupation Sequence Three radiocarbon age determinations from Kuolulo shelter are discussed in full at the end of this chapter. Briefly, the oldest determination, of 460 ? 70 B.P., from 65-70 cm below surface in Unit D20, establishes the initial human use of the shelter in the fifteenth century AD. The earliest use of the site appears to have been confined to the eastern portion of the main floor area, and was thus sampled only in the D20 unit. A charcoal sample from the contact between the cultural deposit and sterile base in Unit M23 yielded a conventional 14C( age of 190 ? 60 B.P. This result was closely matched by a charcoal sample from the same stratigraphic situation in Unit Q26, which yielded a conventional age of 180? 80 B.P. Thus the construction of the inner and outer terraces (which stratigraphically overlie these two charcoal samples), must date to sometime in the seventeenth to eighteenth centuries. The site continued to be utilized into the early historic period, based on the presence of exotic, Euro- American artifacts in the upper 15 cm of deposit. KE'EKE'E NUI ROCKSHELTER (Site OA-D6-58) Site D6-58 lies at an elevation of 85 m above sea level, on the northern, or Ko'olau, side of the Anahulu Stream. As with D6-60, this shelter is situated within the traditional Hawaiian land section, or 'ili, of Ke'eke'e. We therefore dubbed the site Ke'eke'e Nui rockshelter, or "big" Ke'eke'e (in contrast to the smaller shelter, D6-36, which we named Ke'eke'e Iki rockshelter, or "little" Ke'eke'e). Site D6-58 is located just outside the mapped 22 The Rockshelter Excavations SITE D6-60 SQ. M 23 KO'OLAU FACE A 0 A' OVEN STONES 50cm I I I I I Figure 2.14. Plan and stratigraphic section of Feature 10, a large earth oven in the outer terrace of Site D6-60. A- I I 23 Anahiu Figure 2.15. Feature 10 earth oven in Site D6-60 during excavation. Figure 2.16. Feature 10 earth oven in Site D6-60 after sectioning. - - 24 The Rockshelter Excavations Figure 2.17. The M-Trench in Site D6-60 after completion of excavations; view to the north. The ridge of hard-packed clay (Feature 12) can be seen at the far end of the trench, near the base of the large rockfall boulder. boundary of Land Commission Award 4308-2, claimed by the native cultivator Kuolulo, in 1848. As with Kuolulo shelter, the Ke'eke'e Nui site was formed when the stream level was higher than at present and differentially eroded a stratum of less-resistent aa clinker alternating between beds of dense flow basalt. Thus the roof of the shelter is formed by a massive cliff of dense basalt. Unlike Kuolulo rockshelter, however, Ke'eke'e Nui rockshelter lies very close to the stream, where the latter forms a sharp bend just downstream against a sheer rock cliff. The occupation floor of Ke'eke'e Nui shelter is only 2.5-5 m above the normal stream level, and the lowest parts of the shelter floor are thus subject to occasional flooding when the stream level rises (see fig. 2.19). (The highest flood level that had occured within the year or two prior to excavation 25 Anahadu SHELL MIDDEN (gm) M N 0 P Q R BONE MIDDEN (gm) M N 0 P Q R 1.0 21.5 23.2 48.2 67.7 98.745.9 26.6 ............ ............ ..... f f........ ........ 54.6I 133.4 67.7 0.6 10.3 0-25 |...l 26-5o 51-75 76-100 101-125 126-150 CHARCOAL (gm) M N 0 P Q R ALEURITES ENDOCARP (gm) M N 0 P Q R 26 25 24 23 22 0-25 m26-50 51-75 76-100 101-200 201-400 401-500 867.3 596.6 37.3 .-.-........... s o F _ _ _..................... 263.3 406.6 914.51560.1 617228.1 ... , ... _ _ _ '''~~~~~~~~~~.''''''''.. 642_ .. . .. . . ............. ............ 21.2..... .............. ......_ 45.3..... 317.6 125.7 ............ 200.6 0-100 100-300 301-500 501-700 701-900 901-1100 Frequencies of shell and bone midden, charcoal, and candlenut (Aleurites) endocarp in the westem area of was marked by a "strandline" of vegetation debris, 2.85 m above the normal streamflow level.) Upstream from the site (east) is an alluvial terrace or bench, which in the early nineteenth centy was intensively cultivated in taro (Colocasia esculenta) and other crops (see Sahlins and Kirch, in prep.). The small rockshelter, Site D6-36, is situated in the same cliff as Ke'eke'e Nui rockshelter, a short distance to the east (up-valley). The vegetation surrounding Ke'eke'e Nui rockshelter is almost wholly exotic, dominated by Java plum (Eugeniajambos), mango (Mangifera indica), and banyan (Ficus sp.), the roots of the latter forming a cascadse down the rockshelter dripline. A large cycad palm (Cycas sp.) grows just outside the shelter at the downstream end. Ke'eke'e Nui rockshelter has a floor area 22 m long, and ranging from 3 to 6.5 m deep, with an approximate usable floor area under the dripline of 88 m2. Movement in the rear portion of the shelter is restricted due to a low roof, which drops to 1 m or less in height However, the outer 2-3 m under the dripline has a higher ceiling. The large banyan and mango trees that shade the shelter make for a rather gloomy aspect today, but when the shelter was clear of vegetation, it would have commanded a view southward across the valley, toward the Wai'anae valley wall. The proximity to the 26 25 24 23 22 26 25 24 23 22 3F2 37.3 ................... . ... .... . . 12.7 6.0 ............. ............. 3.0...... ............. .................. ::::::::::::::::::: 84.0 323.1 412.9 26 25 24 23 22 Figure 2.18. Site D6-60. .......... ............ ............ ------------ ...... - - ......... ............ .... ...... ............. ............ ........ ... ....I ........ ........... ............ .. ............ 28.1 19.6 54.5 38.3 35.9 13.0 ............ ............. ............... ............ ------------- I & I I 26 40.0 .... :. .: .... .... . ... .. :31.5 0.7 24.2 0.9 The Rock/shelter Excavadons LLI cc LLI /i / i/ o /0 / -J -J E co LO 0 C6 0 C,, 0- E 0 0 C 0 U .-1 c c; Co 0_ ON c co o .: U c) 0 -4 0 c) c 0 s a3 0 co 0 0 4- Co 0 00 CIO Z - w z 0 U, z 0 a: LL w: Hl 27 Anahidu Figure 2.20. Ken Shun standing in the disturbed stone crypt in the eastern portion of Site D6-58. stream, a permanent source of fresh water, would certainly have been an asset. Unfortunately, Ke'eke'e Nui rockshelter had been disturbed in several respects prior to our excavation there. Earlier in this century, a cattle fence had been erected across the valley, with one end running directly into the middle of the shelter. Two large wooden fence posts were still standing in the shelter floor. More recently, the shelter had been the temporary home of a "hippie" recluse, who left behind a scatter of debris, including newspapers (dating to 1971), a rotting mattress, candles, a rusted lawn chair, and other paraphenalia. This most recent occupant of the site had shoveled out a depression in the rear of the shelter, presumably to increase the ceiling height; in the process, the integrity of some of the inner shelter deposits was damaged. The eastern portion of the shelter was furtier disturbed, where a rock crypt had been opened by means of a crude excavation (fig. 2.20), possibly by the "hippie" occupant. (We know from conversations with pig hunters and hikers resident in the Anahulu area that this particular "hippie," who lived in the valley for a year or so, opened and disturbed a number of burial caves.) Despite such disturbance, much of the shelter was in surprisingly good condition when we commenced work, with several surface features relating to the prehistoric occupation of the site. Near the western end of the floor area, just inside the dripline, was an L- shaped terrace facing of angular basalt cobbles, 2-3 courses high. On top of this terrace the cultural deposit had been exposed by water dripping from the ceiling, and several artifacts were collected in this area. The crypt in the eastern portion of the site, disturbed probably in 1971, has already been mentioned. At the extreme eastern end of the shelter is a large rockfall boulder with a grinding surface. South of this is a second, smaller boulder, the entire upper face of which (.5 x 1 m) had been smoothed by grinding, probably of basalt adzes. Outside of the dripline are two additional basalt cobble terrace facings, the highest standing 90 cm. These retain a roughly terraced area south of the crypt and grinding slabs. In addition to these features, a number of large waterworn, stream cobbles were distributed over the site surface. These had presumably been carried to the site for use as seats or anvils. Excavation Strategy As Site D6-58 was the last rockshelter to be investigated during the 1982 Anahulu Project, only a few days could be devoted to its sampling. Our experience with Kuolulo shelter had shown that different portions of an occupation floor might have 28 The Rockshelter Excavations substantially different depositional sequences. Therefore, at Ke!eke!e Nui we decided to space our test units so as to sample several areas, while carefully avoiding parts of the site that had been recently distrbed. We completed the excavation of three 1-m2 units, designated D9, K10, and 011 on our grid. While these provided a reasonable sample of the intemal stratigraphy of the site, they of course provide no details on continuous horizontal activity areas. As will be described below, Ke'eke'e rockshelter provides a well-stratified sequence spanning some 500-600 years, with much potential for elucidating the occupation history of the middle Anahulu Valley. We estimate that at least 80 m2 of undisturbed cultural deposits are present (both inside and outside the dripline) in the site. Thus, Ke'eke'e rockshelter clearly merits a major excavation effort, especially one focused on the horizontal exposure of activity surfaces. A total of 3 m2 was excavated, with a combined volume of approximately 2.0 m3 of cultural deposit. Physical Stratigraphy Site D6-58 presents a rather complex stratigraphic situation, especially owing to the number of features and disturbances in the excavated test units. The stratigraphy of each unit is described below (see chapter 3 for laboratory analysis of sediments). Test Unit 011. The stratigraphy of 0 11 was difficult to interpret due to the presence of an earth oven, several hearths, and a number of rodent burrows that obscured relationships among the main deposits. It was possible, however, to recognize two main cultural deposits, especially in the north face of the unit. The upper cultural deposit, from about 5-35 cm below surface, was dark grayish brown (10 YR 4/2), with a gravelly texture, and very ashy. The lower cultural deposit, from about 35-65 cm, was reddish brown (5 YR 5/34), more silty in texture, and contained much less ash and charcoal flecking than the upper cultural deposit. The stratigraphic boundary between the upper and lower cultural deposits was gradational over about 1 cm, but clearly distinguishable. The Feature 9 ash lens covered the base of the unit, and directly overlay the sterile alluvial sand. Test Unit K10. The stratigraphic situation in K10 was highly complex, as illlustrated in the profiles of the north and south faces (fig. 2.21). The individual stratigraphic components visible in the north face were: (1) upper cultual deposit, dark grayish brown (10 YR 4/2), ashy, with charcoal flecking throughout, and considerable small waterwon stream gravel (1-7 mm size class): (2) white ash (5 YR 8/1) with charcoal flecking; (3) grayish midden deposit very similar to 1; (4) dtin lens of white ash (5 YR 8/1); (5) large pit with brown fil (5 YR 5/2); (6) lens of white ash (5 YR 8/1); (7) dusky red (10 YR 3/34), oxidized alluvium; and (8) basal deposit of reddish brown (5 YR 4/4) sandy alluvium. In the south face, the visible components in the section were: (1) upper cultural deposit, dark grayish brown with significant quantities of charcoal flecking; (2) gray-brown pit fill with mottled flecks of white ash, charcoal, and shell midden; (3) lower cultural deposit, reddish brown (5 YR 5/3-4), with substantially less ash and charcoal than 1; (4) compact white ash with charcoal; (5) reddish-brown silt loam with charcoal; (6) gray ash with charcoal; (7) compact, white ash with charcoal; (8) base of the Feature 5 earth oven, gray ash with charcoal concentrations; (9) thin lens of dark red, oxidized alluvium underlying the Feature 5 oven; and (10) sterile sandy alluvium. The sequence of strata 4-9 are all associated with the Feature 5 earth oven. Despite the complexities of these sections, it was clear that the two main cultural deposits recognized in Unit 011 were also present in KlO.The upper cultural deposit in both units was characterized by a more gravelly texture, and by the presence of substantial quantities of ash and charcoal. The lower cultural deposit contained much less ash and charcoal, as reflected in its reddish brown, rather than dark grayish brown color. In the field, we noted that these differences probably mark a change in the behavioral activities taking place in the rockshelter when these sediments were being deposited. Clearly, there must have been an increase in the number of fires (both for hearths and earth ovens) lit in the shelter during the later period of occupation. The possibility was also raised that the upper cultural deposit marks an increased intensity of occupation on the site, perhaps a shift from intermittent to pennanent utilization. Test Unit D9. The stratigraphy of this test unit differed substantially from that of Units 011 and K1O, due primarily to its lower elevation (the surface of D9 is 25 cm below the surface of K10, and 80 cm below the surface of 011 ), and susceptibility to flooding from the nearby Anahulu Stream. The stratigraphic section is shown in figure 2.22, and the identified strata are as follows: (1) alluvial sand, brown (7.5 YR 5/4), representing a flood deposit; (2) alluvial silt, brown, also a flood deposit; (3) crumbly deposit of reddish brown (5 YR 4/4) sand, containing some cultural material; (4) alluvial silt, sterile (7.5 YR 5/4), representing a flood event; (5) tiin cultural deposit identical to 3; (6) dark reddish brown (5 YR 3/4) midden deposit; (7) dark brown (7.5 YR 3/2) midden deposit; (8) basal, sterile, reddish brown (5 YR 4/4) alluvial sand. It is not possible, without further excavation, to 29 Anahulu SITE D6-58 SQ. K10 WAIANAE FACE 0 50cm I I I I I i SITE D6-58 SQ.K1O KO'OLAU FACE -0 - 50cm 9 50cm Figure 2.21. Stratigraphic sections of Unit K10, Site D6-58. v a 30 TheRockshelter Excavation 31 E U 0 Lo 0 'I 0~~~~~~~~~~~~~~~0 . .W,.......... .... . . . . ... /...... ID '4-. 1' ., .''..'s. . .. _A ..... XX... ............ ~ ~ ~ ~ ~ ~ ~ ~ ~ Anahula correlate either the lower midden deposits, or the thin upper lenses containing cultural materials, with the sections in K1O or 011. Cultural Content and Features Site D6-58 yielded a collection of 478 portable artifacts of indigenous Hawaiian manufacture, of which 336 consist of flakes and cores of volcanic glass. Of particular note are four fishhooks, a basalt adz, a gaming stone, and a tattooing needle. The presence of these and other artifact classes in the upper cultural deposit provide additional evidence of a shift from temporary to permanent occupation of this shelter. These artifacts are analyzed in full in chapter 8. The site also yielded artifacts of Euro-American manufacture. Vertebrate and invertebrate faunal assemblages are analyzed in chapter 4. Ten features were recognized during excavation, as enumerated in table 2.2. Unit K10 contained five features, one of which is a large earth oven visible in the stratigraphic section (fig. 2.21). Unit 011 had four features, including an earth oven. Unit D9 also contained an earth oven, although this was only partly exposed in the excavated area. All other features were small, scoop-type hearths or ash lenses. The Depositional and Occupation Sequence The base of the depositional sequence in all excavation units consists of a stream-deposited alluvial sand, evidently derived from a flood-stage of the Anahulu Streanm No charcoal flecking was noted in this alluvial deposit, which might be taken as evidence that there was no significant burning in the Anahulu drainage basin prior to the occupation of Ke'eke'e Nui shelter. (Alluvial deposits in the valley which have been dated to the period of human occupation generally contain significant quantities of macroscopic charcoal flecking, derived from burning within the catchment area.) In Unit 011, the alluvial sand is directly overlain by Feature 9, a 6 cm thick ash deposit extending beyond the confines of the 1-m2 unit. This deposit probably represents burning of vegetation debris on the floor of the shelter immediately prior to human utilization, and subsequent accumulation of midden deposits. Feature 9 yielded a convendonal 14C age of 600 ? 110 B.P., calibrated to A.D. 1280-1430, with all intercepts on the calibration curve falling in the mid-fourteenth century A.D. This radiocarbon age thus provides an estimate of the initial human use of the site in approximately the fourteenth centy. Both units 0 11 and K10 revealed differentated lower and upper occupation deposits, as described above. The lower deposit is characterized by a sediment with substantially less charcoal and ash than the upper cultural deposit The densities of both vertebrate and invertebrate faunal materials are also significantly higher in the upper cultural deposit (see chapter 4, tables 4A and 4.7). Although our excavated sample of the total occupation area in this site is limited, the stratigraphy suggests a major change in the use and behavioral activities at the site, between earlier and later phases. Tentatively, I suggest that this change was from an earlier pattem of infrequent, temporary occupations in the rockshelter, to a later pattern of more frequent, intensive, and possibly permanent residence. This occupation sequence will be discussed in further detail in chapter 9. The large earth oven in Unit K10, Feature 5, yielded a conventional 14C age of 280 ? B.P., calibrated to AD. 1516-1599 and 1617-1659 at one standard deviation (the two ranges reflect fluctuations in the calibration curve at this time interval). Although the base of Feature 5 lies on the sterile alluvium, the oven was cut from near the base of the upper cultural deposit. Thus the change in site occupation reflected in the interface between lower and upper cultural deposits can be dated to approximately the sixteenth-seventeenth centuries. A third 14C age was obtained from an oven (Feature 10) in Unit D9: less than 90 years B.P. at one standard deviation, or less than 180 B.P. at two standard deviations (and therefore not calibrated). Stratigraphically, the oven is related to stratum 6 (see fig. 2.22) and contains only indigenous artifacts. The oven is overlain by the sequence of alluvial flood deposits, and two thin occupadon lenses (strata 3 and 5), the latter two containing artifacts of Euro-American origin (glass beads, gun flints, and glass sherds). Thus, Feature 9 probably dates to the end of the eighteenth or first few years of the nineteenth century (consistent with the date range of two standard deviations), prior to the regular introduction of imported artifacts to the Anahulu Valley (see Sahlins and Kirch, in prep.). KE'EKE'E IKI ROCKSHELTER (Site OA-D6-36) Ke'eke'e Iki rockshelter is located not more than 100 m east of Ke'eke'e Nui shelter, in the same cliff formation on the northern side of the valley. Access between the two shelters is easy along the base of the steep cliff into which both sites are cut. Because the Anahulu S ntm meanders southwards from D6-58, the stream is 100-150 m south of Ke'eke'e ki shelter. In other respects, the immediate environnents of both sites are very similar. 32 The Rockshelter Excavations TABLE 2.2 FEATURES IN KE'EKE'E NUI ROCKSHELTER FEATURE UNIT DEPTH BELOW TYPE DESCRIPTON NO. SURFACE (cm) 1 K10 17-28 Hearth Small, scoop hearth measuring 13 by 30 cm, maximum depth 11 cm. Consisted of white ash and charcoal. 2 K10 30-35 Ash lens Thin lens-like deposit of ash with some charcoal, 3- 5 cm thick; 25 by 100 cm in plan view. 3 K10 40-45 Hearth Basin-shaped hearth measuring 40 by 55 cm, maximum depth 5 cm. Consisted of compact white ash with some charcoal flecking, and bumed Aleurites endocarp. 4 K10 40-43 Ash lens Small, lens-shaped deposit of white ash with charcoal flecks, underlain by red (burned, oxidized) soil. Irregular shape, with maximum diameter of 30 cm, depth of 3 cm. 5 K10 55-68 Earth Oven Large earth oven, measuring > 1 m in diameter (not fully exposed in test unit). Basin-shaped, with base resting on sterile alluvium. Contained 14 fist-sized, fire-altered volcanic stones. At base of feature is a 3- 4 cm thick deposit of grey ash with charcoal. Underlying alluvium a dark red color (oxidized). 6 Ol 10-24 Earth Oven Earth oven measuring 50 by 60 cm, maximum depth 14 cm. Contained 83 small fire-altered volcanic stones, dog and fish bone, marine shell, charcoal, and ash. 7 Ol 30-41 Hearth Small, scoop hearth measuring 20 by 40 cm, maximum depth 14 cm. Contained charcoal and ash, with some burned bone and shell. 8 Ol 45-55 Hearth Hearth measuring 12 by 31 cm, maximum depth 10 cm. Contained ash, charcoal, and marine shell. 9 011 74-80 Ash Deposit Lens-like deposit of ash and charcoal, 6 cm thick, extending over entire base of Oll test unit. Ash lies directly over sterile alluvium. 10 D9 43-58 Earth Oven Earth oven only partially exposed in excavated unit; maximum depth 15 cm. Contained 37 fist-sized, fire- altered vesicular, volcanic stones. Orange-colored ash lined the base and rim of the oven. Considerable charcoal flecking between oven stones. Some carbonized Aleurites endocarps were present, as well as 3 fragments of a flat, spongy material that may have been Lagenaria gourd. 33 Anahulu Figure 2.23. View of Ke'eke'e Iki rockshelter, Site D6-36. As with the other shelter sites, Ke'eke'e Iki consists of a stream-eroded notch in the cliff face (fig. 2.23). This site, however, is considerably smaller than those described above. The main occupation floor measures 12 m long, with a width of between 2.5 to 5.5 m, and a total floor area of slightly less than 50 m2 (fig. 2.25A). As it is shallow, the shelter is quite light and airy, and the roof is constricting only in the rear 1-2 m of the floor area (fig. 2.25B). The main occupation floor is nearly level, and is retained by a low terrace facing of angular volcanic cobbles that more-or-less coincides with the shelter dripline. The western boundary of the occupation area is defined by several large rockfall boulders over which a low, crude stone wall had been constructed. At the eastern end of the occupation floor, a semicircular stone crypt had been constructed against the rear wall of the shelter. The crypt appeared to have been opened at some time in the past, and was empty. As with the crypt structures situated at the eastern ends of Kuolulo and Ke'eke'e Nui shelters, this structure may formerly have contained a burial. South of the occupation area, and outside of the shelter dripline is a large rockfall from the cliff high above the site. Aside from the disturbance to the crypt, Ke'eke'e Iki shelter was in very good, undisturbed condition in 1982. A hammerstone lay on the surface of the occupation floor about 1.5 m behind the terrace facing. Several large waterworn stream cobbles were also situated within the occupation area. Excavation Strategy The main objectives at D6-36 were to sample the occupation floor between the terrace facing and the rear of the shelter, and to obtain a continuous stratigraphic profile which would provide a temporal sequence for the site. To meet these objectives most efficiently, we excavated a 5-m long trench along the K-line of our grid, running from the rear shelter wall out to the terrace facing (fig. 2.24). This trench yielded a reasonable sample of the occupation deposits within the dripline area, and stratigraphic profiles along both the west and east sides of the trench were recorded. The cultural deposits proved to be relatively shallow in the interior of the shelter, and were deepest immediately behind the terrace retaining wall, where excavation reached 95 cm below surface. In all, a total volume of 1.95 m3 was excavated. Physical Stratigraphy The stratigraphy of Ke'eke'e shelter is uncomplicated, consisting of a single, undifferentiated, 34 The Rockshelter Excavations Figure 2.24. View of Site D6-36 from the west, showing the excavated trench bissecting the main occupation floor and extending out to the terrace retaining wall. cultural occupation deposit overlying and retained by the artifically constructed stone terrace. In the rear portion of the shelter, the cultural deposit directly overlies bedrock. The profile of the western face of the K-trench is diagrammed in fig. 2.26 and a view of the trench showing this face is also given in fig. 2.27. The uppermost 5 cm consisted largely of culturally-sterile aeolian sediment mixed with organic debris which had accumulated after the abandonment of the shelter. Below this the occupation deposit was a dark brown (7.5 YR 3/2), ashy midden. At about 30-35 cm below surface, the deposit graded into a dark grayish brown (10 YR 4/ 2) sediment with more angular gravel, representing terrace fill. A 50-cm deep sediment column was taken from the eastern face of Unit K18. Cultural Content and Features Site D6-36 yielded a collection of 274 portable artifacts of indigenous Hawaiian manufacture, of which 213 (78%) consist of flakes and cores of volcanic glass. The site also yielded artifacts of Euro-American manufacture. These artifacts are analyzed in chapter 8. The assemblages of vertebrate and invertebrate faunal materials are analyzed in chapter 4. Six features were revealed during the excavation of the K-trench, as enumerated and described in table 2.3. Five of these were hearths or charcoal and ash lenses, and one (Fe 4) was a small earth-oven. The Feature 4 oven yielded a radiocarbon sample (see below). The Depositional and Occupation Sequence The human use of Ke'eke'e Iki rockshelter began with the construction of a 2-3 course high terrace facing running along the shelter's dripline, and with concomitant leveling of the main occupation area by filling in the area behind the retaining wall with stones and rubble. Occupation then commenced on the roughly leveled surface, leading to the accumulation over time of up to 35 cm of cultural midden deposit. A radiocarbon sample from the Feature 4 oven yielded a conventional age of 240 ? 50 B.P., and probably dates the initial use of the site immediately following construction of the terrace. This 14C age may be calibrated to either the mid-seventeenth or mid-to-late eighteenth centuries A.D. (due to fluctuations in the calibration curve for this time interval, Stuiver and Becker 1986). The use of Ke'eke'e lki shelter is thus quite late, corresponding to the final one or two centuries of Hawaiian prehistory. 35 36 Anahdu T/ LO -7 co CL 0 00 L) LLI 0? \001 % % C.) 4-4 LLI ui 0 LU F- LL ol 'k. z rA cc 00' ui 42 LLI co O Ir trA 4-o 0 loo, .ol cc 0 of LLI cn O., 0 CL A % x .0- or % ui %.000. 0- % OZ Cli C14 % 0, 14. The Rockshelter Excavations 1: COJ o (0 (1) Figure 2.25B. Profile through Site D6-36 rockshelter. 37 %..-, E CIQ I lp=. I Anahulu E ?01 oJ 0) 0' k I NCO YI co$ i tD _R x 38 0 4) LI) C.) CT 03 C. C.) C) C.) C.) a.) 1-4 C3 C) C.) The Rockshelter Excavations TABLE 2.3 FEATURES IN KE'EKE'E IKI ROCKSHELTER FEATIURE UNlT DEPTH BELOW TYPE DESCRTION NO. SURFACE (cm) 1 K18-19 6-3 Hearth Scoop hearth measuring 55 by 73 cm, containing charcoal and ash, some fire- cracked volcanic stone, carbonized Aleurites endocarps, marine shell and bone. 2 K19 19-22 Ash lens Small lens-like concentration of white ash with wood charcoal measuring 20 by 30 cm, maximum depth 3 cm. 3 K19-20 18-20 Ash lens Thin lens of white ash and charcoal, resting on bedrock. Measured 30 x 50 cm, 2 cm deep. 4 K19 25-40 Earth Oven Basin-shaped earth oven (60 by 65 cm, 15 cm deep) filled with fire-cracked, fist- sized volcanic stones. Large pieces of wood charcoal were dispersed between the oven stones, and a lens-shaped concentration of white ash lined the bottom of the feature. 5 K16 10-29 Hearth Concentration of charcoal and ash (19 cm deep). 6 K17 25-36 Hearth Small concentration of charcoal and ash, measuring 26 x 40 cm, maximum depth 11 cm. The occupation period for Ke'eke'e Iki shelter probably corresponds roughly with the "upper cultural deposit" in nearby Ke'eke'e Nui shelter. RADIOCARBON CHRONOLOGY OF THE ROCKSHELTERS Seven samples were submitted for radiocarbon age determinations, three each from Sites D6-60 and -58, and one from Site D6-36, as enumerated in table 2.4. All samples consisted of wood charcoal, with the addition of carbonized candlenut (Aleurites moluccana) endocarp fragments in the case of five samples. All determinations were made by the Beta Analytic, Inc. laboratory, using the liquid scintillation spectrometer method. All samples were first hand picked for elimination of roodets, then given an extensive hot acid pretreatment to eliminate any contamination by carbonates. Sanples were rinsed to neutrality and submitted to hot alkali solution to extract humic acids, followed by another acid treatment, and final rising to neutrality. All benzene syntheses and subsequent radioactivity measurements proceeded normally (M. Tamers, pers. comm., Sept. 10 and Nov. 8, 1982). Table 2A provides data on 13C, and on the "conventional 14C age" BP. (Stuiver and Polach 1977) for all samples. The calibrated AD. ages shown in the right-hand coluwm of table 2.4 are based on the testrial curve of Stuiver and Becker (1986), and were determined using a revised FORTRAN microcomputer program ("CALIB") provided to the author by M. Stuiver and P. Reimer (cf. Stuiver and Reimer 1986). The calibrated age ranges provided are those obtained from the intercepts of the calibration curve ("Method A"), at one standard deviation. 39 Anahilu o 0%o oo~ V- 00W W 00 00 IDr W) 'lb - o )- CS N O ON 6 0%~~~~~N O~ o 0 o 00 -~~~~~~- o 0 0% 00 0 4l +l +lH + +l 1+ 00 0 N N %m C C 3 3 3t 30 3t 3t 30 t 0 l Q W) > Q 2 g ? 0 i a - 0 4 ~~~~~~~q > a X ?o t ^ W) en~~~~~~~~~~~~~4 ,CW -- c 00 00 s "_ W) 96 00 W' ,n IC~ 0 m- 40 "-, V- n PE m ZY 8 - z z 0 6 z CT E" 0- qA 99 w :: CA m O 0 :9 >4 3 *A 04 .4 ;;N Cu d 4) '4) 10 7) m Lo e0 '4 s 4) 4) 0 0 C.) 4) 0 10 'S. 4) 2 O8 W- 40 WI The Rockshelter Excavations Figure 2.27. View of the K-Trench in Site D6-36 after completion of excavations. The stratigraphic positions of all samples, and their interpretations in reference to individual site depositional sequences, have been discussed above. As a group, the entire suite of samples indicates that Ke'eke'e Nui rockshelter was probably the first to be utilized by prehistoric Hawaiians, in about the fourteenth cenury A.D. Kuolulo rockshelter was in use by the mid- fifteenth century A.D. The smaller Ke'eke'e Iki shelter was not in use until about the mid-seventeenth (or possibly eighteenth) century, when the nearby Ke'eke'e Nui shelter was becoming more intensively occupied. In terms of the archipelago-wide cultural sequence established by Kirch (1985:298-308), these sites were first utilized in the middle of the Expansion Period, a time of major expansion of settlements into leeward areas, and into more ecologically 'marginal' situations (see also Hommon 1976, 1986). Occupation of the Anahulu shelters continued into the Proto-Historic Period (A.D. 1650-1795), during which time these sites may have become more intensively used, even on a permanent basis. Based on their artifact assemblages all shelters appear to have been abandoned as occupation loci early in the Historic Period. REFERENCES CITED Hommon, R.J. 1976. "The Formation of Primitive States in Pre-Contact Hawaii." Ph.D. Dissertation, Univ. of Arizona. (University Microfilms.) . 1986. Social evolution in ancient Hawaii. In P.V. Kirch, ed., Island Societies: Archaeological Approaches to Evolution and Transformation, pp. 55-68. Cambridge University Press. Kirch, P.V. 1985. Feathered Gods and Fishhooks: An Introduction to Hawaiian Archaeology and Prehistory. University of Hawaii Press. Honolulu. Sahlins, M., and P.V. Kirch. in prep. Anahulu: The Archaeology of History in the Early Sandwich Islands Kingdom Stuiver, M., and B. Becker. 1986. High-precision decadal calibration of the radiocarbon time scale, AD. 1950-2500 B.C. Radiocarbon 28(2B):863- 910. Stuiver, M., and H.A. Polach. 1977. Reporting of 14C Data. Radiocarbon 19:355-63. Stuiver, M., and P.J. Reimer. 1986. A computer program for radiocarbon age calibration. Radiocarbon 28(2B): 1022-30. Weisler, M., and P.V. Kirch. 1985. The structure of settlement space at Kawela, Molokai. New Zealand Journal of Archaeology 7:129-58. 41