CHAPTER EIGHT THE PORTABLE ARTIFACT ASSEMBLAGES by Patrick V. Kirch T HE ROCKSHELER EXCAVATIONS yielded a combined collection of 2,388 portable artifacts of indigenous manufacture, summarized in table 8.1. A large proportion of the collection (82%) consists of cores and flakes of volcanic glass; nonetheless, a fairly broad range of Hawaiian material culture types is represented. These are systematically described below, with notes on their stratigraphic distributions in individual sites, and comparisons with excavated assemblages from other sites both on O'ahu and throughout the archipelago. The rockshelters also yielded various artifacts of foreign manufacture dating to the post-European contact or historic period. These are only briefly summarized in this monograph, as they will be the subject of a more detailed study elsewhere. FISHING EQUIPMENT Small quantities of fishing equipment, principally one-piece fishhooks, were recovered from each of the rockshelters. Notably absent were any of the several types of large sinker (for octopus fishing, bottom fish- ing, and for nets) commonly found in Hawaiian sites. One-Piece Fishhooks Twelve fragments of one-piece hooks were recovered, all manufactured of pearl shell (probably Isognomon sp.). These hooks are uniformly small in comparison with the size range of Hawaiian one-piece hooks generally. The most complete specimens have shank heights of 1.7 and 1.2 cm; shank height on a third, nearly complete shank probably did not exceed 1.4 cm. Shank thickness ranges from 0.1-0.3 cm. Line- lashing devices are present on three specimens, all of which are of type HT4 (Sinoto 1962), consisting of an outside knob perpendicular to the shank (fig. 8.1). The finished specimens are all carefully manufactured, evincing considerable skill in the ability to work friable pearl shell at such a diminutive scale. One unfinished pearlshell one-piece hook tab was excavated from Site 60. This specimen indicates the use of the filling-and-notching manufacture technique described by Sinoto (1967). Two-Piece Fishhooks Two two-piece fishhook points were excavated, from Sites 36 and 58. The complete specimen from Site 36 (fig. 8.1) appears to be a small octopus lure point (Emory, Bonk, and Sinoto 1959:28-29) with two prominent notches for lashing on its outer face (fig. 8.4,d). The point is manufactured from the caudal tang of a large acanthurid fish, possibly Naso sp. The basal flanges of the tang have been filed down but not entirely erased. I am not aware of any other two-piece points made from such an acanthurid tang in other excavated Hawaiian assemblages. The hook has a point height of 3.0 cm. The specimen from Site 58 consists only of the tip of what is probably a typical two-piece hook. The point, with a strongly incurved tip, is of mammal bone, and has a width of 0.6 cm and a thickness of 0.35 cm. Anahiu TABLE 8.1 INDIGENOUS HAWAIIAN PORTABLE ARTIFACTS FROM THE ANAHULU ROCKSHELTERS ARTIFACT CLASS D6-60 D6-58 D6-36 TOTALS Fishhooks 5 4 6 15 Line Sinkers 3 3 Adz 1 1 Adz Preform 1 1 Adz Fragments 1 1 2 Adz Flakes/Chips 35 39 15 89 Grindstones 3 1 4 Whetstones 4 1 5 Basalt Files 2 1 3 Polishing Stones 3 3 Coral Abraders 3 7 6 16 Hammerstones 10 2 2 14 Bone Awls 6 2 3 11 Cone-shell Beads 52 4 5 61 Tattooing Needle 1 1 Dog-Tooth Ornaments 1 2 3 Gaming Stone 1 1 Worked Bone 3 1 4 Worked Shell 8 3 2 13 Basalt Flakes 52 60 11 123 Volcanic Glass Cores/Flakes 1,407 336 213 1,956 Coral Manuports 37 13 9 59 TOTALS 1,636 478 274 2,388 Line Sinkers Three small line sinkers were excavated at Site 60. All of these are made from naturally rounded, ovoid stream pebbles of basalt, and have a single narrow, incised groove extending longitudinally around the pebble to facilitate lashing onto a fishing line. The pebbles range in length from 1.4-2.2 cm, and in width from 1.2-1.5 cnL It is probable that these line weights were used in conjunction with the small one-piece fishhooks. Stratigraphic Distribution In Site 36, all of the hooks were recovered from depths between 20 and 35 cnm The three one-piece hooks from Site 58 were recovered from depths below 35 cm in Unit 011 only. The two-piece point was recovered from 45-50 cm in Unit D9. In Site 60, however, the hooks were found in the upper 10 cm of deposit, with the exception of one fragment from 15-20 cm in Unit M25. Two of the three line sinkers were recovered from the upper 5 cm of Site 60, while the third specimen was found between 50-55 cm in Unit D20. Comparsons The one-piece fishhooks from the Anahulu rockshelters are relatively small in comparison with the general size range of one-piece hooks in Hawaiian archaeological sites. Emory, Bonk, and Sinoto (1959:14-5) give a mean shank height of 1.98 cm for Hawaiian jabbing hooks; the Anahulu specimens are all somewhat smaller that this mean value. In a comparative study of three fishing sites (Kirch 1982) I 112 The Portable Artifact Assemblages demonstrated that hook size is direcdy related to certain local environmental conditions, such as the presence of an inshore fringing reef. For example, hooks from two coastal sites in which small inshore reef fish predominated had mean shank lengths of 1.83 and 1.94 cm respectively (1982, table Ill). These contrast with another site associated with benthic bottom fishing, in which shank length averaged 3.4 cnL Were the Anahulu rockshelters situated near the coast, there would be no question that the small fishhooks were intended for taking small inshore reef fish. Their presence in three inland shelters 6 km from the coast, however, leaves this interpretation open to question. Furthermore, since the quantity of manufacture detritus and abrading tools in these sites is very limited, it appears that these hooks were not being produced at the shelters, but were brought to them in finished fornL One possible function of these diminutive hooks may have been angling for the prized freshwater fish called 'o'opu in Hawaiian, and including several endemic species in the families Eleotridae and Gobiidae (Titcomb 1952:122). I have been unable, however, to find any ethnohistoric references to the use of hooks in catching 'o'opu, the usual reference being to nets. Certainly, though, the small size of the one-piece hooks from the Anahulu shelters would be appropriate to the 'o'opu which generally range between 15-30 cm in length. On the three one-piece hooks where the line-lashing device is present this is of type HT4 (Sinoto 1962), a distinctive protruding knob at right angles to the shank. As Sinoto (1962) originally suggested, the dominance of the HT4 type in the "late period" (after about A.D. 1200) has been confmed by subsequent excavations at a variety of sites. The Anahulu sample is thus consistent with this general pattern. TOOLS Adzes One complete adz, several diagnostic fragments, one preform, and 89 flakes or chips from polished adzes are included in the rockshelter assemblages. The whole specimen, from Unit K1O of Site 58, is a typical late prehistoric Hawaiian "quadrangular" adz (the cross- section is actually slightly trapezoidal), with a well- ground face and bevel, and only partially ground tang and sides, leaving considerable evidence of the direct percussion flaking process. Although the adz is relatively small, it has a pronounced tang. It has a total length of 6.1 cm, a cutting edge width of 2.4 cm, shoulder width of 2.35 cm, and shoulder thickness of 1.8 cm; it weighs 46 grams. Figure 8.1. One-piece fishhook fragments from the Anahulu rockshelters. 113 Anahidu The two larger adz fragments, from Sites 36 and 60, are both shoulder fragments from what must have been relatively large quadrangular adzes with pronounced tangs. In both cases grinding did not obliterate the original flaked surface. Both adzes, when complete, were probably in the size range of 10-18 cm. On the surface of Site 60 we collected a small adz preform consisting of a retouched basalt flake. The flake was bifacially trimmed to form a rectangular preform, but was not ground. It has a maximum length of 6.5 cm, width of 2.5 cm, and maximum thickness of 0.95 cm. While larger adzes in Hawaii were generally made from core-type preforms, small, thin adzes are known to have been manufactured from retouched flakes such as this. Relatively little can be said of the 89 adz flakes and chips, as these are largely non-diagnostic and diminutive; their presence in the rockshelters does, however, indicate adz use and/or reworking (such as bevel rejuvination) at these sites. In the few cases where more than one polished surface is present, right angles clearly indicate that most if not all of these flakes were detached from quadrangular sectioned adzes, the dominant form in late Hawaiian prehistory. The size ranges of these flakes, based on the sample of 15 specimens from Site 36, is as follows: length, x=2.17 cm, s.d.=1.22; width, i=1.45 cm, s.d.=0.66 cm; thickness, x=0.58 cm, s.d.=0.20 cm. Stratigraphic Distribution. Adz fragments or flakes were recovered from virtually every excavated level in all three rockshelters. Nonetheless, adz specimens tend to be more frequent in the upper levels of each site. This tendency is most pronounced in Site 60, where two-thirds of all adz flakes were recovered from the upper 20 cm of deposit. In Site 36, 13 of the total of 15 excavated specimens came from the upper 20 cm. This pattern is not so pronounced, however, at Site 58. Comparisons. A marked change in adz variation appears in Hawaiian archaeological assemblages between the end of the Developmental and beginning of the Expansion Periods, ca. AD. 900-1100 (Kirch 1985:184, 304). Earlier assemblages are notable for their range of variation in cross-sectional form and other attributes, whereas later assemblages are highly uniform. Cleghorn (1982) has commented on the virtually monotonous and stereotypic form of late prehistoric Hawaiian adz, with its distinct tang and quadrangular cross-section. The adzes from Anahulu conform perfectly with expectations from this general sequence. Dating to the Expansion and Proto-Historic Periods, they are all of classic quadrangular form. Harmerstones Fourteen hammerstones are included in the collection, with ten of these coming from Site 60. There is considerable variation in these artifacts, which are individually described in table 8.2. Most specimens consist of medium sized, ovoid or slightly elongated waterworn river cobbles of vesicular basalt, identifiable as hammerstones because they display battering, crushing, or pitting on ends, sides, or flat surfaces. Some specimens, however, deviate from this "typical" form. One hammerstone, for example (D6-60-M22-89) is markedly discoidal and of more vesicular stone than most. Two artifacts are pebbles rather than cobbles. While the term "hammerstone" has been used to categorize all of these artifacts, it seems likely that this class includes a number of objects with differing functions. In many cases, the amount of crushing or bruising on the cobble edges is relatively slight, and certainly would not be sufficient to suggest use as a hammer used in the direct percussion flaking of dense basalt. More plausible would be the use of these cobbles for flaking the volcanic glass found in substantial quantities throughout all rockshelters. Furthermore, the presence of pitting or, the flat sides of some of these stones hints that they may have been used as "anvils" rather than "hammers." Schousboe, Riford, and Kirch (1983) suggest that the reduction of Hawaiian volcanic glass assemblages was accomplished using a bipolar technique, requiring both anvils and hammerstones, which would be consistent with this interpretation. The discoidal hammerstone mentioned earlier stands out from the other specimens in this class, and may well have functioned as a hammer for direct percussion of fine-grained basalt adzes; certainly, it is similar to hammerstones from the Mauna Kea adz quarry site (Cleghorn 1982). Stratigraphic Distribution. Hammerstones were recovered both on the surface of sites, and in a variety of stratigraphic contexts throughout the sites. No particular pattern of stratigraphic distribution can be detected, and it seems likely that these artfacts were utilized throughout the period during which the shelters were occupied. Comparisons. Hammerstones have long been an analytically neglected artifact class in Hawaiian archaeological studies. Soehren (1966, ms.) and Tuohy (1965) were among the few investigators to devote some attention to these multipurpose tools. Soehren (1966, Appendix C) proposed a classification of hammerstones with a primary distinction between 114 The Portable Artifact Assemblages TABLE 8.2 HAMMERSTONES CATALOG NO. L W Th REMARKS (cm) (cm) (cm) D6-36-SA-1 D6-36-K19-7 D6-58-K10-361 D6-58-011-21 D6-60-M22-89 D6-60-M24-30 D6-60-M24-53 D6-60-M24-93 D6-60-M24-94 D6-60-M24-162 D6-60-P24-1 D6-60-P25-12 D6-60-P25-194 11.5 9.1 3.4 12.3 8.5 8.7 10.8 10.8 8.0 2.9 10.8 9.3 10.0 7.0 8.9 4.7 5.9 2.4 7.7 8.5 4.0 5.4 pecked finger grips on slides; pitting/ abrasion around perimeter minor pitting on one side; possibly an anvil rather than hammerstone pebble; battering on one end ovoid cobble with battering and flake scars on two ends discoidal shape; highly vesicular; battering/crushing around perimeter ovoid cobble; faint battering and flake scars-one end only 6.8 4.3 5.5 heavily battered/cmshed around perimeter 2.7 4.2 1.7 3.4 7.8 D6-60-Q25-100 elongate, pointed cobble; substantial battering/crushing on pointed end small cobble; pitted one side very small; problematic slightly battered on one end elongate cobble pitted on two ends ovoid cobble pitted on two sides cobble; one side pitted x = 8.93 s.d. = 2.85 x = 5.78 s.d. = 2.40 x = 4.30 s.d. = 1.16 natural stones "selected for their convenience and suitability" and "artifically shaped stones, either made expressly for use as hammers, or other artifact tpes converted to hammerstones" (1966:56). According to Soehren's schema, most of the Anahulu specimens would fall into Type IA, general discoidal or ovoid stones, although there are also examples of IB, elongate or cylindrical stones. Since these artifacts were clearly utilized for a variety of "hammering" functions, however, I believe that a more productive approach will be to focus on the nature of the damage or use patterns, including the type of damage (crushing, flake scars, pitting, etc.), its position and extent on the stone. The material itself, including propertes of density and 115 Anahulu porosity, would also seem to be relevant. For example, most of the hammers identified from the Mauna Kea adz quarry are of highly vesicular, low density basalt, much "softer" than the fme-grained adz basalt being worked. This observation alone might suggest that the hammerstones of relatively dense, waterworn cobbles such as those from Anahulu, were not used in adz manufacture or retouching but for some other purpose. Certainly, a great deal more attention needs to be paid to this artifact class and its functional attributes in the future. Coral Abraders Sixteen pieces of Porites sp. coral displaying one or more clearly abraded facets or surfaces have been classified here as abraders. These include small, often elongate and tapering artifacts commonly termed coral "files" (cf. Emory and Sinoto 1961:53-54), as well as larger pieces with one or two flat, abraded surfaces sometimes termed "rubbing stones." The latter type predominate in the Anahulu assemblages, with only five artifacts which could be termed "files." There are as well a substantial number of apparently unworked coral fragments from all sites (see below); a number of these may be fragments of coral abraders which, however, lack worked surfaces. Some are chemically weathered (with "chalky" surface textures) making it difficult to determine whether they were once used as abraders. Stratigraphic Distribution. Abraders were recovered at different levels in all sites, and no apparent temporal pattern is evident. Basalt Files Two basalt file fragments were excavated from Site 60, and a complete specimen from Site 58. The latter, of dense, fme-grained basalt is a slightly tapering object, with broad surfaces worn nearly flat through abrasion. The edges also show evidence of wear. The file measures 6.9 cm long, 2.1 cm wide, and 0.6 cm thick. This artifact was found at a depth of 45 cm in Unit KIO of Site 58. The pieces from Site 60, both tip fragments, also consist of dense basalt. They were recovered from depths of 3540 cm in Unit M22 and from 9 cm in Unit P26. Comparisons. Files of dense basalt are not common occurences at Hawaiian archaeological sites, but they have been previously reported, for example, from the leeward O'ahu rockshelters (Emory and Sinoto 1961:56), from the Halawa Dune Site on Moloka'i (Kirch 1975:40), and from the Nualolo sites on Kaua'i (Soehren mis.). On Hawai'i Island, most files or abraders of stone are made from the vesicular, scoriaceous lava which is readily available from recent volcanic eruptions. The use of dense basalt, as in the Anahulu specimens and those cited above, may be restricted to the westerly islands, although further distribution studies would be necessary to conflrm this. Grindstones and Whetstones Grindstones are here defined as relatively large stones with one or more grinding surfaces, on which other stone implements (and particularly basalt adzes) are worked and shaxpened. In contrast, whetstones are smaller, hand-held tools and frequently have concave grinding surfaces on opposing sides of a tabular stone. Grindstone fragments were excavated from Site 36 (1 specimen) and from Site 60 (3 specimens). The fragment from Site 36 was used secondarily as a core for the removal of basalt flakes. A specimen from Unit M22 of Site 60, of coarse-grained basalt with olivine phenocrysts, is fractured and blackened from secondary use as an oven stone. Five whetstone fragments were excavated, one from Site 58 and four from Site 60. A basalt whetstone fragment from Site 60 is thoroughly ground on two parallel sides as well as along the adjoining edge. The thee other specimens from Site 60 are of a fine-grained andesite with numerous feldspars; these tabular stones have slightly concave grinding surfaces on both sides. The specimen from Site 58 has been fractured by heat, suggesting secondary use as an oven stone. Stratigraphic Distribution. The grindstones in Site 60 were recovered at depths ranging from 3 to 25 cm; the Site 36 specimen came from a depth of 18 cm. Three of the whetstones from Site 60 came from the upper 5 cm of deposit; the other was excavated at 23 cm in Unit D20. That from Site 58 was found at 15 cm depth in Unit 011. In general, these proveniences suggest the use of grindstones and whetstones only in the later occupation phases of these sites. This pattern may well correlate with the increased frequency of adz use in the upper levels of the rockshelters, as noted earlier, since a primary use of both grindstones and whetstones was probably the sharpening of adz bevels. Comparisons. The Anahulu grindstones and whetstones are comparable to many examples excavated from a variety of sites throughout the archipelago (e.g., Kalahuipua'a, Kirch 1979:167; Honaunau, Tuohy 1965:57; Halawa, Kirch 1975:40; West Moloka'i, Bonk 1954:78-9; O'ahu rockshelters, Emory and Sinoto 1961:64; Nualolo, Soehren ms.). 116 The Portable Artfact Assemblages a 1 2 3 4 5 , ... A A Av . - - Figure 8.2. Artifacts from Site D6-60: a, retouched adz fragment; b, adz flake; c, polishing stone; d, adz preform. Polishing Stones lTree small waterworn stream pebbles of basalt, all from Site 60, have abraded and polished facets and are classified here as "polishing stones." The largest of these measures 4.4 by 3.9 cm, and is 1.3 cm thick; it has been worked on two faces (fig. 8.2,c). An ovoid pebble 4.2 cm long and 1.5 cm thick shows artificial polish along part of its edge. The third specimen, also an ovoid pebble, 3.7 cm long and 1.0 cm thick, has polish on part of one side. The probable use of these artifacts was in polishing or burnishing, possibly of wooden artifacts. Stratigraphic Distribution. The polishing stones were recovered from the following proveniences: 5-10 cm depth in Unit 025, 10-15 cm depth in Unit Q25, and 55-76 cm depth in Unit M22. This sample is too small to speak meaningfully about temporal pattens. Comparisons. Polishing stones have not been widely ed in the Hawaiian archaeological literature, and it is conceivable that these nondescript tools have often been overlooked and discarded during excavations. Bonk (1954:82) notes that many of the West Moloka'i sites contained "smooth water worn stones.. .utilized for burnishing." Tuohy (1965:56) also distinguished 8 stone pebbles from his Honaunau excavations which show "some degree of polish" and which he separated from "stone rubbers." Somewhat surprisingly, Soehren (ms.) does not note the presence of polishing stones in the Nualolo terraces. Awls or "Picks" This category consists of all slender, artificially- pointed objects of bone, which some archaeologists in Hawai'i have separated into "awls" and "picks" (e.g., Emory and Sinoto 1961). This distinction appears to be arbitrary and functionally meaningless and therefore is not maintained here. Awls were recovered from all thee rockshelters. Five of them are of bird bone with one condyle removed and the shaft beveled so as to produce a sharp working end (fig. 8.3,a-b). The other specimens appear to be of mammal bone, possibly dog or pig; four of the six mammal bone awls are only fragments. These objects probably served multiple functions, including the removal of neat from shellfish, the preparation of Pandanus leaves for mat making, and for perforation of various kinds of material. Stratigraphic Distribution. With the exception of one bird bone awl from the lower fill (40-95 cm depth) in Unit K17 of Site 36, all awls were recovered from the upper 15 cm of deposit in each site. Their use was thus largely confined to the later occupation phases. :O 0, 117 ? I Anahdu Figure 8.3. Artifacts from Site D6-60: a,b, awls of bird bone; c, dog tooth ornament; d, unfinished one-piece fishhook. c a . d b S L1 2 3 4 5 A A &- ' ' C ICM Figure 8.4. Artifacts from Site D6-36: a, basalt flake with use wear; b, bird bone awl; c, modified cone shell; d, octopus lure point. 118 The Portable Artifact Assemblages Comparisons. "Bone pickers" were first described by Bonk (1954:109) from his West Moloka'i excavations, who interpreted them as instruments for "extracting the soft parts of mollusks for food." Emory and Sinoto (1961:39) described "picks" from their O'ahu sites, but distinguished them from awls. In assemblages from Halawa, Moloka'i and Kalahuipua'a, Hawai'i, analyzed by the author, the distinction between "picks" and "awls" appeared to be arbitrary and funtionally meaningless, and the term "awls" was applied to all such pointed bone objects (Kirch 1975:40; Kirch 1979: 171). ORNAMENTATION Tattooing Needle Like other Polynesians, the Hawaiians decorated their bodies with finely-executed geometric tattoos, frequently depicted in the sketches and engravings made by the first European artists to visit the islands (Emory 1946). Tattoo needles are, however, a rare occurrence in Hawaiian archaeological sites. The tip of a small tattoo needle was recovered from the uppermost 1 cm of deposit in Unit K10 of Site 58 (fig. 8.5). Made from a bird bone shaft, the needle is slightly tapering and has three very sharp teeth defined by notches filed between them. The needle is distincdy blackened over its entire surface, presumably ink stained. The length of this incomplete needle is 1.3 cm; width measures 0.38 cm, thickness 0.05 cm. Comparisons. Tattoo needles have been recovered from a few excavations, including those at Makaniolu Shelter on 0'ahu (Site 02; Emory and Sinoto 1961), and the K3 and K5 house terraces at Nualolo on Kaua'i (Soehren ms.). The Makaniolu needles are of a compound type quite different from the Anahulu specimen. The latter is, however, very similar to the tattooing combs from Nualolo described by Soehren (ms.). The Anahulu and Nualolo needles also closely match a specimen in the Salem Peabody Museum described and illustrated by Emory (1946:263, fig. 13). Perforated Dog's Teeth Canine teeth of Canisfamiliaris, with a single perforation through the root end, are frequently found in Hawaiian archaeological sites, and three examples are included in the rockshelter assemblages (two from Site 58 and one from Site 60). Such teeth were used in the distinctive dance anklets (Hawaiian kupe'e niho 'ilio) worn by men, and illustrated in the famous engraving of a Sandwich Islands man dancing by Webber made during Cook's third voyage. Buck (1957:553-561) has described Figure 8.5. Tattooing needle from Site D6-58. these anklets in detail. Dog teeth were also strung on a cord to be used as necklaces (Buck 1957:545), although Buck regards such a use as temporary: "they were so strung until a sufficient number had been collected to make the dog-tooth leg ornament characteristic of Hawaii." The Anahulu specimens are typical, with a single perforation drilled through the root from both sides. The drill holes are funnel-shaped and slightly irregular, indicating the use of the indigenous pump drill with a shell or stone tip. On one specimen from Site 58 the area of the root adjacent to the perforation has been abraded smooth on one side only. This is a modification not previously observed on specimens from a large number of other Hawaiian sites. Stratigraphic Distribution. The Site 60 specimen is a surface find, as is one from Site 58. The single excavated example was recovered at a depth of 20-25 cm in Unit K10 of Site 58. Cone Shell Disks One of the most common finds in all of the rockshelters were small, beach-worn disks of the spire portion of Conus shell, many of which have natural perforations in the center (such disks, easily collected on Hawaiian beaches, are popularly called "puka shells" and necklaces strung from them became something of a fad in the mid-1970s). Five of these disks were found at Site 36, four at Site 58, and 52 at Site 60. None of the specimens shows evidence of artificial modification, but it seems likely that they were collected in order to be strung as necklaces, in the manner described by Buck (1957:543, fig. 334). The diameters of these disks range from 0.45 to 1.7 cm. Stratigraphic Distribution. The large sample of cone-shell disks from Site 60 provides the best indication of their temporal distribution. Six specimens are distributed at depths of between 20-60 cm, but the 119 Anahiu majority were recovered between 0-20 cm, and especially in the upper 10 cm of deposit. Comparisons. Cone shell disks are frequent occurrences in Hawaiian archaeological sites, although they have at times not been distinguished from molluscan midden material. Emory and Sinoto observed that in the leeward O'ahu rockshelters "the commonest beads found are cone spires or cone-shell tops, which are plentiful on the beaches" (1961:70). Such beads were also common fimds at the Honaunau sites excavated by Tuohy (1965:80). MANUFACTURE DETRITUS Worked Shell Twelve pieces of pearl shell (Pinctada and Isognomon spp.) with evidence of working were excavated from the rockshelters. Most of these are small fragments with one or two abraded or cut edges. A roughly rectangular piece (1.5 by 1.4 cm, 0.1 cm thick) from Site 58 may be a small fishhook tab. Another specimen from the same site is circular in outline (diameter 0.9 cm, thickness 0.1 cm) and could have been used as an inlay, possibly for the eye of a small wooden figure (cf. Buck 1957:469-478). With the exception of these two pieces, the worked shell material is non-diagnostic. Stratigraphic Distribution. Two pieces of worked shell were recovered at depths of 23 and 45 cm in Site 58 (both in Unit K10). Otherwise, worked shell was recovered from depths of less than 20 cm in all sites. Worked Bone One fragment of worked bone was found at Site 58, three at Site 60; all of these display abrading or cut marks, and probably represent detritus from the manu- facture of bone artifacts. All fragments are of mammal bone, of indeterminate species. The specimen from Site 58 was recovered from a depth of 78, cm in Unit 011 ; of the Site 60 specinens, one was found at 15 cm, while the other two came from the 0-5 cm level. MISCELLANEOUS OBJECTS Gaming Stone ('Ulumaika) A beautifully made example of a bowling disk for the Hawaiian maika game (Buck 1957:372-3) was excavated from Unit K1O of Site 58, at a depth of 28-35 cm. The disk is of fine-grained basalt, and has been highly polished over all surfaces. Its sides have the characteristic convex shape, and are well balanced. This 'ulumaika has a diameter of 7.15 cm, and maximum thickness at the center of 4.35 cm. It weighs 318.5 grams. Modified Cone Shell From a depth of 5 cm in Unit K20 of Site 36 we excavated a Conus shell which has been ground flat on the spire and along the longitudinal surface adjacent to the aperture (fig. 8.4,c). The shell as modified measures 4.1 cm long. Its function cannot be ascertained, although one might presume that it was intended as an item of omamentation. Comparisons. Soehren (ms.) describes and figures several cone shells worked in precisely the same manner as the specimen described above, from sites K3 and K5 at Nualolo, Kaua'i. Based on the flat, ground surface, Soehren interprets these artfacts as "bracelet components," but notes that in the absence of ethno- graphic confirmation, this interpretation is "arbitrary." Oven Stones Four vesicular basalt cobbles with fracturing and fire-blackening are presumed to have been oven stones for use in the indigenous imu or underground oven. The stone from Site 36 was excavated at a depth of 10 cm, while the three from Site 60 were recovered at depths between 20-40 cm. Coral Manuports Small pieces and fragments of coral, including both Porites type and branch coral, were common finds in all sites, with 9 from Site 36, 13 from Site 58, and 37 from Site 60. Some of the Porites fragments may have derived from the fracturing of larger abraders, although they do not bear traces of abraded surfaces. A number of water-rolled branch coral pebbles could have been used as playing pieces in the konane game (cf. Buck 1957:369-372). There is no clear pattem of stratigraphic distribution of these objects, which were recovered at virtually all levels. LITHICS The most frequentdy recovered items in all three sites were flaked basalt and volcanic glass, totalling 2,079 specimens. Basalt Flakes A total of 123 flakes of basalt were excavated, 11 from Site 36, 60 from Site 58, and 52 from Site 60. 120 The Portable Artifact Assemblages TABLE 8.3 VOLCANIC GLASS CORES AND FLAKES SITE CORES FLAKES TOTAL RATIO D6-36 6 207 213 1:34 D6-58 28 308 336 1:11 D6-60 87 1,320 1,407 1:15 TOTAL 121 1,835 1,956 These range from flakes of fme-grained basalt of adz quality (but lacking polished or ground facets) to flakes of rather coarse-grained, vesicular basalt. A number of these can be classified as decortication flakes, and exhibit cortex from waterwom stream cobbles. For the most part, the flakes are small, as indicated by the following summary statistics: length, x=3.59 cm, s.d.=2.2 cm; width,ix=2.59 cm, s.d.=1.75 cm; thickness,Yx=0.88 cm, s.d.=0.56 cm. A small number of flakes have either been retouched or have evidence of use wear; these are individually described below. OA-D6-36-K17-2. Parallel sided blade of basalt with almost glassy texture (6.7 by 3.9 cm). Unifacial use wear in the form of micro flake scars is evident along half of one edge. The flake would be classified as a scraper (fig. 8.4,a). OA-D6-36-K17-4. Roughly rectangular flake of fine-grained basalt (3.9 cm long, 3.0 cm wide, 0.8 cm thick). Both edges along the long axis show use wear in the form of rounding and smoothing. OA-D6-36-K17-44. A rectangular flake of fine basalt (3.65 cm long, 2.85 cm wide, 0.95 cm thick). Micro flake scars along half of one edge suggest use, possibly as a scraper. OA-D6-36-K19-1. Roughly square flake of fine- grained basalt (1.8 by 1.7 cm). Utilization suggested by rounding and polish along the right lateral edge. There is also a small area of polish toward the distal end of the flake scar. OA-D6-58-SA-7. A roughly rectangular flake of vesicular, coarse-grained basalt with some cortex (6.6 cm long, 4.3 cm wide, 1.7 cm thick). Micro flake scars along part of the longest edge suggest use, possibly as a scraper. OA-D6-58-O11-45. A large rectangular blade of basalt (8.4 by 5.7 cm), with step flake scars along one edge. OA-D6-60-D20-53. A small flake of basalt (5.0 by 1.9 cm) with possible modification in the form of the three secondary flakes removed from one end. OA-D6-60-Q25-99. Medium-sized flake of basalt (6.0 by 3.8 cm), with weathered cortex on the dorsal surface. Minor modification or retouch indicated by the secondary removal of three flakes from the dorsal surface, one from the vental surface. No evidence, however, of use wear. Stratigraphic Distribution. In Site 58, there is fairly consistent distribution of basalt flakes from top to bottom throughout the deposit. In Site 36, most of the flakes were recovered in the upper 10 cm, although specimens occurred also in the 10-20 and 3040 cm levels. In Site 60, flakes were distributed throughout the stratigraphic section, but were most prevalent in the 0-10 and 10-20 cm levels. Comparisons. Flaked basalt, though ubiquitous in Hawaiian archaeological sites, has long been a neglected artifact category, which renders inter-assemblage comparisons virtually impossible. Volcanic Glass As is the case in most prehistoric Hawaiian sites, the most abundant artifact class consisted of small cores and flakes of volcanic glass, a low-silica form of obsidian (Barrera and Kirch 1973; Schousboe, Riford, and Kirch 1983). A total of 1,956 of these artifacts were recovered from the excavations (table 8.3). The majority of these consist of flakes, with only about 3-8% of each site assemblage comprising cores. The flake assemblage from Site D6-58 was analyzed for five metric attributes, and the results are reported in table 8.4. The diminutive size of this lithic assemblage is clearly evident in these statistics. Observations were also made on a series of morphological attributes (following the protocol developed by Schousboe et al. 1983), permitting the following general characterization of the assemblage: (1) prepared platforms are extremely rare (1%); (2) bulbs of percussion are dominantly diffuse (8 1%), with low frequencies of salient, sheared, or acuminate bulbs; (3) 121 Anahiu TABLE 8.4 METRIC ATTRIBUTES OF VOLCANIC GLASS FLAKES FROM SITE D6-58 ATTRIBUTE* x s.d. Range N Length 6.55 2.74 2.0-17.5 280 Width 5.41 2.26 1.1-13.4 275 Thickness 1.47 1.07 1.0-6.8 308 Platform thickness .79 .70 0.1-4.8 283 Platform breadth 2.13 1.55 0.1-9.0 272 * All measurements in mm. most flakes have a single dorsal ridge (41%), although up to 4 dorsal ridges may be present; (4) cortex is absent from fully 65% of flakes, and where present tends to be located on the dorsal surface; (5) flake edges are usually divergent (36%) or irregular (26%), with lesser frequencies of parallel, convergent, ovate, or diamond edges; and (6) flake terminations are dominantly feather (52%), with lesser frequencies of hinge, step, snap, and bibulbar types. All flakes from all three sites were also closely examined (under 1OX binocular microscope) for evident signs of use wear. Only nine flakes clearly displayed evidence of use, in the form of micro flake scars, in all instances along a single edge. Eight of these flakes came from Site D6-60, and one flake from D6-58. HISTORIC ARTIFACTS All three rockshelters yielded artifacts of Euro- American manufacture from the upper 5-10 cm of cultural deposit, and from the site surfaces. The most frequent items in these assemblages of exotic artifacts are small glass beads of several types, gun flints (some of which have been reworked), and small sherds of bottle glass. A detailed analysis of these historic artifacts will form part of a larger work on the Anahulu Valley Project (Sahlins and Kirch, in prep.). REFERENCES CITED Barrerra, W., Jr., and P.V. Kirch. 1973. Basaltic glass artifacts from Hawaii: their dating and prehistoric uses. Journal of the Polynesian Society 82(2):176- 87. Bonk, William J. 1954. "Archaeological Excavations on West Molokai," M.A. thesis, University of Hawaii. Buck, P.H. (Te Rangi Hiroa). 1957. Arts and Crats of Hawaii. B.P. Bishop Museum Special Publication 45. Honolulu. Cleghorn, P.L. 1982. "The Mauna Kea Adze Quarry: Technological Analyses and Experimental Tests." Ph.D. Dissertation, University of Hawaii. (University Microfilms, Ann Arbor). Emory, K.P. 1946. Hawaiian tattooing. B.P. Bishop Museum Occasional Papers 18:235-270. Emory, K.P., WJ. Bonk, and Y.H. Sinoto. 1959. Hawaiian Archaeology: Fishhooks. B.P. Bishop Museum Special Publication 47. Honolulu. Emory, K.P., and Y.H. Sinoto. 1961. Hawaiian Archaeology: Oahu Excavations. B.P. Bishop Museum Spec. Publ. 49. Honolulu. Kirch, P.V. 1975. Excavations at Sites A1-3 and A1-4: early settlement and ecology in Halawa Valley. In P.V. Kirch and M. Kelly, eds., Prehistory and Ecology in a Windward Hawaiian Valley, pp. 17- 70. Pacific Anthropological Records No. 24. Departnent of Anthropology, Bishop Museum. Honolulu. . 1979. Marine Exploitation in Prehistoric Hawai'i: Archaeological Excavations at Kalahuipua'a, Hawaii Island. Pacific Anthropological Records 29. Dept. of Anthropology, B.P. Bishop Museum. Honolulu. ____. 1982. The ecology of marine exploitation in prehistoric Hawaii. Human Ecology 10(4):455-76. . 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 Schousboe, R., M.F. Riford, and P.V. Kirch. 1983. Volcanic-glass flaked stone artifacts. In J. Clar and P. Kirch, eds., Archaeological Investigations of the 122 The Portable Arifact Assemblages Mu gane-Wainea-Kawaihoe Road Corridor, Island of Hawaii, pp. 348-70. Dept. Anthropology Report 83-1. B.P. Bishop Museum. Honolulu. Sinoto, Y.H. 1962. Chronology of Hawaiian fish- hooks. Journal of the Polynesian Society 71(2): 162-66. . 1967. Artifacts from excavated sites in the Hawaiian, Marquesas, and Society Islands. In G.A. Highland et al., eds., Polynesian Culture History, pp. 341-61. B.P. Bishop Museum Special Publication 56. Honolulu. Soehren, LJ. 1966. "Hawaii excavations, 1965." Typescript in Dept. Anthropology, B.P. Bishop Museum. Honolulu. . ms. "Archaeological excavations at Nualolo, Kauai." Incomplete typescript report in Dept. of Anthropology, B.P. Bishop Museum. Honolulu. Titcomb, M. 1952. Native Use of Fish in Hawaii. Polynesian Society Memoir 29. Wellington. Tuohy, D.R. 1965. "Salvage excavations at City of Refuge National Historical Park, Honaunau, Kona, Hawaii." Mimeo. B.P. Bishop Museum. Honolulu. 123