CHAPTER THREE A GEOARCHAEOLOGICAL ANALYSIS OF SEDIMENTS FROM THE ANAHULU VALLEY ROCKSHELTERS by Terry L. Hunt T HE PRMARY GOAL OF GEOARCHAEOLOCAL research in Anahulu Valley rockshelter sites is to describe and examine the sedimentary matrix of the sites in order to infer the history of deposition (see Stein and Farrand 1985; Stein 1987). Stein (1985, 1987) has recently discussed general principles of sedimentation as a useful structure for studying archaeological deposits. General principles are based on a "life history" concept where a sediment has (1) a source, (2) a transport history, (3) an environment of deposition, and (4) undergoes post depositional alterations. These principles are used here to interpret the sedimentary matrix and its relation to artifactual and paleoecological data from the rockshelters (see table 3.1, adapted from Stein and Rapp 1985:144). ANALYTIC METHODS During the course of excavations, 1 to 2 kg of sediment were systematically sampled from exposed stratigraphic profiles in at least one location from each site (see chapter 2). This sampling strategy not only included multiple samples of most stratigraphic units, it also enabled various differences within the major strata to be analyzed. An external control sample was also taken so that sediment characteristics between natural and archaeological deposits could be compared (Stein 1985). In the laboratory, sediment samples were allowed to air dry (in the bags from the field). Small sub- samples were taken from the air-dried field samples for various analyses. These sub-samples, ranging from as little as 5 grams up to ca. 40 grams, were made using a Jones sample splitter, thus assuring a non-biased selection of the particles present in the larger samples. A basic goal was one of conservative use of the field samples-thus reserving much of the sediment for analyses not performed in this project or for any necessary re-analysis. Particle Size Analysis Sub-samples for particle size analysis generally require a portion of 20 to 40 grams, depending on the size classes present (Folk 1965). All of the sub-samples for particle size analysis were pretreated to remove organic matter (Folk 1965:17). Prior to pretreatment, the sub-sample was inspected with a low power binocular microscope for materials such as plant macrofossils, landsnails, and artifacts that might be chemically or mechanically damaged or destroyed. Pretreatment for removal of organic matter assures the break up of organically consolidated particles and removal of recent organic (e.g., leaf fragments, etc.) as well as chemical precipitates. The practical effect of pretreatment for organic matter is to keep clay minerals in dispersion and to preclude analytic measurement of debris not thought to relate directly to depositional or post-depositional processes. Jackson (1969) has described the method for removal of organic matter. However, sodium hypochlorite (NaOCl) was used as a much less expensive altenative to hydrogen peroxide, and it has proven just as effective with less danger of violent Ana4Auu TABLE 3.1 A MODEL FOR GEOARCHAEOLOGICAL RESEARCH IN ANAHULU VALLEY Principles of Geological Attributes of Analyses Used in Sedimentation Interpretationl Sediment Anahulu Study2 1) Source defing areas from which textural and mineral texture, including grain sediment derives; maturity; uniformity/ size distributions; point weathering processes composition; mineralogy counts; microartifact counts 2) Transport water, wind, and gravity texture (size distributions and texture, especially grain Mechanism as agents grain morphology and surface size distribution; artifacts structure); structure of the (micro- to non-portable) deposit 3) Environment sites of deposition after texture and structure; texture (and stratigraphic of Deposition transport, e.g., valley geomorphic expressions of context) floors, caves, dunes, etc. the unit 4) Post- lithification; pedogenesis; measuring the additions, pH; organic matter and Depositional family of turbations removals, transfers, and carbonate analysis; color; Alterations transformations within soil texture using chemical and physical analyses 1 Adapted from Stein and Rapp (1985:144); they add archaeological interpretations primarily by adding humans as agents in the life history of a sediment. 2 Analyses relating directly to one phase of life history or attributes of a sediment overlap and have important bearing on other aspects of a sediment. reaction (see Briner 1963; Anderson 1961; Lavkulich and Wiens 1970). Commercial Chlorox liquid bleach (NaOCL) was adjusted to a pH of 9.5 by adding a small (variable) amount of hydrochloric acid (HCI). This procedure is necessary, as a high alkaline solution may attack clay particles (Jackson 1969:37). All of the Anahulu particle size sub-samples received the same basic treatment (following Jackson 1969). Subsequent to pretreatnent for removal of organic matter, the sediments were inspected under a low power binocular microscope in order to ascertain the presence of any remaining organically bound aggregates. This examination revealed a range of fine sand to coarse silt size aggregates comprised of clay and bound by free iron oxides. Such pedogenic aggregation of particles would bias measures of particle size distributions in the direction of large phi classes. As Kunze (1965:574) points out, iron-containing samples are difficult to properly disperse without removal of iron oxides. Thus, a second preteant for free iron oxide removal was undertaken on all Anahulu rockshelter samples using the sodium dithionite-citrate procedure (Kunze 1965:575-76 and references therein). Following the pretreatment for removal of organic matter and free iron oxides, samples were thoroughly rinsed using distilled water and then oven dried (under 100? C). The dry sediment was then added to flasks with 150 mls of (NaPO3)6 and mechanically shaken vigorously for five minutes after which it was left to stand for a minimum of 24 hours to assure that no flocculation would occur. The sample was then wet screened (4 phi screen), separating sand from the fmer fraction. The course fraction was dried and sieved into phi intervals for measurement (grams to nearest .0001) using an electronic balance. The fine fraction was added to a dispersing agent and analyzed using the pipette metiod (phi intervals are sampled based on differential settling imes and measured to .0001 grams; see Stemnberg and Creager 1961 for a comparison with hydrometer method). Statistics for particle size 44 Geoarchaeological Study distributions were calculated and classified after Folk (1974) using a computer program written specifically for this purpose. These distributions were plotted as frequency curves and as cumulative curves (Reineck and Singh 1980:132-34). Such statistics in conjunction with other data form the basis of interpretng "the energy of the depositng medium and the energy of the basin of deposition" (Reineck and Singh 1980:132). In other words, particle size infonrs on the transport mechanism(s) and environment of deposition as determined by available source material. The particle size distributions for the Anahulu rockshelter sites are given below and interpreted in tems of this general sedimentological model (Folk 1965; Folk and Ward 1957; Visher 1969). Grain Shape and Surface Texture The shape (roundness, sphericity) of sedimntary particles ideally records distinguishing features of origin, transport agent, and depositional history (Shackley 1975). These featum are recorded under magnification and can be assigned to ordinal level classes on visual criteria (e.g., Powers 1953). Estimation of grain shape or rounding has proven quite successfuL particularly in the long temporal sequences of French and African rockshelters (e.g., Fairand 1975). Anahulu samples (prepared with grains of 4 phi class) were examined for patterns of grain rounding under varying magnification (40-800X) on the scanning electron microscope (SEM) (see Gillieson 1983). Organic Matter and Carbonate Analysis The loss-on-ignition method is a means to estimate the percentage of organic matter and carbonates of a sediment sample (Stein 1984). The theory of the ignition loss method is that when a dried powdered sample is heated, organic matter will begin to ignite at 200?C and completely burn off when the temperature reaches 550?C. Calcium carbonate (CaCO2) evolves into carbon dioxide at 800?C and is completed at approximately 850?C (Stein 1984). Thus, to measure organic matter and carbonate percentages, the weight difference of an oven dried sample (ca. 5 grams) before and after a 550?C and a 1000?C burn is measured to the nearest .0001 grams. The method used here is described in more detail by Dean (1974) and by Stein (1984), who also discusses applications in archaeology. The Anahulu samples were analyzed for organic matter and carbonate percentages to describe and document pans of potential culural and naul addions and pedogenic transfers within and across strata of the rockshelter deposits. pH Analysis The acidity, neutrality, or alkalinity of sediment is msured in pH levels. Measurements of pH in the Anahulu samples were made on an autmatic Altex 70 pH meter. A minimum of dtree readings was taken once the sample stabilized (20 ml sedinent with 20 ml distilled water), with the pH value calculated as an average. This common method yields data on the chemistry of a depositional matrix and can be used to assess factors of preservation (e.g., Gordon and Buikstra 1981). For the Anahulu samples, pH data were recorded in order to assess any potential of differential preservation and to relate pH to other pedogenic factors. Color Sedimnt color, recorded using the stadadized Munsell system, provides information on source (pimay coloration) and/or may reflect diagenetic causes such as biochemical alterations (Shackley 1981). The color of Anahulu samples was recorded as air dried and moist under laboratory conditions and is included primarily for descriptive purposes. Microartifact Analysis A method designed to detemine quantitatively the percentage of various minerals present in a rock (Galehouse 1971) has recently been extended to estimate the quantity of"microartifacts" in archaeological sediments (Fladmark 1982; Stein 1987). This simple method involves mounting grains of individual phi classes (generally sand-sized classes) on glass slides, then identifying and counting all grains witiin a particular sized area (e.g., one square cenimeter; see Galehouse 1971 for variations in method). Samples from the Kuolulo Rockshelter (Column C) were examined (the 1 phi class) for microartifacts (e.g., lithic microdebitage), including bone, shell, and charcoal fragments, with a minimum of 300 grains counted. From these counts, the frequencies of various classes of material can be esima. These data potentially offer an independent line of evidence for ldnds of human activity associated with the Anahulu rockshelters. RESULTS AND INTERPRETATION Site D6-60, Column A Column A was taken from the central area of the Kuolulo rockshelter, Units M22/M23. The column copWrises 11 samples taken at 5 cm intervals with the column reaching a maximal depth of 55 cm below the 45 Anahadu TABLE 3.2 ANALYTIC DATA FOR COLUMN A (SITE D6-60) Sample Mean s.d. Organic Carbonates pH Color Depth Grain Size Matter % % Dry Moist (cm) (phi) 0-5 3.70 5.20 18.02 10.32 8.18 5YR 3/3 5YR 2/2 5-10 5.49 5.25 15.12 12.61 8.20 5YR 3/2 SYR 2/1 10-15 5.49 4.98 16.52 13.05 8.39 5YR 3/2 SYR 2/1 15-20 4.24 5.15 11.25 17.42 8.45 1OYR 3/2 SYR 2/1 20-25 5.10 5.31 13.60 14.35 8.58 10YR 3/2 SYR 2/1 25-30 5.76 5.05 13.50 13.21 8.52 10YR 3/2 SYR 2/1 30-35 3.39 4.43 15.29 14.43 8.55 10YR 3/2 SYR 2/1 35-40 4.15 5.58 13.83 11.97 8.44 10YR 3/2 SYR 2/1 40-45 2.78 4.76 13.44 10.61 8.50 10YR 3/2 5YR 2/1 45-50 4.85 4.64 12.62 9.96 8.56 10YR 3/3 5YR 3/1 50-55 4.41 4.95 11.03 4.39 8.45 SYR 4/4 SYR 3/4 surface of excavation. Three primary stratigraphic layers were recognized in the field and are described in chapter 2. All the analytic data for Column A are summarized in table 3.2. Representative (upper, middle, and basal samples) particle size distributions for Column A are shown in percent and cumulative curves in figure 3.1. The mean particle size ranges from a minimum of 2.78 phi (40-45 cm) to 5.76 phi (25-30 cm) with standard deviations ranging from 4.43 (30-35 cm) to 5.58 (3540 cm). All of these size distributions reflect extremely poor sorting in terms of the criteria described by Folk (1965:45-6). All the samples from Column A can be described as gravelly mud or gravelly muddy sand. With respect to the particle size distribution model (see Reineck and Singh 1980:132-38), size characteristics of three subpopulations (rolling, saltational, and suspension) are mixed. This fact indicates that: (1) the primary sediment source was most likely poorly sorted, as is typical of immature sedimentological environments such as Anahulu; (2) the most probable primary mode of transport was downslope surficial flow of colluvium, (i.e., not stictly aeolian or fluvial transport) carrying a mixture of gravel, sand, silt, and clay sized particles; a secondary mode of deposition was probably roof-fall with a contribution of angular particles of various size classes; (3) the environment of deposition remained relatively constant and stable; and (4) post-depositional alterations have probably been those associated primarily with pedogenic factors as opposed to sedimentological processes. Comparison of the Column A particle size data with the control sample suggests that the added source of sediment deposition is roof-fall (the gravel constitutent in the samples). The data from other analyses provide results that are consistent with interpretations made on the basis of particle size distributions. Examination of grain shape and surface morphology revealed little variation from one sample to another and systematic analyses were not carried out. Inspection of the particles did show the predominance of angular and non-spherical forms, indicative of an immature age for the sediment. The loss-on-ignition analyses for Column A show that organic matter is an important component in the sediment at almost all depths. The continuing addition of organics at the surface of the deposit (i.e., leaf litter, etc.) is comparable at greater depths by proportions of organic matter that must relate to human sources during the rockshelter's occupation. The lowest percentage of organic matter is at the base of the deposit, congruent with this interpretation. The percentage of carbonate throughout the column is comparable to that of organic mattr, and can be ineted as the similar effects of human agents in introduction of materials-especially shellfish and bone. While there are carbonates found in the non-archaeological control sample, the proportion is lower than that in the cultural deposits. 46 OR-O-0 COL A 0-5 cM Geoarchaeological Study aR-0S-6 CM. A 0-5 CM @ z -2 0 2 4 6 a 10 12 14 --2 0 2 4 6 8 PHI INTERVALS PHI INTERVALS OR-06-60 COL A 20-25 Cm OA-06-6 COL A 20-25 cM 15- ol z z 0. -2 0 2 4 6 8 PHI INTERVALS OR-06- COL R 50-55 CM 10 12 14 PHI INTERVALS OA-0S-W CL. A 50-55 c0 PHI INTERVALS PHI INTERVALS Figure 3.1. Grain size (phi scale) distributions for representative samples. from Site D6-60, Column A. 47 z 10 z w z 20r Anahulu The pH values from Column A are highly consistent at all levels. These moderately alkaline conditions result from the combined factors of natural conditions (compare with control sample pH of 7.65) and human additions of alkaline materials such as shell. The consistency of these values indicates that differential preservation due to variable pH is not a factor in archaeological and paleoecological analyses for this site. Site D6-60, Column C Column C (Unit D20) was taken from a deeper area of the Kuolulo rockshelter deposit than that represented in Column A. This column comprises 17 samples taken at 5 cm intervals with the column reaching a maximal depth of 85 centimeters below surface. All analytic data for Column C are summarized in table 3.3. Representative (upper, middle, and lower samples) particle size distributions for Column C are shown in percent and cumulative curves in figure 3.2. The mean particle size ranges from a minimum of -0.83 phi at the base of the deposit (80-85 cm) to 3.98 phi (35-40 cm) with standard deviations ranging from 4.14 (25-30 cm) to 5.37 (0-5 cm). All of the samples show extremely poor sorting after Folk (1965:45-6). As with Column A the samples from Column C can be described in textural terms that include gravel, sand, and mud combinations. The size characteristics of three subpopulations are again represented and are mixed. The intelpretation made for sources, transport agents (primarily coliuvium with additions from roof-fall), the environment of deposition, and post-depositional alterations is largely comparable for those columns (A and C) of the same site. One difference that can be detected (see figs. 3.1 and 3.2) is the greater contribution of gravel sized sediment at the bottom and top of Column C. This greater proportion of gravel in the deposit contrasts with the sand and silt modes of most of the deposit and indicates a relative change in the importance of roof-fall as opposed to the deposition of mud by surficial flow. In addition to particle size analyses on samples pretreated for iron removal, a second set of samples was analyzed with pretreatment for the removal of organic matter only. The distributions shown in figure 3.2 (dashed lines) reveal the bias that iron fomed aggregates impose on particle size statistics. For example, the effect of iron bound particles is not constant across all the samples, but must vary with pedogenic factors. This is an important observation with implications for all sedinentological analyses performed in Hawaiian or other Oceanic environments. The disaggregation from iron pretreatment is of mostly clay and silt sized particles. Variation in the distributions shown in figure 3.2 may also reflect, to some unknown degree, the unavoidable sampling error between two sediment samples drawn from the same population. The results obtained from other analyses of Colulmn C samples are also comparable to those discussed for Column A. Organic matter and carbonates show similar patterns of variation to that of Column A (suggesting reliable experimental results from strongly similar sediments). As table 3.3 illustrates, pH and color closely correspond from one column to the other. Samples from Column C (representing the deepest deposit sampled for analysis) were examined for variation in grain shape and surficial texture of particles. Four samples selected from the top (0-5 cm), near the middle (25-30 cm and 45-50 cm), and at the base of the deposit (80-85 cm) were prepared at the 1.0 and 4.0 phi intervals for scanning electron microscope (SEM) inspection. The sediments prepared were not subjected to any of the pretreatments done for particle size analysis. SEM examination at magnifications ranging from 1oX to 800X revealed no definitive quantitative differences with respect to classes of grain shape and roundness (e.g., figs. 3.3-3.6). In other words, the four samples from the column could not be objectively distinguished on shape and roundness criteria. Surficial texture has been primarily worked out for quartz grains, and did not apply to the weathered volcanic (virtually quartz-free) sedinents found in the Anahulu deposits. Site D6-36, Column B Column B (Unit K18) was taken from a central area of the Ke'eke'e Iki rockshelter deposit. This column comprises 10 samples at 5 cm intervals with the column reaching a maximal depth of 50 cm below the surface. All analytic data for Column B are summarized in table 3.4. Representative (upper, middle, and basal samples) particle size distributions for Column B are shown in percent and cumulative curves in figure 3.7. The mean particle size ranges from a minimum of 1.13 phi (10-15 cm) to 5.50 phi (3540 cm) with standard deviations ranging from 3.91 (5-10 cm) to 4.99 at the base of the column (45-50 cm). The samples range from very to extremely poorly sorted (Folk 1965:45-6). As with the other column samples, these sediments can be described in textural terms that include gravel, sand, and mud combinations. The particle size distributions of Column B are more clearly bimodal than the other two columns. Gravel prrtons are relatively low, with sand and silt modes. The source of these sediments is primarily surficial flow of sand and silt sized particles. The potential contribution from rockshelter roof-fall at this 48 Geoarchaeological Study 0000~~ 0 0-0 00 0 00 oS Eq~~~~~~~~~~~~ o o N o14 co o o- ov t- ooo oW oq C W > >o ;i. o as eq WI t- t- %O t- mo q v W) ~Q& CSC~ 0000 N00000000 enC1-W t- -- - t--- -- i i ce N6 00 i O 6 C- > 6 W; 0 W; i 6 O i O t W- W- W-0 V0- 4 "- V- W- W- W- P- "- ? t0 b 0 m-r aN 8a v-Is 00i Irn "rI a"q O 4" t-~~ ~ Ne V-4 r-. to 9-4 W -|t t - o t r o e ^ o me-4 N V A e c vE m t q6 "1 -e 4 a oN N oo IV 42 cr G 2 C oE _4 t _ 00 CO aN r} CrEUn t- enso \0 - 00 e 4e ce; cr; C ce; ce; ;(i C-s C4 C4 C on a k Q in oD t o tn o tn o Q o in 0 o- C4 0 e T tn UN m 00 00 S-i A , l i _ A , OS 6 S ? * 6 %A V4V4 m m v v W n o t - 49 (A H ill *i,1 c-q Ca . 0 0 6hi 0 0 hi4 U o'se: 'a C z 0 c0 z V z WI 0 cis I 00 U 2 0C *S .,4. 9 oo4 .o Anahulu maU m-as-o-C O-S Ca Ieuo Is P1ETReAW WLU t-05-O-C 0-5 CM IUN/IC IRO PRETREATMNT 14 z W ..z Wt -2 0 2 4 6 8 10 12 PIll INTERVALS IN U OR-0S-lO-C 40-45 CM IIND ION PRETREATMENT I p I, 14 PHI INTERVALS RCJLU OR-OS-6D-C 40-45 CM IRON IRON RETREATMENT PHI INTERVALS SWILU OR-0S-SO-C 75-60 CM IRONO IRON PRETREATMENT PHI INTERVALS RNfIULU OR-0S-60-C 75-60 CH IRON/NO IRON PETRtEATMENT z I I. z I I 1- a r 0 2 4 6 8 PHI INTERVALS 10 12 14 4 6 8 PHI INTERVALS 10 12 14 Figure 3.2. Grain size (phi scale) distributions of representative samples from Site D6-60, Column C; solid lines show samples pretreated for iron removal, dashed lines indicate samples with no iron pretreatment. I Q I I so Geoarchaeological Study a b Figure 3.3. SEM photomicrographs of 1.0 phi grains from Site D6-60, Column C, 0-5 cm below surface: (a) sand grains at 15X magnification; (b) sand grains at 50X magnification; note grain angularity and surface textures. 51 Anahulu a b Figure 3.4. SEM photomicrographs of 1.0 phi grains from Site D6-60, Column C, 45-50 cm below surface: (a) sand grains at 30X magnification; (b) sand grains at 60X magnification. 52 Geoarchaeological Study TABLE 3.4 ANALYTIC DATA FOR COLUMN B (SITE D6-36) Sample Mean s.d. Organic Carbonates pH Color Depth Grain Size Matter % % Dry Mo is t (cm) (phi) 0-5 2.19 3.81 12.73 11.18 8.15 7.5YR 3/2 lOYR 2/1 5-10 3.36 3.33 12.82 12.57 9.00 7.5YR 3/2 lOYR 2/1 10-15 1.92 5.03 15.17 11.74 9.01 7.5YR 3-4/2 lOYR 2/1 15-20 4.18 4.45 14.25 14.48 9.17 7.5YR 3-4/2 lOYR 2/1 20-25 4.23 4.37 13.37 17.15 9.05 7.5YR 4/2 lOYR 3/1-2 25-30 5.56 4.35 12.97 15.20 9.12 lOYR 4/2 lOYR 3/1 30-35 5.11 4.54 12.69 14.54 9.12 lOYR 4/2 lOYR 3/1 35-40 5.79 4.36 13.34 14.82 9.17 lOYR 4-5/2 lOYR 3/1 40-45 5.47 4.50 12.48 11.79 9.13 lOYR 4/3 lOYR 3/2 45-50 4.93 4.99 9.11 16.00 9.10 lOYR 5/3 lOYR 3/2 Figure 3.5. SEM photomicrograph of 4.0 phi (fine sand) grains from Site D6-60, Column C, 45-50 cm below surface; 60X magnification. 53 Anahiu l __~~~~~ I_ _ *1_~~ Fgr3..SM hooIcrgah f40pi(iesn)gan rmSt D6-6 1 _ounC,8-5c elwsrae (a fin sngrisa Xmanfcto,ntanuatofgan,()fn sadganat2Xmanfctn,oe surface textures. 54 OR-M-S CM. 0 cm I. I Geoarchaeological Study OR-O-U CoS B 0-5 CM * ~~~~100 90 80 70 5 /c 3 SG I _ PHI INTERVALS 0-O6-36 CML B 20-25 CM I. Z, PHI INTERVALS OA-O-36 COL B 20-25 cm PHI INTERVALS OA-0O-36 COL B 40-45 CM PHI INTERVALS OR-06-36 CL. B 40-45 CM 20, 0a bi10 z oIL - -- - - -2 0 2 4 6 a 10 12 =2 0 2 4 6 8 10 12 PHI INTERVLS PHI INTERVALS Figure 3.7. Grain size (phi scale) distributions for representative samples from Site D6-36, Column B. 55 z z L Analudu site is apparently not as great as at Kuolulo rockshelter. The environment of deposition and post-depositional alterations appear to be comparable to Columns A and C, as evident in comparing tables 3.2, 3.3, and 3.4. Site D6-58 Miscellaneous Samples The Ke'eke'e Nui rockshelter (D6-58) was riddled with a complex series of subsurface features. As a result, no uninterrupted column representative of the sedimentological sequence could be obtained. Instead, samples were taken from various depths that reflected the depositional history independent from the constructional activities of prehistoric features. Table 3.5 provides the provenience and analytic data for the six samples analyzed from the Ke'eke'e Nui rockshelter. Representative particle size distributions (fig. 3.8) are typical of those from the columns described above. The one exception is the basal deposit (fig. 3.8, bottom) that has a very strong mode in fine sand and coarse silt-sized sediment. This clearly represents fluvial transport in a depositional environment where finer sediments (fine silts and clays) were carried away in suspension. Subsequent to this depositional event, poorly sorted sediment, comparable to the other contexts analyzed, predominates. Other analyses, even though these samples do not derive from a continuous column sample, provide results comparable to the other sites reported. CONCLUSIONS As Farrand (1985:21) has recently pointed out, rockshelters commonly form sedimentological traps with associated prehistoric human activities. The resulting deposits are typically complex, polygenetic mixtures of immature sediments. The Anahulu rockshelters fit this description well. The general conclusions that can be drawn from analyses of the samples from the three rockshelters include: (1) The primary source of sediment is immature colluvium. (2) This sediment was transported in surficial flow (erosional and redepositional processes). These sediments remain very to extremely poorly sorted in their transport and their depositional context. A secondary source of sediment is roof-fall from the continuous weathering of the basaltic outcrops that form the rockshelters. A third factor recognized in an earlier study (Kirch 1979) is the aeolian deposition of silt and clay sized particles from the adjacent Ewa Plain sugarcane agriculture. Pedogenic factors (not examined in this study) are operative, even over short durations of time, and may account for some variation in the proportion of clay at various depths of the deposit (processes of eluviation and aggregation, e.g., free iron and clay). TABLE 3.5 ANALYTIC DATA FOR SAMPLES FROM SITE D6-58 Sample Mean s.d. Organic Carbonates pH Color Depth Grain Size Matter % % Dry Mo ist (cm) (phi) 10-15 5.88 4.64 13.81 7.40 9.94 7.5YR 4/2 lOYR 2/1 15-20 5.26 4.68 11.96 14.45 8.79 7.5YR 4/2 lOYR 2/1 20-25 5.60 4.71 18.17 12.15 8.43 7.5YR 4/2 lOYR 2/1 35-40 4.98 5.04 10.20 14.18 9.37 7.5YR 4/2 lOYR 2/1 40-45 (pit fill) 4.71 4.55 9.52 17.30 8.94 lOYR 4/2 lOYR 3/1 58-63 4.21 3.19 10.44 4.81 8.57 lOYR 4/3 lOYR 3/2 (basal layer) 56 Geoarchaeological Study M-MS-U KIG E 10-15 CM 044O-511 KIO E 10-15 CM !2 0 2 4 6 a 10 12 14 PHI INTERVALS OR0-58 KIO B 45-50 CM. PIT FILL I.- z A. -2 0 2 4 6 8 10 12 14 PHI INTERVFLS OR-06-5- KIO C 50-1 Ch. SIM LAYER I 0 2 4 PHI INTERVALS O0-8S- K10 B 45S-S CM. PIT FILL 0 2 PHI INTERVALS OR0-5 K10O B 5-63 CM. ORSAL LAYER PHI INTERVALS PHI INTERVALS Figure 3.8. Grain size (phi scale) distributions for representative samples from Site D6-58, Unit K10. 14 z IN I- Q.I 14 57 Anahulu (3) The environment of deposition has remained comparable in the rockshelters and there is no evidence for any major erosion or displacement of material once deposited. (4) Post-depositional changes are more difficult to assess, especially given the poorly sorted sediments comprising all the sequences studied. As mentioned above, pedogenic factors, such as the transfer of clay across a forming soil profile, and chemical reactions, are those processes most visible in analytic terms. This does not preclude the possibility that an array of bioturbations have not affected the deposits to some degree (see Wood and Johnson 1978). The major depositional episode represented in the rockshelters appears to begin with human occupation or use of the valley and continues at a more or less constant rate throughout the archaeological sequence. Thus, human activities must be interpreted as the fundamental cause behind sedimentological changes analyzed at the scale of the rockshelters alone. Removal of forest vegetation would allow for episodes of erosion and redeposition. REFERENCES CITED Anderson, J.U. 1961. An improved pretreatment for minerological analysis of samples containing organic matter. Clays and Clay Minerals 10:380- 88. Briner, G.P. 1963. Survey of clay minerals in some Victorian soils. Proceedings of the Royal Society of Victoria 77(1):191-95. Dean, WE., Jr. 1974. Determination of carbonate and organic matter in calcareous sedimentary rocks by loss-on-ignition: comparison with other methods. Journal of Sedimentary Petrology 44:24248. Farrand, W.R. 1975. Sediment analysis of a prehistoric rockshelter: the Abri Pataud. QuarternayReserach 5:1-26. ___._ 1985. Rock shelter and cave sediments. In J.K. Stein and W.R. Farrand, eds., Archaeological Sediments in Context. Peopling of the Americas, 1:21-39. Center for the Study of Early Man, Orono. Fladmark, K.R. 1982. Microdebitage analysis: initial considerations. Journal of Archaeological Science 9:205-20. Folk, R.L. 1965. Petrology of Sedimentary Rocks. Hemphill Publishing Co., Austin, Texas. Folk, R.l., and W.C. Ward. 1957. Brazos River Bar: a study in the significance of grain-size parameters. Journal of Sedimentary Petrology 27:3-26. Galehouse, J.S. 1971. Point counting. In R. Carver, ed., Procedures in Sedimentary Petrology. pp. 385- 407. Wiley and Sons, New York. Gillieson, D. 1983. Geoarchaeological applications of scanning electron microscopy: some Australian examples. The Artefact 8(1-2):43-54. Gordon, C.C., and J. Buikstra. 1981. Soil, pH, bone preservation, and sampling bias at morturary sites. American Antiquity 46:566-7 1. Jackson, M.L. 1969. Soil Chemical Analysis: An Advanced Course. 2nd ed. Department of Soil Science, University of Wisconsin, Wisconsin. Kirch, P.V. 1979. Late Prehistoric and Early Historic Settlement-Subsistence Systems in the Anahulu Valley, O'ahu. Deparent of Anthropology Report 79-2. B.P. Bishop Museum, Honolulu. Kunze, G.W. 1965. Pretreatment for mineralogical analysis. In C.A. Black, et al., eds., Methods of Soil Analysis, pp. 569-77. American Society of Agronomy, Madison. Lavkulich, L.M., and J.H. Weins. 1970. Comparison of organic matter destruction by hydrogen peroxide and sodium hypochloride and its effects on selected mineral constituents. Soil Science Society of America Proceedings 34(5):755-58. Powers, M.C. 1953. A new roundness scale for sedimentary particles. Journal of Sedimentary Petrology 23:117-19. Reineck, HE., and I.B. Singh. 1980. Depositional Sedimentary Environments. 2nd ed. Springer- Verlag, Berlin. Shackley, M.L. 1975. Archaeological Sediments: A Survey of Analytical Methods. John Wiley and Sons, New York. . 1981. Environmental Archaeology. George Allen and Unwin, London. Stein, J.K. 1984. Organic matter and carbonates in archaeological sites. Journal of Field Archaeology 11:23946. . 1985. Interpreting sediments in cultural settings. In J.K. Stein and W.R. Farrand, eds., Archaeological Sediments in Context. Peopling of the Americas 1:5-19. Center for the Study of Early Man, Orono. . 1987. Deposits for archaeologists. Advances in Archaeological Method and Theory 11:337-95. Stein, J.K., and W.R. Farrand. 1985. Context and geoarchaeology: an introduction. In J.K. Stein and W.R. Farrand, eds., Archaeological Sediments in Context. Peopling of the Americas 1:1-3. Center for the Study of Early Man, Orono. Stein, J.K., and G. Rapp Jr. 1985. Archaeological sediments: a largely untapped reservoir of information. In N.C. Wilkie and W.D.E. Coulson, 58 Geoachaeological Study eds., Contributions to Aegan Archaeology, pp. 143-59. Center for Ancient Studies, University of Minnesota, Publications in Ancient Studies No. 1, Minneapolis. Stemnberg, R.W., and J.S. Creager. 1961. Comparative efficiencies of size analysis by hydrometer and pipette methods. Journal of Sedimentary Petrology 31:96-100. Visher, G.S. 1969. Grain size distributions and depositional processes. Journal of Sedimentary Petrology 39:1074-1106. Wood, W.R., and D1. Johnson. 1978. A survey of disturbance processes in archaeological site formation. In M.B. Schiffer, ed., Advances in Archaeological Method and Theory 1:351-8 1. Academic Press, New York. 59