1. Introduction and Research Questions

This report examines and quantifies the usewear (use-related modification) observed on an assemblage (n=247) of obsidian artifacts collected at the Harney Dune site (35HA718) near Harney Lake, southern Oregon, USA. An evaluation of the site-formation processes at this archaeological site indicates that these tools were made, used and deposited by Wada'tika (Northern Paiute) people (Raymond 1994). Some pothunting and point-collecting has taken place on-site. USFWS investigation of the Harney Dune lithic scatter was carried out partly in response to such looting. Although dating of this surface scatter is problematic, projectile point typology seems to indicate a long period of occupation, from at least 6,000BP to the historic period, when Wada'tika were observed living in the area by Europeans (Raymond op cit) (obsidian hydration dates on these artifacts are pending at this writing). All evidence indicates that during this multi-millennial period, the site area was occupied and used many times, forming a large (7.7km²) palimpsest. A judgemental sample, drawing obsidian artifacts widely from all areas of the archaeological scatter, produced the assemblage exmained in this report (sampling strategy is discussed in section 6).

The research questions guiding this investigation center on the question of the function and design of these artifacts. We do not currently understand how the Harney Dune site was formed. Raymond (personal communication) has forwarded two alternative hypotheses. Hypothesis one states that the scatter is the result of seasonal aggregations of large groups of Wadaítika primarily to exploit seasonal richness in tui chub minnows of Harney Lake. Hypothesis two suggests that the site resulted from smaller populations, also seasonally aggregating at the site, but in lesser numbers and for longer periods, and for a wider range of purposes than posited in hypothesis one.

One method of evaluating these competing hypotheses is to reconstruct the activities carried out on-site (other approaches, such as examining palimpsest formation per occupation episode, are being investigated by Raymond and the myself). To do this, as well as to assist in a synthetic understanding how the site was formed in behavioral context (sensu Schiffer 1972), this usewear study was carried out in order to (a) identify common activities at the site and (b) to evaluate the stone tools of the assemblage in terms of utilitarian design. It must be emphasized that the usewear and design-analysis data will be two of several independent lines of evidence used, ultimately, to reconstruct the range and nature of native activities at Harney Dune.

I return to the evaluation of the expectations of hypotheses one and two in section 8: Conclusions.

2. Usewear Analysis: Theory and Methods

After production, artifacts may be modified by at least two factors. Usewear (also known in the literature as 'edge-wear' and 'microwear') is the modification of an artifact surface resulting from use as a tool (Semenov 1962, Keeley 1980, Shea 1991). Non-functional factors may also modify artifact surfaces. Actions such as trampling (by human and non-human agents) and retouching (by pressure-flaking or other methods) must also be considered in the analysis of artifact post-production modification (Tringham et al 1974).

The understanding of both use-related wear and such factors as trampling and resharpening is currently carried out primarily by an experimental approach. Experimental artifacts are produced and used in a range of activities though to be broadly reflective of the activities of the archaeological group who generated the archaeological assemblage; the wear observed on the expeimental tools is compared with that observed on the archaeological tools; like wear is considered to be a result of like actions, and conclusions are made regarding archaeological function based on the comparison with experimentally-generated wear.

A number of factors may effect the validity of usewear analyses. First, because in the realm of stone tools a given artifact could have potentially been used for any of a great variety of activities, it is important to consider the environment and resources of the archaeological society under study, to constrain the variety of experimental actions carried out. This may be more straightforward than it at first appears: palaeoenvironmental, archaeozoological and archaeobotanical data may be used to reconstruct plant and animal resources most likely to have been exploited in the past. Other items of material culture (as well as features) may be used to infer what sorts of tools and work actions were necessary to produce these tools and/or features. Also, ethnohistoric data may be used to inform on the types of tools and activities carried out in a certain place and time. All of these approaches are used in this report to best model the most likely behaviors of the Wada'tika at Harney Dune.

A second factor to consider in usewear studies is that of the raw material used in the experiments; it should replicate that used archaeologically because different raw materials may have been used for different activities, and because wear is generated at different rates and in different modes depending on raw material. In this study, I examine only obsidian artifacts, and I use only obsidian for experiments.
 
 

3. The Experimental Tool Project and Artifact Examination When the factors noted above have been considered and accomodated in the research program, usewear analysis may proceed. At this stage, usewear analysis has three main goals: (1) the determination of whether or not a given artifact was utilized in the past, and if it was used, (2) the determination of the work action (scrape, cut, shave, etc.) and (3) the determination of what material was worked by the tool (wood, animal tissue, etc.). These basic analytical goals were introduced by Semenov (1964), formalized by Keeley (1980) and have been widely and profitably used in analyses of variety of contexts, from Lower to Upper Palaeolithic and more recent periods.

To achieve these goals, a three-step process is normally employed by the investigator. The general goal is to compare usewear on experimental tools with that on archaeologically-derived artifacts, in an attempt to determine the past use of the artifacts. A two-step, modified version of this analytical procedure was carried out in this investigation; reasons for the modification are reported below.

Step 1: Use of an Experimental Tool Set

The first step is the creation and use of a set of experimental tools broadly technically- and functionally-analagous with those of the assemblage being studied. Technical analogy is achieved by studying the technical characteristics of the assemblage (raw material, core and flaking types, and so on).

In this study, the assemblage was largely composed of flakes struck from moderately-prepared cores. Many of these flakes have a distinctive 'blade-like' shape, being semi-rectangular with straight or moderately curved (convex and/or concave) edges between the proximal (striking) and distal (flake terminus) ends. The bulk of non-blade-like items were also flakes, again detached with a hammer from a moderately-prepared core. Angular and shatter items ('core fragments') are very uncommon.

The technical characteristics of artifact production, then, were easily replicated for the experimental tool set (by A. Raymond): flakes, many of them blades with roughly parallel edges, were hard-hammer-struck from moderately-prepared obsidian core platforms. As on the artifact assemblage, retouch prior to use was not present (except in one case noted below) and unmodified blades were used for the experiments.

Functional analogy is achieved by a contextual consideration of the assemblage. An analyst working with a desert- or arctic-derived assemblage has little warrant to include heavy and extensive wood-working activity in the experimental tool set, while an assemblage from an agricultural community may be expected to contain harvesting tools. By a contextual consideration of the assemblage, then, which includes an evaluation of hafting, tool size and the most likely activities carried out given the prevailing environment, the analyst estimates the activities most likely to have been carried out by the prehistoric population, and in the experimental tool phase such activities are replicated.

I proceed below with a general modeling of working of the most likely-worked raw materials, in this case wood, plant material, and animal tissues (as opposed, for example, to antler, ivory, sandstone, and so on).

Wood-Working

Considering the geography and enviroment of the Harney Basin, and the region in general, we note first that it is an arid desert scrubland characterized by a lack of trees; heavy woodworking of fresh wood, then (such as scraping and whittling), may be considered unlikely to have been carried out by the Harney Dune Paiute. Still, these people would have used some wooden tools (e.g. bows, fire-starting dowels) which may have required maintenence (e.g. re-sharpening, removal of splinters, smoothing). For this reason, some woodworking on dried (as well as a small amount of fresh wood) was carried out in the experimental activity set.

Plant-Working

Some historic photos of Paiute homes clearly show houses built largely of wetland reeds, such as cattail and bulrush; this would suggest a large amount of plant-processing which may have been carried out with the stone tools of this assemblage. The the Wada'tika appear, however, to have used different building materials: in 1826, Ogden noted on the North shore of Harney Dune the use, primarily, of 'worm wood or grass' for domestic structure construction (Ogden 1826 in Davies 1961:21). The ëworm woodí mentioned must refer to the only native wood in the area, greasewood (Sarcobatus spp.), while the grass likely refers to Indian Rice grass (Oryzopsis heminoides) and/or salt grass (Distilclis spp.). There is little cattail or bulrush on the N, NE or E margins of Harney Lake today, and although it may be absent due to exhaustion by the Paiute, it was not prevalent even at the time of Ogden's 1826 visit.

We may extraplolate that cattail / bulrush house construction was relatively rare and should not be much represented in the studied tool assemblage. More likely in terms of plant-processing was the use of blades to cut fresh Indian Rice grass (for house construction activities) and fresh 'waada' (Suaeda depressa), the chenopod of which the seed was one staple of the Wada'tika diet. Additionally, the construction of tools such as fishing nets (an important tool at Harney Lake) would likely have included the cutting of fresh vegetal matter for a variety of net-making and other purposes. For these reasons, the experimental tool set included cutting of fresh vegetal matter, including ragweed (Ambrosia artemissifolia), stinging nettles (Urtica spp) and jointed goatgrass (Aegilops cylindrica).

Animal Tissue

Based on the ecology of the region, the animals exploited by the Wada'tika at Harney Dune would have been dominated by fish (mainly the 15-cm tui chub (Gia bicolor)) small mammals, such as rabbits and ground squirrels, and birds. Therefore, experimental tools were used to butcher fish (small fresh trout), a ground squirrel, fresh larger mammal meat on the bone (beef) and a bird (chicken). Ethnohistoric data suggest that bone and antler artifacts were rare among the Wada'tika (Raymond 1994): for this reason, only a few bone and antler raw materials were worked with the experimental tool set. As commonly practiced aborigonally,bone and antler were soaked in water (to soften them) for a period of 24-48 hours before working.

All of the experiments were carried out to understand the general nature of wear generated, with a focus on reconstructing the work action (e.g. scrape, shave, saw) more than the worked material (e.g. plant, meat, wood). This is because the raw material of the studied assemblage, obsidian, is so even and smooth in microtopography that it does not develop the distinctive polishes and other wear traces which develop on coarser-grained raw materials, such as crypt-crystalline silicates (Kamminga 1979). Those traces which develop on obsidian are normally referable only to work action; worked material is only occasionally identifiable on obsidian with the use of a scanning electron microscope (SEM), but this was beyond the scope of the project. Additionally, the research goals of this project, dealing broadly with expediency and tool design were considered to be adequately addressed by identifying tool actions and the number of utilized elements per artifact. Worked material, then, was not assessed for analytical purposes.

The experimental tool set (n=29 items) is summarized in Table 1 (below) and is discussed further below. Each tool was used for between 20 and 103 minutes, depending on the general task being carried out. Although the range of work durations did generate different degrees of wear, types of wear were redundant by action, and rather uniform ëholotypesí of wear were seen to develop. For this reason, work duration was not considered a confounding variable in this study, where the emphasis is on work actions whose wear characteristics seem to develop rapidly and rather independently of work duration.
 

 

Table 1. Work Actions and Worked Materials of The Experimental Tool Set.

n/a = action and material are not compatible

It should be noted that in addition to using and examining these 29 stone tools, since the beginning of my training in 1989 I have used and examined more than 350 stone tools in a variety of other research projects (e.g. Smith 1992, 1996, 2000). This long-term familiarity with usewear is necessary for the analyst to discern the subtle differences in wear which may be used to classify items on functional grounds. Only long-term training of the usewear analyst can cultivate an appreciation for the variation conditioning the usewear traces which are observed to arrive at probabilistic estimates of utilitarian function.

Step 2: Blind Test of the Analyst's Capabilities

In this phase of usewear study, the analyst is tested for their ability to identify usewear. An accomplice uses tools, records their uses confidentially, and submits the tools to the analyst. The analyst examines the tools and submits answers as to used region, action, and worked materials, and these answers are compared with the records kept by the accomplice. In earlier studies (Smith 1990, 1995, 1996) I have been tested in this way. Table 2 indicates the results of these tests.
 
 
 
 
 
 
 
 
 

Table 2. Percent of Correct Identification of Used Regions and Work Actions in Usewear Analysis Blind Tests.

It is evident that my abililty to identify utilized region and work action of experimental tools is comparable with that of other, published analysts: in 84-88% of cases I have correctly identified the utilized region of a tool (86% of cases on average), and in 62-79% of cases I have correctly identified the work action (70.5% average). For these reasons I felt confident in my ability to discern these two variables in this study, without further blind testing, and no blind tests were conducted in this study.

It is also evident from Table 2 that usewear analysis of lithic implements is an imperfect endeavor; while theory has been relatively stable since the crystallization of this form of analysis (by Keeley) a little over two decades ago, methods have varied, and standardization is not present -- partly due to variable research questions as well as to variable raw materials, ancient activities, and so on. Usewear must be considered a probabilistic science, and as in other realms of anthropological and archaeological inquiry, it is necessary always to use multiple, independent lines of evidence to make the best case for a given argument. Within these boundaries, however, usewear remains our best available method for the determination of the past utilitarian function of stone tools (Grace 1990, Jensen 1988). In the case of this study, usewear analysis is well-suited to answering substantive questions.

Step 3: Examination of Artifacts

In this phase, artifacts are examined macro- and microscopically so that the analyst becomes familiar with the wear types generated by the varying work actions and worked materials. Having examined the experimental tools, a set of phenomena is generated which, depending on the raw material, industry, and so on, will adequately characterize usewear on the archaeological assemblage in terms of the research orientation. For example, while certain characteristics of the polishing of used elements is a necessity to identify worked material, in this study polish was not examined, as worked material was not being investigated comprehensively.

The phenomena examined in this study were microflake scars, rounding, and striations. Among microflake scars, four variables were observed: size (which may be used to discriminate between trampling, pressure-flaking and use), termination type (which may be used to identify the density of the material to which applied presure resulted in flaking), orientation (reflecting some elements of work action) and distribution (which may be used to distinguish between random (e.g. trampling) and nonrandom (e.g. use) wear). Among rounding phenomena, a continuum of rounding of artifact terminal edges from 'none' to 'heavily rounded' was observed: this, as many other variables in usewear study, is not practically quantifiable -- a relative term must be used to describe the degree of rounding, and this evidence must be considered with all other data on to wear observed. Rounding, generally, can indicate the degree to which the tool has been used, as abrasion, over time, generally (and microscopically) reduces angles between linear elements of the terminal edge. Among the striation phenomena the variable observed (beyond presence or absence) was orientiation (used to identify tool motion).

Table 3 (next page) summarizes the variables and states observed in this study, and comments on the general inferences possible for each variable state. It must be reiterated that all data are used in identification of artifact function, because any single variable state could be generated by a variety of equifinal processes.

Table 3. Variables, Variable States and Inferential Suggestions for All Variables Observed in this Study.

Results of the Experimental Study

Table 4 indicates the modal state of the most diagnostic observed variables per work action in the experimental tool set.

 

Table 4. Holotype Wear Modes for Work Actions
Carried Out in the Experimental Tool Set.

In general, the results of the 29 experimental activities generated wear which was consistent with wear generated in other published works on the formation of usewear (e.g. Grace 1989, Keeley 1980, Semenov 1963, Shea 1991, Smith 1990, 1996, Vaughan 1985, Yerkes 1987). While there is much overlap in the wear traces of different actions and worked materials, slight differences in one or two variables are often sufficient to discount a particular work action and confirm a different work action. Also, the presence of wear, and the distinction between work actions (rather than worked materials) were, as usual, quite evident and easily distinguished, particularly because obsidian readily accumulates some types of wear. Based on the unambiguous, redundant and confirmatory (with published results and my own previous experiments) results of these tests, the experimental sample of 29 was considered to generally and adequately approach representation of the variation in activity carried out at Harney Dune.

In all cases of usewear examination, the artifact is examined both micro- and macroscopically. It is critical for the analyst to apply both microscopic wear data with a macrosopic appreciation of the artifact in terms of plausibility: do the wear phenomena observed all suggest a specific use, or are the data conflicting? It may be that microscopically, work action and other variables are very clear, but macroscopically, the artifact is only 1cm in maximum dimension; clearly, something is amiss, and the artifact is probably a fragment of some larger item, or perhaps, was hafted. Such observations were also made in this study.

Artifact examination is carried out with binocular light microscopes, the sample illuminated with a high-power fiber-optic lamp capable of providing an intense and mobile light source. The ability to manipulate a direct, intense light source, to throw certain usewear features (such as striations and microfractures) into shadow, and to 'bring out' faint polishes, is critical, and often more important than the magnification available with the microscope.

This study utilized binocular Wild and Leitz microscopes capable of magnification from 7.5-60x and 10-300x, respectively. The Wild microscope was used to examine items up to 60x, with the Leitz used for magnifications up to 100x and, infrequently, to 300x. This is a relatively low-magnification approach compared to most published usewear studies (see Jensen 1988). This is because high magnifications (>200x) are normally used to identify distinctive polishes which may be used to identify worked materal: this was not a research question in this project, however (see above). In other words, for the variables observed in this project, the 0-100x magnification range was found to be useful and was therefore employed.

Usewear observations were recorded and were considered with other artifact data (shape, size and so on) to make functional inferences. In addition to comments, basic inferences on use/disuse, work action, multifunctionality, and number of utilized elements per artifact are tabulated in Appendix A: Usewear Data per Artifact and Appendix B: Uswear Comments on the Assemblage. Before discussing the results of the usewear study (in section 5), I discuss below the technological aspects of the assemblage to identify design characteristics of the assemblage.

4. Technological Characteristics of the Artifacts

The questions of expediency, curation and function at the center of this study must all be informed not only by a functional analysis of the limits stated above, but also by a limited 'design-theory' analysis (Hayden & Franco1996). This approach, based on the European 'chain-operatoire' or 'sequence of operations' perspective (Grace 1997), examines the design and variability of artifact form as the result of a series of actions in the generic production-use-recycling-storage and disuse-discard behavioral system.

A cursory examination of the design objectives and constraints in such a manner follows below. I examine the nature of flaking and flaking products, the range and nature of wear traces and the occurrence of ëemployed unitsí (EUís or areas per stone item observed to bear usewear). All of these will be seen to be seen to be useful in evaluating use of the tools, in particular attention should be paid to the discussion of the occurrence of EUís per artifact, as this informs directly on questions of expediency and curation.

We may first examine the first stage observable on these artifacts, that of initial core reduction and flake production. In the following tables, I report the number of items which are unambiguously classifiable as to the variable being reported: most items are unambiguous, though some are not classifiable. Table 5 indicates the assemblage composition in terms of reduction technique, on 222 (of 247) items where reduction could be unambiguously identified.

 

Table 5. Assemblage Composition by Flake Reduction Type: n=222 Artifacts.

This pattern is repeated when we examine the 98 utilized items (of 115 total utilized items) for which reduction mode is unambiguous:

 

Table 6. Assemblage Composition by Flake Reduction Type:

n=98 Utilized Items.

Table 5 indicates that most items are the product of formal, prepared-core blade production (45.50%) or blade-detachment from less-formally-prepared cores (31.53%); the remainder of items were generally derived from unprepared cores (19.82%), and some few items are derived of bifacial cores (3.15%). Table 6 indicates that production focused on formal or informal blade reduction methods (57% and 33%, respectively). Occasionally, incidental lithic items, such as 'shatter' or 'core fragments' were produced; this is to be expected in any but the most rigidly-constrained lithic reduction strategy.

An examination of flake types is also useful in identifying modes and variation in production objectives. Table 7 indicates the assemblage composition in terms of flake types for 232 of 247 unambiguously-classified artifacts:

 

Table 7. Assemblage Composition by Reduction Product (flake) Type:

n=232 Artifacts.

We see here that the assemblage is composed mainly of formal and informal blades and blade-like flakes, with a few bipolar and other flake types represented: these proportions are repeated in the assemblage of utilized items classifiable to flake type (n=110):

 

Table 8. Assemblage Composition by Flake Type:

n=110 Utilized Artifacts.

In sum, Tables 5, 6, 7 and 8 indicate that blade production, by formal or informal means, was the main objective of core reduction, and that of the various products of core reduction, blades (formal and otherwise) were used most commonly utilized items. Note that in contrast, among clearly unused items (n=100 of 247), roughly 30% are formal blades, roughly 30% are informal blades, and roughly 30% are 'shatter' or unprepared-core, informal items. Most items, then, used or unused, are blades, formal or unformal; these are long (percussion axis exceeds width by 2 times), with relatively parallel edges (though edges are occasionally sinuous in plan view), wide platforms and, usually, relatively wide percussion-axis termini. Artifact 410 (Figure 5c) is a typical formal blade, artifact 512 (Figure 7b) is a typical ëminimally-prepared-coreí blade-like tool, artifact 140 (Figure 10b) is a typical example of a minimally-prepared flake utilized as a tool, and artifact 603 (Figure 6d) is typical ëshatterí which is not frequently used as a tool in this assemblage.

Artifact size may be considered in functional and design analysis: smaller items may have been hafted and larger items may have been for other activities. Figure 1 indicates the size of used and unused items. The vertical scale indicates size class by maximum dimension (normally, that of the percussion axis, or 'length' from striking platform to flake terminus): class 1=<2cm (not present in sample), class 2=2-3cm, class 3=3-4cm, class 4=4-5cm, class 5=5-6cm, class 6=6-7cm, class 7=7-8cm.

Figure 1. Boxplot of Size Classes per Use Category.

The entire assemblage ranges from 2-8cm in maximum dimension, the average artifact being size class 4.39 (roughly 4.3cm in maximum dimension). Utilized artifacts (n=110) are somewhat larger on average (average size class 4.90 with a trend towards size class 5.5).

It is interesting that unused items (n=100) are generally smaller than used items. In my experiments, particularly with obsidian items which are not backed or blunted to prevent cutting the hands (as in the Harney assemblage, which has only one possible case of backing), I have found that smaller tools, being harder to handle and requiring a more firm, controlled grip, often cut the fingers during work. I suggest the smaller average size of unused items reflects people selecting larger items as tools, because the smaller items are more likey to cut the hands (as noted, backing is largely absent, and only one possible case of hafting (on artifact #512) was observed in the assemblage). It should be noted that unhafted tools are generally more expedient and less valuable than tool bits of compound tools (Keeley 1982, Odell 1984).

Multifunctional items are generally quite large; this represents the use of several employed units (EU's) on a single tool. Since a given EU generally consumes 1-3cm of tool edge, multifunctional items (n=19) or single-function items with more than one EU (n=22) must be slightly larger than other tools to accomodate this extra tool edge: multifunctional items, therefore, have an average size class of 5.42, roughly 5.4cm in maximum dimension. Unifunctional items (those with only one EU: n=97) have an appreciably smaller average size class of 4.9.

In sum, we may say that larger items were favored over smaller items, particularly in the case of multifunctional artifacts, and that hafting and/or backing were not commonly used to employ smaller lithic items.

5. Usewear Characteristics of the Artifacts

It must be remembered in the following discussion that this analysis is only on obsidian tools: it is possible that tools made of chert or basalt, also present in the Harney Dune assemblage, were used for other, specific functions. This is a revisited in section 6, Assesment of Confounding Factors.

Artifact functional raw data and notes are provided in Appendices A and B. Figure 2 (below) indicates the number of artifacts per work action identified in the usewear examination: this mainly represents wear which was unambiguous (n=97 of 115 used items) though a few ambiguous items are included as ëbest guessesí.

Usewear on most ëutilized elementsí (UEís, distinct areas of artifact edges that have been used independently) represents scraping (n=53) and cutting (n=29). Scraping wear was normally consistent with the scraping of fresh or dried wood: harder materials, such as bone or antler, appear to have been very infrequently worked with these tools. The distinctinve end-scraping wear normally associated with hide-working, was absent. The cutting wear observed was generally consistent with the cutting of yielding matter. The lack of the distinct polish which forms quickly (and rapidly becomes visible to the naked eye) as a result of cutting vegetal matter suggests that this cuting was not carried out on plant material. Rather, the curring wear observed on the artifacts most closely resembles that generated on the experimental tools used to butcher fish and mammals, and I suggest that this was the function of most of the Harney Dune cutting tools of this assemblage.

Shaving wear was encountered on a few (n=7) items. Shaving wear is relatively esy to identify, and is generally restricted to the working of relatively fresh wood: it may on occasion be confused with cutting of medium-resistant matter, such as leather, however, and at the fidelity of this report, the issue may not be resolved; these tools may have been used for either activity (or both activities). 'Notch' tools (n=4) are those bearing a distinct notch with usewear consistent with a scraping work action, likeley used to smooth the curve of dowel-shaped (roughly cylindrical) branches. Wear consistent with graving (the use of a projection to apply a static load in an incising manner) was unambigously observed on two items; three other items have possible but indeterminate graving wear. These items, with no evidence of hafting, were likely used on relatively yielding matter, such as fresh wood: graving seasoned wood, and particularly bone or antler, normally requires a decent haft for good effect. Two items bear wear probably, but not certainly, attributable to rotary perforation; as with the gravers, these bear no evidence of the hafting required for perforation of more dense materials such as antler, and I suggest that these were used to bore into yielding wood or perhaps leather.

Figure 2. Count of Work Actions for Artifacts with One Utilized Element.

The results seen in Figure 2 are unambiguous. Most of the utilized Harney Dune tools with one UE were used to scrape or cut; a few other uses are noted. The morphofunctional label 'blade', then, should be evaluated. Of the items with unambiguous cutting wear (n=29), 27 are, indeed, formal blades. Scraping wear is found on 53 items, and 32 of these items (60%) are classified as formal blades; the rest of the scraping wear is found on informal blade-like flakes. For this reason, the term 'blade' while unambiguously identifying these artifacts visually and by flake production type, should be used with caution when making functional dsignations and inferences. It may be said that cutting was primarily done with formal blades, but that formal blade flakes could also be used for scraping.

A further point may be made here regarding questions of expediency and curation. The 97 unambiguously-identified tools comprise 84% of the total assemblage of utilized artifacts (n=115). Most of the tools, then, were clearly used for one work action, and only one UE was exploited in most cases. If we consider the average size class of single-UE tools is 4.9 (roughly equating to 5cm), we may extrapolate roughly 10cm of 'employable edge' per average utilized tool (two 5cm-long, lateral tool edges). An informal impression of utilized element sizes is that they range between two and four centimeters in length, with an extrapolated average of three centimeters of tool edge per utilized element. Most used tool use, then, employed only an average of 3cm (about 30%) of the average 'employable edge'. These are rough figures, and there are certainly exceptions, but the general pattern is unambiguous: by this measure, obsidian toolstone was not being intensely conserved or curated. Tools were not being used to exhaustion, and retouch (found on only one used item) was rare. These facts strongly suggest expediency rather than curation. This is a point that will be revisited below in the discussion and conclusions.

Figure 3 indicates counts of artifact functions on unambiguously multifunctional items (n=18).

Figure 3. Count of Work Actions for Artifacts with

More than One Utilized Element.

Here we see less functional homogeneity than in the single-UE, single-function items. Seven multifunctional items bear scraping and cutting wear of the same general character as scraping and cutting wear described above. Other configurations are illustrated in Figure 2: the main point being that only 18 (15.6%) of the unambiguously utilized artifacts are multifunctional, bearing two or three utilized elements for the normal activities of scaping and cutting, with a few other actions. As noted in Figure 1 (see above) these items are generally larger than single-UE items. Using the same reasoning as presented above in estimating the amount of 'employable edge', here we find that roughly the same proportion (about 30%) of 'employable edge' was utilized per tool.

In sum we may say that although these are multifunctional items with more than one UE, and therefore representing some degree of curation, they comprise only about 16% of the used items in the assemblage, and that even these items are not used to exhaustion. Again, toolstone does not appear to have been extensively conserved or curated by means such as resharpening and/or utilization of all 'employable edge' on a given artifact.

These conclusions regarding the 'non-conservation' of toolstone is substantiated in Figure 3, which indicates the number of UE's per utilized items.

Figure 4. Count of Utilized Elements Per Utilized Artifact.

We see in Figure 3 that of all utilized artifacts (including those with ambiguous phenomena that could not be functionaly classified, but could be classified as 'wear', n=130 items), 104 items (80%) have only one UE, 22 (17%) have two UE's and the remainder (n=4 or 3%) have three or less UE's. This does not suggest extensive curation or conservation of toolstone. It is also important to note that of the 247 artifacts in the examined assemblage, 88 items (those 35.62% which could be unambiguously assigned to flake type) exhibited no usewear, even though the unused items include roughly the same proportions of flake types as the utilized items. As seen in Table 9 (next page), most unused items are formal blades or blade-like flakes detached from informally-prepared cores.

 

Table 9. Assemblage Composition by Flake Type:
n=88 Unused Artifacts.

All of the data above are considered in the synthetic reconstruction of activities and design objectives found below in section 7, Conclusions on Artifact Design and Function. First, however, it is necessary to consider a variety of extraneous, confounding factors which may effect the analysis. This is a step rarely carried out in usewear analyses, which often proceed from observations to conclusions. It will be seen below that such assesments are a necessary step to fortify analytical conclusions.

6. Assessment of Confounding Factors

Although a fusion of usewear- and design-theory analysis provides the best current approach to understanding the function of ancient stone tools, usewear analysis is beset with a number of problems which must be addressed before results are accepted. Just as a variety of wear traces and other lines of evidence are used to make the best case for the function of a given artifact, the potential effect of a variety of taphonomic, C- and N-transforming factors (sensu Schiffer 1972) should be assessed in order to demonstrate the taphonomic integrity of the assemblage and its various observed characteristics.

Table 10 (next page) summarizes factors which I considered to be most likely to have possibly effected the results of this study by (a) promoting incorrect functional assignments or (b) in some way altering non-functional assemblage characteristics, as a whole, from their state at the time of deposition. These factors are assessed separately in the following sections.

 

Table 10. Pertintent Taphonomic Factors and Potential Effects.

Assessing Weathering

Weathering, such as wind- and water-action, can wear away artifact surfaces, and eliminate usewear traces, or at least degrade them to such a degree that they are indeterminate (Hiscock 1985, Schiffer 1972). Fortunately, such weathering is rather easy to observe in this obsidian assemblage. Five items (2.02% of the 247-item sample) (noted in Appendix B) were water-worn, which is not surprising as the assembalge was deposited in a beach area. Some items, then, were at some point inundated, and worn by water action (at least one item of these, however, bore a water-rounded cortex, and was not worn in situ at Harney, but nearer to where it was obtained by the original user).

All indications of the hydrology of the area suggest a low-energy lakeshore environment (Raymond 1994), and we may be reasonably certain that the assemblage (in terms of usewear and spatial distributions) was not significantly effected by this factor.

Assessing Trampling

Trampling of lithic artifacts has been shown to sometimes generate wear which may be confused with wear resulting from use: trampling can modify and move stone tools (Shea and Klenck 1991). Hiscock has shown the utility of a taphonomic understanding of stone assemblages in the same manner as faunal assemblages, and that trampling must be understood as a site-formation agent (Hiscock 1985).

For the analyst, the most important difference between trampling and non-trampling wear is the randomized nature of trampling wear as contrasted with the regularized nature of wear from use. My own experiments with trampling (conducted at Simon Fraser University in 1996 and 1997) as well as the work of other studies (e.g. Hiscock 1985, McBrearty et al 1998, Shea and Klenck 1993): trampling wear is not, generally, patterned; scratches will be in conflicting orientations; micro- and macro-flakes will be of a variety of sizes, orientations and terminations; the distribution of microflaking and other dmage will not be as uniform as on a utilized item; rounding will not occur on the tool edges (though previously-rounded tools may be trampled); damage will not be restricted to the edge of the tool; damage will be found on elements of the tool that, kinetically, ëmake no senseí as a tool when considered macroscopically; and, finally, trampling will often break more delicate items such as the thin Harney obsidian blades.

Table 10 indicates the assemblage composition in terms of artifact integrity (complete, broken or fragmen) for the 245 (of 247) artifacts where this property was unambiguous:

 

Table 11. Artifact Integrity in Used Set, Unused Set and Entire Assemblage:

n=245 classifiable items.

Table 11 indicates that in the entire assemblage, about 51% of items are complete, What this means in behavioral terms is unclear: how many items of a given camp area may be expected to be broken during a given time is unknown. Still, using this figure as a baseline, we see that in contrast, of the used items which were classifiable, most (65%) are complete while17% are broken, and 17% are fragments. Of the unused, classifiable items, less items are complete: roughly equal amounts are broken or complete (about 40% each) and some are fragmentary. If the wear traces on artifacts was normally the result of trampling, and was in fact being misidentifed as usewear, we would expect to find more broken items bearing (false) 'usewear'.

Overall, then, these figures all suggest that the effect of trampling on the tools examined in this study is probably very limited. Since most of the delicate obsidian blades of this study would be broken if trampled, and since many tools of used set are whole (65.9%), it is unlikely that many items of the used set were trampled. Even if the thicker blades were not broken, the brittle and thin obsidian edges should exhibit much more unpatterned, step-termination fracturing than was observed. Finally, trampling wear generates rather distinctive damage (to the analyst), which was only infrequently observed in this study. In conclude that trampling damage to the assemblage is mininmal.

Assessing Multiple Use of One Utilized Element

One unresolved problem in usewear analysis is that of the use of one element of the tool for one purpose followed by later use of that same element for another purpose. Depending on the duration and nature of the work activity, the second wear action may or may not effect, confuse, obscure or even obliterate wear generated during previous work. For this reason, wear analyses must caution that potentially, only the most recent (last) work action is being observed on any given utilized element. Ethnographically, it has been observed among living hunter-gatherers that generalized tools, in terms of design, are most likely to be used repeatedly and for different actions, on the same utilized element (e.g. Hayden 1979). In general, daily life, contemporary stone-tool-users do not appear to always switch from one tool ëtypeí to the next in the course of carrying out some task, as archaeologists might assume (and hope!). Rather, people often use a given chip of core of stone, of generalized shape, for a variety of tasks, employing different elements of the tool edge as needed (Hayden 1979 op cit).

The Harney Dune blades are not such generalized artifacts. In Table 6 and Figure 2, respectively, we learned that the Harney obsidian assemblage is predominantly blade-focused, and that most activity carried out was either scraping or cutting. We also find that most tools are not multifunctional, and most have only one or two utilized elements. Although these tools do not seem to have been highly curated, they were designed with cutting and scraping in mind. None of these seem to be the hallmarks of generalized design or use: the shape of the artifact is normalized towards that of the ëblade idealí, which limits the number and variety of projection and / or other features that may have been used in non-blade work actions; use is restricted in spatial distribution on the tool, and use is restricted, largely, to a few actions. Finally, we should remember that here we examine only the obsidian element of the lithic material culture, and that raw material more suited to multifunctionality would likely be the CCS or the basalt also found in the Harney assemblage (see below). For all of these reasons, I conclude that multiple use of single utilized elements in the Harney obsidian assemblage, though possible, would have had negligible effects on this analysis.

Assessing the Rate and Degree of Wear Formation

A major difficulty still being sorted out in usewear studies is that of the duration of a given work action. With increased duration, normally, comes increased wear, and, of course, short work durations may leave little wear, though the tool was indeed used. A combination of factors may introduce significant difficulty for the analyst: consider a single butchery episode, such as that of a rabbit or bird or fish, taking just a few minutes or less, and working with a largely yielding worked material (flesh and cartilage): wear traces may or may not be generated and/or observable. This has been confirmed in a variety of experimental studies (e.g. Keeley and Toth 1982). Until more experiments are done, this problem is largely unresolvable. In any case, the question is not addressable in this study as the faintest wear, generated in single or very short-duration work episodes, requires high-power microscopy (500x and more) of a variety not available to this analysis. Note however that the work action is often rather more readily identified on less-used tools than worked material. Scraping for a few minutes, for example, generates scraping wear (though faint), rather than cutting wear, and in many cases it is posible to identify gross differences between work actions, the main focus of this study.

For these reasons, I feel that duration of work action does not hinder this study, in terms of identifying gross distinctions between work actions (I do not feel I have mis-diagnosed cutters and scrapers), howevr some work actions, particularly butchery and cutting on other yielding raw materials, may be significantly under-represented in this study. Considering the design of the Harney Dune blades, it may be that a reexamination at higher magnification would indicate more cutting than has been identified at this stage.

Assessing Hafting

The importance of hafting in a number of respects has been mentioned in earlier sections of this report. Keeley has pointed out the principal reasons understanding hafting and its effects on the archaeological record: first, in many tools, the bit is not as valued as the handle, and second, hafted items tend to be found not where they were used, but where they were manufactured and where they were finally abandoned (Keeley 1982). Hafting wear on lithic tools is rather easy to identify, summarized by Keeley (op cit) as 'wear which makes no functional sense' in terms of potential actions carried out with the tool. Thus, striations and microflaking on flake scar ridges (particularly on the dorsal), combined with striae on the ventral (often on the bulb of percussion, if prominent) often indicate contact with a haft; and wear which terminates abruptly where it may otherwise have continued on a utilized element, in conjunction with the traces just mentiond, may often be attributed to hafting.

In this assemblage, such traces were the exception. Only one item, artifact #512, bears dorsal scar ridge damage that is expected of hafting and is difficult to explain in any other way: however, this is not conclusive evidence. In sum, then, hafting of these blades seems extremely rare, and sytematic hafting is not supported by this usewear analysis

Assessing Resharpening

A number of analysts have noted that stone tool resharpening and/or reshaping with a billet, pressure-flaker or even a simple 'swipe' abrasion of a utilized edge against a hammer or abrader may easily al;ter or obliterate traces from wear (Smith 1996, Vaughan 1985, Yerkes 1996). Fortunately, such resharpening often leaves rather distinct traces: pressure-flaking microflake scars are very lage in general and are often so redundantly patterned that they are too uniform to represent wear; also, flake terminations often differ from those of use in a number of ways.

In this assemblage, little resharpening was noted. Most of the mimcroflaking which occurs on the tools appears to be a rseult of use. One large item (artifact #5) bears some pressure-flaking, but this is the exception rather than the rule. In sum, then, I conclude that this factor does not effect the conclusions of this study in any appreciable, systematic manner.

Assessing Sampling

Clearly, the archaeological record represents the sampled, C- and N-transformed precipitates of a once-dynamic system of behaviour (Schiffer 1972, Smith n.d.). The character of artifact deposition, a C-transform, should be modelled when analysis is considered. In this case, we have selected only the obsidian items of the entire archaeological assemblage (which includes faunal material, features, and stone tools of other raw materials). And, of the obsidian, we have selected those items greater than 2cm in maximum dimension, noting that smaller pieces are generally considered too small to be used unless hafted (Smith, personal observbation, Stevenson 1991), and that a microlithic hafting technology is not found in the study area (Raymond 1994). Thus, our sample represents a single raw material, and items of a particular size class, most likely representing tool-size items rather than debitage.

The effect of this sampling is to, in a beneficial way, restrict our research questions and the work load. Items smaller than 2cm in maximum dimension are unlikely to have been tools (though I do suggest a brief examination of a sample of these items), and items larger than the 7-8cm items of the present sample are probably cores. The present sample, then, is of the obsidian items most likely to have been candidates for use, and thus the sampling strategy is understood in terms of its effects on our perception of the assemblage characteristics.

Assessing Raw Material Variation

Different raw materials, of course, in lithic technologies, have different properties, and such differences are often exploited by stone-tool-using peoples: on the Northwest Coast, for example, abraders may be made of pumiceous or tuffaceous rock, while blades may be of obsidian or chert, and net-sinkers of basalt (e.g. Smith and Ames 2000).

Naturally, then, the selection of obsidian items in this sample should reflect only the activities carried out with obsidian tools. The restricted range of activities, mainly cutting (which may be under-represented) and scraping, seen in this assemblage, suggests that other common activities of lithic-technology-using-people (perforation, graving, notching, shaving, sawing) may have been carried out with other raw materials. This does not degrade the functional inferences made in this report, but I do suggest a follow-up examination of the usewear on other raw materials in order to obtain a more complete understanding of the activities carried out at Harney Dune

7. Conclusions on Artifact Design and Function

That most of the potential confounding factors identified above have been ëdispensed withí through understanding fortifies the analytical conclusions which follow. The effects of uncontrolled factors are, at least, understood. Based on the usewear and design-analysis investigations noted above, I suggest the following synthesis.

The functional and technological and design aspects of the assemblage are largely unambiguous. The assemblage generally represents blade production, from cores prepared to a greater or lesser degree. Most tools from this assemblage are blades, either formally or informally produced, although the utilized items tend to be blades or blade-like items, and it seems that this is the holotype shape that was the ideal 'lithic solution' being sought during tool production. The tools are generally of the size held in the hand; hafting and backing are the rare exception and the bulk of tools were hand-held. Of these normally hand-held tools, most were used for a single function: scraping, (probably wood), cutting (probably meat and some vegetal matter) or some other miscellaneous maintenence task (shaving, graving, perfoation, etc.). In most cases, only a third to a half so of the 'employable edge' of an implement was exploited.

All of these data strongly suggest that the assemblage was not extensively curated. Confirming this, I note that there is an almost complete lack of the following indicators of curation; (a) intense exploitation of 'employable edge' per artifact, (b) backing, (c) extensive multifunctionality, (d) size reduction due to resharpening and/or (e) the use of more flakes in the assemblage.

With regard to function, we see that although scraping and cutting are dominant (accounting for 54% and 29% of all usewear, respectively), this does not indicate functional homogeneity: there are significant numbers of other work actions, and even the 54% / 29% split between scraping and cutting is significant. There is little multifunctionality, however, on the same tool.

In short, though the technical objective was of blade production and blade use, for a restricted set of activities, there was neither extensive curation nor polarity in activity. The assemblage seems to be specialized in terms of tool form (blades), but expedient and bi-modal (scraping, cutting) in use, and this use does not proceed to tool exhaustion.

We may crudely evaluate the observations of this report by comparing them with the expectations of the competing hypotheses, in Table 12:

 

Table 12. Evaluation of Competing Hypotheses

On balance, it appears that the function and design attributes of these tools more closely meet the expectations of the first hypothesis presented by Raymond for the behaviours leading to site formation (see section 1. Introduction). The expectations of that hypothetical assemblage are a rather standardized, unifunctional set of tools used expediently as large groups gather seasonally. Though these tools are not unifunctional, there is a restricted range of activities, and the indiustry does focus on blade production.

The alternative hypothesis, that people aggregated in smaller groups with less specific behavioral aims per aggregation episode, and perhaps multi-seasonal habitation, would be expected to generate more general-purpose, multifunctional tools. This is not, apparently, the pattern of the tools of this assemblage, which are 87% unifunctional.

Further evaluations of these hypotheses will be best served by a more rigorous and quantitative evaluation of the material correlates expected of each alternative hypothesis. This report, however, is a step in this direction, and appears at present to support the hypothesis of seasonal aggregations of large groups, focusing on a few activities, at least with the blade-like obsidian tools of the examined assemblage. Further usewear analysis may well assist in evaluating the alternative hypotheses.

8. References

Flenniken, J., and J. Haggerty. 1979. Trampling as an Agency in the Formation of Edge Damage: An Experiment in Lithic Technology.
Northwest Anthropological Research Notes 13:208-214.

Grace, R., 1990. The limitations and applications of functional analysis.
In Graslund, B., H. Knutsson, K. Knutsson and J. Taffinder (eds) 1990. The Interpretative Possibilities of Microwear Studies. Proceedings of the International Conference on Lithic Use-wear Analysis, 15th-17th February 1989: 9-14.

Societas Archaeologica Upsaliensis, Uppsala, Sweden.
Grace, R. 1997. The 'chaine operatoire' approach to lithic analysis.
In Graslund, H. & H. Knutsson (eds).1997. The interpretive possibilities of microwear studies Vol III. Societas Archaeologia Upsaliensis, Uppsala, Sweden.

Hayden, B. & N. Franco. 1996. Evaluating lithic strategies and design criteria.
In Odell, G.H., (ed). 1996. Stone Tools. Theoretical Insights into Human Prehistory.
Plenum Press, New York,: 9-50.

Hayden, B., 1979. Palaeolithic Reflections: Lithic Technology and Ethnographic Excavations Among Australian Aborigines.
Australian Institute of Aboriginal Studies, Canberra.

Hiscock P, 1985. The need for a taphonomic perspective in stone artefact analysis.
Queensland Archaeological Research 2:82-95.

Jensen, J.H. 1988. Functional Analysis of prehistoric flint tools by high power microscopy: A review of West European Research.
Journal of World Prehistory v.2(1),53-88.

Keeley, L. 1980. Experimental Determination of Stone Tool Uses: A Microwear Analysis.
University of Chicago Press, Chicago.

Keeley, L.H. 1982 Hafting and retooling: Effects on the Archaeological Record.
American Antiquity 47:798-809.

McBrearty, S., L. Bishop, T. Plummer, R. Dewar, and N. Conard. 1998. Tools underfoot: human trampling as an agent of lithic artifact edge modification. American Antiquity 63: 108-130.

Odell, G.H. 1984. Prehistoric Hafting and Mobility in the North American Midcontinent: Examples from Illinois.
Journal of Anthropological Archaeology 13:51-73.

Semenov, S. 1963. Prehistoric Technology.
Cory, Adams & MacKay, London.

Shea, J. 1991. The Beahavioral Significance of Levantine Mousterian Industrial Variability.
PhD Dissertation, Harvard University.
UMI Microfilms, Ann Arbor, Michigan.

Smith, C.M. n.d. The Archaeological Record as the Sampled, Transformed Precipitates of an Evolving Productive System.
Manuscript in posession of the author.

Smith, C.M. 2000. Lithic Report: Cathedral Park Alpine Archaeological Site Survey, Canada. In Reimer, R. and C.M. Smith. 2000. Cathedral Park Alpine Archaeological Site Survey. Report on file at Lower Similkameen Band Office, Keremeos, British Columbia, Canada:22-42.

Smith, C.M. 1996. Social Stratification Within a Protohistoric Plankhouse of the Pacific Northwest Coast: Usewear and Spatial Distribution Analysis of Chipped Lithic Artifacts.
Unpublished MA thesis, Department of Anthropology, Portland State University.

Smith, C.M. 1995. A Blind Test for Assesment of Analyst Competence in Lithic Usewear Analysis.
Report on file, Portland State University, Laboratory of Archaeology.

Smith, C.M. 1992. Pilot Study of Usewear Characteristics of the Meier Site Lithic Assemblage.
Report on file, Portland State University, Laboratory of Archaeology.

Smith, C.M. 1990. Usewear Analysis of Experimental and Artifactual Stone Tools.
Unpublished BA (Joint Honours) thesis, Departments of Archaeology and Anthropology, University of Durham, England.

Stevenson, M.G. 1991. Beyond the Formation of Hearth-Associated Artifact Assemblages. in Kroll, E. and T.D. Price (eds) 1991. The Interpretation of Archaeological Spatial Patterning:269-299.
Plenum Press, New York.

Tringham, R. and R.G. Cooper, G. Odell, B. Voytek and A. Whitman. 1974.
Experimentation in the Formation of Edge Damage: A New Approach to Lithic Analysis.
Journal of Field Archaeology 1: 171-196.

Vaughan, P. 1985. Usewear Analysis of Flaked Stone Tools.
University of Arizona Press, Tucson.

Yerkes, R. 1987. Prehistoric Life On The Mississippi Floodplain; Stone Tool Use, Settlement Organization, And Subsistence Practices At The Labras Lake Site, Illinois.
University of Chicago Press, Chicago.