updated 8 June 1997
Comment:
The following paper presents the results of a study conducted by Anthropological Studies
Center personnel several years ago to
reassess the potential for visually segregating obsidian glass groups commonly
recovered in western Great Basin and Central Sierran archaeological sites. The study was conducted as an
adjunct to an archaeological investigation of several prehistoric sites
situated along Highway 395 near Bridgeport in Mono County, California.
Previous studies of the efficacy of segregating western Great
Basin glasses employed only archaeological specimens. While their
accuracy rate was fair, the study samples were overweighted
in favor of the most easily recognized glass, Casa Diablo. Errors occurred most
frequently in the smaller samples of Bodie Hills
items in their study.
In the ASC
study, we decided the first step was to determine whether macroscopic
attributes of non-archaeological samples from each glass group were distinctive
enough to correctly identify the particular glass a large percentage of the
time. The results of this first level of analysis were promising: a couple
people sorted most of the groups with a 90+% success rate which seemed to be
based primarily on the researcher's familiarity with recognizing certain
attributes. Not surprisingly the next step, dealing with archaeological
materials, had a lower success rate.
Reexamining the
Potential to Visually Source
Western Great Basin Obsidians
Sunshine Psota
19 August 1990
Paper presented at Annual Meeting of
Society for California Archaeology, Foster City.
Sunshine Psota
Anthropological Studies Center
Sonoma State
University
Rohnert Park, CA 94928
Abstract:
Previous studies suggest that certain western
Great Basin obsidian sources are more reliably
identified than other sources. Given the intra-source macroscopic variability
and inter-source similarities of these obsidians, tests were designed to access
accuracy rates. Three tests were conducted using a reference collection of five
obsidian sources, then selected researchers using archaeologically recovered
obsidian from the Mono area, further accessed reliability. Conclusions will
focus on results of reference collection tests and XRF assignments to
archaeologically recovered obsidians.
Introduction
In 1989 the
Cultural Resources Facility at Sonoma State University
investigated five sites near Bridgeport,
California for proposed
California Department of Transportation highway improvements. Situated in the
eastern Sierra Nevada, north of Mono
Lake, these sites
included both sparse and extensive lithic scatters,
comprised overwhelmingly of obsidian. Given the size of the archaeological
collection and the expected predominance of various local sources, it was
decided to re-evaluate the potential for visually sourcing western Great Basin obsidians as an economical method for
characterizing prehistoric use of this resource. A multi-phase test, consisting
of both macroscopic characterization of reference materials and comparison with
x-ray fluorescence (XRF) assignment of archaeologically recovered materials,
was established.
Obsidian
sourcing has been an important tool for developing inferences about exchange
patterns and subsistence strategies. There are a variety of techniques
available for determining the geologic origin of obsidian such as neutron
activation analysis and XRF. Although these techniques have been used with
success, labs employing these methods are either not readily available or the
costs for processing large numbers of specimens are prohibitive. Researchers
using magnetic sourcing (McDougall, et al. 1983), and microprobe analysis
(Merrick and Brown 1984) have developed faster and more cost efficient
approaches, but the equipment required is expensive and for various reasons not
always an acceptable alternative. Perhaps the most expedient and economical
technique for identifying large amounts of obsidian is visual sourcing. This
method of sourcing has had variable results. Wickstrom
and Fredrickson (1982) have demonstrated a 90% or better success rate in
identifying four obsidian sources in the North Coast Ranges of California. In Chiapas, Mexico
(Clark and Lee 1984) and southern Italy (Ammerman
1979) researchers have had similar success with visual sourcing. It was hoped
visual sourcing of the major western Great Basin sources situated in the Bridgeport and Long
Valley regions could be
equally reliable.
In visual
sourcing tests employing macro and microscopic techniques using
archaeologically recovered obsidians from the western Great
Basin, Bettinger, et al. (1984) produced
results of 83 to 89% success. Their third test which consisted of 60 specimens
predominantly from the area north of Mono
Lake was most similar to
our sample. It consisted of a sample in which 37% were
Mt. Hicks, 25 % were Queen, and 17% were Bodie Hills; three Casa Diablo and 10 specimens from a
northern source were also included. They interpreted their decreased accuracy
rate of 83% to have been affected by the inclusion of Bodie
Hills. Criteria for accurately sorting Bodie
Hills obsidian from others in this region was not accomplished, and they
suggest that it may not be possible.
In contrast, Hull (1988) restricted
her experiment solely to the degree of translucency exhibited by obsidian
specimens. She found that equating Casa Diablo with the most opaque samples, resulted in a high degree of reliability in
collections from Yosemite
National Park. Both of
these studies also noted difficulty in distinquishing
some Mt. Hicks obsidian from Bodie
Hills and found better success with some sources than others.
Reference
Collection
In the Mono
area four major sources occur in relative close proximity: Casa Diablo, Bodie Hills, Mount
Hicks, and Queen/Truman
Meadows. A non-cultural reference collection was developed by obtaining
un-modified cobbles from prehistoric quarry areas from three locations at Casa
Diablo and Bodie Hills, and from smaller areas at Mount Hicks
and Queen/Truman Meadows. Less economically significant sources such as Mono Glass
Mountain and Pine Grove,
which have been sparsely represented in archaeological collections (Jackson
1989, Turner 1989, and Moore 1989), were not examined due to time and cost
limitations. To compensate for some of the biases of these tests, a non-local obsidian, Fish Springs, was included in the test
samples. Since Fish Springs can contain similar visual attributes as some Bodie Hills and Mount Hicks material, it was included
to determine whether it could be successfully separated out from the other
sources. These sources then represent the same sources used by Bettinger, et al. (1984). It is recognized that the
materials collected do not represent all prehistorically used areas within
these sources, the full range of visual variability found in archaeological
collections, or the only sources prehistorically used.
Test
Preparation and Procedures
At the Bridgeport sites, some
investigation areas characterized by secondary biface
reduction were identified. Using the reference collection cobbles, similar biface reduction activities were duplicated, with each
cobble's materials bagged separately. Three tests were prepared, each
containing 90 specimens. These tests were designed to: (1) reflect the
different locations collected from each source; (2) exhibit the greatest visual
variability within the sample; (3) include varying amounts of cortex on 20% of
each source's sample; and (4) include predominately thinner and smaller flakes
which reflected the archaeological materials. Test specimens ranged in size
from 6 to 20 mm, and in thickness from 0.35 to 27.7 mm.
Prior to
taking the tests, participants familiarized themselves with Bettinger,
et al.'s (1984) descriptions, the reference materials, and descriptions
characterizing the reference collection compiled by Thomas Origer
and Kim Tremaine, both from Sonoma State University.
Seven people, with varying amounts of related obsidian experience, took these
tests. Each specimen was checked for attributes such as color, flow structure,
inclusions, texture, luster, and other characteristics such as cortex. In total
1800 identifications were made. For the first test, Test A,
participants made identifications utilizing a number of approaches, such as
direct sunlight, incandescent, and fluorescent lighting. More standardized procedures
were established for Tests B and C which included using a light table, an
incandescent lamp, review of previous test results, use of Origer
and Tremaine's descriptions while taking the tests, and grouping like attributes with like attributes.
Results of
Reference Collection Tests
Results of the
first reference collection test suggested these western Great
Basin obsidians could be visually sourced. From the results of
Tests B & C it was concluded: 1) participants tended to improve with each
test, with some participants consistently more adept in sorting visual
attributes than others; 2) the amount of exposure to these sources through
other avenues, such as laboratory work, usually yielded better test results;
and 3) smaller debitage tended to be more difficult
to identify, as it tended to contain less diagnostic characteristics. Most
participants achieved a 85% to 100% success rate for
identifying specimens of Casa Diablo and/or Queen/Truman Meadows. Bodie Hills and Mount
Hicks were confused, at
times, by all participants. In the third test, scores reflected a range of
success with two participants achieving a poor accuracy rate, while four of the
most successful participants accomplished an overall success rate of 87%.
Visually
Sourcing Archaeologically Recovered Obsidian
With wide-eyed
optimism, two of the more successful participants were selected to sort the
archaeologically recovered obsidian sample from the Bridgeport sites. The researchers began by
examining two bags of obsidian debitage, containing
approximately 100 flakes. In contrasting the reference collection tests with
the archaeologically recovered obsidian, it was found that patination
can obscure some visual attributes. Also, most of the debitage
was characterized by greater variability than the reference collection and
exhibited some different attributes. The obsidian was sorted into groups
characterized by like attributes and not by presumed source. By sorting like
attributes with like attributes, 10 groups were developed with one defined as
miscellaneous. It was expected that the debitage
sample would be characterized by fewer sources, unlike formal tools which were
expected to comprise greater variability of sources.
Using these
attribute groups, the researchers then attempted to sort 60 diagnostic
projectile points from the sites. Additional groupings were identified upon
discovering that a wider range of visual attributes existed for these tools.
This may in part be related to their greater size. At least 16 groupings were
eventually delineated on the basis of differences in visual attributes. After
independently sorting an additional 273 formal tool specimens according to
these groupings, the researchers agreed approximately 42% of the time.
From these
groupings, 18 pieces of archaeologically recovered debitage,
representing nine categories, were subjected to XRF analysis. This small sample
is biased towards primarily small, thin flakes, which may be the least
identifiable but which constituted the majority of the archaeological materials.
Comparing the XRF results with these assignments was discouraging,
the best overall accuracy rate was 44%. XRF analysis assigned all but one flake
to the Bodie Hills source with the remainder assigned
to Mount Hicks. These results indicate that the
established categories may not represent sources, but only variation within
sources.
Next, the 60
projectile points were visually sourced into the groupings, assigned to a
presumed source, and submitted for XRF analysis. In comparing the XRF
assignments with these identifications, one researcher assigned approximately
48% correctly, while the other assigned 85% correctly. Additionally, the latter
assigned specimens correctly to the Bodie Hills
source better than 97%. Remaining sources assigned by XRF analysis comprise
very small samples which will not be further elaborated.
Conclusions
In summary, an
attempt was made to reevaluate the reliability of visually sourcing
archaeologically recovered obsidian from the western Great
Basin. The reference collection tests were designed to familiarize
researchers with these obsidians, the range of variation, and to provide
preliminary assessments of how reliably sources could be identified in
archaeological collections. Test results indicate a moderate success rate in sorting
the reference collection obsidians, but did not achieve a 90% or better success
as some researchers in other regions have yielded (Ammerman
1979, Wickstrom and Fredrickson 1982, Clark and Lee
1984). Archaeologically recovered obsidian, however, was not as accurately
attributed to sources. Although the accuracy rate for sourcing the XRF assigned
debitage was poor, both participants improved
significantly when sourcing the XRF assigned projectile points. This may be
attributable to experience gained from previous tests and to the larger size of
the materials.
In addition
the following conclusions have been reached: 1) most participants exhibited an
ability to improve significantly with each additional reference collection test
which gave participants important experience for sorting archaeologically
recovered obsidian; 2) projectile points, bifaces,
and large to medium size flakes were more successfully visually sourced,
whereas source identification of flakes smaller than 15 mm tended to be less
reliable; 3) individuals need to evaluate their accuracy rates as some were
better able to visually source obsidian than others, regardless of related
obsidian experience; 4) accuracy rates might improve if microscopic examination
was combined with macroscopic examination (Bettinger
et al, 1984, Roper 1989, and Hale 1989); and 5) as previous studies have noted,
inclusion of Bodie Hills and Mount Hicks adversely
affects the accuracy rate within a sample.
Comparisons of
our success rates to other regional researchers may not be entirely appropriate
given the variable proportions of different obsidians within a geographic
region. Thus, assemblages derived from areas in close proximity to the Casa
Diablo source may be expected to contain high proportions of this obsidian and
visual sourcing results should be more reliable than those for archaeological
sites within the Bodie Hills/Mt. Hicks region. Also,
although one participant's accuracy rate was virtually identical to Bettinger et al.'s (1984), this comparison does not reflect
similar amounts of problem sources as only a small amount of Bodie Hills was represented in Bettinger
et al. collections and the Bridgeport
archaeologically recovered sample contained a high percentage of Bodie Hills
obsidian.
In conclusion,
visual sourcing is viewed as a preliminary means to discriminate between
obsidians and of practical importance when examining large amounts of material.
Accuracy rates for visually sourcing western Great Basin
obsidians will vary geographically due to subsistence strategies and exchange
patterns. Although the best overall accuracy rates for the study region are not
as reliable as rates obtained for other areas, the sample size of materials
tested by XRF analysis is much lower than necessary for an accurate assessment.
More detailed experiments need to be implemented under controlled conditions
with specific objectives in mind. In this manner, visual sourcing of obsidian
materials from specific regions, in combination with selective XRF analyses,
may prove to be a viable cost-effective method for determining the composition
of archaeological assemblages. For this region of the western Great
Basin by these researchers, however, visual sourcing has not had a
high enough accuracy rate to be reliable for large scale use.
References
Cited
· Ammerman, Albert J.
1979 A Study of Obsidian Exchange Networks in Calabria. World Archaeology 11(1):95-110.
· Bettinger, R.L., M.G. Delacorte,
and R.J. Jackson
1984 Visual Sourcing of Central Eastern California
Obsidians. In Obsidian Studies in the Great Basin, edited by Richard E. Hughes, pp.63-78.
Contributions of the University
of California Archaeological
research Facility No. 45. Berkeley.
· Clark, John E. and Thomas A. Lee, Jr.
1984 Formative Obsidian Exchange and the Emergence of Public Economies in
Chiapas, Mexico. In Trade and Exchange in Early Meso-america,
edited by K.G. Hirth, pp.235-275. University of New Mexico
Press, Albuquerque.
· Hale, Mark
1989 Personal Communication.
· Hull, Kathleen L.
1988 Obsidian Studies in Yosemite
National Park:
Preliminary Observations. In The Proceedings of the Society for
California Archaeology, edited by Susan M. Hector, Lynne E. Christenson, G. Timothy Gross, and Martin D. Rosen, pp. 169-188. Volume 1. Society for California
Archaeology, San Diego.
· Jackson, Robert J.
1989 Personal Communication.
· McDougall, J.M., D.H. Tarling, and S.E. Warren
1983 The Magnetic Sourcing of Obsidian Samples from Mediterranean and Near
Eastern Sources. Journal of Archaeological Science 10:441-452.
· Merrick, H.V. and F.H. Brown
1984 Rapid Chemical Characterization of Obsidian Artifacts by Electron
Microprobe Analysis. Archaeometry 26(2):230-236.
· Moore, Joe
1989 Personal Communication.
· Roper, C.Kristina
1989 Personal Communication.
· Skinner, Craig E.
1983 Obsidian Studies in Oregon:
An Introduction to Obsidian and an Investigation of Selected Methods of
Obsidian Characterization. M.S. Thesis, University
of Oregon, Eugene.
· Turner, Arne
1989 Personal Communication.
Return to the top of this document.
Go to ALA-555
Go to ALA-42
Go
to SOL-356
Go to Franz Valley Obsidian
Go to Map of Major Obsidian
Sources in California & Western Nevada
Go to Charts of obsidian
hydration results from Pleasanton studies
Return
to Archaeological
Studies Page
Return
to Home
Page