Sunteți pe pagina 1din 13

Society for American Archaeology

On-Site Artifact Analysis as an Alternative to Collection


Author(s): Charlotte Beck and George T. Jones
Source: American Antiquity, Vol. 59, No. 2 (Apr., 1994), pp. 304-315
Published by: Society for American Archaeology
Stable URL: http://www.jstor.org/stable/281934
Accessed: 03/09/2009 17:21

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=sam.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

Society for American Archaeology is collaborating with JSTOR to digitize, preserve and extend access to
American Antiquity.

http://www.jstor.org
ON-SITE ARTIFACT ANALYSIS AS AN ALTERNATIVE
TO COLLECTION

Charlotte Beck and George T. Jones

Artifact collection has become increasingly expensive, especially with respect to curation costs. Consequently,
many government agencies as well as researchers have turned to an alternative, "on-site" analysis approach in
which many analytical procedures usually reserved for the laboratory are conducted in the field. Using data
gathered at an andesite quarry in eastern Nevada, we evaluate such an approach in terms offeasibility, reliability
of results, and cost, as well as its impact on the surface archaeological record. Although we still believe that artifact
collection is central to archaeological research, we argue that in circumstances where collection is not possible, a
well-designed and systematically implemented on-site approach can yield reliable results, although not without
some visible impact on the surface archaeological record.
La recoleccion de artefactos se ha vuelto cada vez mds costosa, especialmente con respecto a los costos de
curaci6n. Consecuentemente, algunas agencias gubernamentales as[ como investigadores han elegido como al-
ternativa el andlisis "en el sitio, "en el cual muchos de los procedimientos usualmente reservadospara el laboratorio
se conducen en el campo. Utilizando los datos obtenidos en una mininade andesita en el este de Nevada, nosotros
evaluamos esta alternativa en terminos defactibilidad, veracidad de los resultados y costos, asf como su impacto
en el registro arqueologico de superficie. Aunque aun creemos que la recoleccion de artefactos es central en la
investigacidonarqueologica, arguimos que en circunstancias donde la recoleccion no es posible, un andlisis "en el
sitio " bien disenado e implementado sistemdticamente puede dar buenos resultados, pero no sin un impacto visible
en el registro arqueologico de superficie.

As surface archaeology has steadily gained acceptance over the past two decades, so also has there
been an increasing call to perform more analysis during fieldwork. Cost savings and reduced impacts
to the record have led some government agencies to institute "no-collection" policies that apply in
most situations. In the state of Nevada, for example, the Bureau of Land Management (BLM) policy,
although flexible, is that "artifact collection can entail considerable direct and indirect costs both
to the operator and to the BLM and should be avoided whenever possible" (BLM 1985:5). As an
alternative, field workers are encouraged to record artifact information "on-site."'
Some form of on-site analysis, of course, is a routine component of all archaeological reconnais-
sance and survey. Site locations are recorded, deposits are evaluated in terms of estimated artifact
density and temporal affiliations, and so forth. In our research in eastern Nevada, for instance, we
follow the codification procedures of the Intermountain Antiquities Computer System (IMACS),
which requires a considerable number of analytic decisions to be made in the field. Aside from
general information, such as the artifact categories represented (e.g., ceramics, lithics, bone) and
features encountered, the surveyor is asked to make more specific decisions such as lithic flaking
stages present and their relative frequencies, density of artifact types present, and so forth. The
point of gathering such survey information is to eventually provide a data base for each region that
can be used for comparative analytic purposes.
The information contained on the IMACS forms, however, as well as information gained in
similar contexts elsewhere, often suffers due to the variable way in which it was gathered; that is,
because information is not necessarily acquired systematically by equally well-trained personnel,

CHARLOTTE BECK and GEORGE T. JONES * Department of Anthropology, Hamilton College, Clinton,
NY 13323

American Antiquity, 59(2), 1994, pp. 304-315.


Copyright ? 1994 by the Society for American Archaeology

304
REPORTS 305

the data produced may not be comparable across investigators and/or geographic areas. Although
the no-collection policy has been called into question by a number of investigators (e.g., Butler
1979; Schiffer 1975), the reliability of field analyses has not often been systematically addressed in
the literature. For instance, what kinds of data can feasibly be obtained under field conditions and
what sorts of analyses might be precluded by these conditions? Further, what is the degree of accuracy
that can be obtained; that is, can precise, reliable data be obtained in a field situation where time
is limited and certain conditions cannot be as easily controlled as in a laboratory environment?2
Finally, what is the cost; are the savings of an on-site approach sufficient to warrant noncollection?
In examining the pros and cons of an on-site strategy, however, one other factor might be
considered as well-the impact of such an approach on the surface record itself; that is, the potential
to maintain the integrity of the artifact distribution. While it is clear that collection destroys the
integrity of the deposits, what impacts result from procedures involved in on-site recording, e.g.,
laying out a grid, walking through the grid units, and picking up objects? A no-collection policy
may appear to conserve archaeological resources, but on-site analyses may in fact cause greater
disturbance than might be imagined. Such an impact should figure into the replicability of any
analytic program.
The purpose of this paper is two-fold: first, to evaluate an on-site approach, addressing the concerns
outlined above, and second, to evaluate the impact of such an approach on the surface archaeological
record. We use data obtained from a basalt quarry located in the southeastern portion of Little
Smokey Valley, eastern Nevada (Figure 1), which was discovered in 1989 during a cultural-resource
survey (see Price 1989).

PROCEDURES
The Little Smokey Quarry coincides with an extensive alluvial fan containing fine-grained andesitic
cobbles. Price (1989) described several dense concentrations of debris along an alluvial terrace in
the northern sector of the site. These concentrations were reported to contain biface-manufacturing
debris including biface-production preforms, cores, and debitage; therefore we decided to focus our
study on this terrace. Our work began with a preliminary walk-over of the terrace, after which we
selected an area of high artifact density measuring 16 x 30 m (480 m2). A grid of 2-x-2-m units was
placed over the entire area, and then artifact analysis proceeded in two phases.

Phase 1 Analysis
In Phase 1, analysis of unifaces, bifaces, and cores was carried out in all grid units. These objects
were weighed using a battery-operated digital balance (with a capacity of 100 g and accuracy to .01
g), identified as to material, and photographed. In addition, cores were described, noting the number
of platforms and their locations. Measurements of length, width, and thickness were taken on
unifaces. Because the non-debitage assemblage was composed primarily of bifaces, however, we
focused most of our attention on this artifact category.
In addition to obtaining information concerning weight and raw material, we identified bifaces
along several dimensions, including the presence or absence of cortex, stage of reduction, type of
breakage, and degree of symmetry. Two thickness measurements were taken, one about a centimeter
from the base (T,) and the second at the point of maximum thickness (T2). Finally, a plan drawing
was made of each biface.
Plan drawings were made for two reasons. First, length and width measurements could be made
from them at a later date if deemed necessary. Second, the drawings could be used to provide an
independent check on manufacturing-stage assignments. We used a biface classification in which
four criteria serve as the basis for class distinctions: (1) degree of bilateral symmetry, (2) conformation
of the edge profile, (3) scale of flake removals, and (4) presence of haft (see Beck and Jones 1990).
Biface-reduction sequences are often regarded as problematic because application of stage typologies
requires the imposition of structure on what are usually empirical continua (Muto 1971; also see
306 AMERICANANTIQUITY [Vol. 59, No. 2, 1994]

Figure 1. Map of eastern Nevada showing the locations of the Little Smokey Quarry (A) and the Butte
Valley study area (B).

Sullivan and Rozen 1985). Thus it is prudent to evaluate them, as with the results of all stage
classifications, against an independent standard. Elsewhere we have used a biface-thinning index
devised by Johnson (1981) for this purpose, computed as weight divided by plan area (see Jones
and Beck 1990). Thus, the drawings allow the measurement of plan area, a task more easily completed
in the laboratory. A plan drawing was also made of the single uniface encountered.
In this phase of the analysis, one analyst recorded all quantitative and qualitative measurements
on a form while the second made the plan drawings and took photographs. If more than one item
was found in a single grid, which often happened, the first analyst attached small adhesive labels
containing consecutive field-specimen numbers that were recorded on both the analysis form and
drawings. Although these labels were temporary, the drawings could be used to identify which
specimen was assigned which number if the study were to be replicated. The results of the Phase 1
analysis are shown in Table 1.

Phase 2 Analysis
In Phase 2, flake analysis was carried out in four randomly selected grid units, as high flake density
precluded total coverage. All flakes in the grids chosen were weighed and identified along five
REPORTS 307

Table 1. Artifact Representation at the Little Smokey Quarry.

Artifact Count
Tool category
Unifaces 1
Bifaces
Stage 1 35
Stage 2 58
Stage 3 26
Stage 4 7
Out of sequence 3
Missing data 1
Cores 6
Total 137
Flake type
BRF 44
Non-BRF 5
Shatter 123
Total 172

dimensions: type of material, presence of cortex, flake type (biface reduction, non-biface reduction,
fragment), platform type (cortex, crushed, single faceted, multifaceted), and type of termination.
The general results of the Phase 2 analysis are also shown in Table 1.

Collection and Laboratory Analysis


Going into this project we had substantive as well as methodological goals in mind. In addition
to evaluating the accuracy of analyses conducted in the field, we also wanted to acquire preliminary
information concerning assemblage content at this quarry for comparison with data we have collected
elsewhere in the region. We were hesitant, however, to make any collections without having an
overall research design, but after returning from the field we realized that we would be unable to
evaluate our analyses effectively without making collections.3 Thus in 1992 we returned to the quarry
to collect these artifacts. Collection proceed in exactly the same manner as had the on-site analysis,
with one exception. A randomly chosen 10 percent sample of grid units (n = 12) was chosen for
flake collection. The larger flake sample was collected in order to permit not only an evaluation of
our analytic results but also of the adequacy of our earlier small sample (3.33 percent) of four units.
Once in the lab, the 1990 plan drawings were used to identify those specimens collected. To our
surprise, of the 130 bifaces analyzed in 1990, only 113 were relocated in 1992. All of the seven
cores and the single uniface were relocated. In addition, 26 bifaces that had not been located in
1990 were collected in 1992. These results were unexpected since we had assumed that the terrace
surface and thus artifact provenience were relatively stable, an assumption that may have been
unwarranted.
Of the 113 bifaces relocated, 20 were found in different grid units, most of which were adjacent
to the ones recorded in 1990; however, three bifaces were found in very different locations, as much
as 12 m away from their original locations. Displacements of one grid unit might be explained by
a slight difference in the placement of the grid from one field season to the next, but the extreme
displacements cannot be explained so easily. Considered together with the fact that 17 bifaces were
not relocated while 26 additional ones were, these displacements suggest that surface stability and
locational integrity are not nearly so great as we had imagined. We will return to this point below
when discussing the impact of our on-site activities on the integrity of the surface artifact distribution.
Laboratory analyses of all artifacts were conducted along the same lines as on-site analyses, with
three exceptions. First, a microscope was available for closer examination of artifacts; second, a
308 AMERICANANTIQUITY [Vol. 59, No. 2, 1994]

Table 2. Comparisonof Field and LaboratoryQuantitative


Measurements of Bifaces.

Variable/
Statistics Field Results Lab Results Difference

Tl (mm)
Mean 6.79 5.54 1.20
S.D. 1.84 1.59 1.18
Range 3.80-13.40 3.10-11.15 0-5.40
T2 (mm)
Mean 12.46 12.70 .40
S.D. 4.28 4.46 .39
Range 6.95-31.17 3.60-32.50 0-2.60
Weight(g)a
Mean 29.30 30.61 .10
S.D. 20.18 20.55 .40
Range 4.09-88.68 4.18-92.64 0-3.96
a
Weightincludesonly those objectsweighingless than 100 g becauseof the
limits of the battery-operated balance used in the field analysis.

larger-capacity (2,000 g) balance permitted us to weigh objects that were too large for the portable
field balance (see below); and finally, plan drawings were not made, as the artifacts were available
for repeated inspection.

EVALUATION OF THE PROCEDURE AND RESULTS


The use of an on-site analysis program can be evaluated along several lines, including the reliability
of the results,4 the feasibility of particular procedures, and cost. Given that the greatest number of
observations were made on bifaces and that field and laboratory observations can be compared on
a case-by-case basis (flakes were not given individual specimen numbers, thus results cannot be
compared in this manner), we will focus our comparisons of field and lab results on this category
of artifacts.

Evaluation of Results of Biface Analysis


Reliability is achieved if a set of procedures applied to the same data set at different times or by
different people yields the same results. We should not expect, however, that all variables can be
measured with equivalent precision, because some use instruments (with known precision) and
others rely on the discretion of the analyst. Those that should be the most reliable are metric
observations, while those that should prove to be the least so are qualitative variables that require
decisions on the part of the analyst.
Several metric measurements were made on the quarry bifaces, including weight and two thickness
measurements. A comparison of field and laboratory results of these measurements is shown in
Table 2. Although not identical, these results indicate only modest differences between field and
laboratory measurements; the smallest difference is for weight while the largest is for Tl (basal
thickness). The correlations between field and lab measurement of both weight (r = .984) and
maximum thickness (T2) are strong (r = .995) while that for T, is somewhat less so (r = .886). The
lower correlation for T1 is understandable in light of this variable's sensitivity to the location of
measurement; a slight difference in the positioning of the caliper in one direction or another may
yield quite different results.
The remaining four variables -the presence of cortex, type of material, degree of symmetry, and
stage of manufacture-are all qualitative. Distinctions as to presence or absence of cortex and
material type are quite clear-cut. The other two variables, however, are complex, continuous vari-
REPORTS 309

Table 3. Correspondence Between Field and Laboratory Results


of Qualitative Measurements on Bifaces.

Agreements Disagreements
Variable N % N %

Type of material 113 100.00 0 .00


Presence of cortex 104 92.04 9 7.96
Degree of symmetry 75 66.37 38 33.63
Stage of manufacture 84 75.00 28 25.00

ables that have been treated in ordinal fashion and as such, measurement reliability rests on precise
definitions of categories. Nevertheless, such decisions often contain some subjective element and
are by definition less reliable. Table 3 shows the correspondence between field and lab decisions
for these four variables. As might be expected, there is complete agreement concerning the type of
material, as 97 percent of the bifaces are of andesite. There is also high agreement (92.04 percent)
between decisions as to the presence or absence of cortex. Had this variable been related to the
proportion of surface covered by cortex (as is done often in flake analyses), agreement between field
and lab decisions would likely have been somewhat lower (see Sullivan and Rozen 1985).
There was much less agreement concerning symmetry; in only 66.37 percent of the cases did field
and lab assignments agree. Although subjectivity again can be cited as a major cause of the difference
in assignments, this lack of agreement is also likely a result of imprecise definitions of categories.
For example, how much must symmetry deviate from "complete" to become "moderate"? Further,
how much of the biface must be present to make this judgment? As Table 4 shows, 16 (42.1 percent)
of the disagreements arise because the analyst evaluated symmetry in one case but found
the object too fragmentary to do so in the other. Thus, the lesson here is that clear, precise definitions
of categories are very important in rendering reliable results, whether the analysis is being done in
the field or in the lab. This is especially true when dealing with a continuous variable, such as
symmetry, that has been broken into discrete ordinal categories.
Finally, we come to the biface-reduction-stage assignments. Table 5 shows field vs. lab assignments
of bifaces to stage categories. As these results show, assignments to Stage 4 as well as the "out-of-
sequence" assignments did not differ between years; however, 11 items assigned to Stage 1, 16 items
assigned to Stage 2, and 1 assigned to Stage 3 in the field were assigned to a different stage in the
lab. The pattern of difference appears to be directional; in all but two cases lab assignments were
to the next highest stage.
Overall there is a 75 percent agreement between field and lab stage assignments. Such a large
number of disagreements might be expected given the nature of biface-reduction classifications. As
mentioned earlier, such classifications are always problematic, because it is difficult to identify
qualitative or quantitative attributes that successfully break apart a reduction continuum into discrete

Table 4. Field vs. Laboratory Measurements of Biface


Symmetry.

Symmetry Symmetry (Laboratory Results)a


(Field
Results) 1 2 3 4 9
1 1 0 0 1 0
2 1 20 13 7 0
3 0 6 46 5 0
4 1 1 1 7 0
9 0 0 0 0 1
a
Symmetry Codes: 1 = low symmetry, 2 = moderate symmetry, 3 = com-
plete symmetry, 4 = not measurable, and 9 = not applicable.
310 AMERICANANTIQUITY [Vol. 59, No. 2, 1994]

Table 5. Field vs. LaboratoryBiface StageAssignments.

Stage
(taeld (Laboratory Assignment)
(Field
Assignment) MDa 1 2 3 4 9b

MD 0 0 0 1 0 0
1 7 17 10 1 0 0
2 4 0 38 16 0 0
3 5 0 1 20 0 0
4 1 0 0 0 6 0
9 0 0 0 0 0 3
Total 17 17 49 38 6 3
a Missingdata.
b
Out of sequence.

segments that prove to be effective classificatory units. As in the case of symmetry, however, 14
(50 percent) of the disagreements occurred because the biface was very fragmentary (either a small
piece of the midsection or a tip), and a field determination probably should not have been made.
Further, comparative material available in the lab permitted decisions that were more consistent
with those made in previous studies.
The Johnson thinning index mentioned earlier was calculated for both sets of analysis results. As
a biface is progressively reduced, its thinning index will decrease with a reduction in thickness. If
the assignments of bifaces to stages is correct, the average value of the thinning index of each
successive stage should drop. Deviations from this general pattern suggest problems of identification
using the stage typology.5 As can be seen from Table 6, in both sets of results the average index
falls in each successive stage, with the only change being an increase in this average in the laboratory
results over the field results. If the results of the lab analysis were more accurate than those conducted
in the field, it might be expected that the ranges for each stage would show less overlap. This does
not occur; the range for Stage 1 is slightly smaller while that for Stage 3 is slightly larger.
These results suggest that neither lab nor field results for biface class are substantially better or
worse as to their general utility for distinguishing stage of manufacture. As long as these data are
used in a largely comparative visual sense, the low reliability of biface-stage assignments is not of
great concern. Examining the two sets of results with this in mind, the relative proportions of each

Table 6. Comparison of Statistics from the Johnson Thinning


Index for Field and Laboratory Results.

Stage
Statistics 1 2 3 4

Field
N 27 53 21 7
Mean 2.12 1.79 1.39 1.32
S.D. .56 .40 .23 .18
Range 1.18-3.24 1.10-3.39 1.04-1.86 .99-1.55
Laboratory
N 17 48 36 6
Mean 2.31 1.83 1.49 1.38
S.D. .52 .43 .29 .11
Range 1.48-3.24 1.10-3.29 1.04-2.16 1.27-1.55
Note: Statistics reflect results for only the 112 bifaces for which both field
and laboratory results are available.
REPORTS 311

Table 7. Results of Chi-SquareTests of ComparisonsBetween


Field and LaboratoryResults of Flake Analysis.

Calculated Significance
Variable x2 df Level
Type of flake 1.36 2 .52
Presenceof cortex .37 1 .56
Type of platform 8.86 4 .07
Type of termination 4.49 3 .22
Type of material .20 1 .67

stage do not differ significantly overall, as revealed through a Kolomogorov-Smirnov test, which
showed no significant difference even at the .10 level. This same point is also suggested by the results
of the Johnson thinning index. If, on the other hand, much is to be made from quantitative
comparisons, low reliability might well be a serious concern.

Evaluation of Results of Flake Analysis


As we noted earlier, comparisons of field and lab results on a case-by-case basis were not possible,
due to the fact that flakes were not originally given individual field specimen numbers. Thus, in
1992, we decided to collect a larger sample of flakes in order to evaluate the adequacy of our original
small sample relative to the types of analyses undertaken. Of the five qualitative variables measured,
only the assignment as to platform type differed between field and laboratory results (Table 7) (X2
= 8.86; p = .07). This result can be attributed to a difference in the presence of multifaceted platforms
in the two samples. In comparing flake weights, the mean flake weight in the 1990 sample is 7.65
g, while the mean for the 1992 sample is 10.62 g.6 A calculated t of -2.2 (df= 410) shows these
means to differ at the .03 level of significance.
These results suggest that the small sample was adequate for the evaluation of presence of cortex,
flake type, and termination, but may not be adequate to evaluate type of platform or weight.
Regarding the latter,
the presence of a small number of very large flakes in the 1992 sample could
easily drive the mean up, but a comparison of distributions as well as modes (3.58 mm for the 1990
sample and 5.42 mm for the 1992 sample) suggests this is not the case and thus that the small
sample analyzed in the field was not a good reflection of the average flake size. Another possibility
is that new flakes may have been exposed over the two-year interval between the two analyses (see
below), which might account for the differences.

Feasibility
The above discussion suggests that a number of procedures beyond simple counts of items or
identification of general technological categories are feasible without collection. Measurements, such
as length, width, and thickness of items such as cores, bifaces, and unifaces are easily obtained using
calipers and, as shown above, can yield reliable results. Such measurements on flakes, however,
might prove more problematic and time-consuming. We would suggest, instead, the use of a pre-
determined set of size categories; each analyst would carry a key on which the plan area is drawn
for each category and the flake can quickly be identified with the appropriate category. Although
we did not use this approach here, we have done so elsewhere, and it is very effective.
Obtaining weights using a portable digital balance is somewhat more tedious but is, as our study
shows, certainly feasible. The portable balance we used runs on two nine-volt batteries and will run
for between three (using carbon batteries) and 10 (using alkaline batteries) hours on one set. We
had no trouble keeping the balance level, as the surface on which we worked was relatively hard
and flat. In a different situation, however, such as a plowed field, it might be necessary to set the
balance up on a flat surface, such as a piece of plywood, and move the platform as necessary.
Unfortunately, our portable balance did not have the capacity to weigh objects the size of any of
312 AMERICANANTIQUITY [Vol. 59, No. 2, 1994]

Table 8. LaborExpendedin PersonHours in On-SiteVs.


CollectionStrategiesat the Little Smokey Quarry.

Person
Task Hours
On-site strategy
Layingout grid 1
Analysis(includingplan drawingsand photographs)of bifaces,
unifaces,and cores 19
Analysisof flakes 8
Total 28
Collectionstrategy
Layingout grid,collection, and processing(packaging,mailing) 5
Laboratoryprocessing(washing,numbering,accessing) 18
Analysisof bifaces, unifaces,and cores 8
Analysisof flakes 5
Total 36

the cores we encountered (ranging from 373.60 to 1,047.70 g) as well as eight of our larger bifaces
(ranging from 100.41 to 253.78 g), but a balance with a 1,000-1,500 g capacity should accommodate
most objects.
Many qualitative measures, such as those performed on the flakes-type of flake, type of platform,
etc. -are also quite feasible for an on-site approach, especially in view of the information these data
can ultimately provide. Some qualitative measurements, however, such as reduction stage, although
quite feasible, can yield more problematic results, especially because the expertise of field investi-
gators often differs. Thus, some sort of additional procedure, such as plan drawings, that can
eventually be used as a test of the field assignments should also be incorporated into the program.
Although these drawings add considerably to field time, plan drawings can be quite useful since
they can provide additional information at a later date. One of the most common practices, for
instance, is the assignment of assemblages to temporal periods based on in-field examination of
projectile points in the assemblage. It has been our experience that such assignments are often in
error due to the variable experience of field investigators as well as the implicitness of many projectile-
point typologies. Plan drawings of these items provide the possiblity for reexamination as well as
additional measurement back in the lab.

Cost

The cost of our procedures can be evaluated in terms of both labor (e.g., person hours) and money.
Table 8 shows the labor expended in both the on-site and the collection approaches at the Little
Smokey Quarry. As can be seen from this information, the collection approach took an additional
eight person hours to complete, an increase in cost of 28.6 percent. Obviously these numbers cannot
necessarily be applied directly to every situation. There are a number of factors that affect the
amount of field time for either an on-site or a collection strategy, such as the amount of area covered,
the type and size of analytic/collection units used, artifact density, vegetation, complexity of terrain,
training of personnel, and so forth. Thus the relative time spent in an on-site vs. a collection strategy
will vary, depending upon the circumstances.
Monetary costs will also vary according to circumstances, i.e., whether the project is a cultural-
resource-management (CRM) project being run by a small or large company, a research project with
hired personnel, or a research project utilizing a field school. Field costs may be quite high if the
project is being run by a large company with high overhead but lower if a field school is being used.
In comparing on-site with collection strategies, field costs may be lowered in some ways and increased
in others. In an on-site strategy no collection, packaging, or mailing supplies are necessary, thus
reducing costs; but if field time is greater than in a collection approach, as in the present case, salary,
REPORTS 313

per diem, and transportation costs will increase. For example, in a CRM situation field workers
might be paid from $8.00 to $12.00 per hour, depending upon their responsibilities and expertise,
plus a per diem of between $25.00 and $50.00 per day. Thus in the present case, the cost would
range between $311.50 and $511.00 for salary and per diem alone in an on-site approach compared
with between $65.00 and $110.00 in a collection strategy. These figures, however, do not include
benefits, which may run anywhere from 10 to 25 percent.
Lab and curation costs, however, are completely eliminated in an on-site approach. Lab processing
(e.g., washing and labeling) in our case was completed by a student who was paid $4.25 per hour
for 18 hours of artifact processing, resulting in a cost of only $76.50. In a contract situation, however,
processing and analysis may be completed by the same individual, at a wage of between $8.00 and
$12.00 per hour, resulting in a cost of between $144.00 and $216.00 for 18 hours of processing and
between $104.00 and $156.00 for 13 hours of artifact analysis. A graduate assistant, being paid
between $12.00 and $15.00 per hour for analysis, would cost between $156.00 and $195.00 for the
13 hours of analysis. Again, benefit costs are not included in any of these calculations.
Curation costs, although they will vary considerably, can be extremely high. The standard cu-
ratorial charge of the Nevada State Museum, for example, is $1,100.00 per cubic foot of material.
In the quarry case, about 1.15 cubic feet of artifactual material was collected, resulting in a curation
cost of $1,265.00. Although by strict interpretation of the Antiquities Act of 1906 the Federal
government is required to pay curation costs for artifacts collected on federal land (Butler 1979),
this cost is in reality more often
the responsibility of the individual or company doing the work. It
is common in CRM projects going out for bid that curation costs be included in those bids. In
addition, federal permits often contain the stipulation
the that permitee shoulder the cost of curation.
Thus, although on-site procedures may increase field costs, the elimination of lab and curation costs
make them less expensive overall.

IMPACT ON THE SURFACE ARTIFACT RECORD


Although one of the arguments in favor of a no-collection approach is to maintain the integrity
of the surface artifact record (Butler 1979), on-site analysis can have substantial impact on that
record. For instance, when an artifact is picked up for examination, how careful is the investigator
to replace it exactly as it was found? How careful are the analysts when walking around not to step
on artifacts, potentially damaging or burying them under a few millimeters of sediment? Before
addressing these questions we must address the larger question of the stability of the surface record
at this site, given the problems discussed earlier regarding differences in provenience and represen-
tation between artifacts analyzed in 1990 and those collected in 1992.
As a result of work we have done elsewhere in eastern Nevada on late Pleistocene/early Holocene
materials in similar environmental situations (Figure 1), we had assumed that the terrace surface
on which the quarry artifacts lay was relatively stable. In fact, our results support this assumption
to some extent in that 100 of the bifaces and cores analyzed in 1990 were relocated within the same
grid units two years later. Further, 17 of those that had a different provenience in 1992 were found
in grid squares adjacent to their original ones. These objects may have been near the boundaries of
the grid units and thus the slightly different placement of the tapes between the two years could
account for the difference. On the other hand, we may not have been as careful in replacing them
as we originally thought.
The drastically different provenience of the remaining three objects, however, is not so easily
explained; neither is the failure to relocate 17 of the original bifaces nor the location of 26 new
ones. Regarding the change in provenience, this phenomenon might be explained by visitors to the
quarry picking up objects and carrying them some distance before putting them down. This area of
the quarry has, in fact, been shown to visitors on several occasions. It is not likely, however, that
the quarry has been subject to vandalism, as it is located in a relatively isolated valley that sees
very little human activity. It is also unlikely that this movement occurred through water activity,
since the area studied does not occur on or near a slope and the artifacts in question are quite large,
ranging in weight from 27.81 to 48.39 g. Studies focusing on surface artifact movement have shown
314 AMERICANANTIQUITY [Vol. 59, No. 2, 1994]

that large objects tend to be more stable in the horizontal dimension than smaller ones, especially
on relatively flat surfaces (e.g., Wandsnider 1989).
The second phenomenon-missing and additional artifacts-is a different problem altogether.
This suggests that artifacts may become buried under a few millimeters of sediment, thus hiding
them from view. We experienced this in our work in southeastern Oregon (Beck 1984; Jones 1984)
in which sites literally "disappeared" from one season to the next. In all of these cases, however,
sites were in low-lying areas where sedimentation was more likely to occur; there is no indication
that such sedimentation has occurred on the surface of the terrace where our study was focused.
The more likely explanation is excessive walking over the site, which could result in the shallow
burial of some artifacts and the exposure of others. As stated above, the quarry is not likely to have
seen a great deal of human activity given its isolated location and thus we must conclude that our
activities there are at least in part to blame. We were very careful to replace each artifact as closely
as possible in the position in which it was found; we also took care not to step on artifacts if at all
possible. Nevertheless, our activities probably did have an impact on artifact distributions, even
though we took steps to minimize that impact.

CONCLUSIONS
Our study at the Little Smokey Quarry has several implications. First, a number of different
analytic approaches can be feasibly and accurately applied in the field if they are implemented
systematically and consistently. Depending on the types of analyses, costs may be reduced consid-
erably, especially if curation is eliminated. On the other hand, on-site analyses do not come without
a cost to the archaeological record; that is, some degree of disturbance will occur even when measures
are taken to prevent it. Thus the claim cannot be made that a no-collection strategy preserves the
integrity of the record, because in many contract surveys measures are not taken to limit the impact.
The point of this paper has not been to hold on-site analysis up as the approach of choice.
Admittedly, there are a number of advantages to collection over analysis in the field, not the least
of which is the opportunity for reanalysis of assemblages or the application of new analyses to those
assemblages. If, however, circumstances do not permit collection, then a well-planned, systematic
on-site approach can be a viable alternative. The most practical solution in the end might be a
combination of collection and on-site approaches, which would allow a larger number of artifacts
to be examined at a lower cost while also having the advantage of a collected sample of those
artifacts to be studied in the lab.

Acknowledgments. We would like to thankJames Ebert,MichaelSchiffer,Steven Simms, and severalanon-


ymous reviewersfor their helpful comments on previous versions of this manuscript.We would also like to
thankEly DistrictArchaeologistBrianAmme for bringingthe Little SmokeyQuarryto our attentionas well as
providingcertain pieces of equipmentthat allowed the undertakingof analyses duringthe summer of 1990.
MariaNieves Zedefiopreparedthe Spanishversion of the abstract.

REFERENCES CITED
Beck, C.
1984 Steens Mountain Surface Archaeology: The Sites. Ph.D. dissertation, University of Washington. Uni-
versity Microfilms, Ann Arbor.
Beck, C., and G. T. Jones
1990 The Late Pleistocene/Early Holocene Archaeology of Butte Valley, Nevada: Three Season's Work.
Journal of California and Great Basin Anthropology 12:231-261.
Bureau of Land Management
1985 Cultural Resources Survey. General Guidelines. 3rd ed. Reno, Nevada.
Butler, W. P.
1979 The No-Collection Strategy in Archaeology. American Antiquity 44:795-799.
Johnson,J. K.
1981 Lithic Procurement and Utilization Trajectories. Analysis, Yellow Creek Nuclear Power Plant Site,
Tishomingo County, Mississippi, vol. II. Archaeological Papers of the Center for Archaeological Re-
search No. 21. University of Mississippi, University.
REPORTS 315

Jones, G. T.
1984 Prehistoric Land Use in the Steens Mountain Area, Southeastern Oregon. Ph.D. dissertation, University
of Washington. University Microfilms, Ann Arbor.
Jones, G. T., and C. Beck
1990 Biface Reduction and Assemblage Variability in Late Pleistocene/Early Holocene Assemblages from
Eastern Nevada. Paper presented at the 22nd Biennial Great Basin Anthropological Conference, Reno,
Nevada.
Muto, G. R.
1971 A Stage Analysis of the Manufacture of Stone Tools. In Great Basin Anthropological Conference 1970,
edited by C. M. Aikens, pp. 109-118. Selected Papers, University of Oregon Anthropological Papers
No. 1. Eugene.
Price, B. A.
1989 Archaeological Survey of Seismic Test Lines in Chevron Exploration and Production Services' Pancake
Prospect, White Pine, Eureka, and Nye Counties, Nevada. Bureau of Land Management Cultural
Resources Report 6-956. Copies available from Bureau of Land Management, Reno, Nevada.
Schiffer, M. B.
1975 Archeological Research and Contract Archeology. In The Cache River Archeological Project: An Ex-
periment in Contract Archaeology, assembled by M. B. Schiffer and J. H. House, pp. 1-7. Research
Series No. 8. Arkansas Archeological Survey, Fayetteville.
Sharer, R. J., and W. Ashmore
1987 Archaeology: Discovering our Past. Mayfield, Palo Alto, California.
Sullivan, A. P. III, and K. C. Rozen
1985 Debitage Analysis and Archaeological Interpretation. American Antiquity 50:755-779.
Wandsnider, L.
1989 Long- Term Land Use, Formation Processes, and the Structure of the Archaeological Landscape: A Case
Study from Southwestern Wyoming. Unpublished Ph.D. dissertation, Department of Anthropology,
University of New Mexico, Albuquerque.

NOTES

'We use the term "site" to refer to location rather than the more traditional archaeological meaning of "a
spatial clustering of archaeological data" (Sharer and Ashmore 1987:600). In this sense on-site analysis refers
to analysis undertaken at the field locations of artifacts, including those within concentrations as well as isolates.
2 James I. Ebert
(personal communication 1993) has suggested that some conditions are more controlled in
a field situation. For example, if artifact analysis is done in the field over a short period of time and the recorders
are working together, communicating their uncertainties to one another as they go, their results will be more
consistent than if the analysis is done in the lab over a longer period of time by a series of analysts working
alone.
3 Results of our on-site analyses have provided the basis for construction of a research design for the quarry.
Additional collection is planned for future field seasons.
4 Obviously we cannot evaluate reliability here in the strict statistical sense, because (a) the artifacts were
collected two years after the initial analysis, and (b) we did not do repeated tests in the lab. It is our opinion,
however, that these comparisons do yield information concerning the confidence one might have in the results
of certain types of analyses over other types in a field situation.
5 As the projectile point that appears to be the ultimate product of this manufacturing sequence is believed
to have been socketed rather than bound to the shaft, it retains a thick cross section, especially in the haft area.
The consequence is that the thinning index may level out or even reverse once a desired thickness is achieved.
6 Because the
portable balance used in the on-site analysis had a limitation of 100 g, these means reflect only
those items that weigh less than 100 g.

Received April 4, 1991; accepted September 20, 1993

S-ar putea să vă placă și