You are on page 1of 27

Artifact Size and Spatial Process:

Macro- and Microartifacts in a


Mississippian House
Sarah C. Sherwood, Jan F. Simek, and
Richard R. Polhemus
Department of Anthropology, University of Tennessee, Knoxville,
Tennessee 37996-0720

Despite their evident utility in archaeological analysis, microartifacts (those artifacts


smaller than 2 mm) have, as a class, been used only sparingly by archaeologists and
then only as if they were larger artifacts. This article explores the variable information
microartifacts contain, using a case study from the Loy Site, a Mississippian village in
East Tennessee. We show that microartifacts provide different information from that
obtained from macroartifacts. We also demonstrate that comparisons among different
size classes provide new information on dynamic aspects of artifact deposition. Especially
in spatial analysis, comparing distributions of different size artifacts enhances our inter-
pretative capabilities over any single artifact size class. ‘i:1995 John Wiley & Sons, Inc.

INTRODUCTION
Spatial analysis in archaeology has long focused on the identification of
“activity areas,” defined as locations where people performed specific and lim-
ited tasks (Whallon, 1973). Initially, this interest was linked to a desire for
broad, ethnography-like reconstructions of archaeological sites. To this end,
various techniques were employed to examine distributions of artifacts presum-
ably used (e.g., tools and containers) and produced (e.g., ceramic sherds and
debitage) during activities. Because of this interest and the prevailing artifact
collection methods linked to it, archaeologists concerned themselves with those
artifacts large enough to be “used” or “manipulated,” (i.e., artifacts over a few
centimeters in size). More recently, most archaeologists have realized that a
variety of problems afflict ethnographic approaches to the archaeological
record, including biogenic, geogenic, and anthropogenic agents that act over
time to reduce concordance between artifact spatial positions and past activity
areas (Schiffer, 1987). Many of these agents affect not only artifacts large
enough to have been used, but also artifacts smaller than those collected by
our usual recovery techniques. If the complex formation of artifact spatial
distributions is to be accurately explained, then many previously ignored attri-
butes, even classes, of artifacts must be examined.
One such class of materials is “mi~roartifacts,’~
sand-sized particles less than
2 mm, generated and incorporated into the record in great numbers by a variety

Geoarchaeology: An International Journal, Vol. 10, No. 6, 429-455 (1995)


0 1995 by John Wiley & Sons, Inc. CCC 0883-6353/95/060429-27
SHERWOOD, SIMEK, AND POLHEMUS

of formation processes both human and natural (Dunnell and Stein, 1989; Stein
and Teltser, 1989). Microartifacts may enhance or provide new information on
site formation processes and complement larger artifacts. For example, it is
believed that microartifacts are more likely to be embedded into the surfaces
where they are generated and less likely to be moved by curation or cleaning
than are larger artifacts (Fladmark, 1982). Microartifacts have been used in
recent years to address such diverse subjects as reconstruction of depositional
histories, site formation, regional land use, and intrasite activity organization.
Nevertheless, we believe their true value has yet to be realized.
Despite their evident utility in activity area studies, we would argue that
microartifacts have been used sparingly and then only as ifthey were larger
artifacts. In other words, distributions of microartifacts are examined, and
when concentrations are found, they are interpreted as reflecting activity areas.
Micro- and macrodata should be complementary, providing distinct but mutu-
ally illuminating information (Dunnell and Stein, 1989). Yet, spatial compari-
sons between small and large size classes (or “scales”) that might elucidate
site formation variation are not often performed. We believe that a great deal
of information lies precisely in this comparative study. In this article, we
outline attempts to use spatial distributions of both macro- and microscale
artifacts to interpret archaeological spatial formation processes. A Mississip-
pian house floor provides data for our analysis. Through study of the house
floor, we show how comparisons among artifact size class distributions can
provide insights into the varied processes that form the archaeological record.
THEORETICAL CONSIDERATIONS
While Hassan (1978)was probably the first to recognize and analyze artifacts
so small as to be visible only under a microscope, Fladmark’s (1982) study of
microdebitage generated during knapping experiments alerted most archaeolo-
gists to the potential information contained in microartifacts. Despite this
recognition, however, and despite theoretical work placing microartifact analy-
sis firmly within the wider context of artifact studies (Dunnell and Stein, 19891,
analyses of these objects remain rare in the literature. We believe there are
two reasons for this: First, recovery, identification, and quantification of mi-
croartifacts has been a time-consuming and costly enterprise (Stein and Teltser,
1989), and few people are trained to undertake it. Second, while important
theoretical issues were delineated by Dunnell and Stein (19891, these have
failed to impact methodologies to the point where the importance of analyzing
microartifacts has become clear to a majority of archaeologists, especially con-
sidering the costs such study typically entails.
In their seminal 1989 article, Dunnell and Stein outline some of the impor-
tant characteristics of microartifacts that compel their consideration as archae-
ological data of the first order. They note that these small objects conform to
most normal definitions of “artifact,” and therefore constitute recognizable and
often numerically important elements of the archaeological record. Conserva-

430 VOL. 10, NO. 6


MACRO- AND MICROARTIFACTS IN A MISSISSIPPIAN HOUSE

tion of the record alone would seem to be reason enough to collect representative
microartifact samples. Moreover, information content may be different for
microartifacts than for larger artifacts, according to Dunnell and Stein, and
they may be most informative about different things (e.g., particle transport
and site formation processes). Equally important, processes that generate mi-
croscopic artifacts vary depending on material and context (Dunnell and Stein,
1989:34-36). These last two issues, differing information content and differing
formation processes within the microscale, are, we believe, especially important
reasons for undertaking microartifact analysis. Information that may not be
available from other aspects of the record will be lost if microartifacts are
ignored. Dunnell and Stein conclude: “While microartifacts supplement some
information obtained from larger artifacts (and this is the dimension along
which their utility has been judged thus f a r ) , most of the information they
contain complements that obtained from larger objects” (1989:31; emphasis
added). We believe that little has changed since Dunnell and Stein made this
observation 6 years ago.
Indeed, most applications of microartifact analysis treat small artifacts as
if they contain the same information as larger artifacts, that is, supplement
the information provided by bigger pieces. For example, Vance (1986, 1989)
used microartifacts to “test” activity area interpretations based on larger arti-
fact distributions a t the Duwamish Site in Washington and a t Robards Tract
in Missouri; in both studies concordance between micro- and macroscales was
seen as indicating correct activity area identification, while lack of concordance
indicated an incorrect identification. Madsen and Dunnell (1989) used micro-
artifacts to increase artifact sample size in low density plow zone contexts.
Metcalfe and Heath (19901, and Simms and Heath (1990) both employed mi-
croartifacts in order to enhance activity area interpretations within Fremont
Culture house structures in Utah; in both cases, the researchers are drawn to
interpretations of site formation processes by patterning in microartifacts
alone; they do not compare microartifact distributions to macroartifact pat-
terns. Other examples of this approach include Rosen (1989) and Reese (1986).
Thus, with few exceptions, microartifacts have been used as if they were larger
artifacts when large artifacts are rare or when there is a perceived need to
verify patterning observed a t the macroscale.
An important exception to this approach, and one that illustrates how we
believe microartifact analysis can make fundamental contributions to our un-
derstanding of the archaeological record, is Hull’s work on northern plains tipi
rings (Hull, 1983, 1987). Hull compared large lithic artifact distributions to
microdebitage distributions, developing an interpretive model to explain vari-
ous states of spatial concordance between scales. She defined three conditions
as follows:

(1) Primary (activity area) refuse is identified by a cluster of macrodebitage


corresponding to a cluster of microdebitage.

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL 431


SHERWOOD, SIMEK, AND POLHEMUS

(2) Secondary (discard area) refuse consists of macrodebitage with no cor-


responding cluster of microdebitage.
(3) De facto refuse, although difficult to distinguish from primary refuse,
should correspond to a microdebitage high density area while containing
relatively large macroflakes and possibly more tools or tool fragments
(Hull, 1987:773).

While Hull has some difficulty distinguishing her third model condition, we
believe that her instincts are fundamentally sound and that it is interpreting
variable concordance relationships that makes comparison between size scales
useful. We see her interpretive difficulties stemming from unwarranted links
between a specific systemic interpretative framework (Schiffer’sotherwise use-
ful descriptions of refuse deposition behavior) and straightforwardly archaeo-
logical observations (i.e., concordance relationships).
In simpler terms, we would define four concordance relationships between
macro- and microartifacts. (1)Spatial concentrations of microartifacts can coin-
cide with concentrations of large artifacts. (2) Macroartifact concentrations
may occur in areas where microartifacts are rare. These two relationships
correspond to Hull’s first two model conditions, although, for now, we consider
no particular interpretation. (3) Microartifacts can concentrate in locations
where large artifacts are few. (4)Neither large nor small artifact concentrations
are present in a given location.
The straightforward interpretation of these conditions in terms of the con-
texts defined by Schiffer (1976),whether systemic or archaeological, is compli-
cated by the raw material of the artifact class being considered. In short, spatial
coincidence among size classes means different things for different materials.
Thus, Hull’s specific interpretations for lithics cannot be generalized.
Raw materials must be considered because, following material type, two basic
processes of microartifact generation can be defined. One group of materials, for
example, stone and some metals, are relatively stable with respect to size once
they have been introduced into the archaeological record. It is true that certain
strong chemical environments and mechanical processes like plowing can pro-
duce microartifacts of these size-stable materials. But microartifacts generated
in this way come from macroartifacts already in the archaeological record; thus,
for the most part the size of artifacts in these materials is fixed at deposition.
(Chemical and mechanical breakdown of size-stable materials normally occur
over a fairly wide area, moreover, so the influence of these in microartifact
production should be general over the affected surface and can be perceived
in an even, low density “background” distribution of microartifacts.) In most
contexts, both microartifacts and macroartifacts, and their concordance rela-
tionships, can be interpreted in systemic terms (sensu Schiffer, 1976, 1987)
for artifacts produced in stable materials. Thus, even absolute microartifact
abundance can be straightforwardly seen as reflecting intensity of the past
activities that produced the small artifacts.

432 VOL. 10, NO. 6


MACRO- AND MICROARTIFACTS IN A MISSISSIPPIAN HOUSE

A second and interpretively more complicated set of artifacts, are “size-


unstable”-those made of materials that break down systematically in archaeo-
logical context. Friable materials like bone, shell, ceramics, and daub are easily
affected by chemical and mechanical deterioration, and it is safe to assume
that many, if not most, microartifacts of these materials are generated after
deposition. Concordance relations between sizes can be pivotal in understand-
ing the effects of post-depositional processes in this case, since the presence of
microceramics or microshell can only be explained if sources for those artifacts
(i.e., larger pieces) can be located. Microartifact abundance has no straightfor-
ward interpretation here, since it might reflect activity intensity, or (perhaps
just as likely) the length of time larger artifacts have been in their depositional
context and the rate and type of active weathering. With enough time and
weathering, large, degradable artifacts will break down, and microartifact
concentrations represent “shadows” of larger objects deposited by human ac-
tivity.
Thus, microartifacts are sand-sized artifacts that provide different informa-
tion than do larger artifacts. They should, therefore, be routinely compared to
larger artifacts in complementary fashion in order to better understand the
processes that produced a given part of the archaeological record. Such compari-
sons must take into account the spatial distributions within size classes and
must examine concordance relationships between scales. Material characteris-
tics and the process of microartifact generation should be considered in develop-
ing interpretations of observed patterning within and between scales. With
these considerations in mind, we turn to a n example of microartifact analysis
from Structure 3 a t the Loy Site, a Late (Dallas Phase) Mississippian village
in eastern Tennessee.

THE LOY SITE


The Loy Site (40JE10) is located on the first terrace on the east bank of the
Holston River, approximately 50 km north of Knoxville, Tennessee (Figure 1).
Based on its location and contents, the site has been assigned to the Dallas
Phase, A.D. 1300-1600. This phase has been observed extending south from
the lower French Broad and Holston drainage through the Tennessee River
Valley, and southwest into portions of Georgia and Alabama (Polhemus, 1987).
A testing, coring and magnetometer program conducted across the terrace has
defined a town covering an area approximately 120 by 210 m. The site includes
three separate episodes of palisade wall construction and expansion, a platform
mound, a central plaza, and a n estimated 60 structures.

Structure 3
Research a t Loy focuses on the spatial organization of the town with emphasis
on the household. Horizontal artifact distributions are an important focus of
this study. In this regard, we are particularly concerned with distinguishing

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL 433


SHERWOOD, SIMEK, AND POLHEMUS

Figure 1. Location map of the Loy Site (40JE10),East Tennessee.

between areas where activities actually occurred versus areas of discard (i.e.,
primary versus secondary deposits). This concern led us directly to the applica-
tion of microartifact analysis to complement more mundane analysis of larger
artifact distributions.
This study involves the interior floor of a domestic structure referred l o as
Structure 3 (Figure 2). The nature of the lithostratigraphic units associated
with Structure 3 are ideal for our purposes. Because sedimentation was confined
inside a n enclosed structure, microartifacts deposited onto the house floor were

434 VOL. 10, NO. 6


MACRO- AND MICROARTIFACTS IN A MISSISSIPPIAN HOUSE

Figure 2. Plan Map of Structure 3, recorded at base ot' level 2 (floor).

less likely to be dispersed during manufacture and use. In addition, trampling


resulting in compaction could potentially preserve depositional patterns (Hull,
1987; Villa, 1982).
Structure 3 appears to have undergone limited rebuilding or renovation, but
without change to the structure position or orientation. This is important since
Dallas structures were often rebuilt over time with occasional reorientations
correlating with change in the community structure (Polhemus, 1987). On the

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL 435


SHERWOOD, SIMEK, AND POLHEMUS

assumption that specific areas within a structure were designated for particular
activities, reorientation could have resulted in unreadable palimpsests of arti-
fact clusters. This does not appear to be a problem for Structure 3. Like many
late Mississippian structures recovered archaeologically , Structure 3 was com-
pletely destroyed by fire. Total burning appears to have caused the house to
collapse, sealing in the floor deposits below. Preliminary observations on the
scatter of large artifacts across the house floor indicate that the Structure 3
assemblage is a n abandonment assemblage.
The general stratigraphy of Structure 3 includes five lithostratigraphic units:
plow zone, subplow zone, level 1,level 2, and the floor (Figure 3). A plow zone
27 cm deep covers the entire first Holston terrace in the vicinity of the Loy
Site. The first lithostratigraphic unit beneath the plow zone, the subplow zone,
includes a remnant of Mississippian sheet midden overlying a n earlier humus
layer. These two layers were combined in the field and measured approximately
8 cm thick. The humus indicates that a past surface was stable in this area
for sufficient time to allow organic material to accumulate; this buried surface
is nearly 30.5 cm below the current surface. The subplow zone is identified in
the field by organically rich, very dark brown silt loam. The next three units,
which compose Structure 3, lay within a basin created by a n unconformity
separating them from the midden and humus. The deposits in the basin are,
taken together, 24 cm thick at the basin’s center. Level 1, the uppermost unit
of the Structure 3 deposits, is also characterized as a very dark brown silt
loam. It is interpreted as overburden or post-destruction. The dark brown
friable sediment is characteristic of a n A horizon, with inclusions of small
artifacts brought up through bioturbation from the underlying cultural deposits
and moved downward from the plow zone. Level 2, the unit above the floor, is
characterized as a very dark brown silt loam, interpreted as a destruction zone
where sediments are the result of the structure burning and collapsing. The
destruction zone is primarily composed of burned clay daub and charred archi-
tectural material mixed with sediments from above and below. The deepest
unit is interpreted as a floor based on attributes of texture (silt loam with less
sand), the fact that it is lighter in color (dark yellowish brown) and more
compact than the deposits above, and on the association of artifacts with hori-
zontal orientations. In addition, subfloor deposits such as postholes, entryway
trenches, and pits become visible at this level (illustrated in Figure 2). The
floor varies in thickness between 3 and 9 cm. Samples for microartifact analysis
were collected from the floor.
A MODEL FOR DALLAS HOUSE ORGANIZATION
Analysis of Structure 3 follows a model for Dallas houses developed by
Polhemus (1985,1987). According to this model, the household unit is archaeo-
logically represented in the Dallas Phase by a primary structure (winter house),
a secondary structure (summer house), and a n outside activity area associated
with processing facilities (Polhemus, 1985). Structure 3 is identified as a pri-

436 VOL. 10, NO. 6


MACRO- AND MICROARTIFACTS IN A MISSISSIPPIAN HOUSE

m
s
.-M
a

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL 437


SHERWOOD, SIMEK, AND POLHEMUS

PRIMARY STRUCTURE

STORAGE BED SIORAGE


PRIMARY
STRUCTURE 1
f\ f\
PUBLC AREA
(Food Processing
Cooking

BED
PRNATE Light)
ENTRWAV AREA
I (Beds
PUBLIC AREA
Storage
- Non F o o d NW Comer
- F o o d SE Corner
- General
NE and SW Caners)
STORAGE I BED
PRIVATE AREA I
I

Figure 4. Polhemus’ model for the interpretation of the use of space in the minimal settlement
unit within the Dallas Phase (after Polhemus, 1985: Figure 3.6).

mary structure “Type 4a” (Polhemus, 19851, a rectangular domestic unit with
a rigid single set post construction and four main roof supports. The main roof
supports are centered around a prepared clay hearth.
Polhemus’ model for the use of space within a primary structure divides the
structure into two principal parts (Figure 4):

. . . a central “public” floor area around the hearth, and an outer “private”
floor area situated between the roof supports and the exterior wall. The
public area is centered on the prepared clay hearth and frequently includes
surface fired areas probably utilized in food preparation. The private floor
area is usually divided into segments through the use of interior posthole
alignments or prepared clay daub partitions. Segments situated between
main roof supports along the central portion of each wall appear to have
functioned as beds or benches as well as serving as loci for activities such
as flintknapping and bone tool manufacturing. Burials are most frequently
placed beneath or immediately in front of the beds or benches. The corners
of the structure, making up the remaining portions of the private floor area,
were utilized for storage (Polhemus, 198557).

This model is based on ethnographic accounts of similar structures in the


southeast (Williams, 1973),archaeological structure remains, and the pattern-

438 VOL. 10, NO. 6


MACRO- AND MICROARTIFACTS IN A MISSISSIPPIAN HOUSE

ing of macroartifacts across the “floor.” A similar model has been proposed
for late Mississippian, Barnett Phase structures from northwestern Georgia
(Hally, 1983). Most of the archaeological data used to develop the Type 4a
model were collected from structures excavated a t Toqua (40MR6) (Polhemus,
1985),where artifact classes or combinations of classes were identified in clus-
ters inside houses and interpreted as activity areas organized in space. These
clusters consisted of chipped stone or bone debris and shell processing equip-
ment, concentrated in front of the remains of benches that once lined the
structure walls (Polhemus, 1987). Clusters were identified in both the “public”
and “private” areas, but the central “public” area was usually relatively clear
of debris except for the immediate hearth area. Corner areas were used for
storage and were either clear of debris or contained ceramic vessels andlor
food remains (e.g., charred corn).
Expectations
Microartifact analysis provides interesting and informative insights into
the formation processes of Structure 3 relevant to the model for household
organization discussed above. Following the model, there should be a “private”
area reserved for sleeping and individual activities where benches line the
walls, distinct from a central “public” area where hearth-centered activities
take place. The “public” area is a zone of substantial activity, but i t is kept
clear of refuse due to the amount of foot traffic and “public” activities that
occur there. Based on this model, the central “public” area should contain a
few or no macroartifacts; those that are found must have been deposited just
prior or after abandonment of the house. This central area should contain
microartifacts in abundance, especially along the margins between the “public”
and “private” areas. This is where individuals seated on the benches engaged
in manufacture or use of implements, producing microartifacts that were in-
corporated into the floor where they fell. Private areas should see little depo-
sition of any size of artifacts, since few and limited activities presumably
occurred there; moreover, the benches that define private zones would have
served as “suspended screens” preventing deposition onto the house floor. A
few artifacts lost through the cracks and artifacts abandoned with the structure
should be the only ones present in these private areas.
Material differences might be expected to influence the deposition of macro-
and microartifacts even when distributions correspond to model expectations.
Architectural features constructed from daub, for example, might be expected
to yield both large and small pieces as they decompose after deposition. Lithics
should not be subject to in situ breakdown. Thus, any patterns predicted by
the model will vary following material characteristics, as we have discussed
above.
METHODS AND MATERIALS
For our analysis, five material classes were chosen on the basis of the natural
sediment (which affects the visibility of microartifacts and their correct identi-

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL 439


SHERWOOD, SIMEK, AND POLHEMUS

fication) and macroartifact material classes present a t the site. These classes
are: lithics, bone, shell, daub and ceramics (Sherwood, 1991).
Sediment size fractions examined for microartifacts were limited for this
analysis to 0.0 4 (2-1 mm), and 1.0 4 (1.0-0.5 mm). Fired clay material classes
(ceramics and daub) can only be identified and differentiated with confidence
down to 1.0 4. Particles smaller than 0.5 mm begin to break down into their
constituents, either fired clay fragments, shell fragments, or quartz grains,
and can no longer be properly identified. This phenomenon was also observed
and noted by Madsen and Dunnell (1989) and Stein and Teltser (1989) when
attempting to point count fired clay microartifacts. Studies that have analyzed
only microdebitage have been able to work with much smaller phi sizes
(Fladmark, 1982).
To provide background information for interpreting microartifact samples
from the Structure 3 floor, a series of control samples was taken at various
points within and outside of the Loy Site boundaries. First, a series of vertical
bulk samples was collected a t 7.6 cm (0.25 ft) arbitrary levels along two perpen-
dicular transects across the structure. Vertical samples were collected to gather
sedimentological information on the deposits filling Structure 3, and to assess
the grain-size distribution of the observed stratigraphic units. In addition,
control samples were collected off-site to assess the natural background of
sand-size particles that could possibly be mistaken for artifacts under the
microscope. Because identification is such an important aspect of microartifact
analysis, we will briefly discuss criteria for recognizing microartifacts under
magnification later.
Sampling
Our microartifact sampling scheme had to produce samples that could be
compared with macro-artifact distributions across the house floor. Since the
Structure 3 floor was excavated in 76 by 76cm (2.5 by 2.5ft) square units
(Figure 51, our microartifact sampling scheme homogenized the sediment from
the surface of each 76 cm, square assuming that microartifacts recovered in
aggregate are representative of the unit area. The areas of microartifact sample
collection within the excavation units are also illustrated in Figure 5. Samples
never extended to the edges of the 76 cm excavation unit in a n effort to represent
only the unit area and avoid potential overlap with activities represented in
neighboring units. Sample collection proceeded through two steps. First, the
area to be sampled was lightly scraped with a clean trowel to remove any
material that may have been transported into the square by wind or on clothing
following excavation. Approximately 600 g of sediment was then scraped and
collected from the sample unit and bagged directly.
Laboratory procedures were designed to reduce the field samples to sand-size
particles ready for analysis. This process, a variation on basic sedimentological
grain-size techniques, comprised wet sieving after a soaking and agitation
step. Soaking involved distilled water and sodium hexametaphosphate to de-

440 VOL. 10, NO. 6


MACRO- AND MICROARTIFACTS IN A MISSISSIPPIAN HOUSE

Figure 5. Sampling areas for macroartifacts and microartifacts within Structure 3.

floculate the sediment. Sieving comprised washing and wet separation of the
four different fractions using stacked screens. The process includes the discard
of silt and clay, separation of 4 4-2 4 (fine sand), 1 4 and 0 4 (coarse sand),
and >-1 4 (granule and coarser) fractions. In addition, wet sieving resulted
in the total removal of aggregates from the sample and the segregation of the
1 C#J and 0 4.
The coarse sand fraction was examined under the binocular microscope for

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL 441


SHERWOOD, SIMEK, AND POLHEMUS

artifacts. When identified, they were counted using the computer program
MMCOUNT so that reliable samples were obtained in minimal time. Reliable
sample sizes for seven material classes ranged from approximately 1350- 1800
grains (mean 1704). The average counting time was 25 min (Sherwood and
Ousley, 1995).

Material Recognition
Lithic Material
“Lithic” here refers t o rock modified by human chipping efforts. The modifi-
cation is assumed to be the result of stone tool production, wear from use,
and/or trampling by people. Angularity is a feature often used in identifying
microdebitage (Hull, 1983, 1987). However, because angular alluvial quartz
grains compose some of the deposits, microdebitage was identified using the
following combination of attributes: angularity, thinness (often opaque), rem-
nants of conchoidal fracture, absence of frosting (present on most of the alluvial
quartz grains), and a t least some hint of color (usually gray, tan, or even
whitish oolites).

Ceramic Material
Dallas Phase Mississippian ceramics are most often tempered with crushed
shell. Shell tempering is fairly obvious, even at the microscale. In addition,
some ceramic sherds at Loy exhibit voids produced by shell temper elements
leaching from the clay matrix. These voids are lenticular and plate-shaped,
easily identified as leached shell tempered ceramics. Importantly, daub, also
composed of burned clay, has neither of these characteristics. Thus, ceramic
microartifacts are distinguished primarily by shell temper or void shape.

Bone Material
Bone fragments are assumed to have been deposited as a result of cultural
processes like human subsistence, domestic canid behavior, or animals drawn
there because of human food stores and refuse. Bone was not difficult t o identify
with its unique structure and shapes, even when burned. Bone preservation was
variable. The attributes used to identify bone were color, shape, and structure
(cortical or cancellous).

Daub Material
“Daub” refers to the local clay that was collected and used in house construc-
tion, lining hearths and sections of the roof and walls, to make a structure
more secure and fire proof. Very fine quartz sand grains are the most abundant
constituents of macrodaub. Often, Mississippian structures, such as the one
under investigation, burned, so the daub is not only dried but also fired. This
gives it a texture and hardness similar to prehistoric low-fired, coarse-ware

442 VOL. 10, NO. 6


MACRO- AND MICROARTIFACTS IN A MISSISSIPPIAN HOUSE

ceramics. Analysis of an experimental comparative collection showed that


abundant very fine sand inclusions distinguished daub from ceramics. In addi-
tion, microdaub is more easily weathered than ceramic material, resulting in
a higher degree of rounding compared to ceramic microartifacts.

Shell Material
Shell fragments also primarily result from cultural processes. Shell is one
of the easiest materials to recognize under the microscope, and is identified by
its white-to-gray color, distinctive structure, and morphology. Shell material
a t the Loy Site was extremely weathered and friable, breaking apart easily
during excavation and processing to produce microartifacts.

Data Analysis
The volume percentages derived from counting the 0 4 and 1 4 fractions were
used to examine distributions and identify clusters. Using SURFER (Golden
Software, 1989), a high resolution two- and three-dimensional graphics pack-
age, the data were organized by material class t o create isopleth density contour
maps. Macroartifacts (those larger than 1 cm) were counted by excavation unit
and mapped in the same fashion. Contour maps can then be compared between
micro- and macrodistributions and among microartifact classes.

RESULTS
Not all actual distributions observed from Structure 3, divided by material
classes and size, conform to Polhemus’ model predictions. Each distribution is
briefly described and then discussed.

Lithics
Figure 6 shows density isopleth maps for lithics by general size class, macro
and micro. The macrodistribution is focused around the central hearth with a
cluster towards the northern limit of the central “public” area. The microdistri-
bution, on the other hand, clusters in the southern corner to the right of the
entrance partition. In short, each size class has a concentrated distribution,
but concentrations are not concordant. Given that stone is a size-stable material
not much subject to weathering, these distributions can best be accounted for
in reference to distinct behavioral aspects of their generation and deposition.
Anyone who has stood near a flintknapper knows the dangers of flying debitage
and the resulting ground accumulations. The corner containing microdebitage
may have been isolated enough from the rest of the structure to provide
some measure of safety when stone knapping had to be performed indoors;
useful products were then removed to be used elsewhere. Polhemus’ model (see
Figure 4 ) does not provide for this kind of specialized work area within the
structure, and clearly needs modification in this respect.

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL 443


SHERWOOD, SIMEK, AND POLHEMUS

ul 0 n n n
x n 2"
.r
rl
N
N
.r
N
N
N
-
PI
N N
.I
0 0
N

2
n
x
N
2
N
c
M
2
N
a
N
?
0

444 VOL. 10, NO. 6


MACRO- AND MICROARTIFACTS IN A MISSISSIPPIAN HOUSE

Ceramics
Figure 7 shows ceramic spatial distributions. The macroceramics are focused
in the center of Structure 3 with a concentration just north of the hearth. The
microdistribution is concentrated along the walls with the majority along the
western periphery. Thus, size classes do not coincide in space, and their relation
can be characterized as a center-periphery distribution with small artifact
concentrations formed around the outside of a central large artifact cluster.
Microsherd distributions are not continuous even around the house periphery,
so that it is unlikely that they represent a background distribution formed by
the in situ breakdown of larger sherds. Concentrations of microsherds are
evident, with the densest cluster along the southwestern wall. The southeast
quadrant of the house exhibits few if any ceramics. Interestingly, between the
entryway trenches along the southwest wall (where traffic would have entered
and exited the structure) a microceramic cluster appears.

Bones
Figure 8 shows the distribution of whole and fragmentary bone a t macro-
and microscales. Again, there is a center-periphery relation between size
classes (with large bone fragments centered on the fireplace). Note how both
scales coincide with the ceramic distributions, suggesting similar formation
processes for ceramic and bone distributions, perhaps generalized food prepara-
tion. However, in contrast to ceramics, concentrations of microbone fragments
are scattered beneath the bench on the northeastern side of the house close to
the house wall. This difference may reflect in part the inclusion of small mam-
mal bone in the sample. In addition, gnaw marks on some larger bone fragments
in this area implicate dogs as influencing bone distribution patterns, and, dogs
being dogs, they may have transported bones beneath the interior bench for
private dining.

Daub
Figure 9 shows maps of daub distributions. Microdaub is concentrated near
the hearth with lesser scatters close to house walls. Macrodaub fragments are
concentrated on the hearth. Hearth-centered daub probably reflects the collapse
of the house roof with its daub-coated smoke opening above the fireplace. The
in situ disintegration of daub-plastered interior partitions may account for
other microdaub concentrations. In any case, daub concentrations at both scales
are due to similar causes: mechanical break down of architectural features.
These causes are distinct from processes producing other material class distri-
butions.

Shell
Finally, Figure 10 compares shell distributions. In both size classes, shell
fragments are scattered along the western half of the house with a major

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL 445


P
P
UJ

CERAMIC

MICRO MACRO
237.5 237.5

232.5 232.5

227.5 227.5

222.5 222.5

217.5 217.5

212.5 212.5

207 5 207.5

202.5 I I I 1 1 1 I I ( ( ( ~~
,
202.5
3100 315.0 3200 325.0 3300 335.0 3400
RANGE = 0-8514 CI = 1

<
0
!- The RANGE for the micro is reported in volume percentages, the Range for m c r o is reported in total counts. The CI indicatesthe contour interval.
2
0
z
P Figure 7. Ceramic distribution in Structure 3 at the Loy Site.
0)
G)
rn
g
a
0 BONE
I
D
rn
MICRO MACRO
g 237.5 237.5
G)
5
D
z
- 232.5
z
-I
rn
n 227.5
z
D
A
0
z 222.5
P
bc
217 5
n
z
F
212 5

207.5

1 1 1 1 1 1 1 1 1 1 1 1
202.5 310.0 315.0 320.0 325.0 330.0 335.0 340.0
31 I 315.0 320.0 325.0 330.0 335.0 340.0
RANGE = 0-1 1% CI = 2 RANGE = 0-200 CI = 25

RANGEfor the micro is reported in volume percentages, the Range fa macro is reported in total counts. The CI indicates the contour intewl.

P Figure 8. Bone distribution in Structure 3 at the Loy Site.


P
4
P
00
P

DAUB

MICRO MACRO
237.5 237.5

- -

2325 - 2325 -
-
2275 - 227 5
-

222.5 - 222.5
-
:A
-
2175 - 2175 -
- -
2125 - 2125 -
- -
2075 - 207.5 -
- -
1 1 1 1 1 1 ~ , ~ ~ , ,
202.5
3100 315.0 3200 325.0 3300 3350 3400 310.0 315.0 3200 325.0 330.0 335.0 340.0
RANGE = 0-8m CI = 10

<
0
!- The RANGE f a the micro is reported in volume percentages, the Range for macro is reported in total counts. The CI indicates the contour interval.
2
0
z
P Figure 9. Daub distribution in Structure 3 a t the Loy Site.
m
MACRO- AND MICROARTIFACTS IN A MISSISSIPPIAN HOUSE

-2
rl
-
-30
2-
- It
-xo
2
- 0
W
- 0 I
2 0
- II

-28
:lf
- ( L

-3
CI
-
I I I I I I I I I I I I I 2
0
P
N
10
N
r?
2
N
'0
N
N
0
-
.r
N
10
-
N
N
2
N
:2
0
N

1 1 1 1 1 1 1 1 1 1 1 1 l

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL 449


SHERWOOD, SIMEK, AND POLHEMUS

concentration at the southwest corner behind a door partition. Spatial concor-


dance of sizes suggests that (as was the case for daub) different size classes
have similar formation processes causing spatial pattern. If concentrations
near the south wall indicate deposition of material away from central activity
areas, as we have proposed for microdebitage, then sharp-edged, fresh shells
were either processed in this zone or were disposed of here. In contrast to
lithics, microshell fragments cannot be directly used to infer processing loca-
tions, since shell can generate microartifacts through natural disintegration.
To summarize our results, three basic patterns can be defined for the five
artifact material classes. Lithics have a nonconcordant set of clustered distribu-
tions, with microdebitage located in the south corner of the house, perhaps in
an isolated reduction area and macroscale lithics forming concentric rings
around the hearth. Thus, given that these are size-stable artifacts, tool produc-
tion and use occurred in different areas. Large tools may have been deposited
as refuse or stored at the periphery of activity zones. Bones and ceramics have
very similar spatial patterns at all scales; large pieces occur near the hearth
and microartifacts in both classes are found at the wall periphery. Concordance
between ceramics and bones suggests a similar set of formation processes for
these classes; concentrations of larger artifacts near the hearth may indicate
generation in a single activity like food preparation. Daub and shell each
exhibit intraclass spatial concordance, but they resemble neither each other’s
configuration nor patterns noted for other material classes. This suggests to
us that a single distinctive formation process must be invoked to explain
patterning for each of these two classes. Daub concentrations implicate archi-
tectural features and house destruction as important formation processes,
while shell patterns suggest that processing, refuse disposal, and material
character are primary causes for shell distributions. Post-depositional micro-
artifact generation is suggested in both cases.
DISCUSSION
Distinct patterns were identified for lithics and ceramics when divided into
size classes. The first represents a lack of concordance between macro- and
microdebitage distributions. According to our initial expectations, a lithic re-
duction area is identified by the concentration of microdebitage. Perhaps due
to “cultural sorting” (Gallagher, 1977) at least some of the larger debitage was
removed from this work location and disposed of elsewhere. As an aside, based
on the nature of the Dallas Phase lithic technology, debitage dumps should be
extremely difficult to identify. Projectile points and other tools made by chipped
stone technology are small, often produced from secondary reduction flakes.
The primary lithic resource used at Loy (based on the proportion found within
the macroassemblage) was Knox Black Chert, which occurs locally but in
relatively small nodules containing various unconformities in the rock. This
suggests that very few large flakes were included with refuse, since macrodebi-
tage may have been saved for use at a later time and place. The macrolithic

450 VOL. 10, NO. 6


MACRO- AND MICROARTIFACTS IN A MISSISSIPPIAN HOUSE

distribution centers around the hearth. The lack of microdebitage around the
hearth indicates an area where stone tools were used but not made.
For ceramics, macrosherds are found centered on the hearth while the mi-
crosherds are scattered along the walls. These patterns were also observed for
bone. Ceramics and bone are materials affiliated with food preparation and
cooking. Their clustering together, adjacent to the hearth, would suggest that
domestic activity was responsible for their distribution. We believe that the
distributions of macro- and micro-sized ceramics are related to cultural activity
rather than to a noncultural post-depositional alteration, because microsherd
distributions do not shadow or form a background for larger ceramic concentra-
tions; thus, it is difficult to invoke post-depositional weathering as a primary
cause of microartifact production. Moreover, the house floor was fairly protected
from surface processes, such as sheet wash or wind, by the “destruction zone”
and the overlay deposited across the terrace sometime after Mississippian
occupation. Bioturbation probably had some minor effect, but no obvious signa-
tures of post-depositional turbations have been identified.
Two explanations are offered here for the center-periphery pattern charac-
teristic of ceramics and bones. The first is that the benches were not substantial,
meaning there were places in the construction (cracks and holes) where micro-
material generated on the benches could filter downward to be deposited be-
neath the benches. Based on ethnographic descriptions (Swanton, 1946) and
the benches that have been recovered archaeologically in circumstances of
exceptional preservation, this seems highly unlikely since benches were usu-
ally reported to be substantial and well made. Moreover, the kinds of activities
that generate ceramic and bone fragments were probably carried out on the
ground rather than on the sleeping platforms.
The second explanation, and the one preferred in this study, involves micro-
artifact movement, something that some researchers argue does not occur
(Fladmark, 1982). Several recent studies have focused on sweeping a s a clean-
ing method and its effects on artifact distributions across floors where activities
are performed. Most have suggested that on an unprepared earthen floor, such
as Structure 3, sweeping would have little effect on the micromaterial while
it would move the macromaterial (McKellar, 1983 cited in Schiffer, 1987;
Metcalfe and Heath, 1990; Schiffer, 1987; Hull, 1987). Our study, along with
those conducted by Nielsen (1991a, 1991b), indicates this may not be the case.
Recent actualistic experiments conducted by Nielsen (1991a) produced some
interesting results concerning the maintenance of lithic activity areas. “Pick
up” cleaning, where all visible flakes are removed by hand, results in a “com-
plete dissociation of macro- and microartifacts” (Nielsen, 1991a:lO). When a
lithic reduction area was swept, a different pattern occurred:

The smallest flakes [<1 mml were transported with the loose dirt that
covered the surface and ended up uniformly distributed across all samples.
The largest microdebitage (1-8 mm in this case), on the other hand, is too

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL 451


SHERWOOD, SIMEK, AND POLHEMUS

small to be intentionally removed with the broom, and heavy enough to stay
in its original location (Nielsen, 1991a:5).

This has some unsuspected implications for studies that attempt to identify
activity areas using microartifact analysis. The largest microartifacts (2 mm),
as defined by Dunnell and Stein (19891, are subject to movement by sweeping
and are, thus, relatively unreliable indicators of activity location. The next
larger phi size, -1.0 (2-4 mm), may be a more reliable indicator due to its
relative stability during the sweeping process. This observation is supported by
distributions in Structure 3, where refuse material (particularly microdebitage)
was swept from the “public” center of the house towards the wall, beneath the
benches, to avoid foot traffic and remove it from sight.
The cluster of ceramic microartifacts along the southwest wall in the entry-
way has interesting implications considering another actualistic study con-
ducted by Nielsen (1991a). This study attempted “to assess the amount of
microartifacts produced by trampling on different material [ceramics and
lithics]” (Nielsen, 1991a:7). The ceramic results are far more conclusive than
those for lithics. Nielsen (1991a:7) observes:

. . . microsherds of trampling origin are far more numerous than those result-
ing from any case of vessel breakage, and would obscure any pre-existing
trend. Trampling is a more intensive process of ceramic damage than any
form of vessel failure that is likely to occur during use.

Thus, ceramics recovered in our microassemblages may not indicate breakage


location, especially in their distribution along the walls beneath benches.
Again, sweeping is implied, most likely after trampling reduced larger sherds.
The cluster in the entryway could implicate one process with two effects. First,
trampling could be the cause for the increase in the microsherds. Second, as
trampling generates more microsherds, it also increases the likelihood of their
being embedded in the floor, allowing them to elude sweeping processes.
The macroceramic clusters towards the center of the structure are each
associated with a light cluster of microceramics. Perhaps these concentrations
do represent breakage events just prior to or during abandonment. This led to
the deposition of a small amount of microsherds (compared to trampling), and
the larger sherds also remained in their place of impact. In sztu breakdown
through chemical weathering and leaching might also be implicated in these
particular patterns.
Polhemus’ model for the use of space based on macroartifacts and ethno-
graphic observation is supported in its dichotomization between the central
“public” area of the house and the surrounding “private” area. A coincident
dichotomy was observed in both the micro- and macroartifacts and their distri-
bution across the structure floor. The concept of “private” space and the use
and organization of this space in the original model should, however, be refined.

452 VOL. 10. NO. 6


MACRO- AND MICROARTIFACTS IN A MISSISSIPPIAN HOUSE

According to this analysis, the “private” area was the focus of microartifacts
that were probably removed from the central area of the structure and deposited
along the interior walls, most likely by cleaning techniques (e.g., sweeping,
raking). Macroartifacts associated with refuse disposal, those larger items that
would have been collected by hand are not associated with the microdistribu-
tions. This would imply that larger refuse is being removed from the primary
structure to a designated location outside the immediate household. Perhaps
the integration of microartifact analysis into the interpretation of macroassem-
blages with an emphasis on assemblage composition and grain-size distribution
will allow for the fine tuning of the public versus private model and its consider-
ation of the use of space.
CONCLUSIONS
To summarize our results, three basic patterns can be defined for the five
artifact material classes. Lithics have a distinct set of distributions, with micro-
debitage located in the south corner of the house, perhaps in an isolated reduc-
tion area or specialized refuse disposal zone. Large lithics form concentric rings
around the hearth, indicating that tool production and use occurred in different
areas. Large tools may have been deposited as refuse or stored a t the periphery
of activity zones. Bones and ceramics have very similar spatial patterns a t all
scales. Concordance between ceramics and bones suggests a similar set of
formation processes for these classes; concentrations of larger artifacts near
the hearth may indicate generation in a single activity like food preparation,
while microartifact scatters a t the structure wall may indicate sweeping of the
floor to remove activity refuse. Daub and shell each exhibit within-class spatial
concordance a t both size scales, suggesting a single distinctive formation pro-
cess for each class. Daub concentrations implicate architectural features and
house destruction as important formation processes, while shell patterns sug-
gest processing, refuse disposal, and material character as primary causes for
shell distributions. Thus, the various distribution patterns observed here
among micro- and macrodata of different materials imply that different forma-
tion processes affect both different material classes and different grain sizes.
In addition, a t least certain size grades of microartifacts are more easily moved
than previously expected. Cleaning and other formation processes must be
considered when interpreting microartifact distributions. Microartifact concen-
trations cannot be simply viewed as markers of prehistoric production activity
locations.
Our results lead to two sets of conclusions, one concerning the spatial organi-
zation of activities inside Structure 3 and another concerning the utility of our
methods. Within Structure 3, three general spatial zones can be defined. The
principle activity area focuses on the central fireplace. Here, stone tools were
refined, used, and retooled. Animal processing and activities involving ceramics
were also carried out. This area received activity-generated artifacts, but also
saw the deposition of daub when the house collapsed. In this central zone, then,

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL 453


SHERWOOD, SIMEK, AND POLHEMUS

artifact concentrations are due to various causes. A second deposition area is


related to the interior benches that lined the house walls. This area provided
a “suspended” activity platform that resulted in less large artifact deposition,
perhaps because the platform was periodically cleaned of activity debris. In
addition, the bench provided cover for small animals who may have contributed
to bone accumulations in this area. Finally, some specialized activities were
performed out of the way behind the door partitions at the south side of the
house.
Comparisons among artifacts of different sizes have yielded a more textured
view of spatial organization within Structure 3 than would have been possible
using only the larger artifacts. If we had examined only those larger pieces,
we would have come to a rather static view of a habitation centered on fireplace
activities. We could not have recognized distinct zones near the door or under
the bench without microartifacts. Specific microartifact classes may provide a
kind of “shadow” image of events no longer reflected by distributions of larger
artifacts. Using only the smaller artifact fraction would have prevented recog-
nition of the dynamic sequence of deposition, resulting in a n interpretation
every bit as unsatisfying as large artifacts alone. Analyzing concordance be-
tween spatial distributions of both macro- and microartifacts provides more
information than only a single size class, especially since the information
content of each size class is different, and allows a more comprehensive view
of the history of a n archaeological deposit.

The authors wish to thank Dr. J. K. Stein and an anonymous reviewer for their helpful comments
on this article. Dr. Stein also lent her guidance and advice to this project and especially to the
thesis this article is based on. All errors, of course, are our own.

REFERENCES
Dunnell, R.C., and Stein, J.K. (1989).Theoretical Issues in the Interpretation of Microartifacts.
Geoarchaeology 4,31-42.
Fladmark, K.R. (1982).Microdebitage Analysis: Initial Considerations. Journal of Archaeological
Science 9,205-220.
Gallagher, J.P. (1977).Contemporary Stone Tools in Ethiopia: Implications for Archaeology. Jour-
nal of Field Archaeology 4,407-414.
Golden Software Inc. (1989).Surfer Version 4.0.Golden, Colorado.
Hally, D.J. (1983).The Interpretive Potential of Pottery from Domestic Contexts. Midcontinental
Journal of Archaeology 8, 163-196.
Hassan, F.A. (1978).Sediments in Archaeology: Methods and Implications for Paleoenvironmental
and Cultural Analysis. Journal of Field Archaeology 5, 197-213.
Hull, K. (1983).Application of Microdebitage and Spatial Analysis. Master’s Thesis, University
of Calgary, Calgary.
Hull, K. (1987).Identification of Cultural Site Formation Processes Through Microdebitge Analy-
sis. American Antiquity 52(4),772-783.
Madsen, M.E., and Dunnell, R.C. (1989).The Role of Microartifacts in Deducing Land Use from
Low Density Records in Plowed Surfaces. 54th Annual Meeting of the Society for American
Archaeology, Atlanta, Georgia.

454 VOL. 10. NO. 6


MACRO- AND MICROARTIFACTS IN A MISSISSIPPIAN HOUSE

Metcalfe, D., and Heath, K.M. (1990).Microrefuse a n d Site Structure: The Hearths a n d Floors of
t h e Heartbreak Hotel. American Antiquit?/ 55, 781-796.
Nielsen, A.E. (1991a).Where Do Microartifacts Come From. 56th Annual Meeting of t h e Society
for American Archaeology, New Orleans, Louisiana.
Nielsen, A.E. (1991b).Trampling the Archaeological Record: An Experimental Study. American
Antiquity 56(3),483-503.
Polhemus, R. ( 1985).Mississippian Architecture: Temporal, Technological, and Spatial Patterning
of Structures at t h e Toqua Site (40MR6).Unpublished Master’s Thesis, University of Tennessee,
Knoxville.
Polhemus, R. (1987). The Toqua Site: A Late Mississippian Dallas Phase Town, Vols. I and 11.
University of Tennessee Department of Anthropology Report of Investigations No. 4 1. Tennes-
see Valley Authority Publications in Anthropology No. 40. Knoxville: University of Tennessee.
Reese, J.A. (1986).Microarchaeological Analysis of t h e Chinese Worker’s Area a t t h e Warrendale
Cannery Site, Oregon. 51st Annual Meeting of t h e Society for American Archaeology, New
Orleans, Louisiana.
Rosen, A.M. ( 1989).Ancient Town and City Sites: A View from the Microscope. American Antiquity
54, 564-578.
Schiffer. M.B. ( 1976). Behauioral Archaeology. New York: Academic Press.
Schiffer. M.B. (1987). Formation Process of the Archaeological Record. Albuquerque: University
of New Mexico Press.
Sherwood, S.C. ( 1991).Microartifact AnalysisofA Dallas Phase House Floor. Unpublished Master’s
Thesis, University of Tennessee, Knoxville.
Sherwood, S.C. and S. Ousley ( 1995). Quantifying Microartifacts Using a Personal Computer.
Geoarchaeology 10, 423-428.
Simms, S.R., and K.M. Heath (1990). Site Structure of the Orbit Inn: An Application of Ethno-
archaeology. American Antiquity 55, 797-813.
Stein. J.K., and Teltser, P.A. (1989).Size Distributions of Artifact Classes: Combining Macro- and
Micro-Fractions. Geoarchaeology 4, 1-30.
Swanton, J.R. ( 1946). Indians ofthe Southeastern United States. Bulletin 137. Washington, D.C.:
Bureau of American Ethnology.
Vance, D.E. I1986). Microdebitage Analysis in Activity Analysis: An Application. Northwest A n -
thropological Research Notes 20, 179-189.
Vance, D.E. ( 1989).The Role of Microartifacts in Spatial Analysis. Unpublished Doctoral Disserta-
tion, University of Washington, Seattle.
Villa, P. (1982).Conjoinable Pieces and Site Formation Processes. ArnericanAntiquity47,276-290.
Whallon, R. 1973).Spatial Analysis of Occupation Floors I: Application of Dimensional Analysis
of Variance. American Antiquity 38(3), 266-278.
Williams, S.C., Ed. (1973).Adair’s History ofthe American Indians. New York: Promotory Press.

Received January 12, 1995


Accepted for publication May 12, 1995

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL 455

You might also like