Cliff Palace

Material Information

Cliff Palace a critical re-study of an archaeological site
Milo, Richard Gardner
Publication Date:
Physical Description:
xii, 169 leaves : illustrations ; 29 cm

Thesis/Dissertation Information

Master's ( Master of Arts)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Anthropology, CU Denver
Degree Disciplines:
Committee Chair:
Quiatt, Duane
Committee Co-Chair:
Janes, Craig R.


Subjects / Keywords:
Archaeology -- Colorado ( lcsh )
Cliff-dwellings -- Colorado ( lcsh )
Archaeology ( fast )
Cliff-dwellings ( fast )
Mesa Verde National Park (Colo.) ( lcsh )
Colorado ( fast )
Colorado -- Mesa Verde National Park ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references (leaves 163-169).
General Note:
Submitted in partial fulfillment of the requirements for the degree, Master of Arts, Department of Anthropology
Statement of Responsibility:
by Richard Gardner Milo.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
22700953 ( OCLC )
LD1190.L43 1990m .M546 ( lcc )

Full Text
Richard Gardner Milo
B.A., University of Colorado, 1987
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Master of Arts
Department of Anthropology

(c)l990 by Richard Gardner Milo
All rights reserved.

This thesis for the Master of Arts
degree by
Richard Gardner Milo
has been approved for the
Department of
Duane Quiatt
Craig R. Janes


Milo, Richard Gardner (M.A., Anthropology)
Cliff Palace: a Critical Re-study of an Archaeological
Thesis directed by Professor Duane Quiatt
Archaeological sites which have been excavated in
the past may still yield new and valuable information
through careful re-studies. In many cases, however,
such sites have been changed over time by the action of
the physical environment and by repair and
stabilization subsequent to the original excavation.
As well, many sites, such as Cliff Palace in Mesa Verde
National Park, are today better known through the
descriptive and interpretive texts which have grown up
around them than they are through the archaeological
remains themselves. In order to avoid representing the
present condition of a site as reflective of its
original configuration, and to guard against the
reproduction of interpretive assumptions contained in
the analytical and descriptive literature, a re-study
of an archaeological site must be informed by a
critical, historical perspective on previous work at
the site under investigation. Using the example of
Cliff Palace, I demonstrate this kind of critical

approach. I present an historical overview of site and
textual transformations in the archaeology of the Mesa
Verde, conceived as passing through four successive
periods: Reduction, Production, Reproduction and
Deconstruction. I then discuss my method of study,
emphasizing the collection of quantitative, empirical
data drawn from maps of Cliff Palace. An historical
review of descriptive and interpretive texts bearing on
that site follows; and I demonstrate some ways in which
the quantitative data produced by this re-study can be
used to test some of the interpretive assumptions
contained in those texts. I conclude by suggesting
some areas for future research, focussing on questions
which are amenable to study using the data which I have
The form and content of this abstract are approved. I
recommend its publication.
Duane Quiatt

Tables ..................................... x
Acknowledgements .......................... xi
STATEMENT OF PROBLEM ....................... 1
ON THE MESA VERDE............................6
Introduction .............................. 6
Reduction: ca.1300 ca.1906 .............. 7
Production: 1906 1935 .................. 11
Reproduction: 1923 1976 ................ 16
Deconstruction: 1976 Present ........... 19
3. RESEARCH DESIGN AND METHODS ............... 25
Research Design .......................... 25
Context of the Study....................25
Data Requirements.......................26
Sample Selection ...................... 27
Selection of Cliff Palace ................ 28
Data Sources...............................30
Cliff Palace Data

Methods of Study............................40
Site Data................................40
Dimensional Data.........................41
Relational Data..........................44
Additional Data..........................47
Supplementary Sources .................. 47
Computer Coding ........................ 48
Data Limitations............................49
4. CLIFF PALACE.................................55
Introduction .............................. 55
Physical and Temporal Setting ............. 57
Morphology of the Site......................62
Previous Descriptions
of Cliff Palace.............................66
Reduction and Production I:
Chapin, Birdsall, Nordenskiold .... 66
Reduction and Production II:
Jesse Fewkes.............................82
Production after Fewkes ................ 98
Reproduction ........................... 99
Re-Study Data on Cliff Palace..............103
The Data Set............................104
Applications of the Data...................108
Room Size and Use.......................108
The Status of
Speaker-chief's House ................. 124

5. SUMMARY AND CONCLUSIONS.....................128
C. CORRECTIONS TO FEWKES' TEXT.................162
BIBLIOGRAPHY .................................... 163

3.1. Tabulation of Sites in the Sample..............29
3.2. Interior Doors at Cliff Palace ............... 35
3.3. Exterior Doors at Cliff Palace ............... 37
4.1. Corresponding Room Numbers for Data Set
and Maps......................................73
A.l. Room Data for Cliff Palace....................141
A. 2. Kiva Data for Cliff Palace....................150
A. 3. Open Area Data for Cliff Palace...............156

I would like gratefully to acknowledge the
contributions of those people who made my research and
the writing of this thesis possible. Duane Quiatt made
it possible for me to spend three months on the Mesa
Verde in conjunction with the research project which
motivated this study, and without his advice and
encouragement this thesis would not have come to
completion. I am grateful to Jack Smith, Park
Archaeologist at Mesa Verde National Park, for his part
in making my stay there possible, and for making
available to me the extensive materials archived at the
Research and Cultural Resources Management at the park.
My thinking on the subject of this thesis was greatly
influenced by the long conversations which I had with
him during my work there. Ms. Kathleen Fiero, Chief of
Stabilization at Mesa Verde National Park, generously
and patiently shared with me her intimate knowledge of
the archaeological sites on the Mesa Verde, as well as
dozens of photographs of the sites which have been
taken during the course of stabilization work. I must
also express my admiration and thanks to the unknown
artists and cartographers who produced the

extraordinary maps of Cliff Palace which constitute the
principal data source for this study.
I would also like to express my appreciation to
the Jacob K. Javits Fellows Program of the U.S.
Department of Education for their generous support
during my graduate work at the University of Colorado.
All of the photographs in this thesis were
reproduced through the courtesy of the Western History
Collection of the Denver Public Library.
Above all, my deepest gratitude goes to Ms. Robin
Benny, without whose support, encouragement and
consumate editing skills this thesis would never have
been written in its present form.
I, of course, assume full responsibility for any
errors or omissions in the work presented here.

This thesis presents a descriptive re-study of
Cliff Palace, by many accounts the largest of the
prehistoric cliff dwellings in Mesa Verde National
Park. I collected the data on which the description is
based as part of a comparative study of twenty-eight
Anasazi ruins located within the boundaries of the
park. This is the first presentation of that data. My
method of study is an example of a non-interventional
approach to the study of archaeological sites. This
kind of approach has as its purpose the retrieval of
analytically useful data without excavating or
otherwise physically disturbing the site under
investigation. Such approaches are particularly
appropriate in the case of sites which were excavated
and described in the past and which are now protected
as valuable cultural resources. In some of these
cases, such as many of the major cliff sites on the
Mesa Verde, the ruins have received little or no
subseguent analytical attention. Consequently, my
present study has two parallel purposes. The first is
to present the raw data for Cliff Palace, the methods

by which the data were collected, and some examples of
the uses to which the data may be put. In addition, I
discuss the ways in which the sources of data used in
the study contain potential sources of uncertainty and
bias which must be taken into account in such an
investigation. The sources to which I refer include
both the physical sites and the textual materials
maps, descriptions, and interpretive texts which
accompany the sites and which are indispensable to the
researcher. I argue that an uncritical use of either
textual materials or the present configuration of the
site exposes the investigator to the risk of reaching
false conclusions or of perpetuating problematic basic
assumptions. On the other hand, a critical historical
and contextual use of both textual materials and site
data may help to control for bias in both sources, as
well as providing a more secure basis for inferences
about the site as it existed in the past. It is a
major intent of this study to illustrate such a
critical approach.
The inspiration for this paper is Stewart
Baldwin's 1977 re-study of Spruce Tree House, MV-640,
in Mesa Verde National Park. Baldwin argued that re-
studies can "salvage" several categories of information
about previously excavated archaeological sites,

including descriptions of architectural features, the
reconstruction of building seguences,
dendrochronologies, and usage pattern studies
(1977:113). But a review of Baldwin's study and the
sources on which it was based revealed that he was in
error in two important respects. First, he asserted
that excavated cliff alcove sites can be considered
accurately to reflect the original condition of the
sites and that "any stabilization and reconstruction is
distinguishable from the original remains" (loc. cit.).
Second, he placed too much confidence in the authority
of the textual sources which he employed, especially
the park's descriptive and interpretive literature.
Specifically, he neglected to account for the fact that
both the physical sites and their collateral texts
change over time, and that such changes or
transformations are unevenly inscribed in the site
and the literature surrounding the site. This problem
is especially acute in the case of sites like Spruce
Tree House and Cliff Palace, which were excavated and
described under a set of social and scientific
assumptions which today should be regarded as
It is my contention that archaeological re-
studies, such as Baldwin's and my own, can and should

provide two important kinds of information. The first
consists of empirical data suitable for a broad range
of analyses. To Baldwin's list of categories of such
data room descriptions, building sequences,
dendrochronologies and use-pattern studies I add
quantitative data for comparing structures by size and
orientation, and for assessing the spatial
relationships among them. No less important, however,
is an examination of the changes which have taken place
in the site and in the literature concerning the site
since first it came to scientific attention. The kind
of re-study which I advocate here applies synchronic
and diachronic approaches to the same problem; it aims
to establish an historical context for the data
produced by the research. To that end, I shall devote
considerable attention in this paper to a discussion of
physical changes at Cliff Palace, as well as to the
circumstances surrounding the production and
reproduction of the textual sources available for the
site. I wish especially to emphasize the ways in which
interpretive assumptions about the sites have become
fixed in the literature, thereafter to become the
authority for subsequent studies and interpretations.
Widespread interest in many of the major
archaeological sites on the Mesa Verde Cliff Palace,

Spruce Tree House, Far View House and Oak Tree House,
to name a few began in the late nineteenth century.
The texts which were produced during these early years
of scientific work in the American Southwest reflect
certain assumptions about the prehistoric inhabitants
of the region which have been reproduced in subsequent
studies. Many of these assumptions have been inscribed
artifactually on the physical sites by subsequent
restorers looking for guidance to the literature
produced by their predecessors. Thus, the sites
gradually became modified by texts which purported only
to describe them, and to the extent that they have been
so modified, the sites may be treated as extensions of
the texts. The limitations of such a treatment are
obvious; nevertheless, it is appropriate and effective
when the task of research involves working back through
a series of descriptive texts and concomitant
restorative reconstructions to identify processes of
change. In order to provide a systematic basis for
this effort, I have identified four historical periods
in the history of archaeological research on the Mesa
Verde, which I call Reduction, Production, Reproduction
and Deconstruction. In a sense, this is the
epistemological basis for the kind of re-study advanced
in this paper.

It is often useful to conceive of the history of a
discipline like archaeology as comprising a sequence of
stages or periods. Such periods are named and defined
according to common elements of practice, method,
theory or world view. In the case of archaeology in
the United States, periods have been defined for method
and theory in American archaeology as a whole (e.g.
Willey & Sabloff 1974), as well as for archaeology in
single regions, for example Southwestern archaeology
(e.g. Brew 1946), and for archaeological work in
individual areas like the Mesa Verde (e.g. Watson in
Lancaster et al. 1954). Each such periodization is
conceived to make a point, either to demonstrate
historical continuity, or a perception of progress in
terms of the development of method and theory, or both.
In this chapter, I present a periodization of
archaeological work on the Mesa Verde, the purpose of
which is to call attention to transformations of both
the sites and the literature surrounding the sites. I

have identified four periods, and they are named
according to the patterns of transformation to which I
wish to call attention: Reduction, Production,
Reproduction and Deconstruction. In a literary sense,
these names have both theoretical and ideological
connotations. I mean here to apply them in a narrow
sense to a discussion of some of the major ruins on the
Mesa Verde. An explication of their broader
theoretical and ideological implications must be
deferred for the present. The periods so demarcated
actually overlap to a considerable extent, and the
dates which bound each period are therefore somewhat
arbitrary. Nevertheless, the general scheme will be
useful in providing a framework for the description of
Cliff Palace which follows.
Reduction: ca.1300 ca.1906
Physical reduction of archaeological sites on the
Mesa Verde began at the time the cliff ruins were
abandoned by their inhabitants near the close of the
thirteenth century. This period is here considered to
end in 1906, the year in which Mesa Verde National Park
was established to protect the ruins (Haas 1926:167).
During this time, the sites were reduced in terms both
of the integrity of the architectural remains and the

quantity and distribution of the artifacts within them.
It is convenient to distinguish between two agencies of
site reduction: the physical environment, and human
In the broad sense, of course, any kind of site,
whether living or archaeological, is subject to
constant change due to the action of the physical
environment and human activity. As long as a site is
occupied, the changes represent the patterned behavior
of human beings, and it is the reflection of that
behavior in the archaeological record which we seek to
read (Binford 1977:6; Hodder 1986). The occupants of a
site transform it in accordance with their own
purposes, and they actively respond to changes produced
by the physical environment through repairs, rebuilding
and other deliberate activities. Changes which occur
subsequent to the abandonment or de-population of a
site, however, are of a different nature: they reduce
the site. In particular, they tend to obscure the
patterning left behind by the occupants: they are
random with respect to the social and symbolic contexts
in which the original occupants acted. Environmental
degradation goes unrepaired. Post-occupation human
activity which transforms sites includes "casual
vandalism" (e.g. the removal of wood or masonry for use

elsewhere), the retrieval and removal of artifacts
under uncontrolled conditions (especially with economic
motivation), and recent repair and stabilization. The
first two sources of human transformation are typically
prior to the latter, and are often unrecorded in any
source; we may often be aware that such transformations
have occurred, but we cannot say much about their
nature or extent. In a sense, their effect is to
reduce the amount of reliable information on which
restoration and stabilization can be based. Under
these circumstances, restoration and stabilization must
necessarily be guided by assumptions about the original
occupants' subsistence, social or symbolic systems
derived, for example, from ethnographic analogy.
Until approximately 1881, the primary source of
reduction acting on the prehistoric ruins of the Mesa
Verde was the action of the environment, especially
rain, wind and the freeze-thaw cycle. It seems chiefly
to have been the masonry which suffered the greatest
degradation during this first part of the period for,
in 1893, Gustav Nordenskiold wrote of finding numerous,
well-preserved domestic artifacts at Cliff Palace.
Unfortunately, a second source of reduction was already
at work when Nordenskiold wrote: the activity of
European and American explorers and entrepreneurs.

They found a ready market for antiquities among both
private collectors and public institutions. One
particularly fine collection was exhibited in Durango
and Denver, and was subsequently sold to the Historical
Society of Colorado (Nordenskiold 1893; O'Bryan 1950).
The real tragedy of this artifact looting, apart
from the loss of artifacts and their provenience, was
that the architectural remains were regarded as an
impediment to collection. Walls were broken through or
thrown down, their foundations were undermined, and the
floors were dug and re-dug (Fewkes 1911; 1916).
The net effect of this period was the removal of
an enormous quantity of artifacts, as well as the
destruction of much of the architecture and the
undoubted obliteration of many floor features, such as
hearths and similar indicators of room use. These
depredations continued into the late 1920's (Haas
1926:207), but by about 1906, site reduction was being
supplanted by site repair and stabilization. Yet,
although subsequent repair and stabilization work was
conducted with considerable care, the loss of data was
and remains inestimable, in that it is now impossible
to say how many and what kinds of features were
destroyed (Fewkes 1911:11-12).

Production: 1906 1935
In large part on account of the almost complete
loss of material culture from the ruins on the Mesa
Verde, much of what was considered "known about the
prehistoric inhabitants of the Mesa Verde was actually
produced by early workers like Frederick Chapin, Gustav
Nordenskiold and Jesse Fewkes out of a combination of
19th century assumptions about primitive people,
ethnographic analogy, and an uncritical perspective on
the archaeological record. Exhibitions of the
artifacts removed from the sites on the Mesa Verde,
along with lectures and articles in the press by those
who had seen the ruins, sparked considerable interest
in the prehistoric monuments and the people who had
built and lived in them (Fewkes 1911). Mesa Verde
National Park was created by an act of Congress in 1906
with the express purpose of preserving the ruins. The
emphasis at the time was on repairing and stabilizing
the ruins and on explaining the monuments and their
prehistoric inhabitants to the general public; "but
modern scientific methods were not used and the results
[left] much to be desired" from the viewpoint of
archaeology (Watson in Lancaster et al. 1954). While
careful, stratigraphic excavation began to be applied
on the Mesa Verde by 1923, I have extended this period

to include the depression-era stabilization work at
Cliff Palace, Spruce Tree House and Far View House,
since some important and little-known changes were made
to the sites during this time.
Between 1907 and 1915, Jesse Walter Fewkes,
working for the Smithsonian Institution's Bureau of
American Ethnology and charged with the repairs by the
Secretary of the Interior, undertook extensive
excavation and repair at Spruce Tree House, Cliff
Palace, Far View House and Pipe Shrine House. In the
process, he may be said to have produced new sites out
of the old, interpreting the sites in the light of his
own ethnographic work among the Hopi and stabilizing
them in the light of that interpretation. He also
produced a series of descriptive and explanatory texts
in which archaeology and ethnology were intermixed, and
in which a clear and direct correspondence between
prehistoric and modern indigenous people was assumed
(Fewkes 1909).
The repair and stabilization of archaeological
sites is not, in and of itself, production in the
present sense. Rather, production refers specifically
to the introduction of assumptions drawn from tenuously
related sources as, for example, ethnographic analogy,
in the process of repairing a site. Thus, in the case

of Cliff Palace, assumptions about the level of socio-
cultural development of the prehistoric inhabitants of
the Mesa Verde, about their symbolic systems, about
their type of social organization, and about their
relationships to living people were insinuated into the
earliest descriptions of the great ruin.
These kinds of assumption were not problematic for
scholars trained in the 19th century school of
ethnology, which envisioned all cultures as progressing
through a series of stages, from savagery to barbarism
to civilization. John Wesley Powell, then the director
of the Bureau of American Ethnology, wrote of the
comparative ease of studying "the savage and barbarous
peoples" compared to "cultured people" (Powell
1897:LXX). The contemporary American cultural milieu
also provided an impetus for portraying all indigenous
peoples, prehistoric and modern, as belonging to a
common (and primitive) stage of cultural development.
As Curtis Hinsley, Jr., points out:
American museum anthropology served ... as an
important defense against racing change, social
turmoil and ... human variety. [M]useum
anthropology made the ... point [of the supremacy
of civilization] by exhibiting the inferiority of
other people. (1981:83-84)
If the prehistoric Anasazi and the modern Hopi shared
the same level of cultural development, then there was
little reason to be cautious about using the latter as

a direct analog for the former. As Fewkes put it, "The
picture of culture derived from [the ruins] is
practically the same as that of a pueblo like Walpi .
until about fifty years ago. The people were farmers,
timid, industrious and superstitious" (Fewkes 1909:53).
Among the notable interpretations which became
fixtures of the sites and in the literature at this
time was the identification of the circular,
subterranean rooms in the sites with the ceremonial
structures of the Hopi: the kivas. This assumption has
guided interpretations about social organization (e.g.
Rohn 1977), religious practices (e.g. Lancaster et al.
1954), and household organization (e.g. Rohn 1971).
The association had already been suggested by Chapin
(1892) and Nordenskiold (1893), who had called these
structures by the Spanish word estufa ("stove")
(Nordenskiold 1893:16-17). This name had been applied
by Spanish explorers to the rectangular and circular
community and ceremonial rooms of the Hopi because of
the fires kept burning there (Nusbaum 1929). Fewkes
adopted the Hopi word "kiva," but retained and
reinforced the assumption of a ceremonial function for
these structures.

In the long winters the kivas served as the
lounging places for the men who were engaged in
an almost constant round of ceremonies of
dramatic character, which took the place of the
pleasures of the chase. (1909:53)
The social organization of the prehistoric people
of the Mesa Verde was also assumed to be directly
comparable to that of living southwestern Pueblo
people, notwithstanding that social organization and
language were recognized as varying considerably among
the Hopi, Hopi-Tewa and Zuni.
At Cliff Palace, it was assumed that the
prehistoric inhabitants were the direct ancestors of
the living Hopi, that their social and ceremonial lives
were virtually identical with those of the Hopi, and
that the circular, subterranean structures in the ruin
(as well as three substantially different structures)
were homologous with the kivas of the Hopi. These
kinds of assumptions were insinuated into the earliest
descriptions of the great ruin. When Fewkes repaired
Cliff Palace, he interpreted the remains in the light
of those assumptions. But perhaps more significantly,
he wrote the most extensive description yet published
about Cliff Palace. His description, which was in
effect a guide book to the ruin as he had excavated and
repaired it, repeated and elaborated upon the
ethnographic and behavioral assumptions of Chapin and

Nordenskiold. In effect, what Fewkes produced was a
synthetic site comprising, on the one hand, a cleaned
and protected Cliff Palace eerily stark and bare,
devoid of its social blood, and mute and, on the
other hand, a text through which the ruin purported to
speak to the reader. Having been fixed in the textual
canon of southwestern archaeology, these assumptions
were reproduced over and over again even as modern
archaeological methods and techniques were applied to
excavations on the Mesa Verde.
Reproduction: 1923 1976
Although careful, stratigraphic excavation had
been undertaken in the Four Corners area as early as
1915 (Brew 1946), it was not employed on the Mesa Verde
until the early 1920's (O'Bryan 1950). A notable
example of this early work was Nusbaum's re-excavation
at Step House Cave in 1926 (Nusbaum 1981), which
contributed substantially to an understanding of the
Basketmaker to Pueblo transition on the Mesa Verde.
Yet notwithstanding that this and subsequent
excavations were scientifically and methodologically
rigorous, they reproduced the descriptive categories,
explanatory themes and interpretive assumptions which
had been produced in the preceding period. This

reproduction is especially apparent in interpretations
of the ceremonial life and social organization of the
Mesa Verde Anasazi, and it extended even to the
selection of sites for excavation according to the pre-
established explanatory narrative (Lancaster & Pinkley
The continued attribution of a ceremonial function
to kivas is the most obvious example of the
reproduction of early assumptions and interpretations.
The kiva "was the common ceremonial room, the center of
ritual observances for a religious society or a small
village" (Lancaster & Pinkley 1954:53). It was
explicitly not considered to have served any domestic
or mundane functions. This interpretation of the
nature and use of kivas is ubiquitous in both
descriptive texts and scientific excavation reports.
Not only was the basic assumption of a ceremonial
function reproduced, it was frequently justified by
citing "standardized" features (e.g. Lancaster &
Pinkley 1954) or a scarcity of artifacts with clear
domestic functions (e.g. Lancaster & Van Cleave 1954).
So deeply embedded had this assumption about the nature
and function of kivas become that in 1979 Robert Lister
posthumously congratulated Gustav Nordenskiold for
correctly recognizing that "the circular rooms in the

cliff dwelling [were] the earlier counterparts of the
modern Pueblo kiva" (Lister 1979). Only very recently
have these assumptions come under critical scrutiny
(e.g. Cater & Chenault 1988).
The Anasazi continued to be regarded as the direct
ancestors of the modern Pueblo people, especially the
Hopi, and the latter continued to be used as direct
analogs for the former. "In order to answer the
question [of how to interpret the life and customs of
an Anasazi man] we simply go to the descendent of that
man. Without doubt they still have the same customs"
(Watson 1949:59). Having established the prehistoric
kiva as the exact counterpart of the modern kiva, all
of the real (and imagined) social and ceremonial
activities of the modern Hopi were attributed to the
prehistoric Anasazi. In some cases, these authors
reproduced not only the early interpretations of the
prehistoric ruins, but also 19th century attitudes
toward "primitive," "timid," and "superstitious"
indigenous people. The Anasazi were portrayed as
living a "religious and social life [which] was rigidly
regulated" and which revolved around "rigid,
ritualistic ceremonies" (Watson 1949:53).
It would be a mistake, however, to regard the
archaeological interpretations of this period as merely

passive reproductions of prior interpretations. One
notes, for example, a tendency to conflate ethnographic
models of prehistoric behavior with social
constructions drawn from Euro-American culture,
including a Western-style dichotomy between the secular
and the religious. Thus, Lancaster and Pinkley compare
the typical Unit Pueblo to "our own small rural
communities, a few farm houses nestled close to a
country church" (1954:37-38). Alden Hayes, in his
survey of Wetherill Mesa, pictured "the people ...
using the pits [kivas] for family prayer meetings"
(1964:89). Yet Hayes, himself, was one of the first to
call explicit attention to the application of the
category "ceremonial" to any artifact or structure
which did not have an obvious or mundane explanation
(Hayes 1964:113). In so doing, he foreshadowed the
period of deconstruction which was to follow more than
twenty years later.
Deconstruction: 1976 Present
In the strict sense, deconstruction is a method of
literary criticism which has grown up in the context of
post-modern literary theory.

To deconstruct a discourse is to show how it
undermines the philosophy it asserts, or the
hierarchical oppositions on which it relies, by
identifying in the text the rhetorical operations
that produce the supposed ground of argument.
(Culler 1982:86)
The method of deconstruction can be applied to any
discourse centered on a text or texts, from
philosophical and literary discourse to the discourse
of Southwestern archaeology. In that sense, much
recent work in Southwestern archaeology has contributed
to a deconstruction of the authority of traditional
archaeological interpretations and explanatory models.
The philosophical basis of traditional texts has been
revealed as resting on social and scientific ideology,
as for example the contrast between and differential
valuation of primitive and civilized people. As well,
some of the oppositions which are central to
traditional interpretations, such as the dichotomies
between male and female roles or between the secular
and the sacred, have been shown to be specific to
particular cultures and therefore problematic as the
basis for generalization to prehistoric cultures. The
impetus for the deconstructive trend which gives its
name to this period has come both from the development
of archaeological theory since the middle 1960's, and
from the production of large amounts of new data

through extensive excavation and ethnoarchaeological
Within the realm of archaeological theory, I shall
use two examples to illustrate my point. McGuire and
Schiffer (1983) presented a theory of architectural
design which emphasized the interaction of material
factors with social, symbolic and ideational factors in
determining the architectural characteristics of any
particular culture. They proposed that the symbolic
content of architecture increases in response to
greater social differentiation. While provisionally
accepting the attribution of a ceremonial or socially
integrative function to the Southwestern kiva, they
were attempting to formulate a rigorous, middle-range
theory which could be employed to test exactly that
kind of assumption. Patricia Gilman (1987) employed a
world-wide ethnographic survey to investigate the
relationship between pit structures, surface
structures, and the behavioral and subsistence
characteristics of the people who employ them. She
developed a theory of pit and surface structure use
which predicts a seasonal residence pattern among, for
example, the Mesa Verde Anasazi. Both of these
theoretical approaches treat prior assumptions about
social organization and human behavior as hypotheses

amenable to testing; and both call for new data for
hypothesis testing and theory building.
The collection of data applicable to
archaeological interpretation and theory building has
also advanced on numerous fronts since the birth of
and, by some accounts, the death of (e.g. Courbin 1982)
the New Archaeology of the 1960's.
Ethnoarchaeological studies have examined patterns of
artifact disposal among living people (e.g. Binford
1980) and the differential use of activity areas and
sexual division of labor (Kent 1984). These studies
have revealed an unanticipated degree of variability
among superficially similar cultures. They have also
called into question several assumptions which had been
widely used in interpretive analyses. Kent, for
example, found that activity loci are seldom
monofunctional, that artifact deposition patterns are
often equivocal indicators of activities, and that
sexual divisions of space and labor based on Euro-
American standards are not necessarily applicable to
indigenous cultures.
In a more specific vein, the identity of the pre-
fourteenth century kiva as a "specialized ceremonial
room" has been seriously questioned by Cater and
Chenault (1988), Lekson (1989) and others. Recent

excavations at Sand Canyon Pueblo seem to confirm the
skepticism of these authors, for example by recovering
artifacts from kivas which suggest an important and
continuing domestic function for those structures until
at least the close of the thirteenth century.1 One of
the signal outcomes of these lines of research,
considered together, is the demonstration that many
traditional models of cultural variability, social
process and human behavior are inadequate. We
recognize, for example, that human cultures are open,
dynamic systems and that they are formed and
transformed through both ecological and symbolic
processes. We have learned that it is misleading to
rely too heavily on direct analogies between two
cultures, or even the same culture at points in its
history 700 years apart (Baldwin 1977). As a result,
archaeological theory is coming to terms with a new
appreciation of the complexity of cultural process.
All of these recent studies, both theoretical and
empirical, have two things in common. They adopt a
skeptical position with regard to previously
unquestioned assumptions in archaeology, and they call
for new data with which to address new questions. And,
although the decision about what constitutes useful 1
1. Bruce Bradley 1990: personal communication.

data is inevitably to some extent subjective,
quantitative data which can be rigorously analyzed
without recourse to preconceived assumptions or
interpretations are particularly desirable. It is
those kinds of data which I have gathered in the course
of this study.

This chapter describes the research design and
methods employed in the collection of the data on Cliff
Palace and the other sites in my study sample. I
discuss the overall aims of the project in terms of the
theoretical and practical questions it was designed to
address, the selection of the sample studied, and the
methods employed in collecting the data. I conclude
with a discussion of some limitations of the data at
the present time, and suggest the collection of some
kinds of supplementary data.
Research Design
Context of the Study
The research project was conceived within the
context of an ongoing study of climatological factors
which may have conditioned human behavior on the
prehistoric Mesa Verde. For that reason, it was
desirable to collect data sufficient in quantity and
detail to accomplish three related tasks. First, the
data should be adequate to describe and analyze
individual sites in terms of their architectural

characteristics, the use of available space, and the
spatial relationships among structures and open areas.
Second, the data should facilitate meaningful
comparisons between and among sites within a
circumscribed geographical area. This is in contrast
to many comparative studies for which small samples
have been selected within an entire region (e.g.
Baldwin 1982; Lambert 1983). Finally, the data should
permit a search for systematic or patterned
similarities and differences among sites which may aid
in identifying community patterns or regular responses
to environmental factors. This paper presents a
quantitative description of Cliff Palace and is the
first presentation of data from the research project of
which it is a part.
Data Requirements
To meet the requirements of individual and
comparative site studies, it was necessary to collect
two principal kinds of data. Required, first of all,
were reliable measures of size and a consistent
representation of the shapes of individual structures
and open areas, as well as a standardized overall
characterization of the size and shape of the sites.
The latter is of special interest in the case of cliff
alcove sites, which vary considerably in their size,

setting and directional orientation. If there are
consequential environmental correlates to variations in
site size, setting or orientation, these may be more
apparent in cliff alcove sites than they are among open
site habitations. Secondly, relational data were
desirable which could be used to describe the location
of any structure, feature or open space within a site
and the spatial relationships among structures,
features and spaces. Having determined the kinds of
data which would be collected, the next step was to
decide on the composition of the sample and the most
appropriate sources for the data.
Sample Selection
Because the larger study of which my research will
form a part concerns prehistoric human responses to the
environment on the Mesa Verde, all of the sites in my
sample were selected within the boundaries of Mesa
Verde National Park. An initial list of sites within
the park was compiled from the surveys of Hayes (1964),
Rohn (1977) and Smith (1987) by Duane Quiatt, and that
list was reduced to 55 multi-room alcove sites, plus
ten sites on the mesa top, all assigned by their
excavators or surveyors to the Pueblo III period on the
Mesa Verde. Pueblo III spans the period of time
between ca. A.D. 1200 and A.D. 1300. The final sample

was drawn from this second list, thereby ensuring that
all of the sites in the sample are roughly
contemporaneous. The elements of the sample were
selected from all of the major drainages on the Mesa
Verde and include both high and low altitude sites and
both large and small sites. The final composition of
the sample was determined by the availability of the
principal data source: high quality maps for each of
the sites included in the sample. Table 3.1 lists the
twenty-eight sites which comprise the sample.
Selection of Cliff Palace
I have chosen, in this thesis, to confine my
analysis one of the sites in the sample: Cliff Palace.
By considering a single site, I can better call
attention to the problems attendant on an
archaeological re-study than would be the case in a
treatment of a number of sites. Cliff Palace
recommends itself as the focus of this discussion for
several reasons. It the arguably the largest cliff
alcove site on the Mesa Verde, and it is certainly the
best known. There is a substantial body of
interpretive and descriptive material concerning Cliff
Palace, the majority of it produced prior to 1920.
However, the site has received little or no extended
treatment since that time. It is thus an ideal

Table 3.1. Tabulation of Sites in the Sample
Site No. Name TvDe Rooms Kivas Location
MV-397 Cedar Tree Tower mesa top 2 1 Chapin Mesa
MV-499 mesa top 13 2 Chapin Mesa
MV-515 alcove 11+ 3 Soda Canyon
MV-522 New Fire House alcove 19 3 Fewkes Canyon
MV-523 Oak Tree House alcove 50 6 Fewkes Canyon
MV-535 Pine Tree House alcove 17 3 Navajo Canyon
MV-557 Painted Kiva House alcove 13 2 Soda Canyon
MV-615 Balcony House alcove 43 2 Soda Canyon
MV-625 Cliff Palace alcove 149 23 Cliff Palace Canyon
MV-626 Sunset House alcove 30 4 Cliff Canyon
MV-629 Swallow's Nest alcove 16 2 Cliff Canyon
MV-640 Spruce Tree House alcove 133 8 Spruce Tree Canyon
MV-647 Little Long House alcove 18 4 Navajo Canyon
MV-650 Square Tower House alcove 48 7 Navajo Canyon
MV-693 alcove 11 1 School Section Canyon
MV-808 Far View House mesa top 71H 5 Chapin Mesa
MV-820 Coyote Village mesa top 27b 5 Chapin Mesa
MV-1062 Hemenway House alcove 27 1 Soda Canyon
MV-1062a Little Hemenway House alcove 8 0 Soda Canyon
MV-1200 Long House alcove(in prep .) Rock Canyon
MV-1212 Kodak House alcove 47+ 7+ Rock Canyon
MV-1229 Mug House alcove 94 8 Rock Canyon
MV-1233 Jug House alcove 23 3C Rock Canyon
MV-1241 Ruin 16 alcove 42 5 Rock Canyon
MV-1385 Double House alcove 56 5 Bobcat Canyon
MV-1452 Badger House (P-III) mesa top 5 1 Wetherill Mesa
MV-2 3 8 2 alcove 14 2 E. Navajo Canyon
MV-3604 alcove 5 0 School Section Canyon
a. Includes 39 first-story rooms plus upper story rooms.
b. Ground floor rooms only.
c. In addition to the three kivas, one surface room has been identified as
a 'kihu,' or above-ground ceremonial room.

exemplar for the methodological approach to
archaeological re-studies which I advocate in this
Data Sources
It was clear from the beginning that to collect
these kinds of data at the sites, themselves, would be
prohibitive in terms of time and access. The study was
undertaken between January and May, 1989, a period
during which many of the sites on the Mesa Verde are
inaccessible due to weather conditions. Furthermore,
it would be extremely time consuming to obtain all of
the required data for even a single site through on-
site measurement, especially working alone. Finally,
it was most desirable that the research approach be
non-intrusive, or non-interventional, in order to have
the least possible impact on these fragile and
irreplaceable sites. Fortunately, the Research and
Cultural Resources Management Center, Mesa Verde
National Park, maintains a large archive of site maps,
including camera-ready site publication maps, survey
maps, and maps produced within the last few years as
part of a mapping project presently underway in the

Cliff Palace Data
My re-study of Cliff Palace was based in large
part on a set of truly extraordinary maps which were
produced between 1932 and 1934 as part of the
Depression-era construction, repair and stabilization
work underwritten by the Federal Government and the
National Park Service. Maps of Spruce Tree House and
Far View House were produced during the same period.
The identities of the cartographers who worked at Cliff
Palace remain unknown; but the title block on the maps
of Far View House indicate that the mapping was done
either by or under the direction of Stanley Morse and
with the guidance and advice of Jesse Nusbaum and Earl
Morris. Morris, at that time, was directing
stabilization work at at least three major sites on the
Mesa Verde. The maps were apparently taken to the
Museum of New Mexico by Jesse Nusbaum shortly after
their production, and were returned to Mesa Verde
National Park only in 1976 (Peckham 1976) To my
knowledge, this study represents the first effort to
retrieve the information contained in those maps, some
of which is recorded in no other source.
It is, regrettably, impossible for me to reproduce
in this paper the maps of Cliff Palace from which my
research data were gathered. The mapping at Cliff

Palace had not been finished at the time the project
was terminated. The artists (I believe that I can
discern at least two different "hands" in the work) had
completed an entire set of dimensioned working
drawings; sightings and benchmarks on the drawings,
along with traces of clay still adhering to some of the
heavy paper stock, indicate that the working drawings
were produced on the site, probably employing an
alidade and planetable. The cartographers had begun to
trace the working drawings onto blue velum using India
ink, but had not progressed much beyond the outline of
the rear of the alcove, some undimensioned tracings of
rooms and kivas, and plan and elevation drawings of
Kiva W. All of the data were therefore collected from
the working drawings, drawn in great detail to the
large scale of 1/2" to the foot.
Three additional maps were also used in the study
of Cliff Palace. Among them are the maps published by
Baron Gustav Nordenskiold in 1893 and by Jesse Walter
Fewkes in 1911. Their maps are reproduced as Figures
4.5 and 4.7 in the following chapter. The third map
was drawn in 1965, to a scale of 1" = 10', by a
draftsperson identified only as LDA. This map is
neither as accurate nor as detailed as the 1930's maps,
but it is the only available plan of the site suitable

for relational and descriptive purposes. I have
extensively revised and annotated LDA's map, and the
new map produced therefrom is reproduced as Figure 4.1.
The reader should refer to Figure 4.1 with reference to
the tabulated data and the description which follows.
During the data collection process, I made use of
all of the available textural material describing or
relating to Cliff Palace; likewise for each of the
other sites. The textual sources included the official
park survey report; descriptions of the site produced
by W.R. Birdsall (1891) Frederick Chapin (1892),
Gustav Nordenskiold (1893) and Jesse Walter Fewkes
(1911); James Lancaster's stabilization field notes
from 1932 to 1934 at Cliff Palace; park archaeologist
Don Watson's descriptive / explanatory treatments of
Cliff Palace (1949 and n.d.); and mention of the site
in subsequent reports on the archaeology of the Mesa
Verde (e.g. Lancaster et al. 1954; Rohn 1977). It was
through the use of these texts that I became aware of
their production and reproduction over time as a form
of site transformation, the organizing theme around
which this paper is based. Notwithstanding the high
quality of the maps and the available textual materials
for Cliff Palace, however, there remained some
questions for example, details concerning Rooms 1, 2

and 3 and Rooms 89, 90 and 91 which could only be
resolved by inspection at the site itself.
To ensure that the data on Cliff Palace were as
complete as possible, and with the kind cooperation of
the park staff, I spent a total of about ten hours
collecting information at the site. Included in this
information were a tabulation of the location and form
of all extant doorways, notations concerning ambiguous
indications of additional rooms, and the resolution of
some questions concerning the relationships among
terrace levels (e.g. at Kivas T and U). Much of this
information is included as annotations to the map in
Figure 4.2. Tabulations of inside and outside doorways
and their communications are included as Tables 3.2 and
3.3. I should note that the site inspection was
accomplished subsequent to, and with reference to, the
collection of data from the maps and texts. That
collection procedure is the topic of the next section.

Table 3.2. Interior Doors at Cliff Palace
A. Positively Identified Doors
Ser.No. Connected Rooms Remarks
1. Rooms 1 and 3
2. Rooms 1 and 2 These are not true
3. Rooms 2 and 3 doors, representing instead passage over the walls of Room 2.
4. Rooms 3 and 5 Above the rock shelf in Room 3.
5. Room 3 to passage connecting Room 5 with Kiva D
6. Passage from Room 5 to Kiva D
7. Rooms 6 and 9
8. Rooms 11 and 12 Step up from Room 11
9. Rooms 11/2 and 13
10. Rooms 12 and 13
11. Rooms 17/2 and 18/2
12. Rooms 19/2 and 20/2
13. Rooms 28/2 and 29/2
14. Room 35 into Kiva H
15. Rooms 43 and 46
16. Rooms 43 and 44 T-door
17. Rooms 45 and 46
18. Rooms 44 and 45 T-door with bottom filled in.
19. Rooms 47/2 and 48/2
20. Rooms 51 and 52 Chest-high with broad slab sill.
21. Rooms 54/2 and 55/2
22. Rooms 57 and 58
23. Room 65 to space behind
24. Rooms 67 and 69 Step up from 69.
25. Rooms 67 and 68
26. Rooms 71 and 72
27. Rooms 73 and 74
28. Rooms 72/2 and 74/2 Sealed door.
29. Rooms 82 and 84
30. Rooms 87 and 88 Possibly sealed.

Table 3.2. (contd.)
B. Probable Doors
Ser.No. Connected Rooms_____________Remarks
31. Rooms 14 and 15
32. Rooms 26 and 27
33. 625501 to Room 36
34. 625501 to Room 37
35. Rooms 55/2 and 56/2
36. Rooms 73/3 and 74/3
37. Rooms 81 and 82
C. Passages Between Kivas and Rooms or Other Kivas
Kiva A and Room 1
Kiva A and Kiva B
Room 35 and Kiva H
Tunnel through east
wall of Kiva, rising
via steps through a
hatch in the floor
of Room 1.
Short, straight
Door in NW wall of
Kiva into Room 35,
the floor of which
is at the floor
level of the kiva.

Table 3.3. Exterior Doors at Cliff Palace
A. Positively Identified Doors
Ser.No. Connected Rooms/Areas Remarks
1. Room 4 to KBCT Leads into low, smoke-stained chamber in 4.
2. Room 4/3 above KACT T-door.
3. Room 8 (or 7) to KBCT In south wall.
4. Room 11 to KCCT
5. Room 11/3 above Room 14 T-door.
6. Room 11/4 above KCCT T-door. Produced during 1932-1934.
7. Room 11/4 above Room 14
8. Room 12 to KDP/CT Very high sill at Kiva D plaza level.
9. Room 13 to KEP/CT
10. Room Room 17/2 above roof of 16
11. Room Room 18/2 above roof of 24
12. Room 19/2 above KEP/CT
13. Room 21 to KJCT
14. Room 21/2 above KJCT Possible balcony.
15. Room 22 to KJCT Waist high door with projecting stone beneath.
16. Room 22/2 above KJCT Possible balcony.
17. Room 23 to KJCT
18. Room 25 to KJCT Waist high sill.
19. Room 26 to KJCT Waist high sill. Floor of Room 26 is higher than floor of Room 2 5.
20. Room 27 to KJCT Waist high sill with projecting stone beneath.
21. Room 28 to KJCT
22. Room 34 to KHP
23. Room 36 to 625501 An outside door if 625501 wasn't a room.
24. Room 36/2 above 625501
25. Room 37 to 625501 An outside door if 625501 wasn't a room.
26. West east wall of 625501 or wall of 625502
27. Room 38 to KKP T-door.
28. Room 39 to KMP/CT 37

Table 3.3. (contd.)
Ser.No. Connected Rooms/Areas Remarks
29. Room 40 to KMP/CT
30. Room 41 to 41/42P
31. Room 42 to 41/42P
32. Room 43 to KMP/CT T-door, quite large.
33. Room 47 to KMP/CT
34. Room 49 to CT05O/49
35. Room 50 to 50P Waist high with broad slab sill.
36. Room 51 to STREET
37. Room 53 to KPCT
38. Room 55 to KPCT
39. Room 55 to KPCT Sealed door.
40. Room 56 to KPCT
41. Room 57/2 to KQCT
42. Room 59 above COURT High door, probable
ladder grooves in sill.
43. Room 60 to STREET S. end of west wall.
44. Room 60 to STREET N. end of west wall.
45. Room 61 to STREET
46. Room 62 to STREET
47. Room 64 above Kiva R High in west wall.
48. Room 64 to STREET Not a formal door.
49. Room 63 to 67P
50. Room 65 to 67P
51. Room 66 to RRC
52. Room 67 to 67P Step up out of room.
53. Room 67 to RRC
54. Room 68 to RRC Two steps up out of
55. Room 69 to SCE
56. Room 69 to SCPL
57. Room 70 to SCP
58. Room 71 to SCP
59. Room 71 to SCE T-door, sealed.
60. Room 72 to SCP
61. Room 72/2 above SCP T-door.
62. Room 73 to SCE
63. Room 73/3 to roof of
Room 71/2
64. Room 74/3 to roof of
Room 72/2
65. Room 75 to RRC
66. Room 78 to RRC Room 78 is a free- standing storage
67. Room 80 above Room 90
68. Room 86 to KUCT T-door.

Table 3.3. (contd.)
B. Probable and Possible Outside Doors
Number Connected Rooms Remarks
69. Room 20/2 above KEP/CT
70. Room 28 to RRA
71. Room 29 to KJCT
72. Room 48 to KMP/CT
73. Room 63 to STREET
74. Room 71/2 above SCP
75. Room 76 to SCR 76 and 77 are similar to Room 78.
76. Room 77 to SCR
77. Room 81 to 8IP

Methods of Study
As I noted in the preceding section, the final
selection of sites in the sample was determined by the
availability of high-quality maps, the principal data
source for the study. I devised a data collection
instrument with which to record data on dimensions,
relationships and site features in an orderly and
systematic way. A copy of the completed instrument for
Cliff Palace, along with the respective protocols for
Rooms, Kivas and Open Areas is included as Appendix A.
Since the data collection process was quite time
consuming, I determined to extract from each map the
maximum possible amount of information, both
dimensional and relational, using the procedures
described below.
Site Data
Summary data were collected for the site as a
whole as well as for each component within the site.
In addition to the total number of rooms, kivas and
open areas, I recorded the elevation and directional
orientation of Cliff Palace and all of the other alcove
sites. The purpose of a measure of directional
orientation is to assess whether site layouts vary with
respect to the amount of sunlight which they receive.

In most cases, aside from crude compass orientation, it
is the cliff overhang above an alcove site which
determines the pattern of sunlight in the alcove,
especially the amount of sun which reaches the inner
parts of the alcove. The measure of directional
orientation which I used is the compass azimuth
perpendicular to a straight line which approximates the
cliff overhang. In many cases, the termini of the
cliff overhang coincide with the end of the alcove
proper. In the case of Cliff Palace, however, the
cliff overhang does not extend over the rooms and kivas
at the extreme northern and southern ends of the ruin.
Dimensional Data
Dimensional data include basic, linear
measurements and the areas and ratios calculated from
them. Basic dimensional data room lengths and
widths, kiva diameters, etc. were measured directly
from the maps. For consistency, all measurements were
recorded in meters.
Rooms. It should be noted that linear measures of
this kind are important not only as direct indicators
of room size and shape, but also in order to derive
floor areas and the ratio of length to width. It has
sometimes been the practice to calculate floor areas of
rooms by using the lengths of the longest two walls in

a structure. Since perfectly quadrilateral rooms are
the exception rather than the rule in cliff alcove
dwellings, such a practice may misrepresent the true
floor area, which in turn may bias inferences about the
use or uses to which the rooms were put (e.g.
"domestic" vs "storage"). Consequently, the dimensions
recorded for rooms represent the means of two to four
measurements across each axis. At a minimum, the
measurements included both walls along each axis. In
the not uncommon cases in which the walls are not
perfectly straight, additional measurements were taken
to reflect bulges and constrictions. The overall aim
was to describe as accurately as possible the actual,
useable floor area within each room. In a few cases,
rooms are sufficiently irregular in shape that the
linear dimensions should be regarded as summary
descriptions only. In these cases, floor areas were
calculated separately by geometric means.
Kivas. Kivas were treated similarly, the floor
diameters shown representing the mean of at least four
measurements across the geometric center of the kivas.
Mean diameters were then used in calculating floor
areas. All kiva measurements were taken at floor
level, since it seems reasonable to assume that the
desired floor area was the chief determinant of the

overall size of a kiva. A number of sources suggest
that many, if not most, kivas were not constructed with
vertical side walls, but rather in the shape of a bell
or truncated cone (Nordenskiold 1893; Bennett 1932;
Rohn 1971). Consequently, measurements taken at the
level of the banquette may underestimate the true floor
Open Areas. Open areas (aka courts, plazas,
activity areas, etc.) were the most difficult to
measure accurately since they are typically more
irregular than rooms and kivas. It was further
desirable to enumerate open areas in a way which did
not involve assumptions about the possible use or uses
of the areas. For most areas, therefore, the lengths
and widths shown are summary descriptions of the area's
maximum extent. Surface areas were calculated by
decomposing the space into geometric shapes which could
accurately be measured, the results of which were then
summed to arrive at a total surface area. Open areas
above kivas were decomposed into two parts: the area
defined by the outer diameter of the kiva walls (i.e.
the roof top area of the kiva); and the area
surrounding the kiva roof. In order hypothetically to
reconstruct the total area of a 'plaza' above a kiva,
some areas were arbitrarily divided at a point midway

between adjoining kivas, as in the case of the
otherwise continuous open area above Kivas B and C at
Cliff Palace. In some cases, however, the total area
above a kiva is bounded by terraces and rooms in such a
way that to decompose it would be to urge an artificial
distinction, as is the case with the area above Kiva P.
These areas were treated as a single unit.
Relational Data
Relational data, that is, data which record the
spatial relationships among structures and open areas
and their locations within the sites, were collected
using both categorical and metrical measures. Both
methods were used in order to facilitate the use of a
variety of analytical techniques in future research.
Categorical Methods. Rooms, kivas and open areas
were grouped into several sets of mutually exclusive
and exhaustive categories reflecting the relationships
among the structures and features and their locations
within the site. For example, rooms were categorized
by shape (quadrilateral, round, compound, irregular and
highly irregular); by their spatial relationships with
kivas and open areas; by their relative positions
within the site; by their degree of exposure to the
physical environment outside of the alcove; and by the
orientation of their outermost walls with respect to

the back of the alcove and the compass azimuth of the
mouth of the alcove. Kivas and open areas were
similarly categorized. The reader is referred to the
coding protocols in Appendix A for complete
explications of the various categories used.
Metrical Methods. Rigorous investigations of
spatial relationships require data which uniquely
locate each element in the analysis (Whallon 1973,
1974). To satisfy this kind of data requirement, I
located each structure and feature, as well as the
defining boundaries of the sites and their sub-parts,
on a Cartesian, or X-Y, coordinate grid. I
established the convention that the origin of the grid
should lie at the defined southern limit of the site,
or the southern terminus of the line defining the
orientation of the site as discussed under Site Data.
The same line became the Y-axis of the grid. Sub-
components of a site, for example shelves and ledges on
which additional rooms were constructed, were mapped
onto the grid established for the principal component
of the site. The use of this system confers several
advantages in addition to the relative location of
mapped elements: a) the Y-axis of the grid approximates
the cliff overhang above an alcove site; b) the X-axis
is parallel to the directional orientation, or compass

azimuth, of the site; c) individual representations of
sub-components can be overlaid to produce a true,
scaled picture of horizontal relationships among
The actual data points mapped onto the site grid
are the geometric centers of the respective structures
and features. For rooms, the center was located at the
junction of two lines joining the mid-points of
opposite walls. Due to the irregular shape of many
rooms, this method is more accurate than the use of
lines joining opposite corners. For kivas, the data
point was located at the center of that perfect circle
which most satisfactorily included the actual plan of
the kiva. All hearths and fire pits were similarly
mapped onto the grids. Finally, the outline of each
site was mapped onto the grid by recording the
coordinates of each point at which the front and rear
boundaries of the site change their direction. The
same procedure was followed for each sub-component of a
site. All coordinates were recorded in meters from the
origin of the grid.
The coordinate data thus collected permits
computer plotting of each site and the structures and
features within it. Figures 3.1 through 3.4,
reproduced at the end of this chapter, are plots of

Cliff Palace produced with a micro-computer-based
topographic mapping program.1 Plotting each site as it
was completed provided a good check for accuracy of the
data measurements.
Additional Data
In keeping with my determination to extract the
maximum amount of data from each map, I recorded on the
data collection instrument all doorways, hearths,
cists, and details of masonry which had been indicated
by the cartographers. The quality of this kind of data
was not consistent from map to map; for example, the
map of MV-515, a small site not far north of Balcony
House, is comparatively sketchy, while the maps of
Spruce Tree House contain an incredible amount of
detail. Consequently, a number of collateral sources
of information were employed in an effort to make each
site description as complete and accurate as possible.
Supplementary Sources
As the data for each site in the sample were being
recorded, I made constant reference to as much
additional information as was available. The sources
for this information included official park survey
reports, previous descriptions, excavation reports, 1
1. SURFER (tm), published by Golden Software Company,
Golden, CO.

field notes and similar sorts of textual resources. In
addition, I made extensive use of photographs taken
during repair and stabilization work at various sites,
as well as the first-hand knowledge of the
stabilization staff at the park Research and Cultural
Resource Management Center.
Computer Coding
The format of the data collection instrument was
designed from the beginning to facilitate computer
coding of the data for electronic analysis. To that
end, a number of indices were conceived to permit
sorting the data both by categories and by indices of
reliability; I discussed categorical data in the
preceding section. Indices of reliability were
constructed to aid in selective analysis. For example,
the index of structure reliability (SRL in the data
set, Appendix A) distinguishes among rooms which are
securely identified by the presence of all four walls,
rooms for which data was estimated because one or more
walls is missing, rooms not clearly present but
considered probable, rooms which are considered
possible on the basis of circumstantial factors, and
numbered areas which may or may not be rooms at all.
The reader should refer to the data collection

protocols in Appendix A for a complete treatment of the
various reliability indices.
Data Limitations
It is most important specifically to note some
areas in which the nature of the available data sources
limits the completeness of the data set. Perhaps most
significant is the lack of vertical data for structures
and the sites overall. Baldwin (1977) and Rohn (1971)
have noted the importance of room height in determining
the possible function of the rooms. At the same time,
this kind of information is often very difficult to
obtain due to the ruinous nature of many of the
structures and the general loss of floors in many
multi-storied structures. Furthermore, elevation views
are rarely included on the site maps, so that vertical
data is generally completely lacking. Notable
exceptions include Spruce Tree House, for which
extensive elevation views were produced as part of the
maps of the site, and Mug House, for which Rohn
included excellent summary descriptions of each room in
his excavation report (1971). The extant data set
would be greatly augmented by the addition of data for
room heights and alcove height.

In addition, each map represents a graphical
snapshot of a site at a particular moment in time. I
have previously called attention to the transformations
to which most of the major sites on the Mesa Verde have
been subjected, and additional examples will be
presented in Chapter 4. The point I wish to make here
is that these site descriptions, such as the
description of Cliff Palace presented here, must be
taken as approximations to, rather than complete and
accurate reflections of, the sites as they existed in
the prehistoric past. It would be a mistake, for
example, to employ the extant hearths at Cliff Palace
(Fig. 3.4) as a guide to community organization or
activity patterns, since we may be sure that
uncontrolled digging at the site, along with heavy
weathering in the fore-part of the site, has
obliterated a number of these features. Indeed, it is
a major point of this paper that efforts at this kind
of re-study must be tempered by an appreciation of the
effects of transformations over time.

Figure 3.1. Plot of ground floor at Cliff Palace.3
a-. Closed squares are rooms with no story above them.
Open squares have a story above the one figured. Kivas
are represented by bulls-eyes.

Figure 3.2. Plot of second-story rooms at Cliff

Figure 3.3. Plot of third-story rooms at Cliff Palace.

Figure 3.4. Plot of extant hearths at Cliff Palace.*3
b. Closed circles are hearths in rooms. Open circles
are hearths in open areas.- Fire pits in kivas are
represented by bulls-eyes. There is only presumptive
evidence for a fire pit in Kiva L.

I suggest in this paper that a careful re-study of
an archaeological site must look at the site at
different moments in time in order to be aware of how
it might have been transformed. That is particularly
true in the case of sites like Cliff Palace, which were
excavated many years ago. In the absence of subsequent
work at the site itself, this kind of site is generally
known only through the descriptive and interpretive
texts which represent it. Such texts may be informed
by assumptions which become axiomatic but remain poorly
demonstrated. At the same time, the text may acquire
an authority which inhibits the testing of those
informing assumptions, such as the assumption that all
structures which resemble kivas are kivas, and that
their function is known. One way to guard against the
perpetuation of assumptions is to look at the physical
site as it has stood at several moments in time, taking
note of aspects of the site as it exists now which have
been substantially or subtly altered in the

course of repair and stabilization. At the same time,
attention to the descriptions of the site under study
can help further to reveal ambiguities in
reconstruction or interpretation. In this way,
attention may be directed to new hypotheses about the
society of farmers which built their sandstone
dwellings on the Mesa Verde.
I chose Cliff Palace to illustrate this approach
because, although it is arguably the largest cliff site
on the Mesa Verde, it has received very little
attention in the years since Jesse Fewkes excavated and
stabilized it. Because of that lack of attention, the
Cliff Palace which has been used in subsequent
descriptions and for comparative remarks has been the
textual site, not the archaeological site. I have
sought to redress that situation by providing a
quantitative characterization of the site in the form
of the data presented here. I am not, therefore,
advancing another textual description of Cliff Palace.
Rather, I present an historical sample of descriptions
of Cliff Palace to convey to the reader a sense of how
it has been reduced, produced and reproduced over the
past one hundred years. I then present my own,
quantitative data and illustrate two ways in which it
can be used to move beyond description and

interpretation to an empirical inquiry into internal
variability. I also present a revised map of the site
(Fig. 4.1). This is NOT the map, produced between 1933
and 1934, which was used as the source of the
quantitative data in my study; it is based on a map
drawn in 1965, which I have extensively revised based
on data in the 1930's maps and my own observations on
the site. Until such time as the extraordinary 1930's
map is completed and assembled, however, I believe that
Fig. 4.1 is the most accurate and complete map yet
produced of Cliff Palace.
Physical and Temporal Setting
Cliff Palace sits near the mouth of Cliff Palace
Canyon, a north-eastern branch of Cliff Canyon, on
Chapin Mesa, in Mesa Verde National Park (Fig. 4.2).
The north-western counterpart of Cliff Palace Canyon is
Fewkes Canyon, in which are located several other
large, cliff alcove structures. On the promontory
separating Fewkes Canyon from Cliff Palace Canyon sits
the site of Sun Temple, considered by many to have some
special significance on account of its uncommon
construction and symmetry.
The alcove in which the site of Cliff Palace was
built is at the head of a small, subsidiary rincon, and

Figure 4.2. Partial map of Mesa Verde National Park.

consequently it forms an arc somewhat skewed to the
north and opening to the southwest. Like most of the
other large cliff-dwellings in Mesa Verde National
Park, the alcove in which Cliff Palace was built is
formed in the Cliff House formation, a layer of
Cretaceous sandstone and the uppermost of four
principal geological members forming the Mesa Verde.
Ground water seeps down through the porous sandstone
until it encounters an impermeable layer of shale,
whereupon it flows along the plane of the shale, slowly
undercutting the sandstone. At the canyon edges, the
undercutting action of the water causes chunks and then
blocks of sandstone to collapse out of the cliff face,
forming the cliff alcoves.
Cliff Palace Canyon at the site of Cliff Palace is
approximately 200 ft. (61 m.) deep (Fewkes 1911:20).
The canyon bottom supports a thick growth of spruce,
gamble oak, juniper and low shrubs. The mesa top
supports the typical Mesa Verde flora, including pinyon
pine, juniper and sage. It is likely, however, that
during the time Cliff Palace was occupied the forest
had been greatly reduced by clearing for farm land and
for fire wood (Hayes 1964:109).
On the basis of architecture and tree-ring dates,
Cliff Palace is regarded as representative of the late

Pueblo III period on the Mesa Verde, dated from A.D.
1200 to 1300. Rohn names this the Montezuma phase on
Chapin Mesa, in an effort to highlight local sequences
within the broader regional context (1977:243). It
appears that, during the later part of this period,
most of the population on the Mesa Verde lived in the
cliffs, with "[p]ossibly some mesa-top pueblos ...
still occupied." The circumstances of this move into
the cliff alcoves, along with the apparent reduction of
population on the Mesa Verde as a whole, are still very
poorly understood (Hayes 1964; Rohn 1977).
Robinson and Harrill (1974:53-54) report tree-ring
dates for a total of thirty-six samples taken from
Cliff Palace since 1935. The latest date is 1279 from
Room 20, and three cutting dates of 1271 have been
obtained from samples at Kiva K. These appear to
represent the last period of construction at Cliff
Palace; but there is no reason to suppose that the
alcove had not been used and re-used prior to that
time, the remains of earlier structures having been
overlaid by subsequent construction or incorporated
into the later structures. The dates of 1271 and 1279
are among the latest recorded on the Mesa Verde. Thus,
although it is known that the prehistoric inhabitants
commonly re-used wooden beams (Rohn 1971; 1977), and

that beams brought from other sites were sometimes
employed during repair and stabilization work (maps
1933), there seems no reason to doubt that little or no
construction was accomplished at Cliff Palace after
A.D. 1280.
The masonry at Cliff Palace includes many
different types of construction and finish. Among the
masonry types represented are spall and adobe, slab
construction, dry-laid masonry, and chipped-faced
masonry. Prominent, as well, are pecked- and ground-
faced masonry and terrace walls, all of which are
considered to be characteristic of P-III architecture
on the Mesa Verde (Hayes 1964; Rohn 1977). Both
single- and double-coursed masonry are present. Room
sizes and construction are both quite variable at Cliff
Palace; and the size, shape and construction of kivas
is similarly variable.
Morphology of the Site
Cliff Palace has been considered the largest
alcove site on the Mesa Verde since its discovery in
1888 (Nordenskiold 1893), although Long House may well
have been as large. The alcove as defined by the cliff
overhang is approximately 81.3 meters long and 28.5
meters deep at the deepest point. The ordinate of the

coordinate grid on which the site was plotted has as
its southern terminus the point at which the west wall
of Room 1 meets the alcove wall; its northern terminus
lies where the common wall between Rooms 86 and 87 meet
the alcove wall (see Fig. 3.1). The ordinate
represents an abstraction of the cliff overhang above
the alcove. The directional orientation of Cliff
Palace, as defined by the cliff overhang, is 245-30/
from True North.
The cave floor is relatively flat for only twelve
to thirteen meters from the rear, after which it falls
off in a series of rough steps toward the canyon floor.
In several places the topography of the site is
interrupted by large boulders which have fallen from
the cave roof. The ancient builders were obliged to
construct a series of terraces following the slope of
the cave floor in order to provide space for building
and activities. Jesse Fewkes excavated and stabilized
four enumerated terrace levels. The fourth and
uppermost terrace is the cave floor, itself. It is at
least possible that construction originally had been
carried further down the slope, although in that
location it would have been directly exposed to the
elements, especially rain water flooding over the edge
of the overhang (Fewkes 1911:24). Below the retaining

wall of the first terrace, a talus slope of trash and
rubble, now covered with brush and small trees,
testifies to the quantity of masonry which has
collapsed and fallen out of the site. I have found no
record that the trash slope which presumably underlies
the masonry rubble has ever been tested
Most of the existing rooms lie on the third and
fourth terrace levels where, except for those at the
extreme ends of the site, they are protected by the
overhang. From Kiva A at the south, an unbroken
sequence of rooms and room blocks extends across the
rear third of the alcove as far as Room 79 above Kiva
S. The remainder of the rooms are interspersed in
block-like arrangements among the kivas in the outer
two-thirds of the alcove, producing "streets," "courts"
and "plazas" allowing movement among the buildings and
providing space for human activity. In several places
the room blocks approximate compound unit pueblos
surrounding kivas, for example at Kivas J, M, P and U
(cf. Lancaster et al. 1954). Free-standing rooms are
virtually absent in Cliff Palace, a reflection both of
the Anasazi practice of building contiguous rooms and
of the shortage of space in the alcove.

Where the alcove is deepest, the builders left
open areas behind the rear-most rooms. These areas are
typically interpreted as refuse areas or turkey "pens,"
although heavy smoke blackening on the alcove ceiling
suggests the frequent use of fire in those recesses.
Three of the kivas (D, E and V) were constructed
against the alcove wall, and kivas D and E are
surrounded by rooms so that they are effectively
separated from the rest of the site and from the
outside environment. With these and a few additional
exceptions, most of the kivas occupy the outer half of
the alcove where their construction necessitated
extensive terracing and filling. In almost every case,
terrace levels were defined so that the kiva roof was
at the level of the surface, producing the typical
subterranean structure with the probable entrance
through the roof. This pattern is typical of most
alcove sites with kivas; indeed, it may have required
considerably less labor to build up a level surface
around a kiva than to dig one into the floor of an
alcove. The builders of both kivas E and V had
partially to excavate those kivas into the back wall of
the alcove (cf. Fewkes 1911:54,60).

Previous Descriptions of Cliff Palace
Reduction and Production I;
Chapin. Birdsall. Nordenskiold
Cliff Palace was not mentioned in the writings of
the early Spanish chroniclers in the American Southwest
(Fewkes 1911:13). Ranchers Richard Wetherill and
Charley Mason, who guided most of the early visitors to
the site, claimed to have discovered and named it in
December, 1888; but a number of local ranchers told
Jesse Fewkes that they had seen Cliff Palace early in
the 1880's, and they claimed that many of the walls in
the ruin had then stood "much higher" than they were
when explorers began arriving from the East (Fewkes
1911:13). Three of these early explorers Frederick
H. Chapin, Dr. W.R. Birdsall, and Baron Gustav
Nordenskiold published descriptive accounts of Cliff
Palace between 1890 and 1893. None of these men
actually excavated at the site, but their accounts
provide a valuable glimpse of Cliff Palace as it
existed before widespread looting supplanted the
elements as the most severe source of reduction. They
also began the textual production of the interpretive
The first account of Cliff Palace was read before
the Appalachian Mountain Club by Frederick Chapin on

February 13, 1890. He subsequently published his
report as The Land of the Cliff Dwellers in 1892.
Chapin, like his successors, was awed by his first
sight of the great ruin, "like an immense ruined castle
with dismantled towers."
The stones in front were broken away, but behind
them rose the walls of a second story; and in the
rear of these, in under the dark cavern, stood
the third tier of masonry. Still further back in
the gloomy recess, little houses rested on upper
ledges (1892:106).
Chapin was almost certainly not aware of the true
extent of the site, since many of the rooms and kivas
along the front of the site were covered with rubble
from the collapsed rooms above them. "So many walls
have fallen," he wrote, "that it is difficult to
reconstruct the building in imagination; but ... there
must have been many stories." Chapin and the
Wetherills remained at Cliff Palace for most of a day,
"and ransacked the structure from one end to the other"
Chapin had visited the Moqui (Hopi) in Tusayan,
and he was convinced that the inhabitants of the
pueblos there were the direct descendents of the cliff
dwellers: "we have much reason to infer that the
inhabitants of the two classes of dwellings were
identical" (1892:64). He was also the first to equate
the circular, subterranean rooms in the prehistoric

ruin with the estufas (kivas) of the Hopi, apparently
because of the near ubiquity of fireplaces in the
circular rooms.
Chapin reported the dimensions of the site,
measured by Richard Wetherill, as 425 ft. long, 80 ft.
deep and 80 ft. high in the center. He counted 124
rooms on the ground floor (probably including the
exposed kivas), and estimated the original population
at nearly 1000 people. Figure 4.3 is a photograph of
the north side of the central tower, Room 36, taken by
Chapin during his visit to the site in 1890. Notice in
this photograph the remains of a wall pierced by a
high-silled door and abutting the tower. I shall
return to this wall in a subsequent discussion.
Within a year of Chapin's report to the
Appalachian Mountain Club, a second account of Cliff
Palace was published by Dr. W.R. Birdsall (1891) .
Although he regarded the prehistoric Mesa Verdeans as a
race very low on the scale of development, Birdsall's
discussion is notable for his cautious attitude toward
ethnographic assumptions, like Chapin's, which were
already becoming attached to the prehistoric ruins. He
decried the lack of "investigations carried out
according to modern scientific methods for ethnological
and archaeological research," and noted with


disapproval that reports on other Southwestern ruins,
produced in haste after short, cursory visits, "have
served as the basis of original observation and
scientific authority for archaeological writers since"
(1891:610-611). Birdsall was particularly skeptical
about the equivalency of prehistoric kivas and Hopi
kivas. "The fact that they served any particular
purpose in other societies is not proof that they were
put to the same use by these primitive people"
(1891:614). He also questioned whether the prehistoric
people were the direct ancestors of any particular
living group, calling attention to the cultural and
linguistic variety among living Southwestern pueblo
It is ironic that such an eminently scientific
attitude toward the prehistoric ruin and its
inhabitants should have been expressed not by an
archaeologist or ethnologist, but by a medical doctor.
It is perhaps for that reason that he received only
scant acknowledgement from later workers. Certainly,
his cautionary remarks were ignored. The production of
Cliff Palace had already begun, in this sense: the
archaeological site was being treated not as the
architectural and artifactual remains of an unknown
people, but as a ruined Hopi pueblo whose functions and

inhabitants were already known. That production was
furthered by Gustav Nordenskiold, who added the
authority of his own, extensive work and description on
the Mesa Verde.
During the summer of 1891, Gustav Nordenskiold
explored, mapped and photographed a large number of
cliff alcove ruins on the Mesa Verde, including Cliff
Palace. Naming it "probably the largest ruin of its
kind known in the United States," his impression of
Cliff Palace recalls that of Chapin before him:
[W]ith its round towers and high walls rising out
of the heaps of stones deep in the mysterious
twilight of the cavern, and defying in their
sheltered site the ravages of time, it resembles
at a distance an enchanted castle (1893:59).
Nordenskiold contributed detailed descriptions of
many aspects of room and kiva (estufa) construction,
masonry, decoration and condition. His photographs
document the condition of the site as it then existed.
And he published the first widely circulated map of
Cliff Palace, which I have reproduced as Figure 4.5.
On the map, he numbered 102 ground level rooms,
including seventeen kivas (Fewkes reported
Nordenskiold's count as 102 rooms and seventeen kivas

From Nordenskiold's map and the photograph
reproduced as Figure 4.4,^ the extent to which the site
had been reduced by the elements is apparent. (For the
reader's convenience, a list of eguivalent room numbers
according to Nordenskiold's and Fewkes' schemes is
given in Table 4.1. My map, Figure 4.1, employs
Fewkes' numbering scheme.) Masonry rubble completely
covered the floor of the alcove; Kivas A, B, C and F
and a number of rooms lay completely beneath the rubble
slope which continued down into the canyon. The entire
northwest corner of the sguare tower (Nordenskiold's
91, Fewkes' 11) had collapsed, largely into Kiva C. I
call attention to the quantity of rubble in the
photograph because the construction of which it was
once a part is now completely gone. Jesse Fewkes,
repairing the site twenty years later, did not attempt
to rebuild the ruin; he simply removed the overburden.
A large number of artifacts were still present in
Cliff Palace when Nordenskiold worked there, and he
photographed and described many of them. The dry
interior of the cliff alcove had left many artifacts in
a remarkable state of preservation, "even wooden
articles, textile fabrics [and] bone implements"
(1893:10). Unfortunately, he did not record the 1
1. This photograph was taken in 1899, but it is very
similar to Nordenskiold's published photograph.

Table 4.1. Corresponding Room Numbers for Data
Set (Appendix A), Milo's Map (Fig. 4.1) and Fewkes' Map
(Fig. 4.9), and Nordenskiold's Map (Fig. 4.5).
Data Set Fewkes & Milo Norden- skiold Data Set Fewkes & Milo Norden- skiold
625001 1 100 625023 23 80
625002 2 99 625223 (23/2)
625003 3 98 625024 24
625004 4 101 625025 25 79
625204 (4/2) 1 625026 26 78
625304 (4/3) 625027 27 77
625005 5 97 625028 28 73
625205 (5/2) 625228 (28/2)
625006 6 94 625029 29 74
625007 7 102 625229 (29/2)
625008 8 96 625030 30 71
625009 9 95 625031 31 69
625011 11 91 625032 32 ?68
625211 (11/2) 625033 33
625311 (11/3) 625500
625411 (11/4) 625034 34 66
625012 12 92 625035 35
625013 13 90 625036 36 63
625014 14 89 625236 (36/2)
625214 (14/2) 625336 (36/3)
625015 15 87 625501
625215 (15/2) 625502 64
625016 16 625037 37 61
625017 17 86 625237 (37/2)
625217 (17/2) 625038 38 60
625317 (17/3) 625238 (38/2)
625018 18 85 625039 39 59
625218 (18/2) 625239 (39/2)
625318 (18/3) 625040 40 58
625019 19 84 625240 (40/2)
625219 (19/2) 625041 41 41
625020 20 83 625042 42 40
625220 (20/2) 625043 43 39
625021 21 82 625044 44 35
625221 (21/2) 625045 45 34
625022 22 81 625046 46 38
625222 (22/2) 625047 47 36
625247 (47/2)
1. Upper-story room numbers in parentheses are not shown
on the maps, but are used in the description.

Table 4.1. (contd.)
Data Set Fewkes & Milo Norden- skiold Data Set Fewkes & Milo Norden- skiold
625048 48 37 625073 73 17
625248 (48/2) 625273 (73/2)
625049 49 33 625373 (73/3)
625050 50 32 625074 74 15
625051 51 31 625274 (74/2)
625251 (51/2) 625374 (74/3)
625052 52 30 625075 75 22
625053 53 50 625076 76 105
625253 (53/2) 625077 77 104
625054 54 49 625078 78 103
625254 (54/2) 625079 79 13
625055 55 48 625279 (79/2)
625255 (55/2) 625080 80 14
625056 56 46 625280 (80/2)
625256 (56/2) 625081 81 8
625057 57 45 625281 (81/2)
625257 (57/2) 625381 (81/3)
625058 58 47 625082 82
625258 (58/2) 625282 (82/2)
625503 625083 83
625059 59 43 625084 84
625060 60 29 625085 85 6
625061 61 28 625504
625062 62 27 625086 86 5
625063 63 26 625087 87 3
625064 (64) 42 625287 (87/2)
625065 65 625088 88 2
625066 66 23 625288 (88/2)
625067 67 25 625388 (88/3)
625068 68 21a 625089 89 10
625268 (68/2) 625090 90 11
625069 69 24 625091 91
625070 70 20 625092 92
625071 71 18 625093 93
625271 (71/2) 625094 94
625072 72 16 625095 95
625272 (72/2) 625096 96

Figure 4.4. Cliff Palace in 1899, photographed by McKee.

Figure 4.5. Nordenskiold's map of Cliff Palace, 1893

provenience of the artifacts which he removed, most of
which were shipped to his native Scandinavia, where
they are today on display in the National Museum of
One remarkable circumstance which Nordenskiold
noted was the loss from the site of virtually all of
the wooden beams and rafters. Only the wall sockets
remained to indicate the former presence of beams.
Nordenskiold speculated that they had been removed by
the former inhabitants or later Indian visitors, either
for re-use or for firewood. Later, Fewkes guessed that
many of the beams had been removed for firewood by
parties of pot hunters wintering over in the ruin
(Fewkes 1911:24).
Nordenskiold accepted the equivalency of the
circular, subterranean rooms at Cliff Palace with the
estufas (kivas) of the Moqui (Hopi). He extended the
name estufa to two structures "of singular
construction," numbered 44 and 57 on his map and kivas
0 and R on Figure 4.1.2 These structures differ in
substantial ways from more typical kivas: they have no
pilasters, fire pits, sipapus or deflectors. In form,
they feature a rounded lower portion surrounded by a
2. The number "44" was actually omitted from
Nordenskiold's map. The structure is identified as
"Fig.35,36," referring to the accompanying figures in
his text.

high, rectangular, masonry enclosure. Both possess
shafts similar to the vent shafts associated with
conventional kivas, and it was on that basis that
Nordenskiold declared them to be kivas. He suggested
that they might represent a transition to the
rectangular kivas of the Hopi. Such a designation,
however, may very well mask the possibility that they
represent a separate class of structure.
Nordenskiold also made frequent references to the
"fortifications" of Cliff Palace, reflecting the
widespread belief that construction in cliff alcoves
was a response to the constant threat of attack. This
interpretation, produced in the 1890's and reproduced
for eighty years thereafter, seems to have been based
largely upon European notions of warfare and
fortification (e.g. Lancaster & Pinkley 1954:44; Rohn
1977) and an assumption of conflict between mobile and
sedentary populations. It therefore illustrates the
way in which European cultural biases were conflated
with ethnographic analogy during the production of the
interpretive canon surrounding the archaeological ruins
on the Mesa Verde and throughout the Southwest.
Because of his attention to detail and his status
as a scientist (and a European one, at that),
Nordenskiold's work instantly acquired an authority of

lasting influence. He affirmed the primitive nature of
indigenous American social development, stating that
"the former inhabitants of the cliff-dwellings were an
agricultural people on the level of the Stone Age"
(1893:20, original emphasis). He validated the
assumption of cultural and behavioral continuity from
the prehistoric past to the ethnographic present, and
he firmly established the Hopi as the living analogue
for the prehistoric Anasazi. Furthermore, he
reinforced the ascription of a ceremonial function to
the Mesa Verde kivas and broadened the identification
of kivas to include a variety of architecturally
diverse structures. There is, however, one observation
of Nordenskiold's which is worthy of mention even
though it is not specifically concerned with Cliff
Palace, precisely because it has been virtually ignored
by later researchers.
In his description of a kiva at Painted Kiva
House, MV-557, Nordenskiold noted that the kiva had the
shape, not of a cylinder, but of a "truncate cone"
(1893:15). Seventy-eight years later, Arthur Rohn, in
his excavation report on Mug House (MV-1229), remarked
of the kivas there that "[t]he lining wall almost
always sloped inward toward the top" (1971:68). L.E.
Bennett's 1932 map of Spruce Tree House documents the

same shape (i.e. a bell-shape) in all eight of the
kivas at that site (Fig. 4.6). Nevertheless, every
kiva on the Mesa Verde which has been excavated and
stabilized, including those at Cliff Palace, Mug House
and Spruce Tree House, has been reconstructed with
vertical side walls. Thus, at the same time that
interpretive assumptions from ethnography were added to
the archaeological record, a potentially significant
piece of empirical information was deleted from the
archaeological record. Clearly, this is an aspect of
kiva construction which merits further investigation.

Figure 4.6. Kiva profile from Bennett's map of Spruce
Tree House, 1932.

Reduction and Production II:
Jesse Fewkes
By 1888, a booming market for prehistoric American
antiquities was resulting in the wanton pillaging of
most of the major archaeological sites on the Mesa
Verde, including Cliff Palace. When Jesse Walter
Fewkes began repair and stabilization at Cliff Palace
in May, 1909, the site had suffered more extensive
physical damage it had been reduced to a greater
extent than any other cliff dwelling in Mesa Verde
National Park (Fewkes 1911:24).
At the time of the site's discovery by Euro-
Americans, most of the damage had been done by the
action of the physical environment: rain, wind, snow
and the annual freeze-thaw cycle. This kind of
reduction was particularly marked at the northern and
southern ends of the site, where the structures stood
out beyond the protective overhang of the cliff. It is
also possible, as Fewkes himself noted (1911:27), that
many more structures once occupied the terraces across
the front of the site. These, too, would have been
directly exposed to the elements and would have
contributed to the enormous rubble slope visible in
Figure 4.4.
By far the most grievous destruction, however,
took place at the hands of pot hunters beginning at

about the same time as the first published descriptions
of Cliff Palace. In the scramble to obtain prehistoric
artifacts for public exhibition and private sale, the
site was rapidly stripped of the cultural material
which might have shed light on the domestic life and
social organization of the ancient inhabitants. Worse,
the pot hunters had dug over most of the floors,
thereby doubtless destroying important floor features
such as hearths and grinding bins. And, they had
seriously damaged many of the walls, sometimes
breaching them with dynamite to facilitate the removal
of refuse, undermining the foundations, or pushing them
wholesale down the slope (Fewkes 1911:25; Watson
n.d.:12). Some of these breaches are visible in Figure
[T]he vandalism wrought by those who had dug into
it had destroyed much data ... the ruin had been
almost completely rifled of its contents, the
specimens removed, and its walls left in a very
dilapidated condition. (Fewkes 1911:11-12)
Fewkes' institutional mission was to stabilize
Cliff Palace to prevent further deterioration and to
prepare it for display to the public. His personal
mission was to explain the ruin, and to him that meant
explaining the cultural and behavioral continuity of
the site's inhabitants with the living Hopi.

Believing that modern Pueblo culture is the
direct descendant of that of cliff-dwellers, the
writer has not hesitated to make use of
ethnology, when possible, in an interpretation of
the archeological [sic] record. (1911:12)
He approached both aspects of his work with vigor and a
clear sense of purpose.
The repairs at Cliff Palace consisted mainly in
clearing the masonry rubble and stabilizing the
remaining rooms and terrace walls. This included
rebuilding the terrace walls at the northern and
southern ends of the site where they had been most
exposed to the weather and to run-off from the cliff
overhang. In the northern extremity of the site,
especially, many of the original wall lines had been
completely obliterated. Most of the kivas had been
severely damaged by pot hunters and falling masonry.
Fewkes rebuilt the kivas, with vertical side walls, "to
their former height at the level of neighboring plazas"
(1911:25). He found the northwest wall of Room 39
completely undermined except at the corners and, in
rebuilding its foundation from the floor level of Kiva
M, he subtly altered the profile of that kiva
(1911:45,62). Perhaps the most ambitious construction
was devoted to the four-story tower, Room 11 (at the
right in Figure 4.4). The entire northwest corner of
the tower had collapsed, and it was necessary to build
up the corner from a new foundation laid at the floor

level of Kiva C (1911:41). His crew also added the
beams which now tie together the walls of the tower,
although he did not say from what source they were
obtained (1911:42). In order to protect the newly
exposed masonry from the weather, Fewkes capped many of
the walls and plaza surfaces with Portland cement and
built rain gutters to channel away flowing water.
Figure 4.7 shows Cliff Palace as it looked after
Fewkes' excavation and stabilization. The site had
become cleaned, stabilized, and frozen in time.
An example of the way that even minor physical
changes can become a potential source of error in
stabilization and in re-studies is the case of a small
room which stood either to the north or the northwest
of the southern-most tower, Room 36 (near the center of
Fig. 4.4). Figure 4.3 is a photograph taken by
Frederick Chapin in 1890. The view is to the south,
with Room 36 just off center to the right and Rooms 37
and 38 visible to the left of the tower (cf. Fig. 4.1).
To the right of the first-story doorway in Room 36 are
the remains of a wall, bonded to the tower wall,
showing clearly the sill and one side of a well-formed
door. (Note, also, the traces of light colored plaster
above and around the tower door. Other photographs
suggest that a number of doors in Cliff Palace were

Figure 4.7.
Cliff Palace after stabilization, 1911

similarly decorated.) Nordenskiold published a very
similar photograph in 1893 which shows that the wall at
that time remained as Chapin had seen it (Nordenskiold
1893:61, Fig. 34).
Figure 4.8 was taken by McKee around 1899, ten
years before Fewkes7 excavation and repair. By that
time, the mortar bonding the upper part of the former
wall had deteriorated, and the slab which had formed
the side of the door had toppled onto the sill.
Nordenskiold figured on his map a parallel wall to the
west of the one I am describing, and he numbered the
space between the walls as 64 (Fig. 4.5; some of the
masonry may be visible in Fig. 4.8, as well as in
Nordenskiold7s Fig. 34). He did not, however, record
the presence of the door. Notwithstanding Fewkes7
familiarity with Nordenskiold7s report and photographs,
he and his crew evidently took the fallen slab as the
top of the wall. Neither did Fewkes find any traces of
the parallel wall which Nordenskiold had noted. Note
on Fewkes7 map, Figure 4.9, however, the indication of
a wall perpendicular to the first and extending to the
west, in which location it would represent the south
wall of Nordenskiold7s Room 64. Whether or not Fewkes,
himself, directed the inclusion of that south wall on
his map is open to doubt, for there are other

Figure 4.8
Room 36, photographed by McKee around 1899

Figure 4.9
Jesse Fewkes' map of Cliff Palace, 1911.