i
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
T H E U N I V E R S I T Y O F C H I C A G O
O R I E N T A L I N S T I T U T E S E M I N A R S
N U M B E R 1
Series Editors
Thomas A. Holland
and
Thomas G. Urban
ii
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
iii
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
CHANGING SOCIAL IDENTITY
WITH THE SPREAD OF ISLAM
ARCHAEOLOGICAL PERSPECTIVES
edited by
DONALD WHITCOMB
with Case Studies by
JODI MAGNESS, TRACY HOFFMAN,
YURY KAREV, MARK C. HORTON,
and
TIMOTHY INSOLL
THE ORIENTAL INSTITUTE OF THE UNIVERSITY OF CHICAGO
ORIENTAL INSTITUTE SEMINARS • NUMBER 1
CHICAGO • ILLINOIS
iv
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
Library of Congress Control Number: 2004117327
ISBN: 1-885923-34-1
The Oriental Institute, Chicago
©2004 by The University of Chicago. All rights reserved.
Published 2004. Printed in the United States of America.
Series Editors’ Acknowledgments
The assistance of Katherine Strange Burke, Lindsay
DeCarlo, Katie L. Johnson, Leslie Schramer, Alexandra
Witsell, and Ilya Yakubovich is acknowledged in the production
of this volume.
Front Cover Illustration
Egypt Gaziret Quoma
Muslim man who has painted his pilgrimage to the
Haj on his home.
©Hutchison Picture Library/Chris Parker
Title Page and Back Cover Illustration
Ballas style of water pot, excavated at Quseir alQadim
in Egypt and dated to the fourteenth
century. This pot is in the same tradition as the
Nubian water jar in the front cover illustration and
continues to have a widespread distribution in contemporary
Egypt.
Printed by McNaughton & Gunn, Saline, Michigan
The paper used in this publication meets the minimum
requirements of American National Standard for Information
Services — Permanence of Paper for Printed
Library Materials, ANSI Z39.48-1984.
v
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
TABLE OF CONTENTS
PREFACE ....................................................................................................................................... vii
LIST OF ABBREVIATIONS ......................................................................................................... ix
LIST OF FIGURES......................................................................................................................... ix
INTRODUCTION. Donald Whitcomb ........................................................................................... 1
THE CASE STUDIES
1. KHIRBET ABU SUWWANA AND EIN ªANEVA: TWO EARLY ISLAMIC
SETTLEMENTS ON PALESTINE’S DESERT PERIPHERY. Jodi Magness................... 11
2. ASCALON ON THE LEVANTINE COAST. Tracy Hoffman ............................................ 25
3. SAMARQAND IN THE EIGHTH CENTURY: THE EVIDENCE OF
TRANSFORMATION. Yury Karev ...................................................................................... 51
4. ISLAM, ARCHAEOLOGY, AND SWAHILI IDENTITY. Mark C. Horton ..................... 67
5. SYNCRETISM, TIME, AND IDENTITY: ISLAMIC ARCHAEOLOGY
IN WEST AFRICA. Timothy Insoll ...................................................................................... 89
v
vi
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
vii
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
PREFACE
The goal of this seminar was a comparative analysis of different sites and regions, based on
archaeological monuments or artifacts, exploring processes of adaptation or adjustment to local
cultural complexes. Islamic archaeology in Chicago’s Department of Near Eastern Languages
and Civilization may claim to be one of the few doctoral programs that stresses archaeology
rather than art history. This seminar was the first in a new program initiated by Professor Gil
Stein, Director of the Oriental Institute, and we are grateful for his recognition of the potential of
this field of studies. The seminars are intended to result in a monograph and we are grateful to
Thomas Urban, Oriental Institute Publications Office, for overseeing the production of this volume.
A special thanks may also be offered to Ms. Katherine Strange Burke, a doctoral candidate
in Islamic Archaeology, for her analysis of the Roundtable discussions, copy editing the papers
submitted, and page setting the final proofs. Finally we would thank Ms. Nicole Torres for organizing
facilities and hospitality in Chicago.
The participants and their papers were:
• Dr. Tracy Hoffman (University of Chicago), “Ascalon: Domestic Architecture
in a Byzantine-Islamic City”
• Prof. Renata Holod (University of Pennsylvania), “Territory and Text: Reconstructing
Settlement on Medieval and Early Modern Jerba” (to be published
elsewhere)
• Prof. Mark C. Horton (University of Bristol), “Islam, Archaeology, and Swahili
Identity”
• Prof. Timothy Insoll (University of Manchester), “Syncretism, Identity, and Islamic
Archaeology in West Africa”
• Dr. Yury Karev (Russian Academy of Sciences, Moscow), “Samarqand in VIII
Century, the Evidence of Transformation”
• Prof. Jodi Magness (University of North Carolina, Chapel Hill), “Khirbet Abu
Suwwana: An Early Islamic Village Near Jerusalem”
Principal discussants at the Roundtable included:
• Dr. Moain Sadeq (Palestinian Department of Antiquities in Gaza), Prof. Fred
Donner (University of Chicago), and Dr. John Meloy (American University in
Beirut)
vii
viii
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ix
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
LIST OF ABBREVIATIONS
A.H. anno Hegirae, in the year of [Muhammad’s] Hegira
C.E. of the common era
cm centimeter(s)
diss. dissertation
e.g. exempli gratia, for example
fig(s). figure(s)
ha hectare(s)
ibid. ibidem, in the same place
i.e. id est, that is
km kilometer(s)
m meter(s)
p(p). pages(s)
pl(s). plate(s)
LIST OF FIGURES
1. Map showing the spread of Islam and selected archaeological sites ......................................... 9
1.1. Plan of Areas A, B, and C at Khirbet Abu Suwwana ................................................................. 12
1.2. Plan of the building at Ein ªAneva .............................................................................................. 16
1.3. Pottery and stone bowl from Ein ªAneva.................................................................................... 13
2.1. Map of Palestine .......................................................................................................................... 26
2.2. Ascalon in the Madaba map ........................................................................................................ 28
2.3. Excavation areas in Ascalon ........................................................................................................ 29
2.4. Structure One. Ascalon ................................................................................................................ 30
2.5. Structure Two. Ascalon ............................................................................................................... 32
2.6. Structure Three, Phase 3. Ascalon............................................................................................... 33
2.7. Structure Four. Ascalon ............................................................................................................... 35
2.8. Grid 38 bath house. Ascalon........................................................................................................ 36
2.9. Structure Five. Ascalon ............................................................................................................... 37
2.10. Church near Jerusalem gate. Ascalon.......................................................................................... 38
2.11. “Bazaar house”: Fustat groupe I and maison VI. Fustat ............................................................. 45
2.12. “Family house”: Maison III, IV, VII, and VIIbis. Fustat ........................................................... 47
2.13. “Apartment house”: Maison II. Fustat ........................................................................................ 48
3.1. Afrasiab site, general plan ........................................................................................................... 54
3.2. Northern part of the Afrasiab site. “Sacred area” with the complex of the eighth
century and mosques above it, citadel-donjon, and lower terrace of the citadel
with the dΩr al-imΩra from the middle of the eighth century..................................................... 55
3.3. Different structures on the “sacred area.” Afrasiab .................................................................... 57
3.4. DΩr al-imΩra on the lower terrace of the citadel (ca. 750 C.E.). Afrasiab................................. 59
3.5. DΩr al-imΩra on the lower terrace of the citadel (ca. 750 C.E.). Afrasiab................................. 60
3.6. Palace in Pendjikent and Palace in ShahristΩn (Bundjikat) ....................................................... 62
3.7. Palace in Varakhsha ..................................................................................................................... 63
ix
x
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
3.8. Residential quarter in the western part of Afrasiab .................................................................... 65
4.1. Shanga. Aerial photograph of fourteenth century ruins. The site was built over sand
dunes (visible in the foreground). The Friday mosque is located in the center of the
site and the harbor is at the top of the figure............................................................................... 71
4.2. Shanga. Friday mosque, showing the main wall of the prayer hall, with well and
washing area in the foreground. The roof was supported on pillars and made of
thatch and timber, echoing earlier styles of timber mosques ..................................................... 73
4.3. Excavations inside the prayer hall of the Friday mosque at Shanga. The plaster floor
dates to around 1000 C.E., while the earlier stone mosque dates to 900 C.E. The
sand platform fill contains numerous small silver coins, probably carried with
the sand from the town’s trading beach ...................................................................................... 74
4.4. View of the excavations of the timber mosques at Shanga, showing remains of a
ninth century mosque made from small sticks. Note the central posthole. The
other postholes belong to later timber phases ............................................................................. 75
4.5. Plan of the sequence of mosques excavated at Shanga, dating from 750 to 1000 C.E............... 77
4.6. Minuscule silver coins from Shanga dating to the ninth century ............................................... 78
4.7. Staircase minaret at the eleventh century mosque of Kaole (Tanzania). The absence
of tower minarets in East Africa may indicate Ibadi influence in mosque design..................... 80
4.8. Excavations of the Mosque sequence at Ras Mkumbuu (Pemba Island). Below the
floor of a large mosque dating to 1050 are the remains of a small stone mosque,
the north wall of which had been robbed out, but was of double thickness,
suggesting an Ibadi-style mih≥rΩb. Below is an earlier timber building, which may
have been the first mosque on the site dating to the early tenth century ................................... 81
4.9. A simple Ibadi mih≥rΩb in a small contemporary mosque at Tumbe (Pemba Island).
The niche is marked in plaster, but no recess is in the wall ....................................................... 82
4.10. Part of the Mtambwe hoard (Pemba Island). The associated dinars give a deposition
date after 1066 C.E. Over 2,000 silver coins were found, locally minted in the names
of ten local rulers ......................................................................................................................... 82
4.11. The mih≥rΩb at the thirteenth century mosque of Ras Mkumbuu (Pemba Island) is
typical of many examples made from carved coral but with very simple decoration ............... 83
4.12. The very elaborate mih≥rΩb at Kizimkazi (Zanzibar Island) that has a date of A.H. 500
(1107 C.E.) and which is now considered to be largely of this date. It is very much
more complex than many Swahili mih≥rΩbs, and these may reflect doctrinal differences ......... 84
4.13. Baobab tree located at Mbaraki (Mombasa Island, Kenya) that is believed to be a location
of spirits, which has been festooned in rag cloths. In the background is the stone tower
that was attached to a small mosque, and which may have given the location its reputation ... 85
4.14. Siwa (or side-blown horn) decorating the “shrine” at Pujini (Pemba Island) made
in plaster relief. This underground chamber may represent a “spirit cave” ............................... 86
4.15. Deposit within mih≥rΩb of mosque at Chwaka (Pemba Island), comprising a bowl,
covering sherds, a seashell, a hand-grinding stone, and a piece of Chinese pottery ................. 86
4.16. Detail of the mih≥rΩb deposit from Chwaka ................................................................................. 87
5.1. The location of Gao (and other centers) in West Africa ............................................................ 90
5.2. Settlement structure at Gao.......................................................................................................... 92
5.3. View of the Niger River from Gao .............................................................................................. 93
5.4. Selection of imported trade goods recovered predominantly from Gao Ancien ....................... 95
5.5. The palace or rich merchant’s house ........................................................................................... 97
LIST OF FIGURES
1
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
INTRODUCTION
THE SPREAD OF ISLAM AND
ISLAMIC ARCHAEOLOGY
DONALD WHITCOMB
The Oriental Institute of the University of Chicago
On the 12th and 13th of May 2003, an unusual group of archaeologists gathered for a seminar
in the Oriental Institute of the University of Chicago. This assembly was unusual in that
individual specializations varied in geography, from West Africa to Central Asia, and in chronological
focus, from the seventh into the sixteenth centuries. The unifying factor was a common
concern for the regional impact of the spread of Islam and the process of its adoption within each
antecedent culture. One further element was a somewhat unusual attention to archaeological evidence
for interpreting social change. The following introduction attempts to suggest the
archaeological background for this seminar, then introduce some aspects of the case studies presented
here. The comments attempt to give a sense of the Roundtable discussion that allowed the
participants to react to each others ideas and look for common patterns on the subject of social
change.
The spread of Islam is a historical process that has multifaceted interpretations, rarely neutrally
observed either in the seventh century, the thirteenth century, or indeed in the modern
world. Explanations of this process and in particular its successes have been attempted by historians
and more recently by archaeologists. One important example is the impact of Henri
Pirenne’s historical theses and the re-analysis by David Whitehouse and Richard Hodges, and
again by Hodges.1
These latter studies have relied on syntheses of widespread sources of archaeological
investigations. While the “barrier” of the Mediterranean has been effectively
discarded, more subtle conceptual constraints of core and periphery, urbanized and undeveloped,
regionalism, ethnic characterization, and so forth, tend to limit broader understandings. New approaches
to all these aspects are major concerns in modern archaeological research.
This seminar recalls a rather different conference of fifty years ago, organized by anthropologists
and Islamicists from Chicago but held in Germany.2
This meeting was one of a series
of “Comparative Studies in Cultures and Civilizations” organized by Robert Redfield in order to
compare research in different countries toward a common understanding of cultural patterns.
The conference organized by Gustave E. von Grunebaum may seem today heavily philological
and perhaps even “orientalist” in emphasis. This in turn recalls another conference, this time in
London during 1972, which presented a series of comparative studies on “conversion to Islam.”
This conference was quite innocent of archaeology, though its convener, Nehemia Levtzion,
mentions the need for imagination and new methods and immediately suggests the important po-
1
1. R. Hodges and D. Whitehouse, Mohammed,
Charlemagne and the Origins of Europe: Archaeology
and the Pirenne Thesis (London, 1983). See
also, R. Hodges, Towns and Trade in the Age of
Charlemagne (London, 2000).
2. Gustave E. von Grunebaum, “Islamic Studies and
Cultural Research,” in Studies in Islamic Cultural
History, edited by Gustave E. von Grunebaum
(American Anthropological Association, Memoir
76; Menasha, 1954), pp. 1–22.
2
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
tential for archaeological evidence.3
Islamic archaeology has a long tradition of contributions to
art historical studies of the Middle East, as noted below. The wider utilization for Islamic history
is only recently coming to the fore; one might note that none of the archaeological evidence reported
in this volume was available until after the early 1970s. These studies bring innovative
approaches to Islam and the Islamic tradition, beyond the pioneering efforts to understand historical
phenomena through archaeology. A continuing strategy is the juxtaposition of different
fields of study to realize common patterns of information and interpretation.
THE ROLE OF ISLAMIC ARCHAEOLOGY
Islamic archaeology has been practiced for about a century, and more importantly, significant
growth has taken place in this field in recent times. The earliest excavation in each new
region was usually a high-profile, large site of symbolic importance. Samarra remains the archetype
with an underlying purpose of aesthetically significant discoveries; yet even this site has
found significant reinterpretations (alas, without further excavations).4
The next stage in the development
of Islamic archaeology was the multiplication of investigations of specialized sites;
more than common tells, citadels, ports, and especially “desert castles” were examined as classes
of comparable examples. An archaeological alternative to this approach was the regional survey
resulting from the contextualization of elements into a broadly functional logic. The next stage
for Islamic archaeology is a focus on socio-cultural or historical problem solving.
While an indirect relation to archaeological programs is implied in other periods, the
professionalization of the discipline of Islamic archaeology requires a formal consideration of
methodology. Treatment of artifacts using a distinctively archaeological methodology begins
with the unique context of assemblages of abstracted types and employs a comparative mode of
analysis to produce patterns of data sets. Such archaeological patterns are internally generated
but do not usually exist outside of an interpretative framework, the socio-historical matrix. Thus
archaeological procedures are distinctive and capable of producing information and understanding
inaccessible by other modes of analysis. Archaeology is not any study of old or excavated
artifacts, it is a specific procedure with the goal of producing generalizing interpretations on
various scales of abstraction.
Alternative archaeologies may be defined as a search for new contexts.5
Thus for Timothy
Insoll, definition of this field is found in the explicit archaeology of religion, and he attempts to
make the case that material culture may be seen consistently through cultic or spiritual influences
of this alternative archaeology.6
Islamic archaeology is practiced as a historical
archaeology providing vital evidence for the development of society and economy in Islamic
3. Nehemia Levtzion, “Toward a Comparative Study
of Islamization,” in Conversion to Islam, edited by
Nehemia Levtzion (New York, 1979), p. 4.
Levtzion specifically mentions the inscriptions published
by J. Sauvaget, “Les épitaphes royales de
Gao,” Bulletin de l’Institut fondamental d ’Afrique
noire 12 (1950): 318–40.
4. See C. F. Robinson, editor, A Medieval Islamic City
Reconsidered: An Interdisciplinary Approach to
Samarra (Oxford, 2001).
5. Or put somewhat better, “From a methodological
point of view, much of the theoretical debate in
today’s archaeology can be seen as a search for new
contexts” (A. Andrén, Between Artifacts and Texts:
Historical Archaeology in Global Perspective [New
York, 1998], p. 155).
6. Timothy Insoll, The Archaeology of Islam (Oxford,
1999). Insoll’s approach to archaeology has its
roots in the University of Cambridge school of postprocessual
theoretical endeavors; the field of archaeology
of religion is one of these branches. See
also Donald Whitcomb, “Review of The Archaeology
of Islam, by Timothy Insoll,” American Journal
of Archaeology 104 (2000): 413–14.
DONALD WHITCOMB
3
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
contexts. Each project, whatever its intended goals, produces informative associations that may
be applied to relevant textual sources.
The intention of this seminar was not reduction to a sort of essentialism but regional comparisons
exploring Islamic archaeology from different disciplinary perspectives. The
comparative analysis of different sites and regions, based on archaeological monuments or artifacts,
allows an exploration of processes of adaptation or adjustment to local cultural complexes.
Islam may be seen as a religion, political system, and cultural complex, a trinity of inseparable
aspects. The introduction of these variable characteristics of Islam, during its contact with an antecedent
culture and afterwards, resulted in changes in identity approached as a sort of
“cognitive” archaeology. In each specific case, one may assess the nature of the pre-Islamic regional
tradition, the resulting plurality of cultures as a “multi-cultural” society, and finally a
resultant normative condition as a regional or cosmopolitan culture.
THE SPREAD OF ISLAM
The Islamicization of the Middle East (and beyond) has its origins in the Muslim conquest
of the early seventh century. Explanations of this conquest seem to fail modern scholars no less
than the Byzantines and Sasanians who actually confronted this movement. Fred M. Donner’s
extensive analysis moves away from dissatisfactory militaristic explanations toward changes in
“social organization.”7
While the initial successes of Islam became a state acquisition of territory,
the later and long-term adherence to Islam may be sought in social interactions which
resulted in a shift of identity. As Nehemia Levtzion notes, “Islamization of a social or ethnic
group is not a single act of conversion but a long process toward greater conformity and ortho-
doxy.”8
Conformity (and orthodoxy) may find expression in the adoption of architecture and
objects as signs and ultimately symbols of this new identity. Thus variations repeated in vastly
different regions and times may form patterns not completely explained through textual resources
but amenable to archaeological analysis. The purpose of this seminar was to adduce a
very limited number of specific examples toward this patterning in the spread of Islam.
The early period presents a dual problem during the establishment of political control over
pre-existing cultures: first, the self-formulation of Islamic society; and second, formulation of a
multi-cultural society. The resulting configuration is defined as Islamic, both politically and culturally.
In later periods, complexities of Islamic cultures lead to varied experiences of expansion
and domination. Philip L. Kohl’s Islamic archaeology is an archaeology without Islam, a medieval
and pre-modern archaeology of the Middle East.9
This approach stands in direct contrast to
that of Insoll, whose view of Islam is essentially non-historical and tends to offer a static description
of Islamic religion.10
The following case studies are not a search for answers but
examples drawn from archaeological evidence to be utilized with historical information for more
sophisticated narratives and perhaps models and theories of development.
7. Fred M. Donner, The Early Islamic Conquests
(Princeton, 1981), p. 269; an understanding of these
changes in consideration of the character of historical
sources is expanded in Fred M. Donner, Narratives
of Islamic Origins: The Beginnings of Islamic
Historical Writing (Princeton, 1998).
8. Nehemia Levtzion, “Toward a Comparative Study
of Islamization,” p. 21.
9. Philip L. Kohl, “The Material Culture of the Modern
Era in the Ancient Orient: Suggestions for Future
Work,” Domination and Resistance, edited by
D. Miller, M. Rowlands, and C. Tilly (London,
1989), pp. 240–45; he discusses Islamic archaeology
from an “evolutionary focus,” as a backdrop for
Western capitalism and colonialism.
10. Insoll, Archaeology of Islam, p. 13.
INTRODUCTION: THE SPREAD OF ISLAM AND ISLAMIC ARCHAEOLOGY
4
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
THE CASE STUDIES
We begin with Abu Suwwana in one of the most intensively studied regions, through both
regional surveys and excavations. There is modern revisionism: on one side Neil A. Silberman,
in his “thundering hoards,” exposes the fixed models and agenda inherited from Biblical archae-
ology;11
on the other hand, Yehuda Nevo attempts to apply the current debate on the nature of
early Islamic historical sources to archaeological evidence (producing new, agenda-driven dis-
tortions).12
Located on the desert periphery, Abu Suwwana was a rescue excavation of an early
Islamic site, recovering the mosque and residences of a large village. The sophisticated debate
on chronology is possible through the detailed publication of all artifacts, typical of modern archaeology
in Israel. The study of ceramics has advanced significantly with the research of Jodi
Magness, Miriam Avissar, and Yaºel Arnon, all based on controlled excavations of Islamic sites.
Ein ªAneva, though an isolated building, is offered as a comparandum for chronology and rural
architecture. The original implication that the site was abandoned (destroyed) by the “thundering”
Muslim conquest is effectively countered.
Perhaps the most interesting and valuable contribution in this study is the exposition of
modular architecture, a repetition of small parallel rooms. This derives from Mordechai
Haiman’s depiction of “nucleus units” in the Negev during the early Islamic period and, both
agree, reflects a specific social organization, which differs from that implied by dwellings of the
Roman and Byzantine periods in Palestine. The Bedouin may become described as undergoing a
process of sedentarization from transhumance in Arabia; a very similar pattern was explored in
detail at the site of al-Risha by Svend Helms.13
The result is mixed architecture of modular
settlements on the desert periphery in the early Islamic period.14
These villages are suggestive of
the urban changes as a sort of “Polis to Madina” revisited,15
social and legal explanations repeated,
such as the nature of commercialization. These are aspects further explored in the next
paper.
Ascalon (or Asqalon) represents archaeology in an urban setting. As Tracy Hoffman points
out, the normal indications of change, public institutions such as the mosque, are not available
except through limited textual descriptions.16
Rather, residential structures form the primary evidence
of change. There is a continuity in basic structures with some reorganization of interiors,
11. Neil A. Silberman, “Thundering Hordes: The Image
of the Persian and Muslim Conquests in Palestinian
Archaeology,” in Studies in the Archaeology of Israel
and Neighboring Lands in Memory of Douglas
L. Esse, edited by Samuel R. Wolff (Studies in Ancient
Oriental Civilization 59; American Schools of
Oriental Research Books 5; Chicago and Atlanta,
2001), pp. 611–23. Silberman advocates “a more
complex conception of cultural reorganization,”
largely through archaeological research (ibid., p.
616) and, in context of Magness’ paper, a more sophisticated
understanding of pastoralists in pre-Islamic
as well as Islamic society.
12. This discussion is under the title of “revisionism in
the desert” in Donald Whitcomb, “Islam and the
Socio-cultural Transition of Palestine, Early Islamic
Period (638–1099 C.E.),” in The Archaeology of Society
in the Holy Land, edited by T. Levy (London,
1995), pp. 488–501.
13. Svend Helms, Early Islamic Architecture of the
Desert: A Bedouin Station in Eastern Jordan
(Edinburgh, 1990).
14. Whether this locational characteristic is selective or
accident of preservation remains to be explored.
15. This refers to the seminal article by H. Kennedy,
“From Polis to Madina: Urban Change in Late Antique
and Early Islamic Syria,” Past and Present
106 (1985): 3–27. One also finds echoes of an “appropriation
of the land,” suggested by O. Grabar,
The Formation of Islamic Art (New Haven, 1973).
16. This paper is abstracted from her dissertation, successfully
defended shortly before this conference,
Tracy Hoffman, “Ascalon ªArus ash-Sham: The Archaeology
and History of a Byzantine–Early Islamic
City” (Ph.D. diss., University of Chicago,
2003).
DONALD WHITCOMB
5
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
as well as suggestions of neighborhood changes. By intriguing coincidence the urban code of
Julian of Ascalon makes clear that urban changes began well before the advent of Islam. Fustat
houses are used to illustrate how this structural pattern continues in the prototypical “Islamic
city.” The resulting picture is an urban phenomenon which builds, as one might expect, from
pre-Islamic forms but in a new organizational pattern, a model that may be tested on other
Middle Eastern cities. Most importantly, one has a pattern that avoids the assumptions and dictations
of studies of European cities. This new type of medieval city may be seen as determined by
physical, formal attributes, moving beyond the subjective “feel” of the Islamic city, which Doris
Behrens-Abouseif so beautifully describes.17
At the same time that Arabs were settling into Palestinian villages and perhaps rehabilitating
the old houses of Ascalon, their armies set out from the amsar (Islamic urban foundations) in
Iraq in large numbers for Central Asia. They proceeded to colonize and rule these ancient lands
of the East. Samarqand was a capital of the Soghdian rulers, a distinct region located between
the Chinese and Sasanian empires. Yury Karev describes the archaeological evidence for the development
of this frontier of the first Muslim empire. His account is based on his participation in
French-Russian projects at Afrasiab, the site of medieval (and ancient) Samarqand. Despite, or
possibly because of, the exotic location of Samarqand, this project may represent a more traditional
archaeology, large-scale clearance of major architectural monuments on the citadel and
“sacred area." This is a careful analysis of fragmentary historical documentation balanced with
stratigraphic and architectural analyses, which are perhaps even more fragmentary and challenging,
toward the goal of reconstruction of a historical narration of urban development.
As one might expect, Soghdian antecedents are mixed with architecture and artifacts typical
of the western parts of the Muslim empire (e.g., Amman citadel). The view from the citadel is
one emphasizing the political change and implied acceptance by the elites, an aspect particular
to the conquest period and in line with the “official view” found in documentary sources. At
Marv, there are reports of massive Arab settlement in the villages, which may present another
side of this process.18
There is a need to look at regional archaeological evidence with
Samarqand as a part of a larger pattern, a perspective forced on archaeologists dealing with archaeology
in sub-Saharan Africa.
Mark C. Horton’s comprehensive presentation of Swahili culture provides an introduction
to a peripheral region of some archaeological vitality.19
The history of its archaeology is colonial
in two senses: the political milieu of its first practitioners and the dominant model for Swahili
culture, which led to an archaeology with “a very oriental interpretation [focusing] on stone
towns and their architecture,” thought to be importations of Arab and Persian culture. Horton
touches upon current revisionism stressing the dominance and continuity of the Tana tradition,
early Iron Age cultural patterns of the interior of East Africa. This debate has a correlate in the
material evidence of Islamic culture explored by Michael G. Morony;20
this approach often pre-
17. Doris Behrens-Abouseif, “La conception de la ville
dans la pensée arabe du Moyen Âge,” in Mégapoles
méditerranéennes: Géographie urbaine rétrospective,
edited by C. Nicolet (Paris, 2000), pp. 32–40.
18. Paul Wheatley, The Places Where Men Pray Together:
Cities in Islamic Lands, Seventh through the
Tenth Centuries (Chicago, 2001), pp. 187–89, on
Samarqand, see pp. 312–14.
19. One may note a recent synthesis of archaeological
research in East Africa; S. Pradines, Fortifications
et urbanisation en Afrique orientale (British Archaeological
Reports, International Series 1216;
Oxford, 2004).
20. Donald Whitcomb, “Toward a ‘Common Denominator’:
An Archaeological Response to M. Morony
on Pottery and Urban Identities,” in Identity and
Material Culture in the Early Islamic World, edited
by I. Bierman (Los Angeles, 1995), pp. 47–68.
INTRODUCTION: THE SPREAD OF ISLAM AND ISLAMIC ARCHAEOLOGY
6
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
sents a dichotomy between regional traditions and cosmopolitan identifications (perhaps an example
of the Great Tradition).
Departing from this archaeological tradition of African culture and testing these hypotheses
on the role of Islam, Horton analyzes his excavations at Shanga, a large town of the fourteenth
century with a central mosque having antecedents going back into the eighth century. He presents
the complications deriving from his interpretation of the archaeological evidence, on the
nature of buildings, the role in the community, and implications for the arrival and spread of Islam.
One fascinating aspect is the utilization of “the Swahili’s own oral traditions and
chronicles,” and in fact, the ethno-historical considerations of a continuing cultural tradition. He
contrasts this approach to a more abstracted model of conversion presented by Insoll.21
Both scholars are aware of the openings into enlarged debate brought by archaeological research
on the role of Islam. Evidence is revealing, such as the large post marking the mih≥rΩb,
which has correlates in early mosques of the Middle East. More vexatious is the search for evidence
of sectarianism in this Islamic archaeology. Horton uses the example of Ibadism as a
search for evidence of imported rites; this led to a lively discussion in the Roundtable session on
the comparative Ibadi evidence from the island of Jerba.22
Insoll takes the problem of Islamization directly across the continent of Africa to the Gao region
of Mali. He uses this example for a phased conversion model that views the process as one
of syncretism, though he recognizes the negative connotations of this term. While he sees conversion
as dependent on social grouping, he might have dissected the phases as processes of
imitation-hybridization-syncretism methodology in analysis of cultural change.23
He claims the
first to convert are nomad groups. This presents a double irony in that early Islamic tradition
questioned the faith or faithfulness of these tribal groups, and of course their evidence is possibly
the most difficult to detect archaeologically.
Not surprisingly, Insoll concentrates on urban elements, though his fluidity in population
composition may be overstated or peculiar to West Africa. Unlike his earlier account of this process,
he has left out the crucial stage of conversion of ruler (and associated elites) that is found
in so many other instances. Urban complexes are the location of most archaeological evidence,
while the pagani of rural localities are seen, not surprisingly, as the most conservative. His last
group to convert brings us back to rural Palestine and the evidence of early Muslim communities
there. Again his discussion of residual “cults” may find reference in Nevo’s misguided revisionism
(as noted above). One might ask whether the human representations at Jeme-Jenno are so
different from the impulse that led to the frescoes adorning Qusayr ‘Amra in Jordan.
Thus, Insoll’s hypotheses inspire comparative observations and may spark a reconsideration
of implications from the material evidence from the previous case studies. As with the other papers
presented here, there is a patterning through contextualizations, which is shown to be the
fundamental methodology of archaeology. He ends with a note on the “privilege of archaeolo-
21. This is found in his detailed survey, Timothy Insoll,
“The Archaeology of Islam in Sub-Saharan Africa:
A Review,” Journal of World Prehistory 10 (1996):
439–504, and Timothy Insoll, The Archaeology of
Islam in Sub-Saharan Africa (Cambridge, 2003).
22. Renata Holod, “The Medieval and Early Modern
Periods on Jerba: A Preliminary Report on the Jerba
Archaeological Survey,” Africa (forthcoming). It is
particularly to be regretted that her paper at this
seminar, “Territory and Text: Reconstructing
Settlement on Medieval and Early Modern Jerba,”
could not be included here. On North Africa, one
may consult P. Pentz, From Roman Proconsularis
to Islamic Ifriqiyah (Göteborg, 2002).
23. Tasha Vorderstrasse, “A Port of Antioch under
Byzantium, Islam, and the Crusades: Acculturation
and Differentiation at al-Mina, A.D. 350–1268”
(Ph.D. diss., University of Chicago, 2004).
DONALD WHITCOMB
7
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
gists” to have access to information on these processes of change. In an earlier study, he emphasizes
“the privileged position of the Islamic archaeologist as an observer of a living religion.”
This activity is indeed privileged but is usually labeled ethnography; historical archaeology has
important implications for present populations, as explored in great detail by Silberman.24
CONCLUSIONS
Finally one may note that discussion at the Roundtable turned around two concerns: the nature
of questions asked of archaeological data and the relationship of this data to documentary
resources. For instance, some would suggest that the first century be treated as “prehistoric”
since documentation is so unreliable. Others would counter that interpretation of texts is always
important, but this is a critical operation just as is the interpretation of artifacts by archaeologists.
Studies of identity, a prominent concern in modern archaeology, adhere to varied fields of
theory and continue to struggle with the use of textual documentation or ethnological analo-
gies.25
Islamic archaeology needs to establish methodologies comparable to other branches of
modern archaeological study that provide frameworks for structuring material evidence and the
resultant historical narratives. One approach is to begin with the example of Muqaddasi, a tenth
century scholar who stressed geographical organization, not unlike the regional structure of the
seminar.26
Exposure to unfamiliar subjects and historical perspectives were not sufficient to initiate abstract
comparative modeling in a couple of days, though this may be hoped for in the future.
24. Timothy Insoll, The Archaeology of Islam, p. 54.
See Neil A. Silberman, Between Past and Present:
Archaeology, Ideology, and Nationalism in the
Modern Middle East (New York, 1989).
25. S. Shennan, editor, Archaeological Approaches to
Cultural Identity (London, 1989).
INTRODUCTION: THE SPREAD OF ISLAM AND ISLAMIC ARCHAEOLOGY
26. One may agree that “Geographical evidence has a
clear advantage over historical information in that it
is not so susceptible to dispute” (M. Lecker, Muslims,
Jews and Pagans: Studies on Early Islamic
Medina [Leiden, 1995]), p. 147.
8
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
9
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
Figure 1. The spread of Islam and selected archaeological sites
(adapted from D. Nicolle, Historical Atlas of the Islamic World [New York, 2003], pp. 6–7)
10
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
11
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
1
KHIRBET ABU SUWWANA AND EIN ªANEVA:
TWO EARLY ISLAMIC SETTLEMENTS
ON PALESTINE’S DESERT PERIPHERY
JODI MAGNESS
University of North Carolina at Chapel Hill
The history and archaeology of Palestine’s early Islamic cities and towns such as Jerusalem,
Ramla, Tiberias, and Caesarea are relatively well known. However, small villages and farmsteads,
especially in remote rural regions, have attracted less scholarly attention. In this paper, I
consider two early Islamic settlements located on Palestine’s desert periphery: Khirbet Abu
Suwwana and Ein ªAneva (see fig. 2.1 in Hoffman article). First, I examine and correct the
chronology of each settlement and suggest that they date to the eighth to ninth centuries. I then
consider the information that the layout of these settlements provides for changes in the social
structure of some Palestinian villages during the early Islamic period.
KHIRBET ABU SUWWANA
Khirbet Abu Suwwana is the site of an early Islamic village located near Maªaleh Adumim,
to the east of Jerusalem. In 1991, Ofer Sion conducted a rescue excavation at the site.1
The excavation
in three areas on the west (A, B, C) revealed a crowded system of residential units and a
mosque (see fig. 1.1), while in Area D to the east six residential units of high quality were uncovered.
Here I focus on the first of two main occupation periods that were distinguished. This
period, to which most of the ceramic finds and five of the six coins belong, showed evidence of
two sub-phases, which were dated by Sion as follows: Phase 1, from the second half of the seventh
century to the first half of the eighth century (the Umayyad period); during this period most
of the settlement was founded and reached its peak; Phase 2, from the second half of the eighth
century to the first half of the ninth century; during this period the settlement declined. The second
main period, after an occupation gap of 300 years, is dated to the Crusader period and is
characterized by nomadic settlement in the structures.2
The mosque is located on the northern side of Area A, among the twenty-two rooms and
courtyards that made up this residence. It had a doorway in the north wall, with a mih≥rΩb opposite.
Sion estimated that the mosque could have accommodated up to fifty-four worshipers (in
three rows of eighteen each).3
Except for the mosque, the rooms at the site were not plastered.
The stone walls were laid directly on exposed bedrock, following the topography. Most of the
1. See Ofer Sion, “Khirbet Abu Suwwana,” ªAtiqot 32
(1997): 183–94 (in Hebrew with English summary
on p. 50*; the pottery is published separately in English
by Joelle Cohen Finkelstein, “The Islamic
Pottery from Khirbet Abu Suwwana,” ªAtiqot 32
[1997]: 19*–34*). A preliminary report on the excavations
was published by Ofer Sion, “Khirbet
Abu Suwwana,” Excavations and Surveys in Israel
13 (1995): 66–67.
2. Sion, “Khirbet Abu Suwwana,” p. 50*.
3. Sion, “Khirbet Abu Suwwana,” p. 184.
11
12
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
Figure 1.1. Plan of Areas A, B, and C at Khirbet Abu Suwwana (from Ofer Sion, “Khirbet Abu
Suwwana,” ªAtiqot 32 [1997]: fig. 3; reprinted with permission of the Israel Antiquities Authority)
JODI MAGNESS
13
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
floors were of packed dirt and lime, and with one exception, all of the roofs were apparently flat
and of wood. The dwellings consisted of courtyards and spacious rooms, with numerous ovens
and installations.4
Most of the few bones recovered were identified as sheep and chickens, from
floor levels and near ovens. An infant burial found below the floor in Unit D6, about 0.5 m below
floor level is probably Bedouin.5
Sion estimated the population of the settlement during Phase 1 at about two hundred inhabitants,
based on the assumption that five people lived in each room, and the calculation that 40%
of the 100 rooms excavated were used for dwelling purposes. The presence of agricultural installations
and pens indicates that the economy was based on agriculture and animal husbandry.
Many pens were surveyed inside and outside the village; more than ten are scattered at a distance
of up to 250 m from the site. Smaller pens are located within the village. If each herd had
sixty to eighty sheep, the total estimated number of sheep was 600–800. No other early Islamic
sites were found in surveys to the east of the site, but seven were identified in surveys to the
north, and twelve more to the south. Only three others were of the same size as Khirbet Abu
Suwwana. During Phase 2, the settlement at Khirbet Abu Suwwana declined, and the population
decreased by 35%.6
Sion’s chronology and historical conclusions are based on the numismatic and ceramic evidence.
Eight coins were found at the site: one of Anastasius I (498–518); three Umayyad coins
(one of 700–749; a fragment dating to before 749; and a third that is post-reform, of 697–749);
one Umayyad or ªAbbasid coin (about 750); a Byzantine-Arab coin; a Crusader coin; and one
unidentified coin, perhaps of the eighth or ninth century.7
85% of the ceramic material and five
out of six of the coins belong to the first main occupation period, mostly to Phase 1. 87% of the
finds from the rooms in Area A are from the seventh to eighth centuries. According to Sion,
most of the ceramic material from the mosque and the bronze coin (no. 2, Umayyad, 700 –749,
from the floor level in Room 18) indicate that the mosque was constructed in the second half of
the seventh century, while the rest of the finds, which are from Phase 2, date from the eighth to
ninth centuries. Soundings dug to bedrock in most of the rooms in Areas B and C revealed just
one occupation level.8
The floors in L1003 and 1026 (Area B) and L1027 (Area C) were dated
on the basis of the ceramic material from the seventh to eighth centuries. According to Sion,
Unit B2 was continuously occupied until the ninth century, whereas Rooms 1007, 1008, 1010,
1017, 1024, 1026, and 1038 were neglected/abandoned during Phase 2. All of the rooms in Area
C were built during Phase 1; during Phase 2 Rooms 2001 and 2004 went out of use. Thus, most
of the pottery from Area C belongs to Phase 1. Similarly, 67% of the pottery from Area D belongs
to Phase 1, and 22% to Phase 2.
Joelle Cohen Finkelstein, who published the pottery from Khirbet Abu Suwwana, noted the
absence of signs of destruction that could account for the end of occupation at the site. Few complete
or almost complete vessels were found in the excavations, and most of the pottery studied
came from fills. Floors were exposed in some of the loci. Usually, only one floor level was con-
4. Sion, “Khirbet Abu Suwwana,” pp. 183–84.
5. Sion, “Khirbet Abu Suwwana,” p. 50*.
6. Sion, “Khirbet Abu Suwwana,” p. 192. Sion dated
the second sub-phase of the first main occupation
period to the ªAbbasid period and attributed the decline
at this time to the tribal revolts in 744–745.
7. Sion, “Khirbet Abu Suwwana,” p. 50*, 191; the
coins were identified by Rachel Milstein, Gabriela
Bijovsky, and Ofer Sion.
8. Sion, “Khirbet Abu Suwwana,” p. 191.
KHIRBET ABU SUWWANA AND EIN ªANEVA
14
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
structed above bedrock, and sometimes reused. In a few loci an earlier floor level was reached.
Limited soundings or full-scale excavations were undertaken below some of the floors in order
to reach bedrock. However, since in most cases the material from the floor was not separated
from that of the layer between the bedrock and the makeup of the floor, the pottery cannot accurately
date the construction of the floors.9
Finkelstein dated Phase 1 from the mid-seventh
century to mid-eighth century, and Phase 2 from the late eighth to early ninth centuries. A review
of the ceramic material indicates that this chronology is too high. This is most easily
demonstrated by examining the pottery from Phase 2, which is when glazed pottery first appears.
Phase 2: The glazed bowls from Phase 2 are made of the well-levigated, sandy buff to pink
ware and have the slightly curved walls and flaring rim characteristic of early Islamic glazed
bowls.10
The bowls also have the white slip characteristic of Miriam Avissar’s “Fine Glazed
Bowls,” while their exterior glaze is a feature of that same class.11
The splashed monochrome
green and polychrome glazed decoration on the interiors of many of the bowls corresponds with
Avissar’s “Fine Glazed Bowls, Types 6–9.”12
The sgrafitto bowls also belong to this class.13
These early Islamic glazed wares are dated by Avissar to the ninth to tenth centuries, though
sgrafitto bowls did not appear before the end of the ninth century.14
The fact that sgraffito bowls
are the least common type at Khirbet Abu Suwwana15
suggests that Phase 2 should be dated
from the ninth century to the first half of the tenth century. This chronology is supported by the
other ceramic types found in this phase, which include an incurved rim basin of eighth to tenth
century date,16
buff ware (“Mefjer ware”) jugs,17
early ªAbbasid Mahesh ware vessels,18
and
channel-nozzle oil lamps of the eighth to tenth centuries.19
The only cooking vessel illustrated
from this phase is a neckless, globular pot that has a rim with a lid device, horizontal handles,
and no traces of glaze.20
9. Finkelstein, “Islamic Pottery,” p. 19*.
10. Finkelstein, “Islamic Pottery,” 24*, fig. 3:1–2, 4–11
(the bowl in fig. 3:2 has a different profile); idem,
“Islamic Pottery,” p. 31*; see Miriam Avissar, “The
Medieval Pottery,” in Yoqneªam 1: The Late Periods,
edited by Amnon Ben-Tor, Miriam Avissar,
and Yuval Portugali (Qedem Reports 3; Jerusalem,
1996), pp. 75, 78. Some have the thicker walls
characteristic of Avissar’s “Common Glazed
Bowls” (Finkelstein, “Islamic Pottery,” fig. 3:4–7;
Avissar, “The Medieval Pottery,” p. 75), while others
have the thinner walls seen in her “Fine Glazed
Bowls” (Finkelstein, “Islamic Pottery,” fig. 3:1, 8,
Avissar, “The Medieval Pottery,” p. 78). Similarly,
both the disc bases characteristic of Avissar’s
“Common Glazed Bowls” are represented
(Finkelstein, “Islamic Pottery,” fig. 3:9, 11), and
the low ring bases of her “Fine Glazed Bowls”
(Finkelstein, “Islamic Pottery,” fig. 3:10; Avissar,
“The Medieval Pottery,” p. 78).
11. Finkelstein, “Islamic Pottery,” p. 24*; see Avissar,
“The Medieval Pottery,” p. 78.
12. Finkelstein, “Islamic Pottery,” p. 24*, fig. 3:1–2, 4–
7, 9, 11; see Avissar, “The Medieval Pottery,” pp.
78–82.
13. Finkelstein, “Islamic Pottery,” p. 31*, 24*, Group
D, fig. 3:8, 11; see Avissar, “The Medieval Pottery,”
pp. 81–82, Type 7, “Polychrome Splashed
and Mottled Sgrafitto Ware.”
14. Avissar, “The Medieval Pottery,” pp. 78–82.
15. Finkelstein, “Islamic Pottery,” p. 31*.
16. Finkelstein, “Islamic Pottery,” p. 22*, fig. 2:1; see
Jodi Magness, Jerusalem Ceramic Chronology,
circa 200–800 C.E. (Sheffield, 1993), pp. 210–11.
17. Finkelstein, “Islamic Pottery,” p. 28*, fig. 7:5–6, 8–
13; see Avissar, “The Medieval Pottery,” pp.
155–61, Types 2–12; James A. Sauer and Jodi
Magness, “Ceramics of the Islamic Period,” in The
Oxford Encyclopedia of Archaeology in the Near
East, Volume 1, edited by Eric M. Meyers (New
York, 1997), pp. 477–78.
18. Finkelstein, “Islamic Pottery,” p. 27*, fig. 6:6, 8;
28*, fig. 7:1–4; see Donald Whitcomb, “Mahesh
Ware: Evidence of Early Abbasid Occupation from
Southern Jordan,” Annual of the Department of Antiquities
of Jordan 33 (1989): 269–85.
19. Finkelstein, “Islamic Pottery,” p. 30*, fig. 8; see
Magness, Jerusalem Ceramic Chronology, pp. 258–
59, Oil Lamps Form 5.
20. Finkelstein, “Islamic Pottery,” p. 28*, fig. 7:17; see
Avissar, “The Medieval Pottery,” pp. 132–34; it
most closely resembles her Cooking Pots from the
Early Islamic Period, Type 1, which differs in having
strap handles.
JODI MAGNESS
15
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
Phase 1: If Phase 2 at Khirbet Abu Suwwana dates from the ninth century to the first half of
the tenth century, what is the date of Phase 1? The diagnostic types from Phase 1 include bowls
with a mid-seventh to ninth or tenth century range,21
incurved rim basins,22
a buff ware bowl,23
and jars and jugs of buff ware (“Mefjer ware”).24
A metallic buff jug with painted white bands
on the shoulder is a type that appeared at Pella in the late sixth century and is represented in the
746/747 destruction level at that site.25
The storage jars with swollen necks are dated from the
late seventh to ninth or tenth centuries.26
Another type of storage jar characterized by a sharp
ridge in the middle of the neck should be assigned to the eighth to ninth centuries.27
The cooking
vessels from Phase 1 include a casserole with horizontal handles, the deep form and dark brown
ware of which point to a late seventh/early eighth to ninth or tenth century date.28
Another cooking
pot with globular body, no neck, and two horizontal handles has parallels of tenth to early
eleventh century date from Yoqneªam.29
Though some of the ceramic types from Phase 1 have a range beginning in the late sixth to
seventh century, the complete absence of types characteristic of Jerusalem in the mid-sixth to
late seventh centuries (such as Late Roman red wares, Fine Byzantine Wares of mid-sixth to late
seventh century date, and large candlestick oil lamps) provides a late seventh century terminus
post quem for the establishment of the settlement at Khirbet Abu Suwwana. The ceramic and numismatic
evidence thus indicates an eighth century date for Phase 1 and a ninth to early or
mid-tenth century date for Phase 2.
EIN ªANEVA30
Ein ªAneva is a small spring on the south bank of Nahal Zeelim (Wadi Seiyal), about 4 km
north of Masada. The spring’s output was greater in antiquity than today, and evidence indicates
that the water was used to irrigate about ten dunams of agricultural terraces below it and to the
west. In the past, a small building to the west of the spring was identified as a Roman fort
erected to guard the spring during the siege of Masada in 72/73 C.E.
In 1981 Yosef Porath excavated this structure, which turned out to be a small rectangular
dwelling (ca. 6.0 ≈ 15.5 m) with two similar habitation units aligned on the same axis (see fig.
1.2). This axis was oriented north–south, so that the building was aligned with the riverbed.
21. Finkelstein, “Islamic Pottery,” p. 21*, fig. 1:2; 22*,
fig. 2:3, 6, 14; Magness, Jerusalem Ceramic Chronology,
pp. 198–99, Fine Byzantine Ware Bowls
Forms 2A and 2B.
22. Finkelstein, “Islamic Pottery,” p. 21*, fig. 1:7, 10,
fired buff, not dark brown.
23. Finkelstein, “Islamic Pottery,” p. 22*, fig. 2:16,
Mahesh ware?
24. Finkelstein, “Islamic Pottery,” p. 25*, fig. 4:1, 3,
10–11, though not identified as such by Finkelstein.
25. Finkelstein, “Islamic Pottery,” p. 25*, fig. 4:6; see
Alan G. Walmsley, “The Umayyad Pottery and Its
Antecedents,” in Pella in Jordan, Volume 1: An Interim
Report on the Joint University of Sydney and
the College of Wooster Excavations at Pella 1979–
1981 (Canberra, 1982), pp. 146–47, 156.
26. Finkelstein, “Islamic Pottery,” p. 26*, fig. 5:3–5;
Magness, Jerusalem Ceramic Chronology, pp. 230–
31, Storage Jars Form 7.
27. Finkelstein, “Islamic Pottery,” p. 26*, fig. 5:7–11.
28. Finkelstein, “Islamic Pottery,” p. 28*, fig. 7:21;
Magness, Jerusalem Ceramic Chronology, pp. 214,
Casseroles Form 3.
29. Finkelstein, “Islamic Pottery,” p. 28*, fig. 7:18;
Avissar, “The Medieval Pottery,” 132–35, Type 5.
The Yoqneªam examples differ, however, in their
dark red brown ware, and the splashes of glaze suggest
they are later than the Khirbet Abu Suwwana
piece (which is of light orange ware and is apparently
unglazed).
30. See Yosef Porath, “A Sixth–Seventh Century CE(?)
Structure Near ªEn ªAneva,” ªAtiqot 42 (2001):
51*–56*, with an English-language summary on pp.
324–25.
KHIRBET ABU SUWWANA AND EIN ªANEVA
16
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
Each unit consisted of an inner room and an outer room or courtyard, with the doorways opening
from the outer rooms or courtyards onto a path that led to the spring and the riverbed.31
All of
the floors of the building were made of packed earth laid over a fill. Both of the outer rooms or
courtyards had low stone benches built along the northern wall.32
Porath distinguished two occupation levels: Stratum 2 (the earlier phase) and Stratum 1 (the
later phase). A layer of marly dust mixed with pebbles that covered the floors of the Stratum 2
(earlier) building seems to have come from the collapse of the roof. This together with the paucity
of finds from this level suggested to Porath that the original building was abandoned before
it collapsed.33
The Stratum 1 building reused the earlier structure. The northern unit was reoccupied
and the floor levels were raised. The bench in the outer room was put out of use.34
The
southern unit showed evidence of two sub-phases. In the earlier sub-phase (1B), the two rooms
were united into one (L107). During this sub-phase, the building apparently consisted of one
dwelling unit (the northern half of the building with the inner and outer rooms) and a large
courtyard (or pen? = L107). Because no doorway was found in the wall between these two parts
of the building (the dwelling and courtyard or pen), access would have been from the outside.35
In the later sub-phase (1A), L107 (the southern unit) was redivided into two parts, along the
lines of the original plan.36
31. Porath, “ªEn ªAneva,” p. 51*.
32. Porath, “ªEn ªAneva,” p. 53*.
33. Porath, “ªEn ªAneva,” p. 53*.
34. Porath, “ªEn ªAneva,” p. 53*.
35. Porath, “ªEn ªAneva,” p. 53*.
36. Porath, “ªEn ªAneva,” p. 53*.
Figure 1.2. Plan of the building at Ein ªAneva (from Yosef Porath, “A Sixth-Seventh Century C.E.(?)
Structure Near ªEn ªAneva,” ªAtiqot 42 [2001]: plan 1; reprinted with
permission of the Israel Antiquities Authority)
JODI MAGNESS
17
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
Installations were discovered only in the northern unit, which is the better preserved part of
the building. An oven (tabun) was sunk into the floor in the northeast corner of the outer room
or courtyard of the northern unit. A niche that apparently served as a cupboard was built into the
wall at the southwest corner of the same room.37
The floors of the Stratum 1 building (last occupation
phase) were covered with a layer of fine, marly dust covered by large fieldstones. This
indicates that the Stratum 1 occupation, like Stratum 2, ended with the abandonment and eventual
collapse of the building.38
Porath concluded that the plan of the building and its location
next to agricultural fields indicate that it served a “civil,” not military function.39
According to Porath, the few finds recovered do not provide an accurate date since they
have a range from the second to eighth centuries C.E. He tentatively dated the structure to the
sixth to seventh century, suggesting that it served as an agricultural plot for the Byzantine monastery
at Masada beginning in the early sixth century and was deserted “some time after the
Islamic conquest of the country in the mid-seventh century.”40
However, an examination of the
finds indicates that the occupation dates to the early Islamic period.
Only seven finds — six fragments of ceramic vessels and a steatite bowl — are illustrated
from the excavations (see fig. 1.3). The only pieces published from a Stratum 2 context are a
cooking pot and the lower half of a jug.41
The cooking pot represents a type that is well known
from early Islamic contexts, and in my opinion, it should be dated mainly to the eighth century.42
The rest of the published material comes from Stratum 1 contexts. The finds include a deep casserole,
two casserole lids, and the rim and shoulder of a flask.43
Though none of these types is
closely datable, they could all be early Islamic. However, the steatite bowl, which comes from a
Stratum 1A context in L102 of the southern unit, undoubtedly dates to the eighth to ninth centu-
ries.44
The coarseness of the bowl, the ledge handles, and the signs of burning indicate that it was
used as a cooking vessel.45
The occupation at Ein ªAneva should therefore be dated to the eighth
to ninth centuries. As Porath noted, a great deal of effort was invested in the establishment of
this small agricultural holding, including the construction of terraces on the very steep bank of
Nahal Zeelim. The terraces were watered by a channel or aqueduct that brought water from the
nearby spring.46
Interestingly, although we know that the Roman army utilized the water from
the spring at Ein ªAneva during the siege of Masada, the only evidence for permanent settlement
and exploitation of the spring is during the early Islamic period.
MODULAR ARCHITECTURE
Porath noted the peculiar arrangement of identical side-by-side units at Ein ªAneva, each
consisting of an inner room and outer room or courtyard.47
These modular units — which have
been described elsewhere as “nucleus units” by Mordechai Haiman48
— are characteristic of
37. Porath, “ªEn ªAneva,” p. 53*.
38. Porath, “ªEn ªAneva,” p. 53*.
39. Porath, “ªEn ªAneva,” p. 324.
40. Porath, “ªEn ªAneva,” p. 324.
41. Porath, “ªEn ªAneva,” p. 53*, fig. 2:4 –5 (for their
context see table on p. 54*).
42. The cooking pot is illustrated in Porath, һEn
ªAneva,” p. 53*, fig. 2:4. For parallels see Rina
Avner, “Elat-Elot: An Early Islamic Village,”
ªAtiqot 36 (1998): 32*, fig. 13:8–10 (in Hebrew).
43. Porath, “ªEn ªAneva,” p. 53*, fig. 2:1–3, 6.
44. See Jodi Magness, “The Dating of the Black Ceramic
Bowl with a Depiction of the Torah Shrine
from Nabratein,” Levant 26 (1994): 199–206.
45. Porath, “ªEn ªAneva,” p. 54*.
46. Porath, “ªEn ªAneva,” p. 55*.
47. Porath, “ªEn ªAneva,” p. 55*. According to Porath,
it is not characteristic of Palestine.
48. Mordechai Haiman, “Agriculture and Nomad-State
Relations in the Negev Desert in the Byzantine and
Early Islamic Periods,” Bulletin of the American
Schools of Oriental Research 297 (1995): 35.
KHIRBET ABU SUWWANA AND EIN ªANEVA
18
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
some early Islamic villages and farmhouses in Palestine. They are attested, for example, in the
eighth to ninth century village excavated by Rina Avner at Eilat-Eilot and at other villages in the
Eilat region.49
The rooms were constructed according to recurring measurements, with external
dimensions ranging from about three to five meters.50
The dimensions reflect the maximum
length of the wooden beams that could be obtained for roofing. The floors are of beaten earth,
and the walls are constructed of stone or mudbrick on a stone socle. The rooms are arranged as
side-by-side units, sometimes with narrow rooms or courtyards between them. The buildings in
the Eilat area villages and at Ein ªAneva are aligned along the banks of wadis.
At Khirbet Abu Suwwana, we see a variation on this modular arrangement. The dwellings
consist again of side-by-side units, sometimes with rows of single rooms and sometimes two
rooms, one behind another. These units are clustered around a series of courtyards or common
open spaces. Here too the units are aligned according to the topography, running roughly northsouth
along the slope of the hill.51
A number of scholars have commented on the appearance of this type of modular architecture
in early Islamic Palestine. For example, Uzi Avner and Jodi Magness note that, “The degree
of uniformity among the buildings in the six villages [in the Eilat region] suggests a common ar-
49. See R. Avner, “Elat-Elot,” p. 33*; idem, “Eilat,”
Hadashot Arkheologiyot 103 (1995), and 105 (in
Hebrew). For the other villages in the Eilat region,
see Uzi Avner and Jodi Magness, “Early Islamic
Settlement in the Southern Negev,” Bulletin of the
American Schools of Oriental Research 310
(1998): 40. For an example near Sde Boqer, see
Rudolph Cohen, Archaeological Survey of Israel,
Map of Sede Boqer-East (168), 13– 03 (Jerusalem,
1981), p. 51, plan 2 (Nahal HaRoªah).
50. U. Avner and Magness, “Early Islamic Settlement,”
p. 40.
51. Sion, “Khirbet Abu Suwwana,” p. 186.
Figure 1.3. Pottery and stone bowl from Ein ªAneva (from Yosef Porath, “A Sixth-Seventh Century
C.E.(?) Structure Near ªEn ªAneva,” ªAtiqot 42 [2001]: fig. 2; reprinted with
permission of the Israel Antiquities Authority)
JODI MAGNESS
19
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
chitectural experience and social organization.”52
They have suggested that this uniformity reflects
the initiative of a single body that organized or sponsored the settlement of one or more
ethnic groups.53
Haiman points out that the Negev farmhouses also show this combination of
modular units. The farmhouses consist of one or more square buildings, each with one to three
rooms and a courtyard.54
Each of these units seems to reflect a dwelling structure for a family of
four or five individuals. Haiman believes that each farm was jointly owned by a group of families,
analogous to a Bedouin “paternal house.”55
Relatively small, square, or rectangular rooms have always been a characteristic component
of rural architecture in Palestine due to the limited availability of wooden beams for roofing.
However, the modular units of the early Islamic villages and farms are arranged in a different
manner from those of the Roman and Byzantine periods, suggesting a change in the organization
of familial units (and perhaps therefore in the origins of the inhabitants) and in the village structure.
Yizhar Hirschfeld establishes a typology of the private dwellings of Roman and Byzantine
Palestine based on their layout.56
He distinguishes three main types of houses based on differences
in ground plans:
1. The simple house: The most basic and common Roman/Byzantine dwelling type consists
of a one or two room house behind or in front of an open courtyard.57
2. The complex house: This is an expanded version of the simple house by means of wings
or dwelling units built around three or more sides of the outer courtyard.58
3. The courtyard house: This dwelling type has a central courtyard surrounded by rooms on
all four sides. According to Hirschfeld, this house-type was used exclusively by wealthy
families and is usually found in cities.59
The common denominator of all of these dwelling types is the arrangement of rooms around
or along one or more sides of an open courtyard. In villages such as Capernaum, Chorazin, and
Khirbet Sumaqa, the houses are grouped together in insulae separated by streets or alleys.60
Sometimes large houses can take up an entire insula, as at Mampsis.61
The early Islamic villages and farms that we have considered here follow the long-established
local tradition of having one or more rooms built along one or more sides of an open
courtyard. However, they also display significant differences. One of the most striking features
is the repetition of relatively small rooms, all roughly the same size in each dwelling and occur-
52. U. Avner and Magness, “Early Islamic Settlement,”
p. 40.
53. U. Avner and Magness, “Early Islamic Settlement,”
p. 40.
54. Haiman “Agriculture,” p. 41.
55. Haiman “Agriculture,” p. 41.
56. Yizhar Hirschfeld, The Palestinian Dwelling in the
Roman-Byzantine Period (Jerusalem, 1995).
57. Hirschfeld, Palestinian Dwelling, p. 21.
58. Hirschfeld, Palestinian Dwelling, p. 22.
59. Hirschfeld, Palestinian Dwelling, p. 22, 102.
Atrium or peristyle houses, a sub-type of the courtyard
house, are rare in Palestine; see ibid., pp. 57,
85–97.
60. For Capernaum, see Stanislao Loffreda, Recovering
Capharnaum (Jerusalem, 1985), pp. 8–9; for
Chorazim, see Ze’ev Yevin, The Synagogue at
Korazim: The 1962–1964 and 1980–1987 Excavations
(Israel Antiquities Authority Reports 10;
Jerusalem, 2000), p. 5, plan 2; for Sumaqa, see
Shimon Dar, Sumaqa: A Roman and Byzantine Jewish
Village on Mount Carmel, Israel (British
Archaeological Reports, International Series 815;
Oxford, 1999), p. 15, map 3; p. 36, fig. 20. The
streets or alleys are not always laid out orthogo-
nally.
61. See Arthur Segal, The Byzantine City of Shivta
(Esbeita), Negev Desert, Israel (British Archaeological
Reports, International Series 179; Oxford,
1983), p. 63, fig. 19; Hirschfeld, Palestinian Dwelling,
pp. 73–76.
KHIRBET ABU SUWWANA AND EIN ªANEVA
20
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ring throughout the settlement. There is no evidence for a second story level, which is a common
(though not universal) feature of all types of Palestinian Roman-Byzantine dwellings.62
Although
the number of units in each early Islamic house differs, the rooms do not vary greatly in
size and are not differentiated in other ways, such as by interior decoration. Each dwelling consists
of one or more side-by-side rooms (sometimes with two rooms, one behind the other).
These rooms include working areas with tabuns and other installations, which might have been
small open courtyards. The rooms are aligned along an open space that may have provided access
to more than one familial unit. The units can be contiguous or separated by some distances,
but there are no insulae. Hirschfeld has noted that in Roman-Byzantine villages in Palestine, the
poorer artisans and farmers typically occupied much smaller houses on the periphery of the
settlement.63
In contrast, there is no clear differentiation in dwelling size or elaboration within
these early Islamic settlements and no indication that poorer families were spatially
marginalized within the village.
Haiman’s suggestion that the introduction of this type of modular architecture can be understood
as reflecting the structure of a Bedouin paternal house is attractive. The smallest units of
the Bedouin tribal system consist of small individual families or clans that migrate and camp together.
Werner Caskel has observed that, “In this tribal organism there was no official leader, let
alone a hierarchy. A leader can acquire a position of any official character only by being appointed
to, or confirmed in, his office by a non-Bedouin power; otherwise he is only a primus
inter pares.”64
Caskel notes that this tribal organization and its ideological superstructure are
found not only among the Bedouins but also among settled Muslim populations.65
The modular
architecture seems to reflect the kind of social structure described by Caskel. Its appearance may
therefore provide evidence for the settlement of a new or different ethnic group (or groups).
Another interesting difference between the Byzantine and early Islamic villages that we
have considered is the placement of the religious (congregational) buildings. Generally speaking,
in Byzantine villages, the synagogues and churches (aside from monastic establishments)
tend to occupy a central position within the settlement. In contrast, at Khirbet Abu Suwwana the
mosque is at the edge of the village, and nearly all of the mosques in the Negev highlands are located
outside the settlement, often on a hilltop a few dozen meters away.66
Exceptions to this
rule include the mosques in settlements such as Khirbet Susiya, Eshtemoa (Samoªa), and Shivta,
where an earlier church or synagogue was converted into a mosque. Perhaps this too reflects a
different kind of social organization and structure within the village. Placing the congregational
religious structure in the center of the village requires the appropriation of communal property
and therefore some sort of organizing authority. The location of mosques may also represent a
different religious tradition. For example, if we assume that these villages were occupied by
families or clans structured like subunits of Bedouin tribes, which were accustomed to worship
at open-air shrines, it may have been important to place the religious congregational structure
next to or in an open space.
62. See Hirschfeld, Palestinian Dwelling, p. 102.
63. Hirschfeld, Palestinian Dwelling, p. 68.
64. Werner Caskel, “The Bedouinization of Arabia,” in
The Arabs and Arabia on the Eve of Islam, edited
by Frank E. Peters (Aldershot, 1999), p. 35.
65. Caskel, “Bedouinization,” p. 36.
66. See Gideon Avni, “Early Mosques in the Negev
Highlands: New Archaeological Evidence on Islamic
Penetration of Southern Palestine,” Bulletin
of the American Schools of Oriental Research 294
(1994): 83–100.
JODI MAGNESS
21
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
I believe that we see here at a micro-level (i.e., the level of the family and village) the same
sort of changes that have been noted by Hugh Kennedy at the macro-level (the level of the town
or madina).67
The classical cities of the Near East were characterized by an orthogonal layout
with broad colonnaded streets and large open agoras or forums surrounded by monumental public
buildings. In contrast, during the early Islamic period, city streets were frequently converted
into narrow, private, cul-de-sacs giving access to the houses on each side. The open spaces of the
agoras were replaced by suqs, with shops lining narrow streets or alleys. Large markets were
now located outside the gates, where livestock could be brought from the countryside for sale.68
Mosques not only replaced churches as religious buildings, but also agoras, theaters, and civic
basilicas in the functional sense of serving as public congregational structures.69
Kennedy attributes these changes to a number of factors. For example, the system of patronage
changed. Whereas in the classical cities the emperor and his local representatives sponsored
the construction of public monuments, the Muslim state authorities interfered less in the activities
of their subjects. Although the Umayyads constructed some large urban mosques and rural
palaces, they did not spend money beautifying the streets of cities or on public entertainment.70
In addition, whereas Roman law made a sharp distinction between state and private property, in
Islamic law the important unit was the family and its house.71
On the other hand, I disagree with
Kennedy’s assertion that “commerce and manufacture do not seem to have been the most important
factor in the prosperity of [classical] towns.”72
According to Kennedy, classical towns
derived their prosperity from the activities of neighboring landowners who chose to live in them.
In contrast, Islamic towns, with their narrow streets lined with small shops, “brought the focus of
commercial activity firmly within the city walls.”73
Kennedy believes that one consequence of
the change from open colonnaded streets to crowded suqs was to increase the number of retail
shops inside the city.74
In my opinion, the difference between the classical city and the Islamic city is in the location,
not the level of commercial activity. The agora or forum of a classical city functioned as its
commercial center. Vendors set up their stalls and sold their wares in this large, open paved
space, which was typically located in a central spot. Access was provided by broad streets,
which were also lined by shops. The broad streets were necessary to accommodate wheeled carts
and other wheeled traffic in which the goods were transported. Richard Bulliet suggests that
broad colonnaded streets disappeared in early Islamic cities because wheeled transport was replaced
by pack animals. Now camels, donkeys, and mules carried merchandise to small shops
lining narrow streets and alleys. Large animals and fresh produce were offered for sale in large
open areas outside the city walls.75
67. See Hugh Kennedy, “From Polis to Madina: Urban
Changes in Late Antique and Early Islamic Syria,”
Past and Present 106 (1985): 3–27.
68. Kennedy, Polis to Madina, pp. 12–13. However,
Kennedy notes that Muslim rulers adopted orthogonal
town planning when laying out new cities. The
difference is that whereas “in classical antiquity
most cities including the largest and wealthiest were
planned and ordered, in Islamic society they were
not” (p. 16).
69. Kennedy, “Polis to Madina,” pp. 15–16.
70. Kennedy, “Polis to Madina,” pp. 18–20.
71. Kennedy, “Polis to Madina,” p. 21.
72. Kennedy, “Polis to Madina,” p. 23.
73. Kennedy, “Polis to Madina,” p. 25.
74. Kennedy, “Polis to Madina,” p. 25.
75. See Richard W. Bulliet, The Camel and the Wheel
(Cambridge, 1975), especially pp. 224–28.
KHIRBET ABU SUWWANA AND EIN ªANEVA
22
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
Khirbet Abu Suwwana, Ein ªAneva, and other early Islamic villages and farms may provide
evidence at the micro-level for the changes described by Kennedy. The Roman and Byzantine
villages of Palestine had many of the same features of the larger towns and cities but on a more
modest scale. These features include an orthogonal layout (sometimes), differentiation in the
size and ornamentation of private dwellings (reflecting the existence of local elites), and the
placement of market areas and public congregational buildings in the center of the settlement
(due to the patronage or organizational impetus of local elites). In contrast, the early Islamic villages
and farms we have considered here are characterized by modular architecture. Within each
settlement, there is little differentiation in size and ornamentation. This organization seems to reflect
a village structure oriented around the familial unit instead of powerful individuals or
patrons belonging to local elites. The public congregational building, the mosque, was now located
on the periphery of the village or outside it. The open spaces around the mosque might also
have been the site of commerce and other public activities, like the markets outside the walls of
cities where large animals and produce were sold.
Not all of the early Islamic settlements in Palestine — whether established de novo after the
Muslim conquest or occupied continuously from the Roman/Byzantine period — have modular
architecture or show no evidence for social differentiation. For example, an early Islamic house
at Khirbet Susiya has two symmetrical wings with dwelling rooms opening onto a courtyard on
one side and two small shops opening onto an alley on the other side. Hirschfeld notes parallels
for this layout at Roman and Byzantine sites in Palestine.76
In the early Islamic village at
Capernaum there are discernible differences in the sizes of the houses and the sizes of the rooms
within the houses.77
The buildings in the village, many of which had a second story, lined broad
lanes.78
A large public building of unknown function appears to be located inside the village.79
During the second phase of early Islamic occupation, many of the rooms inside the houses were
subdivided or new rooms were added, and some of the streets were blocked.80
A modest dwelling
built in Tiberias during the ªAbbasid period consists of two small, square side-by-side rooms
opening onto a courtyard. Its ground plan looks modular, but unlike the examples from the village
and farms we have considered, this house had a second story level with living quarters.81
A
large urban dwelling of the Umayyad period at Pella contained two peristyle courtyards, attesting
to the continuity of this Roman/Byzantine building tradition.82
The presence of six
residential units described as being of “relatively high quality” in Area D at Khirbet Abu
76. See Hirschfeld, Palestinian Dwelling, pp. 37–38;
idem, “Excavation of a Jewish Dwelling at Khirbet
Susiya,” Eretz-Israel 17 (1984): 168–80. For parallels
at Beth Shearim and Gerasa, see p. 177 nn. 11–
12. For the suggestion that this house was established
in the early Islamic period and not during the
sixth century, as Hirschfeld claims, see Jodi
Magness, The Archaeology of the Early Islamic
Settlement in Palestine (Winona Lake, 2003), pp.
101– 02.
77. See Vassilios Tzaferis, Excavations at Capernaum,
Volume 1: 1978–1982 (Winona Lake, 1989), pp.
5–6 (Area C; perhaps a public building), 10–19
(Area A). According to Jodi Magness, “The Chronology
of Capernaum in the Early Islamic Period,”
Journal of the American Oriental Society 117
(1997): 481–86, Stratum V should be dated ca.
700–750, and Stratum IV should be dated ca. 750 to
the second half of the ninth century.
78. Tzaferis, Capernaum, pp. 15–16.
79. Tzaferis, Capernaum, p. 3 (Area B).
80. Tzaferis, Capernaum, pp. 16–19 (Area A).
81. See Hirschfeld, Palestinian Dwelling, pp. 40–42
and fig. 19; idem, “Tiberias,” Excavations and Surveys
in Israel 9 (1989/1990): 107–09.
82. See Anthony McNicoll, Robert H. Smith, and Basil
Hennessy, Pella in Jordan, Volume 1: An Interim
Report on the Joint University of Sydney and the
College of Wooster Excavations at Pella 1979–
1981 (Canberra, 1982), pp. 123–26; Hirschfeld,
Palestinian Dwelling, pp. 47–49.
JODI MAGNESS
23
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
Suwwana83
suggests that there might have been some degree of social differentiation even in
settlements with modular architecture. In other words, differences in the size, layout, and elaboration
of houses, presumably reflecting social differentiation and the existence of elites within
towns and villages, are discernible in many settlements in early Islamic Palestine.
The newly-established farms and villages with modular architecture — such as Khirbet Abu
Suwwana and Ein ªAneva — tend to be located on Palestine’s desert periphery. The appearance
of modular architecture in these farms and villages presumably reflects the settlement of a new
population with a distinctive social organization and needs. As Hirschfeld notes, the manner in
which private dwellings are designed regulates and even controls the lives of their inhabitants.84
Similarly, at the macro-level Kennedy observes that city planning “was adapted for different
purposes, life-styles, and legal customs. The changing aspect of the city was determined by
long-term social, economic and cultural forces.”85
As we have seen, villages and farms with
modular architecture may best be understood as reflecting a familial and village organization
along tribal lines. Haiman argues that the establishment of new settlements in the desert periphery
of Palestine during the Umayyad period is due to the deliberate, state-sponsored
sedentarization of nomadic tribes.86
Although I am not convinced that this process is the result of
state sponsorship, Haiman is correct that the peripheral desert regions were settled only when
there was a strong and stable government authority.87
In this paper, I have attempted to shift scholarly attention to small settlements in early Islamic
Palestine’s peripheral regions by focusing on the appearance of modular architecture. This
complex phenomenon, which archaeology is uniquely positioned to illuminate, deserves more
systematic treatment. For example, it would be instructive to document all known examples of
settlements with modular architecture and examine their pattern of distribution relative to other
settlements. The social structures of villages and small towns versus large towns and cities in the
Roman/Byzantine and early Islamic periods need to be studied. Factors other than social and political
structures should also be considered in relation to the appearance of modular architecture.
Can this architecture be best understood in light of the settlement of a new population organized
along tribal lines? Hopefully, future studies and new discoveries will provide us with additional
information to help understand this phenomenon.
83. See Sion, “Khirbet Abu Suwwana.”
84. Hirschfeld, Palestinian Dwelling, p. 15.
85. Kennedy, “Polis to Madina,” p. 17.
KHIRBET ABU SUWWANA AND EIN ªANEVA
86. Haiman, “Agriculture,” pp. 46 –47.
87. Haiman, “Agriculture,” pp. 46–47.
24
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
25
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
2
ASCALON ON THE LEVANTINE COAST
TRACY HOFFMAN
The issue of changing social identities in the Early Islamic period can be investigated on
numerous levels; in the case of the medieval port city of Ascalon the question is best approached
through a study of the city’s urban development during the Byzantine and Early Islamic period.1
This approach is rendered difficult by the nature of the archaeological record. Initially investigated
by visitors such as the Comte de Forbin and Lady Hester Stanhope in the early nineteenth
century, Ascalon was first excavated by John Garstang and his assistant W. Phythian Adams in
the early twentieth century.2
Their work uncovered evidence of architecture dating to the Islamic
period and earlier scattered throughout the site. More recent excavations under the direction of
Lawrence Stager have produced significant medieval period material culture ranging from architecture
to ceramics.3
The location and nature of this material makes the site of Ascalon so
challenging and promising for the study of Islamic period cities and their inhabitants (fig. 2.1).
Traditionally, archaeologists have approached the study of Islamic period cities through an
analysis of their institutions or vernacular architecture. Some practical reasons for this approach
can be explained by the quality of preservation and the types of artifacts generally identified, and
subsequently regarded, as providing the most direct evidence for change or development in a
city. The construction of a mosque, for instance, in a city where none previously existed, reveals
information not only about inhabitants of that community but also about the development of the
given site. It also represents the type of visible change that draws scholarly attention. Indeed,
modification on any level of the city center, generally the public heart of a city, often serves as
the starting point for interpreting change generally, and in the framework of this study, transitions
from the Byzantine to Early Islamic period more specifically. The lure of urban institutions
is so powerful that it is seemingly impossible to interpret a city in their absence.4
At a site such
as Ascalon, where the limited evidence for city institutions stems largely from the identification
of where they are not, it is necessary to develop a new methodology for interpreting the city.
At first glance, interpreting the city of Ascalon seems straightforward; the transition from
the Byzantine to Early Islamic period and beyond appears to be well documented in limited but
nonetheless informative references that illuminate important aspects of the city. Some of the
most intriguing and as yet underutilized sources are Byzantine and Early Islamic period mosaic
depictions of Ascalon, such as the Madaba Map and Umm al-Rasas, which show the cardo,
25
1. See Tracy Hoffman, “Ascalon ªArus al-Sham: Domestic
Architecture and the Development of a Byzantine-Islamic
City” (Ph.D. diss., University of
Chicago, 2003).
2. For the publication of preliminary results, see J.
Garstang, “The Fund’s Excavation of Askalon,”
Palestine Exploration Quarterly 21 (1921): 12–16;
idem, “The Excavation of Askalon, 1920–1921,”
Palestine Exploration Quarterly 21 (1921): 73–75;
idem, “The Excavations at Askalon,” Palestine Exploration
Quarterly 22 (1922): 112–19; idem,
“Askalon,” Palestine Exploration Quarterly 24
(1924): 24–35.
3. Lawrence Stager, Ashkelon Discovered: From
Canaanites and Philistines to Romans and Moslems
(Washington, D.C., 1991).
4. The literature on the subject of Islamic cities and
their institutions is extensive and devoid of any consensus
on matters of definition. A good synopsis of
some of the main issues can be found in Timothy
Insoll, The Archaeology of Islam (Oxford, 1999).
26
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
TRACY HOFFMAN
Figure 2.1. Map of Palestine (after Denys Pringle, “King Richard I and the Walls of
Ascalon,” Palestine Exploration Quarterly 116 [1984]: pl. 1)
decumanus, a city gate, multistory buildings, and the main church in the late sixth century (fig.
2.2).5
In these mosaics the public face, the monumental vernacular architecture, of a classical
city is easily recognizable. They also show aspects of the city infrastructure such as streets and
fortifications that in addition to the various buildings depicted may be identifiable in the archaeological
record.
5. Herbert Donner, The Mosaic Map of Madaba: An
Introductory Guide (Kampen, 1992); Michèle
Piccirillo, and Eugenio Alliata, eds., The Madaba
Map Centenary, 1897–1997: Travelling through the
Byzantine Umayyad Period (Proceedings of the International
Conference Held in Amman, 7–9 April
1997; Jerusalem, 1999). A third mosaic depicts
what is identified as the central church of Ascalon,
refer to Roland de Vaux, “Une mosaïque byzantine
à Ma’in (Transjordanie),” Revue Biblique 47
(1938): 227–58.
27
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
6. For a summary of the medieval sources that describe
Ascalon, see Moshe Sharon, Egyptian Caliph
and English Baron: The Story of an Arabic Inscription
from Ashkelon (Corpus Inscriptionum
Arabicarum Palaestinae; Jerusalem, 1994).
7. Muh≥ammad ibn Ahmed Muqaddasi, Ahsan altaqΩsÏm
fi maªrifat al-aqΩlÏm, edited by M. J. de
Geoje (Bibliotheca Geographorum Arabicorum 3;
Leiden, 1906), p. 174.
8. Nasir-i Khusraw, Nasir-i Khusraw’s Book of Travels,
edited and translated by Wheeler M. Thackston
(Costa Mesa, 2001); UsΩmah ibn-Munqidh, An
Arab-Syrian Gentleman and Warrior in the Period
of the Crusades: Memoirs of UsΩmah ibn-Munqidh,
translated by Philip K. Hitti (Records of Civilization,
Sources and Studies 10; New York, 1929), p.
40; William of Tyre, A History of Deeds Done Beyond
the Sea, translated by Emily Atwater Babcock
and August Charles Krey (Records of Civilization,
Sources and Studies 35; New York, 1943).
9. For the history of Ascalon in the medieval period,
refer to Moshe Sharon, Corpus Inscriptionum
Arabicarum Palaestinae, Volume 1: A (Leiden,
1997); see also Pringle, “King Richard I and the
Walls of Ascalon,” pp. 133– 47; J. Prawer,
“Ascalon and the Ascalon Strip in Crusader Politics,”
Eretz Israel 4 (1956): 231–48 (Hebrew with
English summary), and “The City and Duchy of
Ascalon in the Crusader Period,” Eretz Israel 5
(1958): 224 –37.
Descriptions found in texts that constitute the written record on medieval Ascalon also focus
on the city’s public institutions.6
Dating from the tenth century and later these texts describe institutions
that differ from those depicted in the mosaics, referring instead to the city mosque and
suqs rather than Ascalon churches. One of the most well-known descriptions is Muqaddasi’s
commentary in the tenth century on Ascalon’s mosque, thriving markets, strong fortifications,
and unusable harbor.7
A century later Nasir-i Khusraw mentions the city’s multistory buildings.8
Sporadic details about Ascalon in other texts include a brief mention of the city mosque by
Usamah ibn-Munqidh in the twelfth century. Perhaps the most vivid description is William of
Tyre’s recording of the Crusader siege of Ascalon in 1153 C.E., which offers the most detailed
description of the city’s fortifications and the surrounding area. In general, the sources describe a
thriving city with the expected markers, suqs, and mosques of an Islamic city. In many ways the
information is the same as that found in the mosaic depictions; city institutions are enumerated,
even as much of the city remains unilluminated.
These sources add few details to the history of Ascalon as a city that is known to have surrendered
to Mu’awiya on terms (sulhan) in the mid-seventh century.9
Situated on the southern
Levantine coast, a border with the Byzantine controlled Mediterranean, Ascalon, along with
other port cities, was the focus of efforts to settle new inhabitants in and around the city to bolster
its defenses and that of the expanding Islamic empire. In due course sources record the
construction of a mint and a mosque in Ascalon although it is not until the descriptions outlined
above that any real picture of the city emerges. Passing from Umayyad to ªAbbasid and then
Tulunid rule Ascalon’s ties with Egypt were cemented under the Fatimids. William of Tyre’s description
indicates the importance Ascalon held for both the Fatimids and Crusaders during the
twelfth century from which time the city traded hands repeatedly until it was ultimately abandoned
in the mid-thirteenth century. Documentary sources on Ascalon illuminate, in however
limited a way, the most useful aspects of the city for tracing change and continuity, but the
sources do not reveal everything. In fact, the only way to assess the development of the city is
through the archaeological record and what it might reveal about the events, institutions, and inhabitants
as seen in the documentary evidence.
Since 1985 the Leon Levy Ashkelon Expedition has conducted the first large-scale systematic
excavation of the ancient city housed within the remains of the medieval city wall (fig. 2.3).
In an effort to best target the time periods of interest, the Canaanite and Philistine periods, the
expedition has explored and/or excavated the three mounds, designated North, Middle, and
South, as well as the ground level areas that constitute the site of ancient Ascalon. In the process,
ASCALON ON THE LEVANTINE COAST
28
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
a large corpus of Byzantine and Islamic period material culture was collected, establishing a
scattered but viable database for analyzing the city. Conspicuously absent from the archaeological
record, however, is any evidence for city institutions; the very images that pervade the
documentary evidence are absent from a material record that is largely non-public in nature. The
archaeological record can in any case be used to trace the development of the city and how that
process might have affected the city’s inhabitants.
The material record of Ascalon from the Byzantine and Early Islamic periods consists
chiefly of five buildings identified as domestic structures, a church, and a kiln housed in the converted
remains of an earlier bath house. At least two more structures, another bath and a second
kiln, originally constructed in the Roman period continued to be used into the Byzantine period.
An examination of these structures reveals a remarkable degree of continuity as well as visible
change in the housing, organization, and plan of the city. Although non-public in nature the archaeological
evidence for Ascalon provides an opportunity not only to examine this aspect of the
city but also to explore its potential in dealing with the larger issues of urban development and
changing social identities.
The excavated domestic structures cluster on the three mounds with the notable exception of
one building that is located just under the city wall. Starting on the North mound, Structure One
is situated near the city gate and in proximity to two other buildings, a church and a bath (fig.
TRACY HOFFMAN
Figure 2.2. Ascalon in the Madaba map (after Donner, The Mosaic Map of Madaba)
29
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ASCALON ON THE LEVANTINE COAST
Figure 2.3. Excavation areas in Ascalon
30
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
2.4). While the latter two buildings predate the house in their construction, the evidence is inconclusive
as to whether or not the three buildings might have had some overlapping periods of
use. As a building, a number of characteristics or features in Structure One appear to be common
in houses throughout the city. The plan consists of a central courtyard around which three rooms
can be identified. The standard building material for both interior and exterior walls is stone with
both uncut fieldstones and cut ashlar blocks present. The flooring materials range from beaten
earth and plaster surfaces in interior rooms to stone paving in the courtyard. A prominent feature
in this house is a drainage system that includes three intakes and a main channel which starts in
the courtyard then branches into two channels into two of the interior rooms.
One may note that at least part of the structure was built prior to the Byzantine and Early Islamic
periods, a conclusion that is based on stratigraphy and the analysis of ceramics collected
during the excavation of certain walls. The existing segment of the building continued to be used
during subsequent periods of renovation and occupation. Ceramics collected from within the
house and analysis of the architectural remains indicate this structure continued to be used into
the Middle Islamic period and in all probability through the final occupation of the city in the
mid-thirteenth century. The longevity of the use of this structure is noteworthy as is the realization
that the maintenance and any renovation of the building did not result in significant
organizational or functional changes.
TRACY HOFFMAN
Figure 2.4. Structure One. Ascalon
31
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
As an independent building Structure One demonstrates the types of building materials and
features as well as the type of building layout that might be found in the houses of Ascalon. It
further indicates that houses might be located near non-domestic structures such as baths and
churches, and that prolonged periods of use were common for some of the buildings in Ascalon.
In this area of the city then, the North mound, the nature of the architectural evidence illuminates
not only the nature of the city’s housing but also the organization of Ascalon’s neighborhoods.
The evidence from Structure Two, located south of Structure One on the North mound, is
more elusive but deserves some mention because of its location (fig. 2.5). The remains of this
structure, also identified as a house, are fragmentary and consist chiefly of a number of stone
walls that form at least two rooms. Five meters away from the main structure is a plaster floor in
which are two stone-lined fire pits. The stratigraphic evidence indicates that the main building
and this surface were probably used contemporaneously, but it is impossible to determine
whether or not it is all part of one building.
Again, features first seen in Structure One such as stone walls and beaten earth surfaces are
visible in this house. One distinctive aspect of this second house is that it does not have as long a
period of use as Structure One. Although the fragmentary nature of the architecture and material
culture collected from Structure Two makes it difficult to provide a firm chronology, it seems
clear that this house was used only in the Byzantine and Early Islamic periods. Unlike many
other houses in Ascalon, no evidence indicates that it was occupied into the Middle Islamic period.
Any assessment of this house must acknowledge that it does not offer much new material
in terms of the physical structure of the building. The paucity of the building’s remains does not
detract from the significance of its location, and its existence more generally, which serves to
demonstrate the extent of settlement within the city. Together, Structures One and Two indicate
that much of the North mound was settled and that much of that occupation was residential in
nature.
Farther to the south the Middle mound has a third building (Structure Three), one of the
largest and most complex structures dated to the Byzantine and Early Islamic periods (fig. 2.6).
Structure Three reflects many of the patterns identified in the previously discussed houses while
also showing some significant differences. The building materials are consistent with other
buildings: stone walls and beaten earth, plaster, or stone paved surfaces, although some new features
occur such as mosaic floors. This house has an extensive drainage system, which includes a
well, with channels running through both interior and exterior spaces. The layout seems to consist
of rooms around a courtyard although in this case the evidence argues for the existence of
several courtyards and suites of rooms, indicating that this house might have functioned as a
multiunit building. Indeed, the maze of rooms makes little sense as a single structure, but when
viewed as a multiunit house, a form known to have existed in Ascalon, the architecture becomes
more intelligible (see below).
Structure Three provides the most complete evidence for the reuse of a building constructed
in the Byzantine period that underwent subsequent renovation and continued in use into the
Middle Islamic period. The layout of the building changes substantially; part of the building is
abandoned and the remainder of the structure reduced to one living unit. However, little evidence
suggests that the modifications in the layout resulted in any changes in the overall
function of the building. Structure Three appears to have been constructed as a house and to
have been used as such while it was occupied. Interestingly, evidence for a second structure just
to the north of Structure Three hints at the extent of settlement in this area of the city. The façade
of this second building, less than one meter away, is pierced by a number of windows and appears
to have an east–west dimension similar to that of Structure Three. Although the two
ASCALON ON THE LEVANTINE COAST
32
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
TRACY HOFFMAN
Figure 2.5. Structure Two. Ascalon
33
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ASCALON ON THE LEVANTINE COAST
Figure 2.6. Structure Three, Phase 3. Ascalon
34
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
buildings are in close proximity, their constructions and placements are neither haphazard nor
random; in this area of Ascalon the impression is one of deliberate planning rather than indiscriminate
utilization of usable space.
Structure Four is located on the northern edge of the South mound (fig. 2.7). This building
is clearly recognizable as a house with suites of rooms organized around a central courtyard.
Structure Four displays many of the features found in other excavated houses. The building material
is stone, and floors range from beaten earth or plaster to stone pavement and mosaic
surfaces. An elaborate drainage system can also be associated with a small pool located in the
courtyard. As in the case of Structures One and Two, evidence shows that part of Structure Four
was built prior to the Islamic period. Analysis of the ceramics collected from the building’s
walls indicates that all the exterior walls date to the Byzantine period while the interior walls can
be assigned an Early Islamic date. Some Byzantine period interior walls continued to be used in
the Islamic period, indicating at least part of the existing organization of the building was maintained.
This pattern of renovation and reuse is noteworthy as it occurs in other houses at
Ascalon.
The longevity of this house is surprising since a number of new structures were built in its
immediate vicinity. One of these new buildings with a courtyard and a pool of similar construction
utilizes one of Structure Four’s exterior walls. An additional building to the east shows that
Structure Four was probably part of a densely settled neighborhood. The impression of this area
is not the same one of deliberateness or orderliness as was seen in the area of Structure Three.
Rather, the addition of new buildings, using in some cases walls of the existing structure, gives
more of an impression of unregulated growth. Farther to the south, even greater change in this
neighborhood is indicated as a former bath house, used into the Byzantine period, was converted
into a kiln facility in the Early Islamic period (fig. 2.8). The continuity visible in Structure Four
stands in striking contrast to the changes and developments in the buildings around this central
structure.
If Structure Four is located in an area showing the densest occupation in Ascalon, then the
last house, Structure Five, is in the most isolated area (fig. 2.9). Situated in the southwest corner
of the city, just below the mound of the city’s fortifications, the fragmentary remains of this
house are located near the presumed area of the city’s sea gate. The building consists of a series
of walls that bound a courtyard with a well and an elaborate series of drains. In general this
building displays the same building materials and the same features found in previously discussed
houses. The ceramic evidence indicates a relatively short occupation and abandonment
during the Early Islamic period. Structure Five is valuable as evidence for the extent and nature
of settlement within the city walls, offering tangible proof that residential architecture, in at least
some areas, extended up to the city walls and was not limited to the three mounds within the
city. In addition, Structure Five is not the only building built near Ascalon’s fortified city wall.
Just south of the Jerusalem Gate is a Byzantine period church that continued to be used into the
Middle Islamic period (fig. 2.10). Again, some of the greatest significance of this building lies
in its location and what it reveals about the city plan.
This brief overview not only highlights some very important aspects of the housing at
Ascalon but also illustrates development of the city plan. Basic patterns are repeated in each
structure, such as similar building materials, features, and layouts, suggesting reuse without systematic
reorganization during renovation. As many of the houses were built in part or in whole
prior to the Islamic period, similarities and continuity in those aspects of the structures are obvious.
Indeed, the longevity of the excavated houses indicates that in non-public areas of the city
the status quo was maintained to a surprising degree. Partly, the answer lies in expediency; it
TRACY HOFFMAN
35
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ASCALON ON THE LEVANTINE COAST
Figure 2.7. Structure Four. Ascalon
36
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
TRACY HOFFMAN
Figure 2.8. Grid 38 bath house. Ascalon
was easy to occupy standing buildings in good repair and the houses were a recognizable form,
something in which the inhabitants of Ascalon — both old and new — were comfortable.
It would be misleading, however, to characterize the archaeological record as devoid of
change, which is indicated both in the functions of buildings, such as the bath converted into a
kiln, and in settlement, such as the construction of new buildings in the area of Structure Four.
The significant drawback in understanding these changes lies in the nature of the archaeological
record because they are so few and dispersed throughout the city. To make sense of the changes
and to utilize the archaeological evidence to its full potential, additional context is needed so that
these structures can be situated within the cityscape and made more meaningful. The situation of
the structures in the cityscape is particularly important in trying to assess the Byzantine and
Early Islamic period transition and in trying to understand what the city reflects about its inhab-
itants.
That much needed context is found in a little-known legal text, On the Laws or Customs of
Palestine, which follows in a long tradition of texts written to govern both building practices and
37
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ASCALON ON THE LEVANTINE COAST
Figure 2.9. Structure Five. Ascalon
relations between neighbors stemming from such issues.10
This volume is particularly useful in
that it was written specifically about Ascalon. It is generally believed that the author of this
code, an architect by the name of Julian, was born in the city of Ascalon and that he was still living
there when he compiled the legal code during the mid-sixth century.11
It is not known
whether the compilation of the legal code was an officially sanctioned project or if it was an un-
10. Catherine Saliou, “Le Traité de Droit Urbain de
Julien d’Ascalon: Coutumier et Codification,” in La
codification des lois dans l ’antiquité (Actes du
colloque de Strasbourg 27–29 Novembre, 1997),
edited by Edmond Lévy (Travaux du centre de recherche
sur le Proche-Orient et la Grèce antiques
16; Paris, 2000), pp. 293–313.
11. The origins of his family, whether they were pagan
or Christian, are unclear. About Julian specifically,
many scholars believe he had at least some classical
education based on the fact that his text is written in
Greek and that Julian makes reference to several
classical authors. For the most complete discussion
of Julian, his identity, and the dates of his text, see
Joseph Geiger, “Julian of Ascalon,” Journal of Hellenic
Studies 112 (1992): 31–43. For a synopsis of
Julian, which also discusses his origins, see Besim
Hakim, “Julian of Ascalon’s Treatise of Construction
and Design Rules from Sixth Century
Palestine,” Journal of the Society of Architectural
Historians 60 (2001): 4–25.
38
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
official endeavor on the part of a private citizen.12
The motivation behind the writing of the treatise
is equally unclear although scholars agree one reason was Julian’s desire to ensure Ascalon
remained a pleasant place to live. This condition was achieved by minimizing the dangers and or
nuisances that certain activities or neighbors might cause each other and by preserving aesthetically
pleasing aspects of the city, such as views, public art, and gardens.
Absent from Julian’s text is any real reference to the city center or the public institutions of
the city, which suggests another possible explanation for his legal code. While construction and
12. On the subject of legal codes in this period and the
various schools of thought that might have influenced
Julian, see Hakim, “Julian of Ascalon,” pp.
6– 8.
TRACY HOFFMAN
Figure 2.10. Church near Jerusalem gate. Ascalon
39
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ASCALON ON THE LEVANTINE COAST
renovation in public areas of the city remained well regulated, perhaps the same was not true for
such practices in non-public areas of the city. In simple terms, perhaps the text was necessary,
perhaps it met a need generated by the inhabitants of Ascalon who were neglecting traditional
principles of design and construction in the classical city, thereby producing an unpleasant urban
environment. Perhaps the pace and/or nature of change within the built environment, even if
well controlled, was significant enough for a need to reiterate regulations on a more intimate
level than the legal code issued in Constantinople. Julian never states why he compiled the legal
code, but it is nonetheless significant that in the mid-sixth century such a code was written for
Ascalon. Equally significant, its inclusion in the mid-fourteenth century Hexabiblos of Constantine
Harmenopoulos indicates that the laws and stipulations recorded in the text might have
continued to be relevant to urban planning in the Byzantine empire for centuries.13
In the study of Byzantine and Early Islamic Ascalon, Julian’s text is best used not as a legal
code per se but rather as a source of textual data, along the lines of the recovered material record,
for interpreting the city of Ascalon. The most basic level of analysis for which the text can be
used is to build a corpus of the city’s architecture. The process of building a corpus of the city’s
architecture involves analysis of particular features, such as windows, doors, terraces, and sewage
systems associated with those buildings. In discussing the features of various buildings it is
possible to associate certain activities with many of the buildings because many of the structures
are identified by their functions. As a corpus of material these buildings cease to be independent
artifacts and become, rather, key features in the context of the built environment. These descriptions
can be used for a second level of analysis that identifies patterns which not only illuminate
aspects of the city plan and the continuing development of the city, but also lend greater meaning
to the lists of buildings by providing them a context.
Julian of Ascalon organized his treatise according to the four elements, earth, wind, fire, and
water, which he regarded as the source of all conflict between people. To make it more meaningful
for archaeological research and the investigation of Ascalon, the text can be organized into
buildings and/or activities that fall into the following categories: daily life, industry, service,
commerce, residential, and infrastructure. Within each category buildings can be discussed in
terms of characteristic features, location (within the city and in proximity to other structures),
and function. The category of daily life, for instance, includes private bath houses and bakeries.
In both cases Julian discusses hearths and/or fires as features of such buildings. The existence of
hearths in these structures, or any nonresidential building for that matter, plays a large role in determining
their location as does their proximity to neighboring buildings. The nature of nearby
buildings, their height, placement of windows and doors, as well as the prevailing wind, further
influence where a bath or bakery might be built.
The category of industry includes ceramic, gypsum and lime kilns, dyers, glassworks or factories,
oil makers, rope makers, launderers, and finally garum and cheese makers. Again, as in
the case of the bakeries and bath houses, these buildings are discussed in terms of their characteristic
features and their locations within the city. Brothels, taverns, stables, and animal pens are
included under the service heading. With the exception of brothels these buildings or activities
are all allowed inside the city, although they are subject to rules that are designed, as in the case
of industry, to minimize the nuisance posed to neighbors. Taverns, for example, are not allowed
13. Konstantinos Harmenopoulos, Manuale legum sive
hexabiblos: Cum appendicibus et legibus agrariis,
edited by Gustav Ernst Heimbach (Aalen, 1969).
40
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
TRACY HOFFMAN
to have benches outside their doors for fear that patrons might want to be served outside, thereby
posing a nuisance to passersby. Julian offers less information about the features of these types of
buildings, but their inclusion in his text further illustrates the variety of buildings to be found in
Ascalon as well as their locations in all areas of the city.
The next category, commerce, includes warehouses and shops that are discussed not as aspects
of formal institutions but rather as important features of various city neighborhoods.
Warehouses, according to Julian, might be stand-alone structures or part of larger structures possibly
wholly commercial in function or mixed-use buildings that combine commercial and
residential space. Shops, often found along portico-lined streets, could also be found in mixeduse
buildings with shops on the first floor and one or two upper floors of residences. In
discussing shops, Julian addresses the issue of private shops extending their business into the
porticoed walkways in front of them. Julian allows private shops to extend outward with the caveat
that such activities must take place directly in front of a specific shop and may not extend
sideways into the area of other shops.
The next category, residential structure, is of particular importance for interpreting Ascalon
since it is the type of building for which we have the most direct archaeological evidence. The
houses of sixth century Ascalon as described by Julian might be single story family homes or
multistory, up to three or four stories high, condominium buildings. Residences could also be
found in mixed-use buildings where in some cases residences and warehouses might even share
a common courtyard. Julian mentions some of the common features of such buildings including
courtyards, balconies, roof terraces, gutters, cisterns, and latrines in addition to the obvious windows
and doors. Many of these features are identifiable in the archaeological record and
information about those that are not is extremely valuable for better reconstructing entire struc-
tures.
The final category of infrastructure includes streets, porticos, public art, public gardens, and
water and waste drainage and storage systems. Many of these aspects of the city are mentioned
only within the context of Julian’s discussions of specific buildings. It is nonetheless clear that
they were important aspects of city life and that features such as public art and gardens were regarded
as key mechanisms for ensuring that the city remained a pleasant community in which to
reside. All together then the various categories of buildings, infrastructure, and activities found
in Ascalon and presented in Julian’s text provide a fairly detailed picture of sixth century
Ascalon. Importantly, it is a picture that reflects the very aspects of the city also revealed in the
archaeological record. The text complements the material culture record and provides a tool for
not only reconstructing Ascalon in the sixth century but also for tracing how the city continued
to develop into the Islamic period.
A starting point may be to take the buildings and infrastructure outlined in the categories
just explored and situate them within the context of the city. This data can then be examined in
light of the archaeological data and combined to create a more comprehensive context for the
houses and their situation within the city. According to Julian’s text, bath houses could be located
anywhere within the city, including residential areas, so too could bakeries. Julian’s text
allows two important conclusions to be drawn. First, residential areas did not consist only of
houses but also incorporated bakeries, baths, and other activities housed in stand-alone or multipurpose
buildings. Secondly, important aspects of daily life such as food and hygiene were not
limited to one area within the city but rather could be found in dispersed locations. Residents did
not need to go to one specific area of the city to visit a bath or to buy some bread but could expect
to find such services in their own neighborhoods.
41
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ASCALON ON THE LEVANTINE COAST
As mentioned, the activities categorized in this study as industry are varied and include pottery
kilns, gypsum/plaster kilns, lime kilns, dyers, glassworks and glass factories, oil makers,
rope makers, fullers, and finally garum and cheese makers. Many of these industries were regarded
as so dangerous or as presenting such a significant nuisance that they were relegated to
spaces outside of the city. In some cases they were allowed within cities only if absolutely necessary
and if such buildings were situated in the most isolated areas away from other buildings
as much as possible. This information suggests that in the Byzantine period industry was, as
much as possible, situated within more peripheral areas of the site where settlement might not
have been as dense, thereby allowing industry to operate with minimum impact on neighbors.
Julian’s indication that activities traditionally situated outside the city could be moved inside if
necessary might suggest that it was actually occurring, that buildings housing dangerous activities
were being built in the city and that Julian was trying to regulate that process.
The rather general category of service includes brothels, taverns, and stables. Julian’s writing
clearly suggests the latter two activities or functions were to be found within the city,
particularly along streets. In the case of brothels it is not clear whether such buildings could be
found in Ascalon although archaeological evidence suggests that at one time Ascalon was home
to at least one brothel.14
The possibility remains that, while brothels were supposed to be located
only in villages or rural areas, they might have been operating in towns. Julian’s stipulations indicate
certain activities should not be located in a town or that they should be in a particular part
of town. To know how closely and for how long the regulations outlined in Julian’s treatise were
followed is not known.
In writing about commercial activities Julian indicates that warehouses could be independent
buildings or space within a multipurpose building that might combine residential and other
commercial spaces. The text leaves open the question of whether or not Ascalon might have had
a warehouse district or specific commercial district. Shops figure prominently in discussions not
only of streets, which they often lined, but also as ground floor occupants in mixed-use buildings.
Julian’s discussion of shops and warehouses indicates, much in the case of bakeries and
baths, that they could be found in residential and nonresidential areas of the city. Warehouses
and shops could be along major thoroughfares or smaller streets. Of particular note are the implications
for how changes in the use of certain space is becoming more noticeable and, perhaps,
more problematic. In his writing about shops and the use of space in the porticoed walkways in
front of specific shops, Julian seems to indicate that encroachment or at the very least dealing
with the relationship between private and public space was a growing concern in the mid-sixth
century. It is equally clear that if this was a problem it was not limited to one area of the city but
rather could be found anywhere that shops or other commercial establishments might be located.
Residential architecture and spaces figure prominently in Julian’s treatise and a great deal
can be learned about how they fit into the city based on his text. From Julian’s discussion of
various features of single family and multiunit buildings it is clear that multistory buildings and
single story buildings, not to mention single family homes and condominium buildings, existed
side by side, creating neighborhoods with diverse architecture. Houses could be located in
densely settled neighborhoods or in areas where unused lots might also be found. Writing about
city infrastructure, Julian records information about streets, public gardens and plantings, public
14. Excavation uncovered a bath house datable to the
fourth century. Adjacent to that building was a large
sewer in which the remains of more than one hundred
infants were recovered. This discovery led
archaeologists to conclude that the bath house may
have been associated with a brothel and that unwanted
babies were disposed of in the sewer. On
this theory, see Stager, Ashkelon Discovered, p. 51.
42
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
TRACY HOFFMAN
art, and water and waste drainage and storage, indicating these aspects of the city were common
and well regulated. Julian’s discussion of porticoed streets indicates that elements of the classical
city were still very much alive and an important part of the city in the mid-sixth century. The
existence of public gardens indicates that, while Ascalon might have had a densely built urban
environment, that environment also included open spaces. This offers a possible explanation for
the extent of settlement found in the material record in which buildings were found on the
mounds and up against the city walls.
Julian’s treatise describes a diverse, densely built urban environment that maximized the
available land within the city of Ascalon. It is clear that many areas or neighborhoods of the city
had buildings with different functions and it is difficult to discern, based on Julian’s writing, a
strict separation of functions among various zones of the city. The city had diverse architecture
not only in its residential structures but also in its commercial and industrial facilities. Maintenance
of public gardens and art were important, even though land was apparently at a premium,
and considered significant in ensuring that Ascalon was a pleasant city in which to reside. Finally,
it is clear from Julian’s text that villages or rural areas near Ascalon must have had various
industries or services not permitted in the city.
The material record of Ascalon collected over sixteen seasons of excavations not only illuminates
some of the buildings and city infrastructure described by Julian but also provides
evidence for how the city continued to develop. On a general level it is possible to corroborate if
not absolutely confirm some important aspects of the city as described in the sixth century. Because
excavation on the mounds and more interior areas of the city, presumably areas of dense
settlement, has produced little to no evidence of industrial or craft activity during the Byzantine
period, industry may have been relegated to more isolated areas of the city. At the beginning of
the Early Islamic period, the conversion of a former bath house into a kiln facility near Structure
Three indicates some change in the location of industry.
The area of Structure Three also provides evidence for the type of mixed neighborhood described
by Julian, with the house first in the vicinity of a bath house and then a kiln. Even better
evidence comes from Structure One, which is a house located in proximity to a bath and a
church. The material record is clear: Ascalon had neighborhoods with both residential and nonresidential
buildings in the Byzantine period and continued to do so in the Early Islamic period.
The lack of archaeological evidence for porticoed streets, shops, warehouses, and the types of
services described by Julian is best taken as a result of the random nature of excavation and not
as a refutation of the text. Better evidence exists for the residential architecture of Ascalon and
Julian’s text does in fact render some of the architecture more meaningful. It is clear that the majority
of houses are single family homes and that they can be found throughout the city. At least
one example appears to be a multifamily home, Structure Three, in which it is possible to identify
courtyards perhaps associated with suites of rooms. The layout of the building is easily
interpreted as including multiple living units.
While the archaeological evidence of Ascalon reflects many of the patterns identified by
Julian, it also clearly shows the types of changes within the city that Julian may have been trying
to control or minimize. Excavation has uncovered dramatic changes in the function of some
buildings, culminating with the construction of a kiln during the Early Islamic period, in the vicinity
of domestic structures. It is hard to know how extensive this pattern was throughout the
city but it undoubtedly marks a departure from the perceived ideal for city organization. Overall,
a strong sense of continuity remains in the architectural evidence and much that Julian writes
about Ascalon in the sixth century seems to be more or less accurate for the Early Islamic period.
43
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ASCALON ON THE LEVANTINE COAST
The fact that the development of the city is marked by subtle rather then dramatic change is important
for understanding, insofar as it is possible, the city’s inhabitants.
One of the most striking aspects of the domestic architecture of Ascalon is the lack of evidence
for the Islamic conquest; no evidence has been found that indicates the newly arrived
inhabitants brought along new urban ideals that changed the cityscape of Ascalon. Whether this
was because the existing city and its domestic structures were recognizable forms with which the
inhabitants were comfortable, or because population pressures or the availability of building materials
encouraged the utilization of existing structures, is unclear. The overall impression is
simply one of continuity within the architecture and city plan of Ascalon. It would be misleading
to argue that no evidence exists for change. The very maintenance and reuse of structures did result
in some physical changes that can be traced through time. Certainly the ceramic corpus,
which establishes the chronological framework for the city, shows change with the introduction
of new ceramics and the disappearance of old forms. The nature of the corpus from Ascalon limits
its usefulness for exact identifications and dating, however. This is because during the last
century of settlement in medieval Ascalon the city repeatedly changed hands between Muslim
and Crusader forces, with the result that the material record of Ascalon is highly disturbed, with
few of the clean, sealed contexts archaeologists rely on for dating purposes. Although it is disturbed
and devoid of any dramatic changes, the ceramic corpus of Ascalon does reflect a steady
progression of ceramic types from Byzantine wares and early glazed wares to distinctive Middle
Islamic glazed wares such as underglaze painted wares and incised wares.
The evidence from Ascalon, both archaeological and documentary, proves that using domestic
architecture to interpret the city is an effective tool for analyzing urban development. It
provides, moreover, insight to an important part of cities that is often overlooked: private rather
than public architecture. In contrast to city centers or public architecture, which often reflect dramatic
changes, the houses do not. They show, moreover, little evidence for political or military
events, until the end of the city’s settlement, leaving a mark on the city’s residential areas. The
evidence from Ascalon is surely repeated at other sites, with or without documentary material,
where houses figure prominently in the archaeological record. Another ideal site at which to apply
this methodology is Fustat, which has a well-published collection of domestic architecture
and a readily available documentary resource, the Cairo Geniza. In fact, the relationship between
text and archaeology is much clearer at Fustat than it is at Ascalon.
A number of scholars have studied the houses of Fustat, making contributions on a number
of levels, ranging from discussions of construction materials and methods to more broad-ranging
archaeological assessments. Shlomo D. Goitein’s publication of contemporary textual evidence
for the houses of Fustat provides archaeologists with a unique database and an invaluable interpretive
tool for Fustat.15
The documents of the Cairo Geniza act in much the same way Julian of
Ascalon’s treatise does; they serve as a source of raw data about the types of houses that might
be found in Fustat and they provide a means to assess the city plan using those structures. The
best sources of information on houses in the Cairo Geniza are deeds of purchase specifically and
15. Shlomo D. Goitein, “A Mansion in Fustat: A
Twelfth-Century Description of a Domestic Compound
in the Ancient Capital of Egypt,” in The
Medieval City, edited by Harry A. Miskimin, David
Herlihy, and Abraham L. Udovitch (New Haven,
1977), pp. 163–78; idem, “Urban Housing in Fatimid
and Ayyubid Times,” Studia Islamica 47
(1978): 5–23. For the most complete discussion on
houses and related topics by Goitein, see his A
Mediterranean Society: The Jewish Communities of
the Arab World as Portrayed in the Documents of
the Cairo Geniza (Berkeley, 1983), and refer to
Volume 4: Daily Life, pp. 47–150.
44
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
TRACY HOFFMAN
legal documents more generally. Such documents usually include information on the location of
the property, its name, its characteristic parts, boundaries, details about proprietors and proprietorship,
price, and any special conditions associated with the property. Although Goitein does
not identify constituent elements for an identifiable basic house form, he identifies a typology of
three house forms using descriptions of domestic architecture in the Cairo Geniza that can be applied
to the structures excavated by Bahgat and Gabriel and Scanlon.
Goitein labels the first house type the “bazaar house.”16
This type of house consists of stores
or commercial space on a ground floor that is topped by one or two upper floors of residences. In
this type of structure separate entrances for the shops or commercial space and the residences are
common. It is difficult to identify this type of building because preservation at most sites is limited
to ground floor wall levels and nowhere is this truer than at Fustat.17
Maison VI and Groupe
I have ground floor plans that suggest they might be examples of Goitein’s “bazaar house.” The
most striking example, Groupe I, identified as a single building, displays a complexity in form,
layout, and possibly function that is not found in any of the other structures excavated by Bahgat
and Gabriel (fig. 2.11).18
The structure, which is bounded on three sides by streets, includes a
number of distinctive functional areas that are easily discerned in its organization. Two primary
means of access into Groupe I are found on the north side of the structure. The first is a doorway
leading from the street into a vestibule that provides access to a house with all the requisite features
identified by Bahgat and Gabriel, the focal point of which is Cour C. A second courtyard,
Cour D, is surrounded by a suite of rooms; the exact function of this unit, whether it is part of the
Cour C unit or an entirely separate unit, is unclear.
Just to the east of the doorway leading into the Cour C house is an alley labeled impasse that
provides access to two other areas within Groupe I. At the south end of the alley a doorway leads
into a vestibule from which two distinct houses, one oriented around Cour B and a second oriented
around Cour A, can be accessed. Again, both of these units have the requisite features
identified as key elements of houses in Fustat. At a minimum three residences, possibly four, can
be identified within Groupe I, but the additional spaces within this building allow it to be characterized
as an example of the “bazaar house.” Back in the alley a doorway on the east side
provides access to a colonnaded room that has a distinctive layout unlike those found in the residential
units. Missing in this space is a central courtyard, an iwan, or even a hall, suggesting this
space did not function as a house. The layout certainly does not fit the model of the “centralcourtyard
house” developed by Bahgat and Gabriel. An explanation for this space might be that
it was nonresidential space attached to the house. It could have been commercial or craft space
for the home’s owner. Even better evidence for commercial space in Groupe 1 comes from the
north side of the building where evidence of two shops is found along the street. These one-room
shops are clearly part of the larger structure and indicate that Groupe I was probably a mixed use
building with both residential and commercial space. In the absence of published material culture
or any interpretive discussion of this structure by the excavators, any conclusion about its
exact function must remain hypothetical. When the evidence is examined in the context of documents
from the Cairo Geniza, however, a direct correlation seems to exist between Goitein’s
“bazaar house” and the layout and function of Groupe I.
16. On house types, see Goitein, “Urban Housing in Fatimid
and Ayyubid Times,” pp. 14 –15.
17. Some exceptions include a number of sites in
Egypt, such as Karanis, where evidence for upper
stories is found in the material record.
18. Ali Bahgat and Albert Gabriel, Fouilles d’al Fustat,
publiées les auspices du Comité de conservation
des monuments de l’art arabe (Paris, 1921); Albert
Gabriel, Les fouilles d’al Foustat et les origines de
la maison arabe en Égypte (Paris, 1921).
45
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ASCALON ON THE LEVANTINE COAST
Figure 2.11. “Bazaar house”: Fustat groupe I and maison VI. Fustat
(after Bahgat and Gabriel, Fouilles d’al Fustat)
46
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
TRACY HOFFMAN
According to Goitein, the most frequently mentioned type of house in the Geniza documents
is the “family house.” This type of house might be a single structure or a compound containing
several structures belonging to one family. Based on information about such houses Goitein was
able to produce a schematic plan of a standard “family house” that is remarkably similar to some
of the houses actually uncovered during the excavation of Fustat.19
Of the eight houses published
by Bahgat and Gabriel at least three, Maisons III, IV, and VII, appear to correspond to this type
(fig. 2.12). The first example, Maison III, is an individual structure organized around a central
courtyard. This courtyard, which has a pool and fountain construction similar to that found in
Maison VI, is the focal point of the structure and is surrounded by rooms on all four sides. Access
into the structure is provided by two entrances, both of which pass through either a hallway
or a vestibule before leading into the courtyard.20
The organization of Maison III clearly fits the “family house” model identified by Goitein.
So too does Maison IV, which is also a solitary structure organized around a courtyard with
rooms on two sides. This house also has two entrances, both from the same street on the north
side of the building, that provide access into the interior of the house through long hallways. The
parallels between these houses that have courtyards with pools or fountains that serve as a focal
point, suites of rooms around the courtyards, and multiple entrances are notable. While Maison
VII also fits the “family house” model, some distinctions exist between it and the previous two.
First, it is not a stand-alone house but rather appears to share a common wall with at least one
neighboring house. Maison VII does have a central courtyard, with rooms on all four sides, but
does not have a fountain or pool. In addition, the house has only one entrance, which leads from
the street into a small vestibule that provides access to the courtyard. The range of forms and organization
of these houses, within which both similarities and distinctions are visible, reflects
the variety of descriptions for such structures in the Geniza documents.
The third house type identified by Goitein is the “apartment house.” This type consists of a
building three or more stories high. According to Goitein, evidence for this type of house comes
from terms such as “middle and upper floors” used to describe such houses. No obvious examples
of this type of structure were found at Fustat, in large part because preservation at the
site does not allow for the identification of upper floors. It is necessary to look for this type of
house in the available ground plans; one structure might be an example of a multiunit building if
not an “apartment house.” The building in question is Maison II, in which the basic plan appears
to include two distinct living units (fig. 2.13). The building is accessed by a single entrance from
the street that leads into a large room or enclosure rather than a vestibule. From that room three
doorways open into different areas of the structure. To the north (room g) passing through a series
of small rooms or halls, access is gained to Cour B, which is surrounded on three sides by
rooms. The organization of this part of Maison II has all the basic elements of a house as defined
by Bahgat and Gabriel: courtyard, hall, and iwan.
Back in the enclosure, on the west side a doorway opens to a short flight of stairs providing
access to a hallway (o) which in turn provides access to Cour A which has rooms on at least two
sides. This area of Maison II also has the requisite elements that allow it to be identified as a
house and it seems possible that in this structure are two living units for two families. This interpretation
is somewhat strengthened by a more careful comparison between the actual and
19. Goitein, “A Mansion in Fustat,” p. 165.
20. One interesting feature of this house not found in
others is a garden that Bahgat and Gabriel identified
on its northeastern side.
47
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ASCALON ON THE LEVANTINE COAST
Figure 2.12. “Family house”: Maison III, IV, VII, and VIIbis. Fustat
(after Bahgat and Gabriel, Fouilles d’al Fustat)
48
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
TRACY HOFFMAN
reconstructed architectural plans provided by Bahgat and Gabriel. On the reconstructed plan a
doorway (s) connects the two courtyards linking the two apparent housing units into one complex.
On the actual plan, however, a wall in place of the doorway may suggest that the two areas
were not directly connected and that they might be two housing units within one structure.
Maison II could include two housing units, which makes it identifiable as a possible example of
an “apartment house” or multiunit building.
Goitein labels the fourth type of house a “multistoried house,” of which only one example is
mentioned in the Geniza documents. This house has a ground floor and nine upper stories.
Goitein is hesitant to identify it as an actual house type and with good reason. The word used for
story, tabaqa, is also used to refer to an apartment forming part of a story. It is possible, therefore,
that this house with nine stories is simply a house that has nine upper apartments on an
undisclosed number of stories. As mentioned in the discussion of apartment houses and how
Figure 2.13. “Apartment house”: Maison II. Fustat (after Bahgat and Gabriel, Fouilles d’al Fustat)
49
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ASCALON ON THE LEVANTINE COAST
they might be identified, the quality of preservation in Fustat does not allow for the identification
of this type of house in the material record.21
The range of forms identified in Goitein’s typology includes single and multifamily homes
as well as buildings that combined commercial and residential space.22
From information in
Geniza documents, these buildings were found side by side in neighborhoods with diversified architecture
and organization. Both these patterns echo descriptions of the houses and city plan of
Ascalon as described by Julian of Ascalon in the mid-sixth century. That similarity raises some
interesting questions, one of the most intriguing of which is whether or not structures combining
commercial and residential space can or should be interpreted as insula. Goitein addresses this
question in an article on urban housing in Fatimid and Ayyubid times acknowledging that
Ahmad Fakhri, an Egyptian archaeologist, “… categorically denied that insulae existed in
Egypt.”23
That denial rings false in light of the “bazaar house” type identified by Goitein and
suggests that some reassessment is necessary.
Any resolution on the existence of insula in cities such as Fustat or Ascalon is beyond the
scope of this paper, but such discussions demonstrate the potential for examining houses not
only as individual structures, but also as important facets of the built environment that contribute
to the study of cities as much as studies of institutions or city centers. In addition, the study of
houses at these sites provides an important tool for examining social identity, to see how inhabitants
are reflected in the domestic architecture of a city. At Ascalon it is possible to show that
whatever the makeup of the population in the Early Islamic period, the inhabitants lived in and
used as much of the city as had the population in the Byzantine period. In the case of Fustat, the
evidence indicates that diverse inhabitants lived in the same type of housing, and architecturally
little distinguishes between houses built and lived in by Jews, Muslims, or Copts. Of course, the
addition of other types of material culture can make the identification of a house’s inhabitants a
little clearer, but even ceramics with an Arabic inscription or a cross cannot be regarded as proof
positive for the religious identity of those residing in the house.
Towards an understanding of cities in the Byzantine and Early Islamic periods, the analysis
of non-public areas can provide a valuable counterpoint to the study of a city center. As the function
of public buildings change, and as houses are built on what were once thoroughfares, signs
of dramatic change are left behind. By examining public and non-public areas together it should
be possible to elaborate a more nuanced analysis of a city’s evolution/growth. The development
of archaeological applications for documentary evidence, such as Julian of Ascalon’s text and
even the mosaic depictions of Ascalon, further expands the potential for utilizing diverse databases
to explore many aspects of cities, their architecture, and how they evolved during the
medieval period.
21. Whether or not such buildings existed in Fustat is a
question of some interest. At least one medieval
source does describe houses five and six stories tall
(Khusraw, Book of Travels, pp. 59–61).
22. Readily identifiable in the material excavated by
Bahgat and Gabriel, the same typology of houses
can be applied to Scanlon’s later excavations. In the
case of this material, however, the identifications of
specific housing forms is more challenging due to
the preservation of the structures uncovered by
Scanlon.
23. See Goitein, “Urban Housing in Fatimid and
Ayyubid Times,” p. 15.
50
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
51
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
3
SAMARQAND IN THE EIGHTH CENTURY:
THE EVIDENCE OF TRANSFORMATION
YURY KAREV
Russian Academy of Sciences, Institute of World History, Moscow
French-Uzbek archaeological Mission in Samarqand
The spread of Islam proceeded differently in different areas; the intensive internal evolution
of the Muslim community meant that this process varied greatly at different periods of time. The
status of the non-Muslim country to be conquered was particularly important. Besides the religious
motivation — affirming the new vision of the universe — two major factors determined
the political approach of the caliphate rulers to expansion from the seventh to eighth centuries.
First was the wealth of the population and the economic interest it represented, and second, the
dominant cult in the region. The position towards pagan religions was certainly most severe and
used to justify, in case of military resistance, the holy war (ghazw) “without mercy,” carried out
until the enemy was totally defeated. At the same time, destruction from war was not extended
blindly for the simple reason that the old infrastructure was, at least in the beginning, indispensable
for establishing control over the local population and collecting taxes. Only pagan temples,
idols, and sacred books were in some cases, but not always, deliberately destroyed and burned in
order to proclaim the ultimate goal of the mission. Despite the resulting shock brought by the
collapse of the old religions, the local population was involved in the long process of integration
into the new state system and the formation of a unified, if very varied, world.
During the sixth to seventh centuries, the countries of the MΩwarΩºannahr (Transoxiana)
were extremely rich and utterly pagan (exception made for the Christian and Jewish communities).
Besides that, they were in permanent rivalry with one another. The geopolitical position of
Samarqand, the capital of Sughd, the core region of the MΩwarΩºannahr, determined the particularities
of the culture of the country. The two main cultural influences came, on the one hand,
from China, and on the other, from Iran. But the importance of the Turkic political power in the
region in the seventh to eighth centuries should not be neglected. Samarqand, like some other
urban centers in MΩwarΩºannahr, played a crucial role in international trading between East and
West. An important part of the benefits accruing to the city was the creation of conditions for the
prosperity of the people, and as a result, the intensive development of all sorts of arts and crafts
ranging from decorative mural paintings to objects for domestic use. A distinctive Sogdian style
of art was elaborated before Islam that included different local and transformed foreign ele-
ments.
One of the most significant impacts on the local history of Samarqand is due to the Arab
invasion. The multiple episodes of military campaigns in KhurΩsΩn and MΩwarΩºannahr from the
end of the seventh to the first half of the eighth century in chronicles like TaºrÏkh ar-rusul wa-lmul„k
of aø-ØabarÏ are not only very detailed but occupy a major place in the author’s year-byyear
description of the history of the caliphate. This is due, obviously, to the predominant role of
the oriental direction in the external policy of the Umayyad state at that time. There is no room
here for detailing the historical description, but one point should be elaborated on: during the
51
52
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
eighth century (particularly between 705 and 752) the struggles between two communities were
extremely severe. All the campaigns of Arab commanders like Qutayba b. Muslim or SaªÏd b.
ªAmr al-H˘arashÏ, and later Ab„ Muslim, were characterized by the persecution of the rebellious
Sogdian leaders, the dihqans (dahΩqÏn) and their kings, and were accompanied by heavy casualties,
especially on the other side. Nevertheless, the resistance of the local rulers, with some
exceptions, was extremely persistent. In reality there was a long process of imposed redistribution
of wealth.
As an example, consider the following lines from a Middle Persian poem lamenting the
Arab conquest:
5 “With one blow that band (of Arabs) has enfeebled our religion and killed the princes. We are
laid low and they <lord it over us> like devils.”
7 “By force they have taken men’s wives from them, their private property, their orchards and
gardens.
8 They have imposed the poll tax and apportioned it upon our heads. They have demanded the
basic rent and a heavy impost as well.”
12 “Lo, we shall pull down their mosques and set up again the sacred fires …”1
An independent view of the situation held by opponents of the Muslims is preserved in very few
non-Christian sources. Whatever the period of this poem and the region of its composition, we
can use it in order to illustrate how the Iranian-speaking population from Ctesiphon to Samarqand
perceived the Arab conquerors. Fortunately, the Chinese chronicles have preserved some letters
from the local rulers to the T’ang Emperor, all of them composed in similar terms.2
Samarqand was taken in 712 by Qutayba b. Muslim after a long siege and many fierce
battles. A treaty was concluded between the Arab general and ikhshÏd Gh„rak, the king of
Sughd, involving an immense sum in the way of contributions. The occupation of a part of the
city by the Arabs was not included in the treaty, but at the last moment Qutayba changed his
mind and forced the ikhshÏd to leave the capital. Gh„rak chose Ishtikhan, situated northwest of
Samarqand, as his residence.
Following these events, Samarqand enters a new period of its history. One of the most
important problems for archaeologists consists in knowing what kind of material remains exist
on the site from the époque and what archaeological evidence could (or could not) reflect the
changing population and its interactions. The best place to carry out this study would presumably
be in the zone protected by the citadel of the city, the seat of the new power in a hostile environ-
ment.
The first Arab building in the city, according to the sources, was the mosque erected by
Qutayba b. Muslim on the site of the fire-temple immediately after the capture of Samarqand in
712. “Samarkand was a place of pilgrimage (h≥a©©) for the heathens, the Friday mosque that is
now in Samarkand was a heathen place of worship (muªbad) and a temple of idols (butkhΩna)
in ancient times.”3
Qutayba has ordered gathering the idols together and despoiling them. An
enormous pile of idols (“like a qaœr” castle) was set on fire by Qutayba’s own hand.4
The sense
1. François de Blois, “A Persian Poem Lamenting the
Arab Conquest,” in Studies in Honour of Clifford
Edmund Bosworth, Volume 2: The Sultan’s Turret:
Studies in Persian and Turkish Culture, edited by
Carole Hillenbrand (Leiden, 2000), p. 92.
2. Edouard Chavannes, editor, Documents sur les TouKiue
(Turcs) occidentaux (Recueillis et commentés
suivi de notes additionnelles par Edouard
Chavannes; Paris, 1904), pp. 203–05.
3. Vasilij Bartol’d, Turkestan v epoxu mongol’skago
nashestvija, Volume 1: Texty (St. Peterburg, 1898),
p. 49.
4. Ab„ Jaªfar Muh≥ammad b. JarÏr aø-ØabarÏ, TaºrÏkh
ar-rusul wa-l-mul„k, Volume 2, p. 1246; Ah≥mad b.
Yah≥ya al-BalΩdhurÏ, KitΩb fut„h≥ al-buldΩn, p. 421.
YURY KAREV
53
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
of this act was obvious — a humiliation of the pagan idol cults and a manifestation of the power
of Islam.
Frantz Grenet, referring to the plan of the temples in Penjikent, supposes that this act took
place in the large courtyard in front of the temple.5
The problem is that the excavation in this
zone has still not yielded any architectural remains that could be interpreted with confidence as
those of a temple or an early mosque. There is nevertheless a part of an edifice with a corner
tower that could be the remains of a temple (see below). In any case, for the first mosque, we
cannot be sure whether it was a new building deliberately constructed on the destroyed temple or
a fire-temple transformed into this mosque.
The works of local history, the KitΩb al-Qand and Persian Qandiya, mention at least two
other mosques in the city constructed by the ªulamΩº who came along with Qutayba, Muh≥ammad
b. WΩsiª and ad≥-D˘ah≥h≥ak b. MuzΩh≥im. As in the case of the Friday mosque of Qutayba we have
no idea about their plan or architectural style. According to aø-ØabarÏ, a part of the population, in
response to the call of the KhurΩsΩn’s governor al-Ashras b. ªAbd AllΩh in A.H. 110/728–729
C.E., started to convert to Islam and to construct the mosques.6
We know also that one group of the learned men (ªulamΩº) was charged by Qutayba with
determining the correct orientation of Samarqand towards Mecca (qibla direction) in order to
build the mih≥rΩbs (mah≥Ωrib) of the Friday mosque as well as of the other mosques in the city. In
other words, a special commission was organized, including the most authoritative ªulamΩº who
came with the Arab army and whose names were preserved in KitΩb al-Qand.7
So, on the one hand we have these clear indications of the construction activity of the
Muslims in Samarqand, and on the other the great scarcity of archaeological data from the first
part of the eighth century. During the 720s and 730s the Arabs were constantly threatened by the
revolts and riots of the Sogdians and by the incursions of the Turks. Once, around 734–735, they
probably lost the city for a while. Thus, the general situation did not stimulate the colonists to
build much. At the same time, the Sogdians were not ready to adopt the innovations brought by
the Muslims since their force of resistance was not completely exhausted.
On the other hand, with the installation of the new colonists in the conquered area the
process of integration was under way. Many Sogdians became the mawlΩs (mawΩlÏ) of the
members of different Arab tribes and participated in military operations with the Arab commanders.
As everywhere in the conquered countries the process of Islamization was much more
intensive in urban centers than in rural areas. The Arabs brought and established new political
institutions, which found their expression in new forms of architecture. This last statement is
based on the results of our excavations in Samarqand during the last twelve years.
NEW ARCHAEOLOGICAL DATA
This archaeological expedition was organized by Paul Bernard, Frantz Grenet, and
Muxammadzhon Isamiddinov and began work at Afrasiab, the ancient Samarqand, in 1989.
Many archaeological elements have been discovered from the Islamic period, dated between the
5. Xasan Axunbabaev and Frantz Grenet, “La mosquée
cathédrale: Fouilles de la mission franco-soviètique
à l’ancienne Samarkand (Afrasiab): Première
campagne, 1989, par MM Paul Bernard, Frantz
Grenet, Muxammedzon Isamiddinov et leurs
collaborateurs,” Comptes Rendus de l’Académie des
Inscriptions et Belles-Lettres (Paris, 1990): 370.
6. Aø-ØabarÏ, TaºrÏkh, p. 1508.
7. Na©m ad-DÏn ªUmar b. Muh≥ammad an-NasafÏ, alQand
fÏ dhikr ªulamΩº Samarqand, edited by Y„suf
al-HΩdÏ (Tehran 1999), pp. 178–79, 603.
SAMARQAND IN THE EIGHTH CENTURY
54
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
eighth and the thirteenth century, when the city was destroyed by Genghis Khan’s troops. This
research builds upon over a century of field research by successive Russian and Soviet expeditions
to this famous site. Since 1991 the author has been co-director with his Uzbek colleague
Dr. A. Atakhodjaev of the excavations on the lower terrace of the citadel of Samarqand. From
the year 2000 up until this latest season in 2002, the ongoing research on the site has been made
possible thanks to the Max Van Berchem Foundation.
The site of Afrasiab occupied an area of almost 220 ha from the moment of its foundation in
the sixth century B.C. (fig. 3.1). The core of the site is situated in the north and formed by the
citadel and the so-called sacred area where the main religious buildings were situated (as far as
we can infer from the remains of the mosques of the eighth and thirteenth centuries; fig. 3.2). It
is here, in the northern part of the site, that most of the archaeological activities of the mission
have taken place.
YURY KAREV
Figure 3.1. Afrasiab site, general plan (based on 1885 plan of Vasiliev and Kuzmin). Drawing by
G. Lecuyot, “Fouilles de la mission franco-soviètique à l’ancienne Samarkand,” Comptes
Rendus de l’Académie des Inscriptions et Belles-Lettres (Paris, 1990): 357
55
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
The first important architectural remains that we could relate to the Arab construction activities
in the city have been found in this “sacred area” (fig. 3.2: Area 1). Despite what could have
been expected, the building discovered can hardly be interpreted as a mosque. The huge complex
being excavated by Frantz Grenet and Igor Ivanickij in the mosque area of Afrasiab belongs to
the new type of architecture and indicates some important changes in its planning (fig. 3.3).8
First of all, its plan is possibly organized around a central courtyard (ca. 50 ≈ 45 m) although
only the part to the east was excavated and remains of the building were badly damaged both in
ancient times and by the excavators at the beginning of the twentieth century. The complex is
oriented on cardinal points; its size is about 115 ≈ 84 m. External walls are more than 4 m wide,
strengthened by massive half-towers and corner towers. All the walls of the building were erected
using the same technique— two courses of mudbricks (50 ≈ 31 ≈ 14 cm) were placed between
two courses of pisé. This technique fits comfortably with the Sogdian tradition.
The complex certainly had two distinctive parts, opposite one another, on the north and the
south. The part to the south is much more comprehensible from an architectural point of view. A
distinctive unit, which can be called a bayt, with a separate disposition was enhanced by a
probable axial corridor from the west and by two perpendicular corridors from the north and
east. The unit itself was organized around a central courtyard or hall of about 19 ≈ 11 m. All the
rooms of varying interior dimensions were paved, probably in the second period, by thick burned
SAMARQAND IN THE EIGHTH CENTURY
Figure 3.2. Northern part of the Afrasiab site. (1) “Sacred area” with the complex of the eighth century
and mosques above it, (2) citadel-donjon, and (3) lower terrace of the citadel with the dΩr al-imΩra
from the middle of the eighth century (Franco-Uzbek archaeological mission in Samarqand;
different plans unified by Cl. Rapin)
8. The description of the complex is based on the article
published in Russian: Franc Grene [Frantz
Grenet] and Igor Ivanickij, “Dvorec omeyadskogo
namestnika pod mechet’ju abbasidskogo perioda na
Afrasiabe,” in Arxeologija, numizmatika i épigrafika
srednevekovoj Srednej Azii, Materialy nauchnoj
konferencii, posvjashchennoj 60-letiju so dnja
rozhdenija d.i.n. B.D. Kochneva (15. XII. 2000)
(Samarqand, 2000), pp. 58– 62. I would like to express
my gratitude to Frantz Grenet for his permission
to publish the most recent field plan of the complex
(fig. 3.3).
56
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
bricks (44 ≈ 38 ≈ 6 cm), which are in fact the earliest example of large-scale brick production of
this type. The Sogdians usually used mudbricks and pisé (pakhsa), with the exception of some
special constructions requiring burned bricks, such as bathrooms. The baked bricks have preserved
rectangular stamps with Arabic inscriptions, probably the marks of the manufacturer; for
the most part these are not deciphered with confidence. In 1904 Vasili Barthold saw bricks with
the inscriptions “ikhshÏd” and “Ishtikhan,” the name of the city situated to the northeast of
Samarqand, where the ikhshÏds were compelled to move after 712. The proposed reading is still
hypothetical and inspires polemics. The most distinctive unit in the northern half of the complex
is the rectangular room, 20.0 ≈ 11.5 m, the proportions of which are very close to that of the
reception halls of Sogdian architecture. The ceramics in the rubble layer inside the rooms can be
dated from the end of the eighth century to the beginning of the ninth and belong to the third and
last period of occupation of the complex. The terminus ante quem non is defined by the finding
in the east wall of a coin of Turghar (738 to ca. 750 C.E.), the last king, an ikhshÏd of Sughd who
had the prerogative to mint his own coinage.
The interpretation of this building is complicated by two problems. Apart from the fact that
the “sacred area” is very badly damaged and extremely difficult to excavate, the complex has
been excavated only partially and its known parts seem not to have been erected at the same
time. The plan looks like a block of different composite units protected and reinforced by a
massive external enclosure. The presence of the walls and towers built on almost the same level
on opposite sides of the huge area is, indeed, the principal criterion for the interpretation of these
architectural elements as a unified complex. On the one hand, the plan and the size of the
complex apparently do not belong to the local tradition, to judge by its latest stage from the
samples from Afrasiab itself and other archaeological sites in Sughd (see below). At the same
time, the construction techniques mentioned above, the ramp leading to the second floor in the
south bay, and particularly the lack of symmetry, as well as evident irregularity of the general
setting and plausible use of previous constructions, all reflect the transitional period between two
traditions. In the archaeologist’s opinion, the building, probably the first administrative palace,
was constructed by the governor of KhurΩsΩn Naœr b. SayyΩr (738–748) in the 740s, who succeeded
in making peace with the Sogdians and calming the situation for some years.
It is important to note that the orientation of the complex follows the general disposition in
the previous periods of walls in the “sacred area.” A mosque built after 770 shows clear evidence
that the qibla direction was different (fig. 3.3), even though the northern part of its complex has
completely disappeared under the later mosque of the early thirteenth century. The southern part
was certainly leveled at the beginning of the ninth century, when the mosque was enlarged to the
west.
The question remains on the situation of the first mosque of Qutayba. One may expect its
orientation towards qibla should be the same as with the later mosques. As we know from other
examples, the mosque could be included in the palatial building (Qaœr al-H˘ayr ash-SharqÏ,
Khirbat al-Minya, Ukhayd≥Ïr among others), but in the case of the public Friday mosque the
latter was usually erected outside of the dΩr al-imΩra, being nevertheless placed close to it (alK„fa,
ªAn©ar, and so forth).
It should be remembered that in the old cities in general and much more in the fortified
zones like a citadel space was always limited, even if the previous buildings had been leveled. In
ªAmmΩn, for instance, the mosque (the orientation of which is different than that of the rest of
the edifices) was built in the center of the “L”-shaped citadel hill, where the space allowed a
most convenient disposition outside the dΩr al-imΩra limited by the external walls of the narrow
YURY KAREV
57
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
SAMARQAND IN THE EIGHTH CENTURY
Figure 3.3. Different structures on the “sacred area.” Afrasiab
58
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
northern part.9
Thus, economy of space should play an important role. In other words, a new
building could be constructed by the Muslims with the same orientation as the with the ancient
one, an exception being when the mosque was included as an irregular element in the new
structure. If this was the case in Samarqand, the first mosque, of a relatively small size, could be
constructed close to the eastern side of the palace, as already suggested by Frantz Grenet. Nevertheless,
the zone to the north and northeast of the plateau should not be entirely excluded (fig.
3.2: north of Area 1). In any case, the archaeological data shows the Arabs started to construct
new buildings and change the situation in Samarqand.
Usually the royal palaces were constructed near to the donjon (called keshk), on the lower
terrace; that is precisely the case of the building discovered and excavated since 1991.10
The
citadel at Samarqand can be divided into two parts: the donjon situated on the highest point of
the site (fig. 3.2) and a lower area to the east of this castle (fig. 3.2: Area 3). This type of
structure, with a “two-part citadel,” is well known from other sites in Central Asia such as
Pendjikent. The plan of this complex represents an important phase in the architectural history of
the region. It consists of two different parts (figs. 3.4 –5). The first, major one is a rectangular
structure 65 ≈ 55 m, organized around an inner central court. The second part includes a relatively
narrow oblong court, located between the western wall of the palace itself (wall of the western
row of rooms), and another external wall, situated 17 m westwards. The use of massive octagonal
columns (2 m diameter) in both courts appears to be one of the most distinctive features of
the monument.
In the first case, an arched gallery surrounded (on all four sides) an interior courtyard
around which all rooms of the eastern part of the building were grouped. In the second case, the
columns were initially planned to support a covered arched gallery built in front of the western
wall of the building along the side of the “exterior” courtyard space, opposite the donjon. The
colonnade of the outer courtyard also had two half-pillars — the pilaster attached to the southern
wall, which was excavated, and by deduction a symmetrical one against the northern wall. The
building had three towers on the eastern side, two in the corners and one in the center (fig. 3.5).
Each tower is 3.5 m in diameter, a size which indicates that they served a decorative rather than a
defensive function. The walls were built in seven or eight courses of mudbricks (47 ≈ 23 ≈ 7–8
cm) alternating with blocks of pisé (200 ≈ 86 ≈ 80 cm). The main gate was situated on the south
side opposite the iwΩn on the north side of the palace. Two massive pillars (3.5 m large each)
formed an impressive portal 5 m wide outside the building and about 11 m deep. On the west
side two symmetrically disposed narrow corridors gave access to the main building (fig. 3.5).
According to the archaeological evidence, the whole area was leveled (the buildings from
previous periods were almost completely destroyed) and the surface prepared for the construction
of a new building. It is thus possible to conclude that a radical reorganization of the architectural
space of the citadel took place in the middle of the eighth century. The building itself was
designed with such precision that it was possible to reconstitute the unexcavated parts by simply
applying the principles of axial symmetry. The fact that this enormous new building should have
occupied the greater part of the lower half of the citadel clearly reflects its dominant function.
9. Antonio Almagro, “El Alcázar Omeya de Ammán,”
in El esplendor de los Omeyas Cordobeses, la
civilización musulmana de Europa Occidental:
Exposición en Madinat al-Zahra, 3 de mayo a 30 de
septembre de 2001, edited by Junta de Andalucía
(Granada, 2001), p. 50.
YURY KAREV
10. Yury Karev, “Un palais islamique du VIIIe siècle à
Samarkand,” Studia Iranica 29 (2000): 273–96. For
the most recent data, see idem, “Nouvelles recherches
dans la citadelle de l’ancienne Samarkand,”
Bulletin de la Fondation Max van Berchem 15
(2001): 1–4.
59
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
There can be no doubt that it was built on the citadel of the city in order to house the administrative
and political apparatus of the caliph’s representative and should therefore be identified as
the governor’s palace, the dΩr al-imΩra.
This palace was never finished according to its initial plan since there is no evidence of
intensive occupation of the building during this first phase. Indeed, no remains of stucco or wall
paintings have been found, yet it is logical to suppose that elements of architectural decoration
such as these would have existed in a building of this type. Clearly some event must have
disrupted the execution of the initial plans of the architect and led to the construction work
coming to a halt. The first phase of construction can be dated, thanks to monetary finds in the
foundation level under the walls (the coin of Turghar, ikhshÏd of Sughd), from the 740s to the
750s.
After a while, the authorities that were ruling the city decided to finish the building with the
means available at that time. These means were clearly not comparable to those of their predecessors,
either financially or intellectually, and led to a considerable simplification of the initial
plan; it is during this constructive phase that baked bricks were used for the first time, both in the
masonry of the walls enclosing the outer and inner colonnades and in the paving of many rooms.
This second period of construction, separated from the first by a short interval, can be dated
between the second half of the 750s and the 770s (thanks to the first coins of Ab„ Muslim from
Samarqand — beginning of the 750s, at those minted by al-Ashªath b. Yah≥yΩ [A.H. 144/760–
761 C.E.] and al-˝unayd b. KhΩlid at Bukhara [A.H. 151/ 768 C.E.]). The third period can be
SAMARQAND IN THE EIGHTH CENTURY
Figure 3.4. DΩr al-imΩra on the lower terrace of the citadel (ca. 750 C.E.). Afrasiab. Plan by E. Kurkina
60
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
dated to the end of the eighth and the first half of the ninth century (by “black dirhams,” with the
name of the caliph al-AmÏn (A.H. 193–198/809–813 C.E.). Finally, we date the fourth period to
the middle of the ninth century, in other words to the beginning of the Samanid period. The
scope of these modifications enables us to suppose that the building survived in a progressively
more and more distorted form up until the tenth century.
HISTORICAL CONTEXT
Archaeological excavations are particularly important because of the silence of the historical
sources about so many aspects of the history of the area. In this context, the discovery of a
building such as a great palace or temple is all the more helpful. The size and the type of plan
used for such structures can be precious witnesses of politically significant events that are not
described in historical sources or only mentioned in cursory fashion. The complete analysis of
the sources was made in a special article on the matter;11
the following text presents the essential
conclusions.
With all necessary precautions, the newly discovered administrative complex in Samarqand
can be interpreted as a creation of the leader of the ªAbbasid movement Ab„ Muslim. He had an
11. Yury Karev, “La politique d’Ab„ Muslim dans le
MΩwarΩ’annahr: Les nouvelles données textuelles et
archéologiques,” Der Islam 79 (2002): 1– 46.
Figure 3.5. DΩr al-imΩra on the lower terrace of the citadel (ca. 750 C.E.). Afrasiab.
Plan and reconstruction by E. Kurkina
YURY KAREV
61
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
obvious predilection for construction activity. Various Arab and Persian sources mention the
remarkable buildings erected on the order of Ab„ Muslim in at least three major cities of
KhurΩsΩn: a dΩr al-imΩra and a new congregational mosque in Marw, a mosque in NÏshΩp„r, and
a huge wall in Samarqand. A number of Arab authors (aø-ØabarÏ, al-MaqdÏsÏ, an-NasafÏ, Ibn alAthÏr)
say that Ab„ Muslim built the wall of Samarqand in A.H. 134/ 751–752 C.E. or one year
later. He also built gates in the wall, towers, and loopholes (or merlons). Why did Ab„ Muslim
build these walls, and why did he remain in Samarqand for at least one-and-a-half years and
probably more than two years?
The first reason was that it was necessary to crush the last remaining velleities/vestiges of
independence of the local kings. This was precisely the moment when, as we know from the
historical sources, eleven of the kings of Transoxiana attempted for one last time to liberate
themselves from Arab power by presenting a request to the Chinese emperor to attack the “Black
Clothes,” that is, the ªAbbasids. Many kings were executed, such as Qutayba b. TughshΩda, the
king of Bukhara, and IkhrÏd, the king of Kish. The sources are unanimous on one point — it was
certainly not only punishment of the rebellious “old-fashioned” kings but a large purge of the
noble class of the country. The second cause was the expansion of the Chinese T’ang empire,
which culminated in the defeat of the Chinese army by ªAbbasid general ZiyΩd b. SΩlih≥ in the
battle of the Øalas River in July 751. Ab„ Muslim was occupied with the preparations for this
campaign in Samarqand. The third cause was a consequence of the foregoing victory. After the
Øalas River Battle “Ab„ Muslim planned to undertake incursions into China and he made preparations
in view of this.”12
The style and the methods employed by Ab„ Muslim must have led to
a significant change in the city and it is possible that he even considered Samarqand as a basepoint
on the road to conquering China.
ARCHITECTURAL CONTEXT OF THE BUILDING
At Samarqand there is no doubt that this palace, built in the tradition of Middle Eastern or
Iranian architecture, was designed by an architect from the region. The plan has nothing in
common with royal buildings of local tradition. We have a relatively good knowledge of Sogdian
architecture of the pre-Islamic period thanks to the excavation of three palaces, among others: at
Bukhara (Varakhsha), Pendjikent, and ShahristΩn (Bundjikat), the capital of Ushrusana.
The palaces of Pendjikent and ShahristΩn are similar (fig. 3.6); the organization of their
inner space belongs to a single artistic school. They are composed of a compact series of ceremonial
rooms, which are not symmetrically disposed. The main, largest room is a rectangular
throneroom (22.80 ≈ 12.25 m at Pendijkent and 17.65 ≈ 11.77 m at ShahristΩn), situated in the
western part of these buildings. In this room a large and deep niche/loggia (4.50 ≈ 4.75 m at
Pendjikent) was built in the southern wall and was clearly meant to contain the throne. To the
east of the throneroom there was a square ceremonial room (of about 10 ≈ 10 m) with four
columns. Both palaces are characterized by the magnificence of their decoration. It is interesting
to note that even the orientation of the main rooms is the same at Bundjikat and at Pendjikent. In
addition, in both cases access to the throneroom is from the northern side via a long corridor
oriented east–west. (The corridors play an important role in the plan of the buildings.)
12. Muøahhar b. ØΩhir al-MaqdisÏ, Le livre de la
création et de l’histoire de Motahhar ben TΩhir elMaqdisÏ,
attributed to Ab„ Zayd Ahmed ibn Sahl
SAMARQAND IN THE EIGHTH CENTURY
Al-BalkhÏ, volume 6, edited and translated by
Clément Huart (Paris, 1919), pp. 74–75.
62
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
Figure 3.6. (a) Palace in Pendjikent (Raspopova, Zhilishcha Pendzhikenta, p. 168) and (b) palace in
ShahristΩn (Bundjikat) (Negmatov Negmatov, “Issledovanija v Severnom Tadzhikistane v 1970 g.,”
Arxeologischeskie raboty v Tadzhikistane 10 [1970], Moscow 1973, fig. 11)
YURY KAREV
a
b
63
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
These two palaces may be considered as part of the same architectural tradition as the
Sogdian house; they mark the “limit” of the development of this type of architecture but are not
fundamentally different in any way.13
This is certainly linked to the level of social, economic,
and political relations of the society, and in particular to the relatively high status of the free
Sogdian citizen and to the limited nature of the king’s supreme power. This type of palatial
complex was widely distributed throughout MΩwarΩºannahr, for example, in ShΩsh (Ak-tepe
Junusabadskij).
The palace of the BukhΩrkhudΩt at Varakhsha follows slightly different architectural principles
(fig. 3.7); the proximity of Iranian architectural traditions can be seen, for example, in an
iwΩn with imposing columns (2 m in diameter), which clearly served as a throneroom. However,
even if the palace of Varakhsha is slightly more regular, it does not have anything to do with that
of a planned symmetrical complex organized around a central courtyard. It is plausible, however,
as Aleksandr Naymark suggests,14
that the rebuilding of this palace after the Arab conquest can
explain such non-Transoxianian elements in the plan (massive columns) and stucco decoration.
It is clear that the kings of the principalities of Transoxiana did not attempt to imitate the
imperial style linked to the organization of the royal court (both administrative and ceremonial),
which is characteristic of the palatial architecture of Sassanid Iran, for example. There, the
SAMARQAND IN THE EIGHTH CENTURY
Figure 3.7. Palace in Varakhsha (V. A. Shishkin, Varakhsha, Moscow, 1963, fig. 16)
13. Valentina I. Raspopova, Zhilishcha Pendzhikenta:
Opyt istoriko-social’noj interpretatsii (Leningrad,
1990), p. 182.
14. Aleksandr Naymark, “Returning to Varakhsha,”
The Silk Road Foundation Newsletter 1/2 (2003):
9–22.
64
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
domed throneroom was placed in a central (axial) position, whereas its access was marked by
the presence (in most cases) of an iwΩn, facing the inner courtyard; for example, the palaces of
ImΩrat-i Khosrow (at Qaœr-i ShÏrÏn) and that of FÏr„zabΩd. This same tradition survived the
Arab conquest and is also characteristic of the palaces of the Umayyad and especially ªAbbasid
periods.
SOME REMARKS ON THE ORIGIN OF THE PLAN
A clear correspondence exists between the plan of the governor’s residence at Samarqand
and the Umayyad castles of BilΩd aå-ÅΩm, situated in the now desert regions of Syria and Jordan
(Qaœr al-H˘ayr al-GharbÏ and al-SharqÏ, ˝abal Says, Umm al-WalÏd, and many others), as well as
with the Umayyad palace of Jerusalem and the bayts (buy„t) of the dΩr al-imΩra at ªAmmΩn. A
contemporary parallel to the dΩr al-imΩra at Samarqand is furnished by the bayts of the palaces
of Raqqa (Palace C), where a similar row of massive portico pillars leads out onto a closed
courtyard (east and west palaces).
We can consider (given the contemporary state of archaeological research) that the dΩr alimΩra
at Samarqand is one of the first administrative buildings in Transoxiana in the Early
Middle Ages built around a central courtyard surrounded by a series of symmetrically disposed
rooms. Even if it did not exert a direct influence on later buildings, it marks the beginning of a
new architectural tradition, of which the lower palace of Khulbuk in Khuttal, dated to the ninth
century, and the caravanserai of Paykend, at the end of the eighth/beginning of the ninth century,
are good examples, as are the later caravanserais and madrasas.
The palace in Samarqand is at the moment one of the oldest known examples of civil
“Islamic” architecture in an area stretching between Iran (Mesopotamia) and MΩwarΩºannahr.
The dating of this monument is based on precise stratigraphic data. Its importance for the study
of the Early Islamic period in KhurΩsΩn, a key region of the ªAbbasid state system, is fundamental
since administrative buildings of this period have not yet been found in large oriental towns
such as Bukhara, Marw, Balkh, or NÏshΩp„r. Indeed, we do not have many examples from the
Near East (ªAmmΩn, Jerusalem, and Raqqa).
Apart from two administrative buildings in the northern part of the Afrasiab giving evidence
of important innovations due to the arrival of the new colonists, the city itself was exposed to the
characteristic transformations. As shown by the excavations of Galina Shishkina in the western
part of Afrasiab since the mid-eighth century, the urban fabric was beginning to change.15
The
courtyards in a dense residential quarter became a significant element, something which was not,
or very rarely, present in the urban fabric of pre-Islamic times (fig. 3.8).
The mid-eighth century is the turning point for the city: architecture aside, we can mention
the reoccupation of the quarters abandoned after the Arab conquest, new forms of ceramics (this
requires a special study), large-scale production of baked bricks, and particularly the first mint in
Samarqand on behalf of the ªAbbasids; from A.H. 143/760–761 C.E. fals were minted in
Samarqand by governor DΩº„d b. KurΩz (GurΩz). The following issue of al-Ashªath b. Yah≥yΩ in
A.H. 144/761–762 C.E. preserved, with the traditional ªAbbasid formula and the name of alMahdΩ,
a tamga, the symbol of the ikhshÏds of Samarqand as a last sign of respect towards the
ancient dynasty losing political control over the country.
YURY KAREV
15. Galina V. Shishkina, “Gorodskoj kvartal VIII–XI
vv. na severo-zapade Afrasiaba,” in Afrasiab 2, edited
by Iakh’ia G. Guliamov, pp. 117–56 (Tashkent,
1973).
65
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
CONCLUSION
MΩwarΩºannahr never had the traditions of an empire state as such, a fact due to its geopolitical
situation between two imperial neighbors, China and Iran. Neither China nor Iran ever held
this region for a long period of time. It goes without saying that their influence was notably
present there, but nothing was imposed by force. From the eighth century on, the situation was
completely changed. In Central Asia, the Arabs succeeded where other empires, such as that of
the Sassanids, had failed. They brought the elements of Iranian and Near Eastern tradition into
the region directly. In a certain manner, the Arabs carried to MΩwarΩºannahr the old-fashioned
SAMARQAND IN THE EIGHTH CENTURY
Figure 3.8. Residential quarter in the western part of Afrasiab (Shishkina, “Gorodskoj,” fig. 1, p. 119)
66
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
imperial traditions absorbed during the first century of Hegira by the new Muslim community.
The impact of the invasion was sufficiently deep to determine the post-conquest political and
cultural evolution of the Central Asian region.
Sughd was thus involved, by force, in a process of integration into the largest empire at that
time. This does not mean that everything was imported or that all local things were neglected; on
the contrary, Sughd and other regions of MΩwarΩºannahr played a role as the breeding-ground
for the formation of the new culture traditionally called “Muslim.” The palaces of Samarqand
mark the beginning of this process.16
YURY KAREV
16. I would like to express my gratitude to M. S. Stride
and M. J. Dallett for their kind help in correcting the
English text of this article.
67
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
4
ISLAM, ARCHAEOLOGY, AND
SWAHILI IDENTITY
MARK C. HORTON
Department of Archaeology and Anthropology, University of Bristol
The coast of eastern Africa is the home to a remarkable Islamic society, which is known by
the general term Swahili.1
This society, which extends along some 3,000 km between southern
Somalia and Mozambique, as well as the offshore islands of Pemba, Zanzibar, Mafia, the
Comores, and parts of northern Madagascar, has many definable features in common. These include
a common adherence to Sunni Islam and a single language, kiSwahili, as well a common
kinship structure based on patrilineal descent groups, and at least in the past, a key role as intermediaries
in the international trade of the western Indian Ocean. Another important feature of
the Swahili is their urban culture, with stone houses, mosques, and tombs, complex systems of
governance, and what is now recognized as a rural hinterland that supplied the towns with trade
goods and food supplies.
While traditional Swahili society survives today in pockets along this immense coastline,
the archaeological evidence shows that it was once much more extensive. There are approximately
400 sites that contain some form of masonry construction, mostly in the form of mosques
and tombs, and about fifty sites that contained stone houses. But this is only part of the story, as
recent archaeological survey has shown. For example, in an area of northern Pemba Island,
where there were two “stone towns,” a 2% sample produced over forty sites — mostly villages
with a small number of mud and thatch houses.2
It appears that ancient populations levels were
perhaps even higher than the present-day numbers, which on Pemba is around 150,000.
It is inevitable that our understanding of the Swahili has been richly colored by the ethnographic
record gleaned from the places Swahili culture has survived. This has been in the Lamu
archipelago, and to some extent on Zanzibar, Pemba, and Mafia. Lamu and the nearby town of
Pate are classic “towns” with many stone houses, narrow streets, numerous mosques, and a developed
urban Islamic culture.3
Accounts of the past, largely gleaned from the patrician
inhabitants, recall a golden age of prosperity, largely supported through domestic and plantation
slavery, in which the privileged classes lived a life of opulence based on a rich imported material
culture. This urban culture developed complex cultural expression through Islamic practices of
piety and charity, and in the purity of the women and houses, and the honour and reputation of
the men.4
But in many ways this is a constructed past that never quite existed, or if it did, only
for a short period in the late eighteenth to mid-nineteenth century.
1. Mark C. Horton and John Middleton, The Swahili:
The Social Landscape of a Mercantile Society (Oxford,
2000); A. H. J. Prins, The Swahili-Speaking
Peoples of Zanzibar and the East African Coast: Arabs,
Shirazi, and Swahili (London, revised edition,
1967).
2. Jeffrey Fleisher, “Viewing Stonetowns from the
Countryside: An Archaeological Approach to
Swahili Regional Systems, AD 800–1500” (Ph.D.
diss., University of Virginia, 2003), pp. 107–49.
3. John Middleton, The World of the Swahili: An African
Mercantile Civilization (New Haven, 1992)
provides a detailed account of the Swahili viewed
from within a largely patrician perspective.
4. Horton and Middleton, Swahili, pp. 204 –05.
67
68
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
In reality Swahili culture and identity is much more diverse and complex. The “Lamu”
based version of Swahili culture is probably very much the exception rather than the rule. For
example, widespread slavery seems to have developed in the late eighteenth century in response
to Omani settlement on the coast, and the expansion of European slaving interests (mostly
French and Portuguese) into the Indian Ocean. Before this time, there is very little evidence that
the Swahili kept slaves, beyond the occasional domestic.
Nineteenth century slavery also altered the nature of the traditional rural economy, where,
before the plantations the “towns” were placed at the center of a large rural conurbation of settle-
ment.5
These towns, often located on small islands, produced their own food, often requiring the
inhabitants to travel to the mainland or main island to farm. But they also were dependant on a
network of Swahili villages that were entirely rural in character, with no stone architecture or
sense of town planning. They can be called “Swahili” in that they were inhabited by Muslims,
with kinship ties to the main towns. They produced their own food (shellfish and fish, cereals,
mostly millets, sorghum and rice, as well as cattle, sheep, and goats) and traded surpluses to the
towns in exchange for imported goods and craft items. In some cases the hinterland to the town
could be a considerable distance away, as in the case of northern Pemba, which supplied food
(mostly rice) to Mombasa. Political control of these Swahili villages by the towns varied considerably;
many had economic and religious links to the main towns but had very weak political
ties. In other cases the towns set out to control their hinterland, often with mixed success.
However, the conurbations also contained other groups that were linked culturally and economically
to the towns, but who were not necessarily Muslim, or even kiSwahili-speaking. This
is best seen on the southern Kenyan coast, where the diwans of Vumba controlled a large territory
from the mid-eighteenth century, which included a series of Swahili villages (known as
“Shirazi” towns, which once had their own rulers but were now dispossessed by the Vumba), as
well as non-Muslim farmers such as the Digo and the Segeju.6
These non-Muslims had a watani
or joking relationship with the Vumba and supplied agricultural goods as well as military protection
to the polity. While militarily weak, the Vumba were able to control these different groups
through their access to trade goods and their ritual status as sharifs and were considered to possess
Islamic authority, which could be interpreted as magical powers, for example, over rain or
healing.
THE CONTRIBUTION OF ARCHAEOLOGY
Documents and traditions often give a static account of a Swahili society, with unchanging
social and religious institutions — a society pickled in aspic, viewed through a nineteenth century
golden age. In reality most historians and anthropologists concern themselves with the last
200 years and have little interest in the previous 2,000 years.
Archaeologists have been prominent in reconstructing the origins and early history of
Swahili society because of the quality of surviving material evidence and the lack of other materials
with which to work. The historical sources are partial, with only short descriptions from
5. Horton and Middleton, Swahili, pp. 136–37; 165–
72; Fleisher, “Viewing Stonetowns.” In Swahili,
Horton and Middleton called these rural settlements
commoner towns, but in reality they are really villages
and in retrospect this is probably a better term.
6. Horton and Middleton, Swahili, pp. 166–68; W. F.
McKay, “A Precolonial History of the Southern
Kenya Coast” (Ph.D. diss., Boston University,
1975); Prins, Swahili-Speaking Peoples, p. 94.
MARK C. HORTON
69
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
outsiders who had scant understanding of how Swahili society functioned.7
The Swahili’s own
chronicles and histories, with the exception of one paraphrase recorded by the Portuguese, all
date to the later nineteenth century, and although clearly an oral record of earlier generations, are
often little more than a listing of the rulers and their activities. Oral traditions can be helpful but
are even more recent and can be subject to much interpretation, sometimes wildly speculative.8
Historical linguistics suggests cultural developments though the study of vocabulary, but there is
always a lack of chronological precision in these analyses.9
However, the record of archaeologists in reconstructing early Swahili society has itself been
controversial. Archaeological research has been undertaken on the eastern African coast since
1948. James Kirkman pioneered a series of excavations at Gedi and the surrounding areas, setting
up a National Park there for the colonial government. Kirkman was also invited to
undertake a short excavation on Pemba Island, looking for Qanbalu, the town visited by alMasudi
in 916; he later excavated at Kilwa at the request of Gervase Mathew and Sir Mortimer
Wheeler, who were keen to locate Roman trade outposts in Africa (as Wheeler had discovered in
India).10
In 1957 Neville Chittick, who had previously worked in Sudan but actually trained as a
lawyer, was appointed as the first conservator of antiquities in Tanganyika. In 1960 a British
Institute was established (under Wheeler’s support) and Chittick became its second director in
1961, where he remained until 1982, undertaking massive excavations at Kilwa as well as in the
Lamu archipelago.11
The Kilwa project was an exercise in historical archaeology, where the various
versions of the Kilwa Chronicle were tested against the archaeological and numismatic
evidence.12
Under Chittick and Kirkman, the archaeology of the coast developed a very oriental
interpretation. The focus was on stone towns and their architecture, which were assumed to have
been founded by overseas merchants (mostly Arabs and Persians), who brought a degree of civilization
to the African coast. Gedi was rechristened the “Arab city of Gedi”; Shirazi settlers, who
ultimately came from the northern coast of the Persian Gulf, founded Kilwa and Manda.13
This
approach had little time for rural settlement, mud architecture, or indeed “African” pottery, when
faced with quantities of Islamic and Chinese imports and well-preserved stone buildings.14
Curiously
neither Chittick nor Kirkman were particularly interested in mosques either. Despite
working at Gedi for close to twenty years, Kirkman failed to notice the ruins of the largest
7. Most of the sources are conveniently collected and
translated in G. S. P. Freeman-Grenville, The East
African Coast: Select Documents from the First to
the Earlier Nineteenth Century (Oxford, 1962).
8. For example, James de Vere Allen, Swahili Origins:
Swahili Culture and the Shungwaya Phenomenon
(London, 1993).
9. Derek Nurse and Thomas Spear, The Swahili: Reconstructing
the History and Language of an African
Society, 800–1500 (Philadelphia, 1985); Derek
Nurse and Thomas J. Hinnebusch, Swahili and
Sabaki: A Linguistic History (University of California
Publications in Linguistics 121; Berkeley, 1993).
10. James S. Kirkman, The Arab City of Gedi: Excavations
at the Great Mosque, Architecture and Finds
(London, 1954); idem, “Excavations at Ras
Mkumbuu on the Island of Pemba,” Tanganyika
Notes and Records 53 (1959), pp. 161–78; idem,
“Kilwa – the Cutting Behind the Defensive Wall,”
Tanganyika Notes and Records 50 (1958), pp. 94–
101.
11. H. Neville Chittick, Kilwa: An Islamic Trading City
on the East African Coast, 2 volumes (Memoir of
the British Institute in Eastern Africa 5; Nairobi,
1974); H. Neville Chittick, Manda: Excavations at
an Island Port on the Kenya Coast (Memoir of the
British Institute in Eastern Africa 9; Nairobi, 1984).
12. H. Neville Chittick, “The ‘Shirazi’ Colonization of
East Africa,” Journal of African History 6/3 (1965);
Chittick, Kilwa, pp. 235–45.
13. Chittick, Manda, pp. 219–20.
14. Mark C. Horton, “The Asiatic Colonisation of the
East African Coast, the Manda Evidence,” Journal
of the Royal Asiatic Society 1986/2 (1986): 201–13.
ISLAM, ARCHAEOLOGY, AND SWAHILI IDENTITY
70
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
mosque on the site.15
Chittick’s work at Manda, which he claimed as the first landfall of the
Shirazi settlers, did not include any investigations of the ruined mosque, which while fifteenth
century, has earlier building underneath it.
By 1980, some fifteen years after the nations of East Africa had gained their independence
from the colonial powers, the “colonial” interpretation of the Swahili was looking somewhat
shaky.16
Historical linguists had been at work on the structure of kiSwahili and showed that it
was not a Creole language based on a mixture of Arabic and Bantu, but a full member of the
North East Coastal Bantu group languages, with surprisingly few Arab (and virtually no Persian)
loanwords.17
Historians were looking at the oral traditions and saw that they could not be
taken literally; indeed several were the product of nineteenth century “arabization” of the
coast.18
The view that civilization could only come from the orient was no longer tenable as a
valid archaeological (or indeed historical) interpretation.19
In many ways, the Swahili coast represented
a parallel case to Great Zimbabwe, where similar interpretations hid the clear African
context of the site.
SHANGA
It was within these debates that the excavations began at Shanga in 1980 and continued until
1988.20
Shanga was a typical Swahili stone town located in the Lamu archipelago (fig. 4.1).
Chittick surveyed it in 1964, but undertook no further work, concentrating his activities at nearby
Manda, which he believed to have been founded by Persians from Siraf. Unlike Manda,
Shanga had its complete fourteenth century urban layout preserved, with some 200 houses, three
mosques, and around 500 stone tombs. The final population was around 3,000. Many of these
houses had plans that were similar to “ethnographically” observed Swahili houses of the eighteenth
and nineteenth centuries, including decorated wall niches. The nearby Swahili town of
Siyu had among its inhabitants a group that claimed to have been the survivors of Shanga who
had settled there when it was overrun by Pate. There was no doubt that Shanga was a fully
Swahili stone town.
The detailed layout of the town allowed us to target excavations to see how old Shanga really
was and how its plan developed over time. A series of excavations identified upwards of
twenty phases of occupation, with a total depth of 4 m of stratigraphy. The bottom levels, dating
to about 750 C.E., contained only posthole structures, several of which were circular huts. Of
particular interest was the local pottery, representing about 97% of the total, which had close
similarities to Iron Age pottery from the interior. One site showed particularly close similarities,
15. Stéphane Pradines, “Islamization and Urbanization
on the Coast of East Africa: Recent Excavations at
Gedi, Kenya,” Azania 38 (2003): 180–82.
16. H. Neville Chittick, “The East Coast, Madagascar,
and the Indian Ocean,” in The Cambridge History of
Africa, Volume 3, edited by Roland Anthony Oliver
(Cambridge, 1977), pp. 183–231, especially p. 218
is probably the last expression of this colonial view
of the Swahili.
17. Nurse and Hinnebusch, Swahili and Sabaki.
18. Randall L. Pouwels, Horn and Crescent: Cultural
Change and Traditional Islam on the East African
Coast 800–1900 (Cambridge, 1987).
19. Thomas Spear, “Early Swahili History Reconsidered,”
International Journal of African Historical
Studies 33/2 (2000): 257–89, covers many of the
developments during this period.
20. Mark C. Horton, Shanga: The Archaeology of a
Muslim Trading Community on the Coast of East
Africa (London, 1996); idem, “The Early Settlement
of the Northern Swahili Coast” (Ph.D. diss., University
of Cambridge, 1985).
MARK C. HORTON
71
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
Wenje, on the Tana River, so we defined its ceramics as “Tana tradition,” a term that now has
widespread acceptance.21
The Shanga sequence showed that the Swahili could be traced back to the eighth century
and that there was a continuous evolution of ceramics, architecture, and indeed culture. Combined
with what is known from other sites, this provides a 1,200 year sequence of Swahili
cultural history to the ethnographically observed present.
One criticism of the Shanga project is that it was focused too much on a single site, and the
opportunity was not taken to undertake survey of the Lamu archipelago and adjacent mainland.
While numerous mud and thatch houses were excavated, the focus of the research remained on a
stone town and its immediate environment. Recent research has moved away from this approach
to locate and excavate sites that largely contain mud and thatch houses and to recover the hinterland
of major sites, showing that they form part of a complex settlement pattern. Locating these
sites and their ceramic signatures has provided a different way of looking at Swahili origins.22
21. The connection between the coast and the interior
for these ceramics was first recognized in D. W.
Phillipson, “Some Iron Age Sites in the Lower Tana
Valley,” Azania 14 (1979): 155–60. The pottery is
also referred to as “triangular incised ware” because
of the dominant motif; Felix Chami, The Tanzanian
Coast in the First Millennium AD: An Archaeology
of the Iron-working, Farming Communities (Studies
in African Archaeology 7; Uppsala, 1994). Some
writers refer to it as triangular incised ware/Tana
tradition (TIW/TT).
22. Jeffrey Fleisher and Adria LaViolette, “Elusive
Wattle-and-Daub: Finding the Hidden Majority in
the Archaeology of the Swahili,” Azania 34 (1999):
87–108; Fleisher, “Viewing Stonetowns.”
ISLAM, ARCHAEOLOGY, AND SWAHILI IDENTITY
Figure 4.1. Shanga. Aerial photograph of fourteenth century ruins. The site was built over sand dunes
(visible in the foreground). The friday mosque is located in the center of the site and the
harbor is at the top of the figure
72
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
EARLY COASTAL SITES
These early Iron Age communities seem to have been maritime from an early date. Not only
do we have the evidence of the Periplus of the Erythraean Sea (ca. 40 C.E.) of the local inhabitants
using sewn boats, especially around the island of Menouthias, but now also a number of
finds of Early Iron Age Kwale pottery from Mafia (associated with a third century carbon-14
date), as well as Kwale and Koma Islands, showing that early Bantu-speaking farmers were also
fully able to cross the Mafia channel from the continent.23
Curiously, there is no transition to
Tana tradition on Mafia Island, suggesting that these communities died out by the sixth century.
There has been some uncertainly about the typological status of Tana tradition pottery. Initially
it was thought to belong to its own localized ceramic tradition, based along the northern
coast, from where Swahili culture was meant to have spread out along the whole eastern coast
(according to both the traditions, and the historical linguistics). However, surveys showed that
by the eighth century it already had a very widespread distribution, as far south as Mozambique,
and as far north as near Mogadishu, as well as to the offshore islands of Zanzibar and Pemba.24
It
was also found inland as far as the Rufigi River and up to 300 miles inland. Often associated
with ironworking and located in areas suitable for farming, it seemed likely that this was the
main type of early Iron Age pottery of coastal eastern Africa. Detailed statistical analysis of the
forms and decoration now show that Tana tradition develops from the early phases of Iron Age
pottery dating to the beginning of the first millennium.25
The typological shift from Kwale to
Tana dates to about 500 C.E. and surprisingly few Kwale sherds have been found in Tana sites.
It seems that the Tana sites represent a very different kind of Iron Age society to that of the
earlier Iron Age. The coastal sites are clearly exploiting the maritime resources on a massive
scale, with extensive fish and shellfish remains. They seem to be involved intensively in shell
bead production and ironworking, with some copperworking. The sites can be large, especially
on the islands of Zanzibar and Pemba, where deposits can cover up to 30 ha (the earlier Kwale
sites cover only a few hectares). Imports, mostly pottery, but also glass, and a few glass and
stone beads are found, indicating that these places are participating in long-distance trade. The
sites with Tana pottery in the interior seem to have a similar range of craft activity, although
relying more heavily on hunted and agricultural resources, while imports are far scarcer.
By the eighth century the eastern African coast was inhabited by quite large populations,
especially in particular areas such as around the islands of Zanzibar and Pemba and adjacent
mainland, in the Lamu archipelago, the Kilwa archipelago and mainland, and on the Comores.
These communities were engaged in the trade of the western Indian Ocean but seem to have
been equally successful exploiting maritime resources and maintaining connections with the hin-
terland.26
23. Lionel Casson, The Periplus Maris Erythraei: Text
with Introduction, Translation, and Commentary
(Princeton, 1989); F. A. Chami, “The Early Iron
Age on Mafia Island and Its Relationship with the
Mainland,” Azania 34 (1999): 1–10.
24. Horton and Middleton, Swahili, p. 42; Chami, The
Tanzanian Coast.
25. Richard Helm, “Conflicting Histories: The Archaeology
of Iron-Working, Farming Communities in the
Central and Southern Coast Region of Kenya”
(Ph.D. diss., University of Bristol, 2000).
26. Horton and Middleton, Swahili, pp. 42–46; on the
Kilwa hinterland, S. Wynne-Jones, personal communication.
The newly discovered site of Tumbe on
northern Pemba covers at least 30 ha; Fleisher and
LaViolette, “Elusive Wattle-and-Daub,” p. 103.
MARK C. HORTON
73
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ISLAMIC ARCHAEOLOGY IN EAST AFRICA
Modern archaeological research has tended to examine the African context and origins of
the Swahili and has focused upon these non-Islamic Tana tradition sites. The role of Islam in
Swahili origins has often been confused with ethnicity; Islam = foreignness, or even Islam = Arabs,
and attempts to identify early African Islam on the coast has diminished local cultural
achievements. Islam to many coastal archaeologists and historians was of little importance in the
formation of Swahili society — all the elements were already there in those Tana tradition settlements
of the coast.27
What, however, remains without dispute is that the acceptance of Islam by the coastal societies
represents the point at which they become different from the Iron Age communities of the
coastal hinterland and interior. Without exception every pre-nineteenth century mosque or tomb
lies within 1,000 m of the seashore, and Islam drew a very clear divide between the coast and
interior, which before the ninth century enjoyed a degree of cultural unity.
Shanga provides the best archaeological evidence for the origins of Swahili Islam (fig. 4.2).
The survey of the town showed that the settlement was arranged around a rectangular enclosure,
with a well at its center. This well dated to the very beginning of the site and continued in use
throughout its existence. The main mosque was attached to this well and so was also located in
the middle of the town. The study of the fabric of the mosque showed that it was a very complex
multiperiod building dating from about 1000 C.E. The floor rested on a platform of white sand
27. For example, the otherwise excellent Spear, “Early
Swahili History,” barely mentions the impact of Islamization
on coastal communities. See also
Chapurukha M. Kusimba, The Rise and Fall of
Swahili States (Walnut Creek, 1999) for an account
of the Swahili that omits virtually all reference to
Islam.
ISLAM, ARCHAEOLOGY, AND SWAHILI IDENTITY
Figure 4.2. Shanga. Friday mosque, showing the main wall of the prayer hall, with well and washing area
in the foreground. The roof was supported on pillars and made of thatch and timber,
echoing earlier styles of timber mosques
74
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
MARK C. HORTON
Figure 4.3. Excavations inside the prayer hall of the Friday mosque at Shanga. The plaster floor dates to
around 1000 C.E., while the earlier stone mosque dates to 900 C.E. The sand platform fill contains
numerous small silver coins, probably carried with the sand from the town’s trading beach
carried from the beach (fig. 4.3), containing datable imports as well as locally minted but minuscule
silver coins, giving two names, Muhammad and Abd Allah, and Arabic mottos.28
Below the platform were the remains of an earlier stone mosque, in this case with a square
prayer hall and a southern room and a western addition. This mosque has a mih≥rΩb but had a
28. Horton, Shanga, pp. 84–86; the earliest well layer
was found in phase one (ibid., p. 117). For the study
of the mosques, see ibid., chapters 9–10, and coins,
ibid., chapter 18.
75
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
different qibla line to the later mosque; it dated to about 900 C.E. This was not the earliest
mosque on the site but below it were a sequence of seven earlier timber mosques, several with
two phases, and again taking a different qibla line (fig. 4.4). The floors were made of a local
mud carried from the mangrove swamps, and the walls were also of mud, supported on either
posts or sticks set in closely set lines. In some cases the roof was supported on a central
ISLAM, ARCHAEOLOGY, AND SWAHILI IDENTITY
Figure 4.4. View of the excavations of the timber mosques at Shanga, showing remains of a ninth century
mosque made from small sticks. Note the central posthole. The other postholes
belong to later timber phases
76
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
posthole. Narrow southern rooms were a particular feature of several of the mosques. One of the
mosques comprised only a spread of small pebbles, and no apparent structure. A sequence of
carbon-14 dates gives a sequence from the later eighth century to the late ninth century, suggesting
that each phase lasted about fifteen years. The conclusion is that Shanga was Islamic from its
beginning or very shortly afterwards.
Needless to say this evidence has been controversially received. Allen refuses to accept anything
earlier than 900 as possible for the existence of Islam on the coast, as this does not fit into
his perceived notions of Swahili history.29
Peter Garlake seems unhappy with my interpretation,
as the area excavated was too small (although the full plans of the buildings were recovered),
the buildings do not face Mecca, and they lack mih≥rΩbs.30
But projecting mih≥rΩbs only become
common in the mid-ninth century (e.g., the first timber mosque at Shanga that has a possible
mih≥rΩb, has a large post on the qibla wall dating to the late ninth century). The orientation can
be carefully explained through a changing geographical knowledge of the relative locations of
East Africa and Arabia. The pattern of consistent correction over time, plus the discovery of Islamic
burials on the earliest qibla line demonstrate that this is where they thought the direction
of Mecca lay. The clinching evidence comes from the use of the cubit, which is precise, and is
518 mm for the timber mosques, 540 mm for the first stone mosque, and 534 mm for the final
mosque — a sequence that reflects usage in the Islamic world as a whole (fig. 4.5).31
Others have suggested that the Shanga mosques were built for the use of visiting merchants
and not for the local inhabitants, who it is supposed continued to follow traditional religious
practices. So to Michael Pearson, the mosque was “for the use of visiting or even resident Arab
merchants; the local population was converted later”;32
to David Whitehouse, the mosque was
used by “an elite minority of Muslims from Western Asia.”33
However, this is unlikely for a
number of reasons. Firstly, the mosque was located in the very center of the site, in what has
been reconstructed as central enclosure, largely empty of habitation. It was carefully placed over
a large tree stump that had been deliberately burnt, while the natural sand contained flecks of
charcoal, indicating that this area had been cleared of vegetation. This area may have been a
sacred grove that was cleared to accommodate the new mosque. A mosque for visitors would be
located at the edge of the settlement, not in the center, and not at such an important location.
But there is other evidence, which comes from the two earliest burials, dating to the early
ninth century. One was of a child, and the second is likely also a child. The silver coins that
seem to have been locally minted in the ninth century give the names Muhammad and Abd Allah,
which are often associated with recent converts to Islam (fig. 4.6). A small seal stone found
in these early levels also proclaims the Islamic nature of the inhabitants.34
There does, however, remain a problem, as the small timber mosques had a capacity of between
ten and twenty-five worshippers, while estimates of the total population of early Shanga
would have been several hundred. So it has been assumed that the mosque served only an elite
group of Muslims. Mosque capacities of about 10% of the total population is nonetheless a con-
29. Allen, Swahili Origins, p. 36, note 4.
30. Peter S. Garlake, Early Art and Architecture of Africa
(Oxford, 2002), p. 170.
31. Horton, Shanga, pp. 227–28.
32. Michael N. Pearson, Port Cities and Intruders: The
Swahili Coast, India and Portugal in the Early Modern
Era (Baltimore, 1998), p. 15.
33. David Whitehouse, “East Africa and the Maritime
Trade of the Indian Ocean, AD 800–1500,” in Islam
in East Africa: New Sources: Archives, Manuscripts
and Written Historical Sources, Oral History, Archaeology,
edited by Bianca S. Amoretti (Rome,
2001), pp. 411–24, p. 417.
34. Horton, Shanga, p. 118; for the seal ring, see ibid.,
p. 357, fig. 275g.
MARK C. HORTON
77
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ISLAM, ARCHAEOLOGY, AND SWAHILI IDENTITY
Figure 4.5. Plan of the sequence of mosques excavated at Shanga, dating from 750 to 1000 C.E.
78
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
sistent figure on the Swahili coast. The final mosque at Shanga held about 300 worshippers (309
sq. m) but the total urban population was about 3,000. In Lamu, the Friday mosque could contain
about 500 worshippers, but at its peak the town had a population of around 18,000. In 1985,
the stone town contained 4,563 inhabitants, and this must be close to the non-slave population in
the nineteenth century.35
Clearly every adult male does not attend the Friday sermon nowadays,
nor did so in the past.
The archaeological evidence suggests that Islam was widely followed by the Shanga population
in the late eighth century and was not confined to an elite ruling (or mercantile) group.
This was a Muslim settlement made up from African converts — converts who seem to have
accepted Islam within a decade or so of the settlement’s foundation.
THE ACCEPTANCE OF ISLAM
The process of conversion of the coastal communities remains controversial, as this is something
that can also be inferred and partly reconstructed from the historical sources, although
often crucial pieces of evidence are missing to make up the complete picture. The archaeological
evidence does allow in certain circumstances for us to go beyond the simplistic recognition of
Islam, to more specific identification of Shiºite, Sunni, or Ibadi presence, groups with whom the
Swahili had active trading connections from the eighth century. This does provide a way of understanding
some of the historical evidence within its Islamic context. This I have done
MARK C. HORTON
35. Francesco Siravo and Ann Pulver, Planning Lamu:
Conservation of an East Africa Seaport (Nairobi,
n.d.), p. 63.
Figure 4.6. Minuscule silver coins from Shanga (diameter 9 mm) dating to the ninth century
79
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ISLAM, ARCHAEOLOGY, AND SWAHILI IDENTITY
elsewhere in detail,36
where my discussion has been much criticized by historians and archaeolo-
gists.
Timothy Insoll dislikes any attempt to develop these interpretations or my attempts to make
sense of the complex historical record, preferring a rather simplistic “phased” model of conversion,
which disregards the Swahili’s own oral traditions and chronicles, external sources written
in the Islamic world, or indeed a considerable amount of modern historical scholarship on the
subject. Recently Justin Willis has echoed these sentiments as “too much willingness to pursue
entertaining speculations based upon minimal evidence, notably regarding the alleged presence
of Shiºism along the coast in the early Islamic period.”37
On the other hand, the well-read African
Islamic historian Randall Pouwels accepts that the “likelihood that heterodoxy existed in
these regions at this time seems high; therefore these accounts have some ring of truth about
them.”38
The main difficulty is that the Swahili have largely followed orthodox Sunni practices since
at least the fourteenth century.39
In the nineteenth century the Omani Arabs introduced Ibadism,
and Indian traders introduced Ismaili and Bohra Islam, but this had little impact beyond their
own communities. The strength of Sunni practices was established by a complex pattern of migration
from around 1200 of sayyids and sharifs from southern Arabia. Mogadishu was one
important center, with the Qahtani Waºil and Fakhr ad-Din clans established from the early thirteenth
century. The Mahdali clan, who originated from the Wadi Surdud, were at Kilwa by 1280,
when they became sultans but were also in Barawa, Lamu, Pate, and Mombasa. Ibn Battuta, who
visited eastern Africa in 1331, noted that Mombasa and Kilwa followed Sunni-Shafi Islam.40
Another group of sharifs settled in Pate, Lamu, and the Comores in the late sixteenth century,
followed by others to Pemba and Zanzibar Islands.
This was not a large-scale migration of Arabs to East Africa, but often single individuals
who settled and established substantial lineages. But the impact was the removal of sectarian and
non-Islamic practices and the development of the coast as a center of Islamic learning and sanctity.
It is not surprising that there is little evidence of early Ibadi and Shiºite Islam in present day
Swahili Islam.41
It is also likely that many of the Swahili chronicles were composed within these
Sunni communities so that accounts of early heterodoxy are garbled and confused. This is a
complex early history that archaeological evidence is now beginning to recover.
A good example of how archaeological evidence can reconstruct religious affiliation can be
seen with the Ibadis. Mosques constructed by Ibadi settlers from Oman in nineteenth century
Zanzibar contain very simple mih≥rΩbs, with little decoration, set within the thickness of the qibla
wall.42
The theological justification for this is that within the Ibadi rite, no one person should
lead the prayers (excepting the elected Imam in Muscat), so the mih≥rΩb space should not allow
36. Horton and Middleton, Swahili, pp. 47–71; Mark C.
Horton, “The Islamic Conversion of the Swahili
coast 750–1500,” in Islamic in East Africa: New
Sources, edited by Bianca S. Amoretti (Rome,
2001), pp. 449–69.
37. Justin Willis, “Review of The Swahili, by Mark C.
Horton and John Middleton,” Azania 38 (2004):
207–08.
38. Randall L. Pouwels, “The East African Coast, c.
780–1900 CE,” in The History of Islam in Africa,
edited by Nehemia Levtzion and Randall L. Pouwels
(Athens, 2000), pp. 251–72, p. 256.
39. Horton and Middleton, Swahili, pp. 68–70.
40. Freeman-Grenville, Select Documents, pp. 27–32.
41. Thomas Spear, “Review of The Swahili, by Mark C.
Horton and John Middleton,” International Journal
of African Historical Studies 33/3 (2000).
42. A. Sheriff, “Mosques, Merchants and Landowners
in Zanzibar Stone Town,” Azania 27 (1992): 1–20;
Helen Little, “The Nature of the Zanzibar Mosque”
(Ph.D. diss., University of Brighton, 1995).
80
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
MARK C. HORTON
this to happen. Similar concerns result in their mosques rarely containing minbars or tower minarets,
which are still rare in coastal Swahili architecture (fig. 4.7).
A number of these features can be found in early mosques. The ruined mosque at Sanje ya
Kati near to Kilwa, a site dating to the eleventh–twelfth century has the mih≥rΩb set within the
thickness of the wall built to double thickness to accommodate it.43
Two examples are also
known from Pemba Island. At Ras Mkumbuu, an excavated early mosque of the tenth century
(sealed below an eleventh century mosque) had a very similar plan (fig. 4.8), with a double
thickness qibla wall, although the area of the mih≥rΩb itself had been robbed out.44
Excavations in
2004 at the north Pemban site of Chwaka located a twelfth century mosque sealed below a fifteenth
century building that also contained a mih≥rΩb niche contained within the thickness of the
wall. Perhaps the most interesting example is at Tumbatu, where the Friday mosque has a large
conventional mih≥rΩb (fig. 4.9) but with a subsidiary mih≥rΩb at the end of an added (probably in
the early thirteenth century) eastern side chamber, which was set within the thickness of the
wall. This suggests the presence of a minority community of Ibadis. The historical record provides
a possible circumstance for this. At exactly this time the Ibadi community at Kilwa seems
to have lost influence, and the sultan of Kilwa, Hasan bin Sulaiman (who may have been an
Ibadi), was exiled in the “land of Zanzibar” for fourteen years. The side chamber may have been
specially built for the refugee ruler and his followers.
Figure 4.7. Staircase minaret at the eleventh century mosque of Kaole (Tanzania). The absence of tower
minarets in East Africa may indicate Ibadi influence in mosque design
43. Neville Chittick, “Kilwa, a Preliminary Report,”
Azania 1 (1966): 1–37, especially p. 30.
44. Horton and Middleton, Swahili, p. 66; Mark C.
Horton, Zanzibar and Pemba: The Archaeology of
an Indian Ocean Archipelago (London, forthcom-
ing).
81
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ISLAM, ARCHAEOLOGY, AND SWAHILI IDENTITY
The presence of Shiºites is more difficult to locate. There seems to be few specific features
in mosque architecture to go on. There are two styles of mih≥rΩb that seem to be broadly contemporary
with one another and may reflect differences in religious doctrine: one that is very
elaborate and highly decorated with inscriptions and carved coral (and later Chinese bowls), and
a second that is plain with minimal decoration, which may be associated with Sunni-Shafi com-
munities.
Coins possibly provide an alternative indicator with their names and mottos, and these have
been minted since the ninth century (fig. 4.10). The Mtambwe series are of interest as several of
the names are of Old Testament origin, a possible indicator of Shiºite associations, while one of
the early Kilwa coins has a motto that translates as “Trusts in the Lieutenant of God” — a clear
Shiºite meaning. The small silver coins from early Shanga have technical similarities to both the
coins produced by the (Shiºite) Zaidites of southern Arabia and to those made by (Sunni) Amirs
of Sind and the nearby (Ismaili) Fatimids of Multan.45
In reality, very little is known of these
minuscule issues, which have been rarely recovered from excavations and may well have been
even more widely produced around the trading communities of the western Indian Ocean.
It is all too easy to dwell on the various comings and goings of different religious groups on
the Swahili coast, without appreciating the underlining continuities that were certainly present.
There is a strong suspicion that religious doctrine was strongly allied to political action (as
seems to have been the case in early thirteenth century Kilwa), which is recorded in the various
chronicles and traditions. But the actors were all Swahili Muslims, who played out (and indeed
continue) the controversies and conflicts of the Islamic world within their own micro-world of
coastal Islam. The real archaeological message is of continuity from the conversions from eighth
century onwards. So mosques retain the same basic proportions as the early timber mosques of
Figure 4.8. Excavations of the mosque sequence at Ras Mkumbuu (Pemba Island). Below the floor of a
large mosque dating to 1050 are the remains of a small stone mosque, the north wall of
which had been robbed out, but was of double thickness, suggesting an Ibadi-style
mih≥rΩb. Below is an earlier timber building, which may have been the
first mosque on the site dating to the early tenth century
45. H. W. Brown, “Coins of East Africa: An Introductory
Survey,” Yarmouk Numismatics 5 (1993): 9–
16; “Early Muslim Coinage in East Africa: The Evidence
from Shanga,” Numismatic Chronicle (1992):
83–87; “The Coins,” in Horton, Shanga, pp. 368–
77; Mark C. Horton, H. W. Brown, and W. A. Oddy,
“The Mtambwe Hoard,” Azania 21 (1986): 115–
23; Nicholas M. Lowick, “Fatimid Coins of
Multan,” Numismatic Digest 3 (1983): 62–69.
82
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
MARK C. HORTON
Figure 4.9. A simple Ibadi mih≥rΩb in a small contemporary mosque at Tumbe (Pemba Island).
The niche is marked in plaster, but no recess is in the wall
Figure 4.10. Part of the Mtambwe hoard (Pemba Island). The associated dinars give a deposition date
after 1066 C.E. Over 2,000 silver coins were found, locally minted in the names of ten local rulers
83
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ISLAM, ARCHAEOLOGY, AND SWAHILI IDENTITY
Shanga. Southern rooms, another early feature, continue in use in contemporary mosques. An
echo of the timber and thatch origins of Swahili mosques is reflected in wall pilasters, unnecessary
thatched roofs, and internal round timber columns. Mih≥rΩb designs dating to the twelfth
century inspire examples in the eighteenth and the nineteenth century as well as contemporary
buildings (figs. 4.11–12). Minarets are still absent on mosques, except where money from the
Gulf has been used to impose foreign styles — styles that are often poorly received locally.
SWAHILI SPIRITS
As the sharif movement swept away heterodoxy, so also were traces of surviving non-Islamic
practices removed. Within the archaeological record there is very little trace of syncretism of
African religious practices within coastal Islam. Indeed most of the evidence for such practice
comes from modern contexts, where spirits play an important part in Swahili religion (fig. 4.13).
There are two main categories of spirit, mizimu and majini. Mizimu are considered more African
and linked to landscape and ancestors. They are given small offerings and it is hoped that they
will help the owner or farmer, or at least not make her or him ill. Majini are more Arabian and
linked to Islamic systems of belief, however loosely. Mapepo are one type of majini, which can
have personal names and can possess people and have associations with ruined settlements, especially
mosques and tombs. Most Swahili do not consider that their belief in spirits is
incompatible with the Islamic faith.
It appears that belief in spirits expanded in the mid-nineteenth century in response to the
settlement of slaves from the African interior, and they have become much more widespread
since the abolition of slavery in the late nineteenth century. Particular locations are well known
Figure 4.11. The mih≥rΩb at the thirteenth century mosque of Ras Mkumbuu (Pemba Island) is typical of
many examples made from carved coral but with very simple decoration
84
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
MARK C. HORTON
for spirit activity and perhaps the most well known is Pemba Island.46
It is here that early evidence
for spirits can be suggested.
Figure 4.12. The very elaborate mih≥rΩb at Kizimkazi (Zanzibar Island) that has a date of A.H. 500
(1107 C.E.) and which is now considered to be largely of this date. It is very much more
complex than many Swahili mih≥rΩbs, and these may reflect doctrinal differences
46. Linda Giles, “Spirit Possession on the Swahili
Coast: Peripheral Cults or Primary Texts?” (Ph.D.
diss., University of Texas at Austin, 1989); Horton
and Middleton, Swahili, pp. 70–71, 190–94. For an
early description of spirits on Pemba, see William
H. Ingrams, Zanzibar: Its History and Its People
(London, 1931), pp. 435–40.
85
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ISLAM, ARCHAEOLOGY, AND SWAHILI IDENTITY
At the fifteenth century enclosed site of Pujini, excavations by Adria LaViolette studied a
man-made cave, entered down steps. A carved siwa or side-blown horn marked the entrance
(fig. 4.14).47
When it was excavated in the early 1900s, the fill was described as “full of bones
of oxen, broken pottery sherds, remnants of wood fires, and two fragments of lamps of coarse
earthenware.”48
LaViolette suggests that this was a spirit cave, specially constructed within this
palace-like complex. Spirits were often believed to live in underground caves, which are common
in the limestone formations of the coast and islands, and this seems to be the most
convincing explanation for the Pujini shrine.49
In 2004 we were able to excavate an intact deposit that was placed within the mih≥rΩb of a
ruined mosque at Chwaka in northern Pemba (fig. 4.15). The deposit was placed on the plaster
floor, sealed by loam deposits,50
and comprised of three elements: a cooking bowl, which had
two large pieces of two other vessels covering it, nearby was a hand-grinding stone, and a seashell.
In the same deposit was a sherd of Chinese pottery, taken from the bowls that decorated
the mih≥rΩb (fig. 4.16). This gives an interesting date as all the bowls had been removed by 1920
and the mih≥rΩb was filled with rubble.51
The style of the local pottery that makes up the deposit
Figure 4.13. Baobab tree located at Mbaraki (Mombasa Island, Kenya) that is believed to be a location of
spirits, which has been festooned in rag cloths. In the background is the stone tower that
was attached to a small mosque, and which may have given the location its reputation
47. Adria LaViolette, “Swahili Archaeology on Pemba
Island, Tanzania, Pujini, Bandari ya Faraji and
Chwaka, 1997/1998,” Nyame Akuma 44: 59–65.
48. Francis B. Pearce, Zanzibar: The Island Metropolis
of Eastern Africa (London, 1920), p. 381.
49. Ingrams, Zanzibar, p. 437.
50. I am very grateful to Drs. Fleisher and LaViolette
for inviting me to participate on their project at
Chwaka/Tumbe in July 2004, and for allowing me
to mention this discovery in advance of their own
publications.
51. Pearce, Zanzibar, p. 397.
86
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
MARK C. HORTON
Figure 4.15. Deposit within mih≥rΩb of mosque at Chwaka (Pemba Island), comprising a bowl, covering
sherds, a seashell, a hand-grinding stone, and a piece of Chinese pottery
Figure 4.14. Siwa (or side-blown horn) decorating the “shrine” at Pujini (Pemba Island) made in plaster
relief. This underground chamber may represent a “spirit cave”
87
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
ISLAM, ARCHAEOLOGY, AND SWAHILI IDENTITY
Figure 4.16. Detail of the mih≥rΩb deposit from Chwaka
is probably seventeenth or eighteenth century. This pot contained the spirit (hence it had to be
covered and contained) and this is part of the tradition of the burial of fingo pots found all along
the coast, which contained prophylactic magic to protect special places such as doorways or
gateways. The shell is an offering made to a sea spirit, while the grinding stone referred to a
domestic spirit. The offering was placed with a mih≥rΩb recess that was similar to a cave; the
mosque itself must have gone out of use and was already much ruined.
ISLAMIC ARCHAEOLOGY IN EASTERN AFRICA
Recently, Timothy Insoll has encouraged us to study the archaeology of Islam, both globally
and within sub-Saharan Africa, and certainly his compendium of material allows us to look at
Islamic practice within the wider study of material culture and society. Yet there remains the
considerable difficulty faced by Africanists as to whether we should separate out Islamic societ-
88
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
MARK C. HORTON
ies from their surroundings in either space or time. In the case of the Swahili, this is a very real
problem; some, including many Swahili themselves, would claim that their Swahili identity was
only formed when they accepted the Islamic faith. Archaeologists and historians prefer to make
no such distinction, seeing a seamless transition from pre-Islamic times, and claim that Islam
had little impact on a society that already had many features akin to Islam, such as the belief in a
single supreme god, and that conversion “demanded little intellectual change — something closer
to a reorientation of spiritual solemnization than to a conceptual shift.”52
One wonders whether the real reason behind this view is that it remains very difficult to
recognize Islamic or indeed non-Islamic societies through the archaeological record. Some sites
are clearly non- (or pre-) Islamic as they are simply too early in date to be Islamic, while others
(e.g., Unguja Ukuu on Zanzibar) spanning the sixth–ninth centuries are likely to have undergone
conversion, but this is not really visible in the material culture of the sites. It may be possible to
infer changes in pottery types, for example, the shift away from large jars that may have been
used for making beer, to smaller bowls and tablewares,53
but at present the location of mosques
and burials remain the only reliable indicator of conversion.
There is also a similar issue when we view these societies spatially. By focusing on the Islamic
communities alone, we tend to draw a distinction from those that remain non-Islamic.
However, there have been complex relationships within the hinterland that have existed for at
least a millennium and probably from the first settlement of the coastal region by farmers 2,000–
3,000 years ago. When we ignore these relationships we fall into the sort of narratives that the
colonial historians were writing about the civilized coastal communities and the impenetrable
“nyika” forest that blocked all communications with the interior. But this was only a barrier to
Europeans trying to build railways and roads, and the new archaeological surveys are showing
just how much connections did exist along the entire coastline.
The conclusion that we need to make is that Islam was important, but that when we study it
in sub-Saharan Africa we have to do so within the framework of a complex understanding of
African society as a whole. To archaeologists, this challenge is particularly acute.
52. Pouwels, “East African Coast,” p. 254; Horton and
Middleton, Swahili, pp. 176–78.
53. Fleisher, “Viewing Stonetowns,” pp. 264–65, 416–
18.
89
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
5
SYNCRETISM, TIME, AND IDENTITY:
ISLAMIC ARCHAEOLOGY IN WEST AFRICA
TIMOTHY INSOLL
School of Art History and Archaeology,
University of Manchester
INTRODUCTION
This paper seeks a brief introduction to the research this author has been completing in the
West African state of Mali over the course of the past decade. However, what is presented is not
intended to be a fieldwork report, as the primary results have already been published extensively
elsewhere.1
Rather, the focus is upon presenting the relevance of these results within their wider
context, both in West Africa, and to a lesser extent, sub-Saharan Africa as a whole.2
That the archaeological study of Islam should involve more than a focus upon the explicit
indicators of the religion, for example, mosques, inscriptions, and artwork, should now be regarded
as commonplace. The influence of religion can be all-encompassing and can potentially
influence all aspects of life, including diet, dress, domestic architecture, landscape, and settlement
form.3
This statement obviously does not apply to Islam alone but can be applicable to
many other religions as well.4
Yet the potential of religion as a primary structuring agent for the archaeological record is
often ignored. Two factors can be suggested for the reticence in recognizing religion as a potential
superstructure into which all other aspects of life might be placed. Firstly, it can be
suggested, a problem lies with the term “religion” itself. Archaeologists appear frightened of using
it as a descriptive device, and hence recourse is made to “ritual” where such material is
considered, “ritual” being the archaeologists’ favorite catchall category for “odd” or otherwise
not understood behavior.5
Hence the implications of “religion” as a term appears little understood,
be it in application to Islam or otherwise. But besides the definitional conundrum that it
generates — when is archaeological material “religious” as opposed to “ritual” in nature, for example
— it can be further suggested that the frequent absence of religion in archaeological
interpretation, certainly as an all-inclusive structuring agent, is also perhaps a reflection of many
1. Timothy Insoll, Islam, Archaeology and History, A
Complex Relationship: The Gao Region (Mali) ca.
AD 900–1250 (Cambridge Monographs in African
Archaeology 39; British Archaeological Reports,
International Series 647; Oxford, 1996); idem, Urbanism,
Archaeology and Trade: Further
Observations on the Gao Region (Mali): The 1996
Fieldseason Results (British Archaeological Reports,
International Series 829; Oxford, 2000);
idem, “The Archaeology of Post Medieval
Timbuktu,” Sahara 13 (2002): 7–22.
2. See also Timothy Insoll, The Archaeology of Islam
in Sub-Saharan Africa (Cambridge, 2003).
3. Timothy Insoll, The Archaeology of Islam (Oxford,
1999).
4. See Timothy Insoll, editor, Archaeology and World
Religion (London, 2001).
5. Timothy Insoll, Archaeology, Ritual, Religion (London,
2004).
89
90
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
archaeologists’ worldview themselves; often largely a secular one. Hence in turn this might be
projected onto the past, even if inappropriate.
Thus with regard to Islamic archaeology, the resulting interpretation might acknowledge religion,
Islam, as significant but reduce its importance to the types of overtly “religious” material
culture already described, rather than providing the holistic examination that Islamic belief and
practice frequently demand. In support of this one need only draw upon the frequently cited
maxim, “Islam is more than a religion but a way of life.” Recovery of the latter was precisely the
focus of the archaeological research completed in Mali: the potential impact of Islam as a strucTIMOTHY
INSOLL
Figure 5.1. The location of Gao (and other centers) in West Africa
91
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
turing agent for all facets of life, but with the allied aims of also assessing, largely via archaeological
evidence, how Islam has helped in the creation of overall social identity in this region,
and crucially, fused with African traditional religions in a syncretic process to create African
Islam(s).
THE ARCHAEOLOGICAL PROJECT
Archaeological excavations and survey have been focused upon Gao, a major city on the
bend of the Niger River in the Sahel region in the east of Mali, and to a lesser extent, the city of
Timbuktu, situated some 10 km north of the Niger River on the fringe of the Sahara Desert (fig.
5.1). Both cities were extensively involved in trans-Saharan trade in what would be called in
European contexts, the medieval period, but is more correctly referred to in African contexts as
the Middle and Late Iron Age. For reasons of descriptive simplicity the term “medieval” will be
used here.
SETTLEMENT
Whereas Timbuktu is a name familiar to many, perhaps as a synonym for the remote and
mysterious, even if they are unsure whether it is a real place and where exactly it is located, Gao
is little known outside of West Africa. However, in terms of historical importance it can be argued
that Gao is much more significant than Timbuktu, being the capital of the Songhai empire,
the last of the three great medieval West African empires of Ghana (ca. eighth–eleventh centuries
C.E), Mali (twelfth–fourteenth centuries C.E), and Songhai (fifteenth–sixteenth centuries
C.E). Yet the origins of Gao predate its high point under the Songhai empire. It seems in fact to
have been founded in the sixth–seventh centuries C.E, thus before Muslim merchants and missionaries
began to reach the Western Sahel, a process beginning in the late eighth century when
historical sources record contacts between the Ibadi imamate of Tahert (in modern Algeria) and
Gao.6
Consequently the city was indigenously founded, probably by what are best termed the
proto-Songhai,7
contrary to earlier interpretations that might have seen innovations such as urbanism
as externally derived from North Africa, and transmitted across the Sahara via Muslim
traders.8
This notion of indigenous origins is a factor of importance in reconsidering the whole
concept of Islamic identity in the region, in its indigenous nature, or rather how Islam was indigenously
adapted to fit local requirements; it is seemingly a crucial element in the initial
acceptance of Islam not only in this region, but in a pattern certainly mirrored through much else
of sub-Saharan Africa as well.9
At Gao, archaeological evidence for conversion to Islam prior to about the beginning of the
tenth century is absent. Thereafter there is an increase in archaeological evidence attesting to
contacts with the Muslim world. Items such as glazed pottery, glass vessels, glass beads, and
6. T. Lewicki, “Part 1: The Ibádi Community at Basra
and the Origins of the Ibádite States in Arabia and
North Africa, Seventh–Ninth Centuries. Part 2: The
Ibádites in North Africa and the Sudan to the Fourteenth
Century,” Cahiers d’Histoire Mondiale 13
(1971): 51–130.
7. Insoll, Islam, Archaeology and History.
8. See, for example, Raymond Mauny, Tableau
géographique de l’Ouest africain au Moyen Age,
d ’après les Sources écrites, la tradition et
l ’archéologie (Mémoires de l’Institut français
d’Afrique noire 61; Dakar, 1961).
9. Insoll, Sub-Saharan Africa.
SYNCRETISM, TIME, AND IDENTITY
92
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
brass metalwork, goods obtained via Muslim controlled trans-Saharan trade directed through
North Africa and Egypt are all found at sites such as Gao.10
Secondly, the first direct evidence
for the presence of Muslims in the Western Sahel is found after this date; evidence already mentioned
such as mosques, Muslim burials, and the remains of domestic structures that might have
been built according to Muslim social requirements.11
Initially, settlement at Gao appears to have been separated; Muslims at the tell site of GaoSaney,
non-Muslims at Gao proper, specifically at the sites of Gadei and Gao Ancien placed
some 7 km away from Gao-Saney and situated close to the Niger River (figs. 5.2–3). But the
existence of dual settlements persisted after Islam had spread to the inhabitants of Gao Ancien in
the late eleventh–twelfth centuries and was possibly due to security concerns — keeping potential
nomad raiders at a distance from Gao Ancien, certainly a factor of later consequence.12
TRADE
A reason for these safety concerns would have been, conceivably, protecting wealth gained
through trans-Saharan trade. Trade was the agency, as noted, by which Islam was initially transferred
to the Western Sahel, either by the merchants themselves or by missionaries
accompanying, preceding, or following them. A wide range of evidence has been recovered
from the excavations at Gao attesting to the operation of this trade, the bulk of which is from
Gao Ancien. Besides allowing an insight into potential trade partners with which the merchants
of Gao were connected on the other side of the Sahara, this evidence also indicates that the trade
seems to have been far removed from a colonial-type system run by groups of North African
merchants. Because, as with much of the other evidence recovered, it reinforces the picture of
the indigenous nature of the city, its control, and its operation.
10. Insoll, Islam, Archaeology and History; idem, “Islamic
Glass from Gao, Mali,” Journal of Glass
Studies 40 (1998): 77–88; idem, Urbanism.
11. See Insoll, Archaeology of Islam.
12. Timothy Insoll, “Iron Age Gao: An Archaeological
Contribution,” Journal of African History 38/1
(1997): 1–30.
TIMOTHY INSOLL
Figure 5.2. Settlement structure at Gao
93
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
Possibly the most spectacular evidence for the export trade found was a cache of over fifty
hippopotamus tusks uncovered in a context dated to the mid-ninth century in Gao Ancien. The
hypothesis has been advanced that these tusks represent a consignment of ivory placed on beams
within the pit in which they were found, and which was awaiting shipment to the ivory workshops
of North Africa but was never sent for reasons that remain unclear.13
A substantial ivory
trade certainly existed between West and North Africa but is little mentioned in the Arabic
sources, possibly due to disapproval from more orthodox Muslims, who likewise condemned the
use of feathers, horns, hoofs, or tusks derived from animals that were not ritually slaughtered.14
For example, the North African jurist Ibn Abi Zayd al-Qayrawani (d. 996) specifically mentions
that the use of elephant’s tusk is expressly disapproved.15
Nevertheless, large quantities of ivory
were certainly used in the workshops of the Maghreb, Islamic Spain, and Egypt, and it is mentioned
in Arabic herbals (such as Ibn al-Baytar and Al-Antaki) as having a variety of beneficial
qualities.16
Evidence for the extensive gold trade that is recorded in the Arabic historical sources is
more elusive, which is not surprising considering its ease of recyclability as a material of enduring
value. In Gadei a small gold bead was found, while the discovery on the surface at Gao
13. Timothy Insoll, “A Cache of Hippopotamus Ivory at
Gao, Mali, and a Hypothesis of its Use,” Antiquity
69 (1995): 327–36.
14. Nehemiah Levtzion and J. F. P. Hopkins, Corpus of
Early Arabic Sources for West African History,
translated by J. F. P. Hopkins (Fontes Historiae Africanae,
Series Arabica 4; Cambridge, 1980), p. 55.
15. John O. Hunwick, “Comment by John Hunwick,”
Saharan Studies Association Newsletter 2 (1994): 11.
16. I am grateful to a reviewer of this paper for pointing
this out.
SYNCRETISM, TIME, AND IDENTITY
Figure 5.3. View of the Niger River from Gao. Photograph by Timothy Insoll
94
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
Ancien of a gold mithqal coin of North African origin dating from 952–975 has been reported.17
Equally, evidence for the historically documented slave trade has also proven elusive. Much
more abundantly testified are items sourced from North Africa, the Near East, and in certain instances,
even further afield. These include glass and glazed pottery, primarily recovered from
Gao Ancien and dating from the eleventh to twelfth centuries.18
Other trade commodities found
include hundreds of glass beads, carnelian beads, cowry shells, and brass ingots (fig. 5.4), the
latter a material not produced in sub-Saharan Africa and hence a good indicator of long-distance
trade.
Overall, the picture of trade, though skewed in favor of more durable commodities for obvious
reasons of archaeological survival, is broadly in agreement with that recorded in the Arabic
historical sources.19
If this evidence were to be considered on its own or solely in association
with the fired-brick architectural tradition recorded in Gao Ancien, then it might be suggested
that Gao was a colonial entrepot run purely to satisfy the requirements of a presumably foreign
elite. However, the approach that has been adopted here, that is, treating Gao in its entirety and
maximizing what can be gained from all aspects of the archaeological record, indicates that such
a premise is flawed, as noted earlier. This was, admittedly, a trade controlled by elites, as might
be expected in the capital of what was to become the Songhai empire, but one controlled by a
local elite, many of whom had converted to Islam.
INSCRIPTIONS
Initially, Muslim merchants appear to have occupied Gao-Saney, a site complex comprising
a tell and associated Muslim cemeteries and tombs, which has not been as intensively investigated
as Gao Ancien.20
The first direct evidence for the presence of Muslims in this region was
found here and consists of various inscribed Muslim tombstones that date from ca. 1100–1300.
Five of these stelae found at Gao-Saney were imported ready-carved from Muslim Spain in the
early twelfth century. These bear both names and dates, and the marble used seems to be from
the vicinity of Almeria on the southern Spanish coast.21
Yet besides providing further evidence for the operation of trans-Saharan trade between Gao
and Almoravid-controlled Spain, these imported stelae also indicate that they were used for purposes
other than commemorating the dead because they seem also to have been used to proclaim
the new found faith of Islam. We can discern this as three of the kings commemorated on the
stelae in the cemetery at Gao-Saney, including two of the imported Spanish examples, were recent
converts to Islam. Their new identity, and indeed their piety, were clearly shown by
17. J. Latruffe, “Au Sujet d’une Piece d’Or Millénaire
Trouvée à Gao,” Notes Africaines 60 (1953): 102–03.
18. Insoll, Islam, Archaeology and History.
19. See Levtzion and Hopkins, Corpus.
20. C. Flight, “Gao 1972: First Interim Report: A Preliminary
Investigation of the Cemetery at Sané,”
West African Journal of Archaeology 5 (1975): 81–
90; idem, “Gao 1974: Second Interim Report. Excavations
in the Cemetery at Sané” (Birmingham
[manuscript on file, Centre of West African Studies,
University of Birmingham], 1978); idem, “Gao
1978: Third Interim Report: Further Excavations at
Sané” (Birmingham [manuscript on file, Centre of
West African Studies, University of Birmingham],
1979).
21. Jean Sauvaget, “Les épitaphes royales de Gao,” Bulletin
de l’Institut français de l ’Afrique noire 12
(1950): 418–40; M. M. Vire, “Notes sur trois
épitaphes royales de Gao,” Bulletin de l ’Institut
français de l’Afrique noire, Series B 20 (1958):
368–76; P. F. de Moraes Farias, “The Oldest Extant
Writing in West Africa: Medieval epigraphs from
Issuk, Saney, and Egef–n–Tawaqqast (Mali),” Journal
des Africanistes 60 (1990): 65–113.
TIMOTHY INSOLL
95
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
successively adopting the name of the Prophet and of the first two caliphs, Abu Bakr and
Umar.22
Neither were stelae confined to Gao-Saney. Gao Ancien was also ringed with Muslim cemeteries,
and the inscriptions recovered from these have also provided important information on
Islamization processes within the region. At one of the cemeteries, Gorongobo, to the northwest
of Gao Ancien, several Muslim tombstones inscribed in Arabic and dating from between 1130
and 1306 were recorded. One tombstone dated to 1210 appears to bear the female Songhai
names of either Waybiya or Buwy, depending on the reading. This is of significance as it is first
and foremost a Muslim tombstone, bearing a local name, and female as well, and conclusive
proof for local conversion to Islam by the early thirteenth century.23
This further supports the
interpretation proposed earlier that we are witnessing indigenous control of trade; trade adapted
to suit a local elite, many of whom were Muslim.
This “indigenization” of Islam was also apparent in various other aspects of the archaeological
evidence recovered; evidence, moreover, indicating that the impact of Islam was being felt to
varying degrees in many aspects of life. The archaeology indicates a picture of complexity, as
evident in the internal and intra-site patterning, reflecting the fact that what was being recovered
was the residue of both past communities and individuals with varying degrees of adherence to
22. C. Flight, “Thoughts on the Cemetery at Sané” (Birmingham
[unpublished paper from the Centre of
West African Studies, University of Birmingham],
n. d.), p. b:1.
SYNCRETISM, TIME, AND IDENTITY
Figure 5.4. Selection of imported trade goods recovered predominantly from Gao Ancien.
Photograph by Timothy Insoll
23. P. F. de Moraes Farias, “Appendix 2. The Inscriptions
from Gorongobo,” in Insoll, Urbanism, pp.
156–59.
96
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
Islam, or even none at all. Thus this reconstructed picture differs somewhat from the types of
broad brush approach adopted historically, which might in passing describe Gao as Muslim by
the eleventh century, but in so doing fail to recognize the inherent complexity therein.
ARCHITECTURE
The architecture uncovered, for instance, was representative of these distinctions; a picture
of “domestic” Gadei as opposed to “cosmopolitan” Gao Ancien could be reconstructed. In the
former, part of a roundhouse built of liquid mud (or banco) was uncovered dating from the eleventh
to fourteenth centuries.24
This type of structure is so far absent in Gao Ancien, where
buildings universally employed the right angle and are of fired brick, stone, and mudbrick.25
Whether the presence of the roundhouse can be associated with the persistence of traditional religion
in this quarter is unclear (and unlikely). But such a question can be posited, as in many
areas of sub-Saharan Africa there appears to be a correlation between roundhouses and traditional
religion, and square or rectangular houses and Islam. Bearing in mind the important caveat
that such generalizations inevitably find exceptions to religious explanation, Muslims can live in
roundhouses and non-Muslims in architecture employing right angles!
Yet the architecture in Gao Ancien differed considerably, and in style (but not materials),
was much more reminiscent of the type of structures encountered at other trade centers in the
Western Sahel, such as Tegdaoust and Koumbi Saleh,26
which were predominantly Muslim. A
large mosque with a fired-brick mih≥rΩb was previously recorded within Gao Ancien,27
while our
more recent excavations uncovered part of a palace or rich merchant’s house also built of fired
brick (fig. 5.5). Elements of a further structure, an aisle, might have been part of another
mosque. The remains of a substantial stone-built wall and gatehouse might have once encircled
the central citadel, also dating from the twelfth to thirteenth centuries.28
Such features were lacking
in Gadei, again reinforcing this picture of complexity.
PERSONAL POSSESSIONS AND DIET
Within the area of the roundhouse described above, objects with Muslim associations potentially
suggest syncretic processes were in operation or that simple architectural correlations and
religious types are flawed, as they probably are. For example, a large wooden bead from a set of
prayer beads (what is sometimes erroneously referred to as a Muslim “rosary”) was recovered
from a context dated to between the mid-eleventh and fourteenth centuries.29
Similarly, and also
from Gadei, the remains of what appear to have been an amulet cover were found, a copper casing
containing the remains of fibrous matter that might have been the Muslim prayer or
invocation contained therein.
24. Insoll, Urbanism, pp. 15–17.
25. Insoll, Islam, Archaeology and History.
26. Jean Devisse, Abdallah O. Babacar, et al.,
Tegdaoust, Volume 3: Recherches sur Aoudaghost:
Campagnes 1960–1965, enquêtes générales
(Éditions Recherche sur les Civilisations 25; Paris,
1983); Jean Polet, Tegdaoust, Volume 4: Fouille
d’un quartier de Tegdaoust (Mouritanie orientale)
(Paris, 1985); Sophie Berthier, Recherches
archéologiques sur la capitale de l ’empire de
Ghana: Étude d’un secteur d ’habitat à Koumbi
Saleh, Mauritanie: Campagnes II–III–IV–V, (1975/
1976–1980/1981) (Cambridge Monographs in African
Archaeology 41; British Archaeological Reports,
International Series 680; Oxford, 1997).
27. Raymond Mauny, “Notes d’archéologie au sujet de
Gao,” Bulletin de l’Institut français d’Afrique noire,
Series B (1951) 13: 837–52.
28. Insoll, Islam, Archaeology and History; idem, Ur-
banism.
29. B. Roy, “The Beads,” in Insoll, Urbanism, p. 106.
TIMOTHY INSOLL
97
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
Dietary patterns also differed across and between the sites as represented by aspects of the
faunal and botanical remains recovered. A varied picture of resource exploitation was indicated
by the faunal remains from Gadei, with both wild and domesticated species present. Of especial
interest was the presence of dog remains in contexts dated between the early/mid eleventh and
late sixteenth centuries, that is, certainly after Islam was the majority religion in the city. This
was interpreted as food refuse owing to the context in which it was found 30
and suggests that a
mixed community was resident in Gadei, both Muslim and non-Muslim, or alternatively, Muslims
who continued to eat species forbidden under Islamic dietary law.31
By contrast, dog
remains were lacking in Gao Ancien, perhaps to be expected in this evidently more Islamized
quarter. Otherwise, the patterns manifest in the species present are broadly comparable to those
at Gadei, with both wild and domestic specimens found.
Besides the mammals, fish, shellfish, and birds indicated that a variety of environments
were being exploited. Fish and shellfish, though little differentiated between Gao Ancien and
Gadei, were obtained from streams, from main channels of the Niger River, pools on the floodplain,
fast and slow flowing water, and swampy and reed-filled environments.32
The picture
obtained from the botanical remains recovered from Gadei and Gao Ancien was also broadly
similar, samples largely coming from the later occupation levels and providing evidence for rice,
dates, local fruits such as the jujube, along with pearl millet, watermelon, and cotton.33
In terms
of connections with Islam, these are indirect, but cotton is generally regarded as an introduction
30. C. Stangroome, “The Faunal Remains from Gadei,”
in Insoll, Urbanism, p. 56.
31. See Insoll, Archaeology of Islam.
32. Nicky Milner, “The Marine and Freshwater Molluscs,”
in Insoll, Urbanism, pp. 36–38; H. Cook,
“The Fish Bones from Gadei,” in Insoll, Urbanism,
pp. 38–44.
33. Dorian Q. Fuller, “The Botanical Remains,” in
Insoll, Urbanism, pp. 28–35.
SYNCRETISM, TIME, AND IDENTITY
Figure 5.5. The palace or rich merchant’s house. Photograph by Timothy Insoll
98
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
in the Islamic period, that is, after the tenth century,34
whereas Ziziphus, or jujube stones, are
known to have been later used as a raw material in the production of Islamic prayer beads.35
Interestingly, on the basis of the cooking equipment found in Gadei and Gao Ancien such as
utilitarian pottery, stove fragments, and strainers, MacLean36
suggests that a “wet” cuisine was
preferred in Gao, and that kitchen mobility, a characteristic of Songhai cooking today, was also a
factor several hundred years ago.37
From this evidence it has been possible to reconstruct something
of the resultant cuisine, and it appears that many of the bones were subjected to “heavy”
chopping, possibly indicating boiling as a cooking method. In other words, this evidence appears
to suggest that the maintenance of traditional African cuisine was being upheld, though the diet
was largely structured by Islamic law. Again, another indicator of indigenous adaptation of elements
of Islamic belief and practice is seemingly apparent in the archaeological record.
SYNCRETISM, TIME, AND IDENTITY
This provides a convenient juncture to return to a brief consideration of the creation and
maintenance of Islamic identity both in the Gao region and in its wider context. A key concept
that has already been mentioned as of importance here is syncretism. This is the blending or fusing
of different religious traditions or elements that can emerge as a practical mechanism for
reconciling time, for instance. Although sometimes condemned as a contentious term implying
“inauthenticity” or “contamination,” 38
for our purposes in considering religion as change, “syncretism”
is preferable to alternatives such as “creolize” or “hybridize.”
The need for syncretic process to reconcile time reckoning can take many forms; for example,
an agriculturally aligned seasonal system must confront and integrate with a very
different religious system.39
This can occur by conversion to Islam with the imposition of a new
calendar, “arranged, without intercalation, to be independent not only of the old Arabian lunar
year but especially of all solar reckoning which was traditionally linked to the structures of agricultural
society and religion.” 40
This does not imply that West African subsistence farmers
necessarily used formal calendars, but if agriculture is really a “ritual revealed by the gods or
culture heroes,” 41
then the abandonment of associated seasonal, temporal, and ancestral frameworks
(the latter possibly key) will be difficult, or alternatively adjustments might be made to
allow the continuation of old and new combined.
This would appear to be what occurred with conversion to Islam in the Sahel region of West
Africa where it is possible to suggest a model of phased conversion allied with syncretic adaptation
based in part on the archaeological data just presented.42
Within this region the earliest
converts to Islam would seem to have been the nomadic populations, precipitated in part through
34. Andrew M. Watson, Agricultural Innovation in the
Early Islamic World: The Diffusion of Crops and
Farming Techniques, 700–1100 (Cambridge, 1983).
35. Fuller, “Botanical Remains,” p. 30.
36. M. R. MacLean, “The Locally Manufactured Pottery,”
in Insoll, Urbanism, p. 77.
37. M. R. MacLean and Timothy Insoll, “The Social
Context of Food Technology in Iron Age Gao,
Mali,” World Archaeology 31 (1999): 78–92.
38. Rosalind Shaw and Charles Stewart, “Introduction:
Problematizing Syncretism,” in Syncretism/Anti Syncretism:
The Politics of Religious Synthesis, edited
by Charles Stewart and Rosalind Shaw (London,
1994), p. 1.
39. Insoll, Archaeology, Ritual, Religion.
40. F. Denny, “Islamic Ritual: Perspectives and Theories,”
in Approaches to Islam in Religious Studies,
edited by Richard C. Martin (Tucson, 1985), p. 71.
41. Mircea Eliade, The Sacred and the Profane: The Nature
of Religion, translated from the French by
Willard R. Trask (New York, 1959), p. 96.
42. Insoll, Islam, Archaeology and History; idem, Ur-
banism.
TIMOTHY INSOLL
99
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
their early exposure to Muslims by acting as their guides in trans-Saharan trade. Equally, the
ease with which they converted is not solely explained by notions of familiarity but also, perhaps,
through the lesser degree of upheaval involved in nomadic conversion than that suffered
by agriculturalists, for instance.43
Hence factors such as the ease of worship that Islam enjoys
would have been significant, allied with a potential by lesser importance ascribed to physical ties
to the land, and in turn to the degree of ancestral significance lent to the land as well. In other
words, the bonds were more easily broken and syncretic mechanisms reconciling the old and the
new were not so essential.
This would certainly seem to be mirrored within the site patterning evident in parts of the
Gao region. For example, a map of Muslim cemeteries in the Niger Bend was prepared by de
Gironcourt.44
It is apparent from this map that the majority of the cemeteries containing Muslim
inscriptions are to be found clustered along the Niger River, predominantly along the left bank,
as well as along the Tilemsi Valley running north from Gao, and to a lesser extent the paleotributaries
of the Niger River.45
This cemetery patterning appears to follow the main axes of
communication and of trade, that is, those that run through nomad territory,46
and could indicate
both mortality rates among merchants and other Muslim travelers and that nomads were among
the first to convert to Islam in this region. As the data presented in the map are “raw” data, a mix
of early and more recent material, this must remain a hypothesis.
The second group to convert to Islam within the Western Sahel seems to have been elements
of the urban population, and again a practical explanation can be proposed to account for this.
Specifically, that they might have benefited from preferable trade conditions with Muslim coreligionists,
or alternatively that Islam had an appeal within the urban environment through its
ability to provide cohesiveness due to the notion of community (ummah) that underpins it. This
is a factor of potential significance in overcoming ethnic differences that were perhaps more
manifest in towns, settlements with a predilection to throw together a variety of different ethnic,
social, and other groups.47
Such an interpretation would appear to be supported within Gao,
however, these urban centers were amorphous forms that attract and release elements of their
populations all the time. Their population, theoretically at least, could be forever altering, even if
their core remains the same. This means that both within West Africa, and throughout much of
sub-Saharan Africa, the ascription “Muslim city” may be tenuous at best. Islamic practice in urban
centers such as Gao does not seem to have depended on the development of syncretic
mechanisms to reconcile the old with the new, at least it was not quite such a pressing concern in
this context, as it was for the third group to be considered, the sedentary agriculturalists.
The last group to convert to Islam in parts of West Africa were the bulk of the population,
the sedentary agriculturalists, and within this context the notion of syncretism is vital. This apparent
tardiness in conversion may be interpreted as related to conceptual changes described
above, that is, the collision of different calendrical and temporal systems which, more than prescribing
when crops might be sown or harvested, provided the whole structural framework
connected with the lynchpin of African traditional religions, the importance of the ancestors.48
SYNCRETISM, TIME, AND IDENTITY
43. See, for example, Nehemiah Levtzion, “Rural and
Urban Islam in West Africa: An Introductory Essay,”
Asian and African Studies 20 (1986): 7–26.
44. Georges R. de Gironcourt, Missions de Gironcourt
en Afrique occidentale, 1908/1909–1911/1912:
Documents scientifiques (Paris, 1920), p. 161.
45. Insoll, Islam, Archaeology and History, p. 13.
46. It is unlikely to be unduly influenced by de Gironcourt’s
survey methodology, as his coverage of the
area was very thorough.
47. See Insoll, Sub-Saharan Africa.
48. See, for example, John S. Mbiti, Introduction to African
Religion (London, 1975); Benjamin C. Ray,
African Religions: Symbol, Ritual, and Community
(Englewood Cliffs, 1976).
100
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
TIMOTHY INSOLL
Ancestral bonds and frameworks linking men to the land were negotiated primarily through
structured relationships, “whether with other living people, or with the spirits of the dead, or
with animals, or with cleared land, or with the bush.” 49
These were beliefs manifest through
what Ranger terms “cults,” as in the maintenance of cults of the land, for example. The existence
of the whole ancestral framework of belief and associated practices meant that conversion to Islam
broke with or altered the balance that sustained the conceptual system, as Rene Bravmann50
argues.
Hence, even though Islam might be well established within the urban environment at centers
such as Gao and among nomad groups in the surrounding region, its impact within the remainder
of the rural environment was frequently minimal even several hundred years later. This can be
indicated archaeologically in various ways, as in the persistence of non-Muslim burial practices,
such as the continuation of a tradition of urn burial in a contracted position accompanied by
grave goods such as iron bracelets and ankle rings51
found at the site of Toguere Doupwil in the
Inland Niger River Delta area of Mali, farther west along the Niger River. This was evidence
dated to the fifteenth century, and thus long after conversion to Islam had occurred in urban centers
such as Timbuktu, which is closer to the Inland Niger River Delta than Gao. Similarly, the
continuation of production of anthropomorphic and figural terra-cotta statuettes, contrary to Islamic
proscription on the replication of figurative imagery, is found up to and even beyond a
similar date. At the urban center of Jenne-jeno, for example, also in the same region of Mali,
over seventy animal or human representations have been recovered; the function is interpreted in
various ways, including for protection, and in “ancestor worship.”52
However, these interpretations
are based upon ethnography, oral tradition, and parallels with material from elsewhere in
the region.
Where conversion did take place, syncretism of Islamic and traditional religions frequently
occurred, seemingly as a mechanism for reconciling issues such as the collision of frameworks
of time and their associated implications for conceptions of land, its links with people and ancestors,
issues of possession, fertility, and the like. Although relevant archaeological data are still
sparse for the Western Sahel from the fourteenth century onwards, large-scale conversion
amongst sedentary agriculturists seems to have been limited until more recently. In some instances
it occurred only after the collapse of the great states such as Songhai. As Levtzion
notes,53
when the great states disappeared and the urban foundation of the religion crumbled in
the seventeenth and eighteenth centuries, this meant that Muslims moved into the countryside,
“and won adherents among peasants and fishermen, who had hardly been influenced by Islam
before.”
Again if we consider the Gao region, though it has to be acknowledged that the data are
sparse, these general interpretations are ostensibly supported. Various settlements south of Gao,
for example, sent various commodities up the Niger River, the most important of which was
gold obtained from the Sirba Valley in Niger.54
Close to this place an Islamic “frontier” appears
49. T. O. Ranger, “African Traditional Religions,” in
The Study of Religion, Traditional and New Religions,
edited by Stewart Sutherland and Peter Clarke
(London, 1991), p. 109.
50. René A. Bravmann, Islam and Tribal Art in West
Africa (African Studies Series 11; Cambridge,
1974).
51. Rogier Bedaux, “Mali,” Nyame Akuma 8 (1976): 41.
52. R. J. McIntosh and S. K. McIntosh, “Terracotta
Statuettes from Mali,” African Arts 12/2 (1979): 52.
53. Nehemiah Levtzion, “Rural and Urban Islam in
West Africa: An Introductory Essay,” Asian and African
Studies 20 (1986): 15.
54. See Jean Devisse, “L’Or,” in Vallées du Niger, edited
by Jean Devisse (Paris, 1993), pp. 344–57.
101
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html
SYNCRETISM, TIME, AND IDENTITY
to have existed,55
indicated by sites such as Bentiya (Kukiya) and Egef-N-Tawaqqast, sites
where Arabic funerary inscriptions dating from the thirteenth to fifteenth centuries have also
been recorded.56
Evidence for Islam dating from before the fifteenth century is lacking farther
south of this point, both direct, such as mosques and burials, and indirect, such as trade goods
sourced from the Muslim World.57
Even close to Gao, an early center of Islam with its mosques,
Muslim burials, even rectangular architecture, survey evidence58
is rare or late in date, excluding
that interpreted as nomad associated, as already described. This is a pattern that appears to be
repeated elsewhere in the region, but the overall absence of archaeological research outside of
the main urban centers limits what can be inferred from this sort of evidence.
CONCLUSIONS
Although hampered by a lack of archaeological research in general in the Western Sahel
(and throughout West Africa), and certainly outside the major historical urban centers such as
Gao and Timbuktu, archaeological patterning of some social interest can be seen. The complexity
redefines organization of urban composition and trade for acceptance of and adherence to
Islam, and the impact of Islam upon social identity.
Conversion to Islam by the inhabitants of Gao occurs, and in a far from uniform process,
and is structured by socio-economic affiliation, whether nomads, city-dwellers, or sedentary agriculturalists.
Equally this did not entail a whole-scale abandonment of older traditions or
traditional religions, as with the tardiness evident in Islamic conversion amongst the sedentary
agriculturalists; likewise social aspects such as continuing adherence to traditional cuisines and
diets, as might be expected. The latter is more unexpected in having greater fundamental implications
in terms of breaking, rather than merely stretching, Islamic religious requirements.
This should certainly not come as a surprise. Throughout sub-Saharan Africa the agency of
syncretism has been adopted as a mechanism to reconcile Islam with older traditions in an ongoing
process of reconstructing social and religious identities. We as archaeologists are privileged
to be able to gain an insight into such processes at work and thus should strive to maximize both
recovery and interpretation from the rich archaeological record that Islamic sites frequently
yield. In so doing we can more fully assess the notion of changing social identities across the
geographical diversity that is the Islamic world.
55. John O. Hunwick, Shariªa in Songhay: The Replies
of al-MaghÏlÏ to the Questions of Askia al-H˘Ωjj
Muh≥ammad (London, 1985).
56. Farias, “Oldest Extant Writing,” pp. 105–06; N.
Arazi, “An Archaeological Survey in the Songhay
Heartland of Mali,” Nyame Akuma 52 (1999): 38–39.
57. Insoll, Sub-Saharan Africa.
58. S. Dawa, “Inventaire des sites archéologiques dans
le cercle de Gao: Mémoire de fin d’études.” (Ph.D.
diss., École Normale Supérieure de Bamako, 1985);
Insoll, Islam, Archaeology and History, pp. 11–15;
Arazi, “Archaeological Survey,” p. 36.
102
oi.uchicago.edu/OI/DEPT/PUB/SRC/OIS/1/OIS1.html