127 8 3MB
English Pages 158 [172] Year 1979
CONVERSION T O ISLAM IN T H E MEDIEVAL PERIOD
Conversion to Islam in the Medieval Period An Essay in Quantitative History
RICHARD W. BULLIET
HARVARD UNIVERSITY PRESS Cambridge, Massachusetts and London, England 1979
Copyright © 1979 by the President and Fellows of Harvard College All rights reserved Printed in the United States of America Library of Congress Cataloging in Publication
Data
Bulliet, Richard W Conversion to Islam in the medieval period. Includes index. 1. Islamic Empire—History. 2. Muslim converts. 3. Islam—History. 4. Islamic Empire—Social conditions. I. Title. DS38.3.B84 909'.09'7671 79-14411 ISBN 0-674-17035-0
To my wife, Lucianne Cherry Bulliet
Acknowledgments
Many people have contributed in one way or another to the formulation of the ideas put forward in this book, and a few have rendered particularly valuable service in the shaping of concepts. I wish to acknowledge with thanks the advice of Donald Olivier, Michael Bonine, and Lucy Bulliet. My special gratitude goes to my parents, Mildred and Leander J . Bulliet, for helping me understand the mathematical aspects of the project. T h e writing of this book was greatly facilitated by a fellowship from the John Simon Guggenheim Foundation.
Contents
1. Introduction 1 2. Regional Variation in Islamic History 7 3. The Curve of Conversion in Iran
16
4. Conversion as a Social Process 33 5. The Development of Islamic Society in Iran 6. The Curve of Muslim Names 64 7. Iraq 80 8. Egypt and Tunisia 92 9. Syria 10. Spain
104 114
11. The Consequences of Conversion Notes 141 Index
155
128
CONVERSION T O ISLAM IN T H E MEDIEVAL P E R I O D
1. Introduction
This book is about Islamic social history. It flies the colors of an essay because it represents an initial, tentative effort—not at describing Islamic social history, which has certainly been attempted before, but at formulating an integrated conceptual approach to the subject. It attempts to tie together the histories of parts of the medieval Islamic world that are more commonly treated as discrete entities by means of a number of sometimes quite speculative hypotheses based upon and inspired by a close examination of certain quantifiable aspects of medieval Arabic sources. Many of the conclusions reached through these hypotheses, such as the suggestion of a causal relationship between the conversion of a majority of a region's population and the dissolution of central Islamic government in that region or the explanation of endemic factional strife in certain areas as a virtually inevitable conflict between the descendants of early converts and the descendants of later converts, will require further corroboration. Nevertheless, the heart of the essay lies in the overall conceptual approach, and the primary effort has been placed there rather than upon the elaboration of its various ramifications. T h e approach is predicated upon the notion that there is a direct and fundamental relationship between conversion to Islam and the development of what may be called an Islamic society. When in the second half of the seventh century A.D. the Arabs conquered the Persian empire and half of the Byzantine empire, they did not bring with them the religion that is described in general books on Islam. They brought with them something far more primitive and undeveloped, a mere germ of later developments. Similarly, the society in which Islam grew 1
Conversion to Islam in the Medieval Period and developed was not exclusively the society of the Arabian desert: it was that of the conquered lands as well. Yet the society of these conquered lands was certainly not an Islamic one to begin with. There were from the time of the conquests onward Muslims in these lands, but they at first represented one small element, albeit the ruling element, within a territory that was dominated numerically by adherents of other religions. As time went on, of course, the relative place of this society of Muslims vis-à-vis other religious groups changed, as did the composition of the society of Muslims. These changes came about through conversion. T h e altered composition of the Muslim population in turn affected the course of development of the Islamic religion. NonArab converts and their descendants made contributions in every area of cultural life under the aegis of Islamic rule, and the customs and outlook of the non-Arab peoples in general became accommodated in various ways within the developing institutions of the Islamic state and religion. Yet as long as the Muslim population remained a minority or constituted only a bare majority of the entire population of a region, the society of that region as a whole was not an Islamic society, nor the culture of that region an Islamic culture. Nor would it be appropriate to consider the Muslim population of such a region to constitute in itself an Islamic society without taking into consideration its position as a minority or bare-majority group embedded in a larger population containing rival religious groups. What was required to make the society of the Middle East and North Africa as a whole a single Islamic society was, first, the completion of the conversion process at least to the point at which internal threats to the dominance of the Islamic religion became inconceivable and, second, the elaboration and spread of a more or less uniform set of social and religious institutions. That the Middle East and North Africa did, in fact, develop into a single Islamic society to a very great extent is undeniable; it is the central historical fact that underlies most discussions of Islam. Yet it is an error of historical hindsight to regard this resultant uniformity as the product of a uniform development. 2
Introduction
It is the premise of this book that the two necessary steps in the development of a unified Islamic society, namely, conversion and the elaboration of Islamic social institutions, took place at different times in different parts of the vast territory conquered by the Arabs. There is no reason to suppose that all the conquered territories converted to Islam at the same rate. Indeed, since the lands involved were inhabited by peoples with varying religious, linguistic, and social identities, it would be more reasonable to assume the opposite. If, then, the development of religious and social institutions was to some degree dependent upon the accomplishment of a significant amount of conversion, it stands to reason that such institutional development would have occurred at different times in different places according to regional differences in the rate of conversion. Or, alternatively, institutional developments taking place in areas converting relatively early could have spread into other regions as they successively reached similar stages of conversion. This book attempts to elucidate this complex relationship between conversion to Islam and the development of Islamic society by first establishing a hypothetical timetable of conversion for six major parts of the medieval Islamic world and then comparing those timetables with the course of historical development in each area. T h e findings for the six areas are finally compared with one another to reveal common features or consequences of the conversion process. Ultimately, there may be no alternative quantitative outline of Islamic conversion against which to test the conclusions reached in this manner, and they will have to be tested instead against the established facts of Islamic political, social, and religious history. However, even without conclusive independent corroboration, the methodology employed here may be deemed at least heuristically valuable if it helps elucidate the general history of medieval Islamic society and provides new insights into it. Beyond the area of Islamic social history, there are several other purposes that this book may serve. First, it attempts to deal quantitatively with the phenomenon of mass religious or ideological conversion. Study of this subject has been hitherto, 3
Conversion to Islam in the Medieval Period almost of necessity, characterized by impressionistic historical judgment. After all, how can one tell, particularly for the medieval period, when or at what rate the faceless masses converted to a new religion or adopted a new idea? T h e routes of missionaries may be traced and the creation of new bishoprics recorded, but there is no way of knowing what percentage of the population the missionary contacted, much less converted, or what proportion of a community's inhabitants belonged to the first bishop's flock. For the Islamic world, the history of popular conversion is even less explored than it is for Christian Europe. Although interest in the subject has grown in recent years, most scholarly effort is being devoted to areas of relatively recent Islamic penetration, such as West Africa and Indonesia. 1 T h e great conversion experience that fundamentally changed world history by uniting the peoples of the Middle East in a new religion has had few modern chroniclers, the reason being that conversion plays so slight a role in the narratives of medieval chroniclers. Without data it is difficult to write history, and medieval Islam produced no missionaries, bishops, baptismal rites, or other indicators of conversion that could be conveniently recorded by the Muslim chronicler. In this book I have attempted to compensate for this lack of primary data by analyzing changes over time in the patterns of first-name giving in medieval Islam. I have tried to show that changes in name patterns accurately reflect the general course of religious conversion and that the course thus revealed ties the phenomenon of religious conversion into a broad current of theorizing about the ways in which people adopt new things. T o the degree that this linkage is a sound one, conversion to Islam can be considered as but a single example of a pattern characteristic of mass ideological change. T h e hypotheses advanced about Islamic conversion may thus be of significance for students of other proselytizing religions or of mass ideological change. T h e second additional purpose this book may serve is to encourage experimentation with quantitative methods by historians of the Middle East. T h e controversy over the value of 4
Introduction quantitative methods for historical study has not yet struck the field of Middle Eastern history. Since not much effort has been made to apply such methods, there has been little to discuss. It may be anticipated, nevertheless, that some students of the Middle East will be either offended or depressed by the numerous graphs and calculations upon which this book depends. For these individuals, two considerations may possibly relieve some of their distress. First, the conclusions reached in this book probably could not have been reached by any other historiographical approach. Graphs and figures are not used to belabor the obvious or to corroborate previously accepted ideas; they give rise to fresh ideas that may serve to stimulate further research even if they do not ultimately prove to be correct in all particulars. And second, although all graphs and tables have an aura of precision, this can easily be misleading. What is important in this book is the soundness of the arguments arising from the data rather than their graphic presentation. Other or more abundant sources of information might alter the latter in some particular without invalidating the former. For those who are more receptive to the idea of quantitative history, it is hoped that the methodology explored here will encourage similar efforts. T h e history of the Middle East is in need of the methodological insights that have been developed by historians of other periods and places. Materials exist for quantitative studies in Islamic history; what is lacking is the perception that such an approach can yield something of value. A final purpose of this book concerns the potential usefulness of medieval Muslim biographical dictionaries. This genre of literature, which has been my scholarly concern for many years, holds the promise of greatly revising our notions about Middle Eastern history. Utilizing the data contained in these compilations is not an easy task, however. Several approaches, both quantitative and nonquantitative, have been advanced by myself and others in earlier publications. This book continues the process of methodological inquiry. However, the type of analysis presented here—the examination of patterns of name giving— has implications beyond the genre of biographical dictionaries 5
Conversion to Islam in the Medieval Period strictly speaking. Almost any extensive list of first names extending over a long period of time has the potential to be used for quantitative analysis. This holds true for European and American history no less than for Middle Eastern history. In a separate study of changes in the first names of Turkish parliament members in the nineteenth and twenties centuries, I have attempted to demonstrate that the sensitivity of naming to currents of social and political change is still a fact. 2 Since the analysis of naming patterns is underexploited as a tool for writing history, it is hoped that the following demonstration of its usefulness will encourage further exploration of its possibilities.
6
2. Regional Variation in Islamic History It has been advanced as a reasonable assumption that timetables of conversion differed from one region of the Islamic world to another and that these regional differences were reflected in political and institutional developments that depended to some degree upon the conversion process. T h e purpose of this chapter is to present a quantitative demonstration of the existence of significant variations in regional importance throughout a part of Islamic history. T h e complex causes of these variations do not yield to simple analysis, but an initial look at the type of data used to reveal them and at the overall pattern thus revealed will serve to indicate a place to start in a region-by-region study of conversion. T h e reader has already been warned that this book is heavily laden with graphs. T h e first graph, however, should give rise to no anxiety in the breasts of the mathematically disinclined. It shows by fifty-year periods what proportion of biographies from a large biographical compilation emanate from different regions of the Islamic world. Leaving aside for the moment the question of the source utilized and the character of the biographies, one can readily note the correspondence between the changing proportions on the graph and the generally accepted political history of Islam. Iraq's substantial importance under the Abbasid caliphate, for example, is quite clear as is its rapid decline following the Mongol invasions of the thirteenth century and its virtual obliteration after the time of T i m u r around A.D. 1400. Similarly, the great swell in the proportional share of Iran coincides exactly with the heyday of the independent Iranian dynasties with decline setting in only under the late Seljuqs 7
Conversion to Islam in the Medieval Period Maghrib. West_ East
^
9 Syria: S Coast a ] ' Inland
80 al-Jazira
Arabia A H 75 125 A.D. 6 9 5 743
175 791
225 275 325 8 4 0 888 937
375 4 2 5 475 525 575 625 675 725 775 825 875 925 975 985 1034 1082 1131 1179 1228 1276 1325 1374 1422 1471 1519 1568
Graph 1. Proportional representation of regions in Shadharät
adh-dhahab.
when security and governmental order are well known to have plummeted. Again, the Mongol invasions mark a further decline. In Arabia the often noted depressing effect of the tribal migrations that followed upon the Islamic conquests and of the movement of the caliphal capital outside the peninsula is indicated by the near total absence of any Arabian representation on the graph between the ninth and the fourteenth centuries. After that, however, there is a noticeable revival that coincides with the installation of the Rasulid dynasty in Yemen and later with the political emergence of the Sharifs of Mecca. Turning to Syria, the most remarkable feature is the tremendous growth beginning with the consolidation of power 8
Regional Variation in Islamic History u n d e r the Ayyubid dynasty b u t waxing even stronger u n d e r the succeeding Mamluks. I n Egypt m u c h the same pattern prevails with the a p p a r e n t difference that the end of M a m l u k r u l e a n d the transition to O t t o m a n r u l e after 1516 coincides with a d i m i n u t i o n of importance that is m u c h more substantial than that in Syria. Finally, at the top of the graph, Spain waxes u n d e r the caliphate of Cordoba in the tenth century a n d retains its prominence into the Almohad period when it gradually declines as Morocco a n d Algeria (Maghrib West) wax in t u r n . As for Anatolia or R u m , its rise coincides precisely with that of the O t t o m a n Empire. T h i s rough correspondence between regional variations in a collection of biographies a n d the e b b and flow of historical prominence and political power d u r i n g the first ten centuries of Islam naturally raises a question as to what sort of biographies were used. As is generally known to specialists on Islam a n d Middle Eastern history, one of the characteristic cultural products of Islamic society f r o m very early u n t i l recent times has been the literary genre of the biographical dictionary. 1 Extant biographical compilations, published and unpublished, contain many h u n d r e d s of thousands of separate entries. Some works concentrate u p o n members of separate worthy professions— poets, doctors, mystics, a n d the like—others u p o n cities or regions, and others u p o n individuals of religious eminence. Some compilations are arranged alphabetically by first names, others by names known as nisbas signifying a person's tribal, religious, geographical, or occupational affiliation, and others by date of death. T h e variety and range of biographical dictionaries is enormous, and their utility as sources of quantifiable data on Islamic history is only b e g i n n i n g to be explored. 2 G r a p h 1 was derived f r o m a single compilation of biographies, I b n al-'Imäd's Shadharät adh-dhahab fi akhbâr man dhahab [Nuggets of gold in the stories of those who have passed away]. 3 I b n al-'Imâd died in 1687, which means that his compilation, t e r m i n a t i n g in the Muslim year 1000 (A.D. 1592), contains none of his contemporaries and was made entirely f r o m written sources. For every year f r o m 1 to A.H. 1000 he gives biographies 9
Conversion to Islam in the Medieval Period of individuals who died in that year. After the first Islamic century, when totals are q u i t e small, the n u m b e r of biographies given per twenty-five years ranges f r o m approximately one h u n d r e d to more than three h u n d r e d . Because of the structure of medieval Arabic names, most of these biographies reveal the place of origin of the subject. Arabic names, whether b o r n e by ethnic Arabs or by non-Arabs, were typically composed of several distinct parts: 1) the ism or first name, which was given at birth, (2) the kunya, a n a m e beginning with Abü, m e a n i n g "father of," or U m m , m e a n i n g " m o t h e r of," which was theoretically acquired after one had become a parent b u t which was very f r e q u e n t l y given at b i r t h for euphonic or other reasons unrelated to potential parenthood, (3) the nasab or genealogy, which was a series of isms, kunyas, or other names strung together with the word ibn, "son of," or bint, " d a u g h t e r of," (4) the nisba, a n a m e most f r e q u e n t l y ending in -I, which signified some sort of affiliation (for example, al-Baghdädi = person f r o m Baghdad) and often served for several generations as a family surname, and 5) the laqab or or honorific name, u n d e r which may conveniently be l u m p e d all sorts of honorary titles, nicknames, and epithets. An example of a medieval Arabic n a m e using all these parts in the order laqab, kunya, ism, nasab, nisba would be B u r h ä n ad-DIn A b ü A h m a d M u h a m m a d b. Yüsuf b. al-Hasan al-Misri. T h e category of n a m e that is of interest at the present m o m e n t is n u m b e r four, the nisba. An individual might have more than one nisba or n o n e at all, and the nisba might be inherited or acquired d u r i n g his lifetime for any n u m b e r of reasons. Hence, it is impossible to rely entirely u p o n the literal m e a n i n g of these names. Still, it is q u i t e a p p a r e n t f r o m an examination of thousands of biographies that the great majority of geographical nisbas relate directly to the individual's place of origin. Sometimes, when there are two or more such nisbas, specific mention is m a d e of which of them refers to place of birth, which to later place of residence, which to place of ethnic origin, a n d so forth. In going through the Shadharät adh-dhahab f r o m the year 100 to the year A.H. 1000, a total of 6113 biographies were f o u n d 10
Regional Variation in Islamic History in which a place of origin could be ascribed to the individual. In the great majority of cases this ascription was made on the basis of a geographical nisba combined with a reading of the biography for corroborative evidence, and in a minority of cases ascription was made on the basis of an explicit statement of place of birth. While the task of ascription was usually a straightforward one, it frequently was not. A variety of problems arose: for example, cities of the same name from different regions had identical nisbas, or the spelling of geographical names was inconsistent, or ethnic identity was not easily distinguishable from place of birth. Once the ascription was made, the names were aggregated in twenty-five-year groups in each of eleven geographical regions: (1) the Arabian peninsula, (2) Iraq as far north as Takrït on the Tigris and 'Änah on the Euphrates, (3) al-Jazira or Mesopotamia as far west as R a q q a and as far north as the T a u r u s mountains, (4) Iran, including the Iranian-speaking areas of Kurdistan, Transoxania, Afghanistan, and Azerbaijan; (5) inland Syria, including parts of modern Syria, Jordan, and T u r k e y east of the Jordan-Biqä'-Orontes valley, south of the Taurus, and into the desert as far east as R a q q a and Palmyra, (6) coastal Syria from Tarsus to al-'Arïsh, including the Çukurova in T u r k e y and the Jordan-Biqä'-Orontes valley, (7) Egypt, (8) the eastern Maghrib comprising Libya and Tunisia, (9) the western Maghrib comprising Algeria and Morocco, (10) alAndalus or Muslim Spain, and (11) R u m or those parts of the Byzantine Empire progressively conquered by Muslims after the battle of Manzikert in 1071. For graphing purposes, the twenty-five-year periods were further consolidated into fifty-year periods. Al-Jazira was treated as a subcategory of Iraq; the two parts of Syria and the Maghrib were separated by dotted lines. Finally, the terminal dates of the fifty-year periods were all put back twenty-five years on the assumption that whatever gained for an individual sufficient fame or notoriety to warrant his inclusion in the biographical dictionary would normally have transpired well before his death although it was under his date of death that he was recorded. T h e twenty-five-year setback, in other words, was 11
Conversion to Islam in the Medieval Period an effort to change known death dates into approximate "prime of life" dates. Many objections may be raised against constructing a graph in the manner described and then stressing its correspondence with the ebb and flow of historical prominence and political power. After all, the graph actually shows nothing b u t the course of regional representation in a single seventeenthcentury biographical dictionary. Yet the obvious fact remains that the graph does appear to correspond rather closely to the generally accepted course of Islamic history. T h e changing proportions for the different regions are not merely coincidental, as can be determined from the nisbas themselves. W h e n a region has a very high representation, many nisbas pertain to specific villages and quarters of major cities. For example Karkh, a popular quarter of Baghdad, appears in the nisba al-Karkhï when representation from Iraq is high. When the proportion is smaller, the name of the major city itself is a common nisba. In the example given, a later resident of Karkh would appear as al-Baghdädi. Finally, when the proportion is very low, the nisba will frequently be derived from the entire province, that is, al-Baghdädi becomes al-'Irâqï. Since the nisbas are part of the individuals' actual names, this progression cannot be due to the editorial activity of the compiler, Ibn al-'Imäd, or of the earlier compilers he drew from. It must reflect the actual plenitude of individuals from different regions at the point or points of original compilation. T h e question, therefore, is not whether but why the distribution of biographies of eminent Muslims in this single work reflects the general course of Islamic history. T h e answer involves a n u m b e r of factors, the first of which pertains to the sources from which Ibn al-'Imäd compiled his work. T h e ten thousand or so biographies from which the 6113 used for the graph were taken were culled from a huge library of earlier biographical compilations. Ibn al-'Imäd directly cites virtually every biographical dictionary of importance. H e does not specify his procedure for selecting individuals for inclusion from the vastly greater n u m b e r of biographies he must have read through, but the existence of biographies of the same per12
Regional Variation in Islamic History son in several works was clearly one important index of eminence. Moreover, many of the works he used were themselves editorial selections of the most eminent individuals found in the sources available to the earlier compilers. I n other words, biographical dictionary compilation in Islamic history has been a continuous process of collection, collation, and culling out. I b n al-'Imäd is the beneficiary of many earlier generations of editors of biographies. Editorial responsibility and sensibility is not his alone, nor does it belong solely to members of his political-religious group, the Hanbali law school, or to people from his area of origin, Syria. A specific comparison of the relative proportions of biographies in Shadharät adh-dhahab from Arabia, Iraq, Iran, Syria, and Egypt with the proportions for the same areas contained in an identically structured work by adh-Dhahabï, Kitäb al-'Ibar fï khabar man ghabar [ T h e book of significant points in the tale of those who have gone by], shows an average variation of less than two percentage points even though the n u m b e r of biographies per fifty-year period differs by as much as 130. 4 T h i s close correspondence is instructive because adh-Dhahabï was a Shafii and therefore in a religious camp opposed to that of I b n al-'Imäd, and he died in A.D. 1348, 339 years before the latter. Of course, the possibility of a certain degree of editorial bias within the entire biographical tradition of which I b n al-'Imäd is the inheritor cannot be ignored. For example, low representation from Egypt during the two centuries of Fatimid rule might suggest an anti-Shii bias. B u t by and large, biographical compilations grew out of the study of the traditions of the Prophet and were thus inclined to be nonpartisan. 5 Moreover representation from Iran during the rule of the Shii Buyids is very high. Yet even if a continuous and very consistent process of biographical editing did occur, why should it reflect the political history of Islam? Here other factors must be considered. T h e subjects of the biographies are eminent Muslims, mostly religious figures but also including many government officials, poets, philosophers, and rulers. T h e latter groups would understandably be distributed according to the vagaries of political history because they were normally denizens of princely courts 13
Conversion to Islam in the Medieval Period
or recipients of court patronage. T h e more illustrious, powerful, and wealthy a ruler, the more extensive and impressive his state apparatus and the greater the amount of patronage at his disposal. For the religious figures, however, the ulama who make u p a majority of the 6113 biographies, direct government employment or patronage was considerably less important. A connection with political history could nevertheless be argued on the following grounds: since the Islamic religion did not develop a formal ecclesiastical structure with an established hierarchy, the prevalence of learned religious figures in a particular time and place was not so much a function of office-holding and organizational structure as of a cultural climate conducive to intellectual endeavour. T h e elements of such a climate included large urban populations, important educational institutions, a cosmopolitan atmosphere, sufficient prosperity to support a nonproductive class, and so forth. These elements would not be typical for all religions, but they are the most recurrent ones for medieval Islam. In the same way, although they are not always connected with political prominence, they usually are: medieval Islamic governments tended to cause rapid growth in capital cities and provincial governing seats. They also acted as magnets for international trade and as sponsors of religious and educational institutions through pious endowments. Accordingly, it should not be surprising to find the largest number of ulama concentrated at any given time in the wealthiest and most powerful states and, in particular, in their capitals and major administrative centers. Yet as plausible as this line of argument might sound, strong objections to it can be raised. Leaving aside the apparent anomaly of low proportional representation for Egypt during the Fatimid period which, as already mentioned, may result from an anti-Shii bias in the sources, it is difficult by the above reasoning to account for the very low representation of Syria during the first fifty-year period when it was the capital province of the caliphate; it is almost as difficult to account for Syrian prominence during the last two centuries when it was politically subordinate either to Egypt or to the Ottoman Empire. Looked 14
Regional Variation in Islamic History at more closely, in fact, the graph appears to represent the ebb and flow of Sunni religious prominence more than it does that of political power. On the one hand, this fact is quite understandable since the primary data relate mostly to religiously important individuals; but on the other hand, the undeniable but somewhat less precise correlation with overall political history hints at a possible linkage between religious eminence and political power on grounds somewhat different from the direct equation of the two mentioned above. T h e question suggested by the graph is why the importance of a region in Islamic history generally parallels the region's political importance but in specific instances differs from it to a significant degree. T h e r e are a number of possible approaches to answering this question, but two in particular are useful in indicating a place to begin a region-by-region study of conversion. One approach is demographic and departs from the simple notion that a region with a large population will produce more outstanding individuals than a region with a small population and be able to muster greater political power as well. T h e other approach is religious-historical and hinges upon the idea that both religious eminence and political prominence may be directly related to the percentage of Muslims in the population of a given area. T h e fact of one area's becoming prominent on the graph earlier than another would thus be related to a more rapid rate of conversion in the former area.
15
3. The Curve of Conversion in Iran If an intellect capable of significant achievement in the religious sphere appears once in every five thousand Muslim births, it seems reasonable to assume that a region with a million Muslims will produce two hundred noteworthy scholars compared with the two produced by a region with a Muslim population of ten thousand. This view, of course, represents a simplistic approach to the question of regional variation in Islamic history, since the field of endeavor and level of achievement of an individual in his lifetime are influenced by many factors besides his innate intellectual endowment. Yet the approach obviously has some validity. If we look at graph 1 in these very gross terms, the general validity of this quantitative relationship between Muslim population and religious achievement suggests Iran as a point of departure for the study of Islamic conversion. As mentioned in chapter 2, one likely contributor to the change in proportional representation among the various regions is demographic fluctuation. In the case at hand, demographic fluctuation can be approached regionally, or it can be approached religiously. Either way, it must be approached gingerly. As to the regional method, for example, it is widely assumed that from the twelfth century onward for a period of several centuries Iraq, Iran, Syria, and Egypt experienced significant population losses.1 Without knowing the relative extent of these losses, however, it is difficult to assess the demographic contribution to the increase in the religious prominence of the latter two regions at the expense of the former. Yet other cases are quite straightforward on a regional demographic basis. In the first fifty-year period shown on the graph, Arabia has over four times the importance of Egypt, but no demographic historian would ever venture the assertion that 16
The Curve of Conversion in Iran the overall population of Arabia even approached the population of Egypt d u r i n g this period or any other. Clearly, what is involved here is not regional b u t religious demography. T h e implication of the broad relationship between population and religious prominence in this case is that Arabia probably had substantially more Muslims than Egypt at that particular period. T h e r e is no surprise in this since it has never been claimed that the bulk of the Egyptian population had converted to Islam at this early date or that the A r a b settlement in Egypt after the initial conquest was particularly heavy, b u t this extreme example serves to illustrate the influence of religious demography on the proportions shown by the graph. Needless to say, religious demography depends not only u p o n the b i r t h and death rates of religious groups, which can vary appreciably, b u t also u p o n the net gain or loss of members through conversions f r o m one religion to another. T h e first factor is imponderable at such a great distance in time, b u t the second is not. Since it is known that the vast majority of the population of the lands conquered by the Arabs converted to Islam at some time or other, it must be assumed that the increase through conversion of the Muslim population in a given region shows u p to some extent on graph 1. If conversion proceeded at the same rate in every region, the only evidence on the graph would naturally be the decline of Arabia relative to the conquered territories as a whole. However, the assumption of an equivalent effect of conversion on the religious demography of every region is quite u n w a r r a n t e d if for n o other reason than that there are different timetables of conquest for different areas. It is m u c h more reasonable to assume at least some difference in the timetables of conversion f r o m region to region. If different regions had different timetables of conversion, and if at least some of the relative change in the religious prominence of the regions is a t t r i b u t a b l e to the increase in the Muslim population through conversion, then what should appear on the graph is a more rapid increase in the relative importance of earlier-converting regions over later-converting regions. Looking at graph 1 with this expectation in m i n d , we 17
Conversion to Islam in the Medieval Period note that Iran increases in importance earlier and faster than any other region. Does this imply that Iran converted earlier than the other regions represented? Perhaps. At least it suggests that we should look at the evidence for conversion in Iran with this thought in mind. For evidence relating to the conversion of Iran to Islam, we turn again to the genre of biographical dictionaries discussed earlier. Here we are greatly aided by the fact that Iran adopted the Arabs' religion but not their language. Arabic was used as the language of religion by converts to Islam, but it was not the language of everyday speech. T h i s bilingualism posed a dilemma for converts trying to decide what names to give their children. An examination of a body of some six thousand biographies contained in Iranian biographical dictionaries—4299 of people from the city of Nishapur in the northeast who died before A.D. 1130 and 1635 of people from Isfahan in the west who died before A.D. 1000—reveals a number of significant facts about Iranian Muslim naming customs.- Most obvious is the almost total absence of Persian names among the subjects of the biographies, a group drawn primarily from the religiously eminent upper class, what may be termed the patrician class. 3 Yet despite the nearly universal custom within this class of giving male children Arabic names, many Persian names do appear in the biographies. They appear in the patricians' genealogies or nasabs which are normally given for two generations and frequently for more. It appears that Persian names were more acceptable among the Muslim ancestors of the patricians than among the patricians themselves. One explanation of this fact could be that the patricians' forebears were frequently people of lesser religious eminence and hence felt less constrained to use the language of the Muslim religion in naming, but a tabulation of where in each genealogy each Persion name occurs suggests a different explanation. Over two-thirds of all Persian names initiate the genealogical sequence in which they occur. An example might be Muhammad the son of Ahmad the son of Rustam, Rustam being a Persian name. Moreover, in certain cases it can be positively 18
The Curve of Conversion in Iran
confirmed that the Persian name that initiates the sequence is that of the first family member to become a Muslim. 4 This predominating pattern indicates that a common practice in Iran was for the first Muslim member of an Iranian family to give his children names drawn from the Arabic onomasticon, a practice that was then continued in the following generations. Furthermore, it indicates that whenever a Persian name occurs at the beginning of a genealogical sequence, it is the name of the first member of the family to convert to Islam. Naturally, this interpretation does not take into account other naming patterns within Muslim families. For example, if the initial convert changed his own name to an Arabic name, which was probably the normal practice, 5 and the family reckoned its genealogy from him, there would be no way of distinguishing such a genealogy from that of a genuine Arab family. Similarly, the practice of fabricating Arabian genealogies is known to have been followed to some extent, and in most cases there is no way of detecting such fabrications. However, the obvious avoidance of Persian names within Iranian Muslim families during the first five Islamic centuries makes it quite unlikely that many genealogies containing Persian names were fabricated. T h e likelihood thus seems quite high that the subgroup consisting of genealogies that commence with a Persian name can provide reliable data on the approximate date of conversion of specific Iranian families. Of the overall group of 5,934 biographies, 469 genealogies fall into this subgroup. In each case, the subject of the biography can be assigned to a chronological period, the termini of each period being based upon dates of death and deduced from internal evidence in the biographical dictionaries (table 1, column A).6 Since the type of names in question are isms, corresponding to our first names, given shortly after birth, these chronological periods based upon dates of death must be adjusted to reflect dates of birth instead. This is accomplished by subtracting a standard figure for average lifespan (table 1, column B). T h e figure that has been used is seventy years, a figure arrived at by averaging all the lifespans specifically reported in the biographies from Nishapur. 7 19
«
Λ
ΐ
Οί S
S
o
•5 t
φ ι— ^ o
a
« Ό
a
-C m
Ό a a t-l 3
Cu
be
f a «
a § Λ ΐ -S « ^ h W^ QJ o e υ sc
- t m
n> " • ^ Ό
Γ-
' S a . «
m
fee o
2
υ
60 JS
« υ
rt
O κ
— o 42 U
•β
tO
CO
ΙΟ
00 φ
ιΟ 00
ι ΙΟ Οί
I ΙΟ tO ^
ο CM cm I ΙΟ 00 ι
Ό 00 (Μ I Ο (Μ Μ
φ ίο (Μ I σι ι—I (Μ
σ> ι—I SO I Φ ΙΟ
ι—ι CM
ι σ> CM
σ>
CM Φ
CO
1 CM Φ
CM CO
to c*-
I> σι
I CM •Π
1 CM CO
00 σι ι Ο0 CM
σ>
to I—1 CM CO to
Ι I
I I
φ φ ι
ο I—I
σ> σι ι ιΟ Φ
ΙΟ r -7 I ο ο
I
"c G I I
Ά
-
I 1
ΓιΟ 1 co
00 ΓI Φ CM
CO CO I σ>
CM ^
to
1 00 IO
CO 1-1
σ> ο CM I φ CO
CO φ CM 1 00 to
y
ι O
c
I
σι CM
c
S
r.
Μ
I
^ ιΟ CM I to 00
00 ο ^7 I φ ΙΟ
« t¿
o
ι •φ
«ο
J3 S
rt •c!
'
ιΟ
σ>
¿
«
c
c
-C
ΙΟ 00
be
-ö
o
ο "φ
g
ΙΟ
-7 ι
•φ
ι—I tO
•φ ι Φ
S
u ω
-rr· Λ
3
1—1
f—ι
σ>
00 CM
to Φ
'S C M (Λ H-
CO
7 CM σ>
o CM 1 φ n
Ι Ε— CM 1 CM O CM
CL,
f i h w J¿ ·
ε «5
3
tO to
ςθ ο
3
i)
CM CO 1
to •o CO 1
CM
=o CM
O CO
Φ Φ CM 1 O O CM
IO to CM 1 φ φ CM
00 1—1 CO 1 IO to CM
Φ
¿
IO CO CO 1 Φ
CM
CO
00 00 CO 1 io CO CO
CO CM 1 o
CM 1 tO
CO
Φ o
io CM Φ
O to
IO CM IO ι
Φ
J
IO CM "7 ι
CM
o 00 ι to CM
IO CO CM 1 00
CO t i CO CM
T3 o
« Λ Λ
ν
s
«
·Ü —.
•5
3
«
Ja
O O CM 1 o CO
2cd &
en IO j
IO IO
ι IO o
Φ · ι—I CM I O to
1—1 1—1
σι to CM 1 lO CM
IO •Φ CO 1 O I CM
O
Η
a 2 « ï o · T3 . Χ , a „ n i . •S 3 W «-» ~ o
O o CM I IO
O t^ CM 1 o o CM
é
CM CM I 1 IO t—
Φ 00 CM 1 O CO CM
σ> CO CO I io 00 CM
lO 1 o φ co
3 & η
ο
χ
ο
^
ir,
G
¿
to Φ
I IO CM
° ·
„
3
5
^
>
>
>
>
m
z
z
z
z
z
z
z
z
z
The Curve of Conversion in Iran T h e next step in assigning a hypothetical period of conversion to each Persian name initiating a genealogy is to construct chronological periods for each generation, working back from the generation of the subject of the biography. T h i s is accomlished by subtracting from the dates previously established a standard figure representing the average length of a generation. T h e figure that has been used is thirty-four years, which was derived by examining reconstructed family trees from Nishapur in which death dates of fathers and sons were recorded. Thirtyfour years is the average length of time that a son survived his father and must, therefore, reflect the average length of a generation. It should be pointed out that a generation is not the length of time from a man's birth to his first experience of fatherhood, but rather the length of time from his birth to the midpoint of his child-engendering years. T h e result of this subtraction process is to establish for each generation in each genealogy a hypothetical chronological period (table 1, columns C - Η ) . T o illustrate, we may take the biography of someone who died between 460 and 525 according to the Muslim calendar. Subtracting seventy years for the average lifespan, he should have been born between 390 and 455. T h e n , subtracting thirty-four years for a single generation, his father should have been born between 356 and 421. Subtracting another thirty-four years, his grandfather should have been born between 322 and 387, and so on for each generation. Obviously, no individual family can be expected to fit exactly into this framework, but the results should be approximately accurate for a large number of families. After this hypothesizing of dates is completed, it at last becomes possible to assign a specific, if highly generalized, period of conversion to each Persian name in the subgroup of genealogies beginning with such names. Since the individual with the Persian name is the putative original convert in the family and was therefore not born a Muslim, one last chronological adjustment is needed to arrive at the individual's age at the time of conversion. Although concrete data are lacking for calculating an average age at time of conversion, two facts allow a reasonable estimate to be made: first, the individual's father never 21
Conversion to Islam in the Medieval Period
became a Muslim or else, according to the assumptions made earlier, he would have been included in the genealogy; and, second, the individual converted at a young enough age to father a Muslim son. These two facts imply that the convert was sufficiently mature and independent to depart from family tradition and paternal authority and sufficiently young to be starting a family. All in all, the age of twenty-five seems like a plausible guess. Thus, the assumed period of conversion that has been assigned to every Persian name in the 469 genealogies commencing with such names is the period of birth assigned to that particular generation plus twenty-five years. T h e purpose of creating these categories and assigning individuals to them is, of course, to reduce promising but somewhat intractable data to a form amenable to quantitative analysis. Yet even then the task of analysis is not a simple one. Since the 469 biographies do not all come from the same period, one cannot simply find which percentage of converts belongs to which generation; and since the basic chronological periods are set by the biographical sources, they are of irregular length, and the periods of the prior generations overlap inconveniently. Moreover, the number of biographies differs greatly from one period to another. T o order the data in a regular manner, the Muslim calendar has been divided into twenty-five-year periods, and the number of names falling into each period has been tabulated (table 2). If a period of conversion falls entirely within a single twentyfive-year period, the total number of conversions in that period has been ascribed to those twenty-five years. If, as has usually happened, the conversion period extends over two or more twenty-five-year periods, the names have been apportioned according to what percentage of the conversion period falls into each of the twenty-five-year periods involved. This procedure has been followed on the presumption that the conversions were evenly distributed throughout the conversion period. T h e totals obtained in this way for each twenty-five-year period may be converted into percentages of the entire sample of 469. Graph 2 depicts these percentages; graph 3 shows the same data on a cumulative basis. These two curves, the first bell22
T h e Curve of Conversion in I r a n
A.D 6 4 6
670
695
719 743
767
791
816
840
864
888
913 937
961 985 1010 1034 1058
Graph 2. Bell curve of Iranian conversion.
Graph 3. Cumulative S-curve of Iranian conversion. 23
Conversion to Islam in the Medieval Period T a b l e 2. Number of converts per twenty-five-year period. City and period«
Generation»
Number in 1group 25
2650
5175
76100
C
7
—
—
ΝIV
D
8
—
—
1
3.5
ΝIV
E
2
—
.5
1
.5
ΝIV
F
2
NV
C
12
—
—
—
—
—
ΝV
D
12
—
—
—
—·
—
ΝV
E
9
—
—
—
.5
NV
F
2
—
—
.5
Ν VI
C
5
—
—·
—
Ν VI
D
13
Ν VI
E
6
—
—
—
—
—
Ν VII
C
5
—
—
—
—
—
Ν VII
D
44
_
_
_
_
_
_
Ν VII
E
19
_
_
_
_
_
_
Ν VII
F
5
—
—
—
—
—
Ν Vili
D
5
—
—
·—
—
—
—
Ν Vili
E
6
—
—
—
—
—
—
Ν Vili
F
7
—
—
—
—
—
—
Ν VIII
G
4
—
—
—
—
—
—
1
—
126150
ΝIV
1
—
101125
_
_
_
3
3
3.5
—
— _
_
_
—
_ — 5
5
1 _
—
3.5 .5
—
—
—
_
_
_ 2.5 —
2
Ν VIII
H
2
—
—
—
—
—
ΝIX
C
2
—
—
—
—
—
—
ΝIX
D
2
—
—
—
—
—
—
Ν IX
E
12
—
—
—
—
—
—
NIX
Η
2
NX
C
4
NX
D
6
NX
E
15
24
2
T h e Curve of Conversion in Iran
151175
176200
201225
226250
251275
276300
—
-
301325
326350
351- 376- 401375 400 425
1
3
7
7
2 —
— —
— 13
3.5 —
— 4.5
4 —
—
—
—
—
—
—
—
—
2.5 9
2.5
20
—
—
_
— —
20
1
1.5 -
—
2
1.5
2
.5
—
_
2
5.5
—
2
1
— 4
— 6
—
_
_
—
s
10
5
2.5 —
—
3
2.5
4
1
_
—
.5
_
_
_ _
_
_
_
_
_
_
_
_
_
_ —
—
—
—
—
—
—
—
—
—
—
—
—
—
.5
1.5
—
—
—
—
—
5.5
5.5
4 25
_
_
_
_
_
_
_
_
_
_
_ _ 2 1.5
2
2 .5
—
Conversion to Islam in the Medieval Period Table 2 (continued) Generation 8
Number in group
NX
F
7
NX
G
3
Is I
C
2
—
Is II
C
5
—
—
Is II
D
3
—
.5
Is II
E
1
Is III
C
9
Is III
D
10
—
—
—
Is III
E
8
—
—
Is III
F
1
Is III
G
1
Is IV
C
15
Is IV
D
47
City and period"
Is IV
E
18
Is IV
F
8
Is V
C
6
Is V
D
32
Is V
E
32
Is V
F
17
Is V
H
Total
125
.5
2650
1
5175
76100
101125
—
—
—
2.5
2.5
—
.5
—
—
—
—
—
1 —
2
.5
—
126150
4
—
.5
2
4.5
3.5
2.5
3.5
2
.5
.5
—
—
—
.5
—
—
—
—
—
4 —
—
—
—
4.5
8
4
1
—
—
—
3
3
—
—
.5
1
1
469
2
18
30.5
4
4.5
9
.5 43
Percentage of 469
.4
.95
1.9
3.8
6.5
9.2
Cumulative percentage
.4
1.35
3.25
7.05
13.55
22.75
a. Data from table 1.
shaped and the second S-shaped, are extremely interesting and tie the phenomenon of Islamic conversion in Iran directly to a substantial body of theoretical and empirical literature dealing with the diffusion of innovations. Some fifty years ago a conceptual transfusion was made from 26
The Curve of Conversion in Iran
151-
176-
201-
226-
251-
276-
301-
326-
351-
376-
401-
175
200
225
250
275
300
325
350
375
400
425
—
—
—
—·
—
—
—
—
— 1
4
2.5
2.5
1
2
.5
1
— 21.5 5.5
5.5 21.5
—
—
—
2.5 —
—
—
3
7
—
—
10.66
10.5
10.5
5.5
5.5
2
h[r,
82.5
77.165
1
10.66
2
10.66
8
62.165
2
—
—
31.165
1
—
17.5
10
9
13.8
17.6
16.4
13.3
6.6
3.7
2.1
1.9
36.55
54.15
70.55
83.85
90.45
94.15
96.25
98.15
3.5 .75 98.9
2.5
2
.5
.4
99.4
99.8
the area of biology into that of sociology. Laboratory experimentation had revealed that the pattern of growth of a population of fruit flies in an environment with a fixed and limited amount of food conformed to what is known as a logistic curve. T h i s curve, which takes its name from a logarithmic term in its equation, describes a stretched-out letter S.8 In the case of fruit flies, 27
Conversion to Islam in the Medieval Period this shape summarizes a population growth pattern characterized by slow initial growth during which reproduction takes place at a maximum rate but population size is limited by the paucity of reproducers, a very steep rise in the middle as the reproductive base expands, and gradual diminution at the end as the limit of the food supply is approached thus lowering the rate of reproduction. T h i s concept of a logistic growth curve of population was transferred, not illogically, to the field of human demography. 9 From demography, the laboratory-derived equation was subsequently applied to other problem areas, such as business market projections. But there were limitations to its utility. In the demographic field the Malthusian simplicity of the concept vitiated its usefulness. T h e subsistence base of human populations was not as inherently limited as that of fruit flies in a bottle. Analogous problems arose in other areas as well. Eventually, discussion of the logistic curve disappeared from most textbooks on statistics except as a historical curiosity. 10 T h e logistic curve was not altogether abandoned, however. Geographers and sociologists adopted it enthusiastically for studying technological and cultural diffusion, some going so far as to regard it as a fundamental natural law governing such phenomena. 1 1 T h i s enthusiasm arose from the observation that in a great many cases the logistic curve proved to be a rough approximation of the actual diffusion process. T h e adoption of all sorts of new techniques, from hybrid corn and weed killers to kindergartens and language laboratories, fit the S-curve pattern. T h e demographers' problem of an indefinite subsistence base did not affect the geographers and sociologists. If a new technique was coming into use in a given area, the "subsistence base" became the totality of individuals who had previously used the outmoded technique. According to the scheme, a few innovators would first adopt the new technique, then it would catch on with a bandwagon effect causing the steep rise in the middle of the curve, and finally there would be a steadily diminishing number of new adopters as the potential market for the new technique became saturated. In making this comparatively successful conceptual transfer 28
The Curve of Conversion in Iran of the logistic curve from the study of fruit flies to the study of innovation diffusion, however, the implication of the steep rise in the middle part of the curve necessarily changed. T h e simple reproduction rate explained this part of the curve in the case of the fruit flies; but in the case of individuals adopting a new device or technique, reproduction played no significant role. Instead, what was postulated was a notion of contagion by transfer of information. If one individual adopted a new technique, he might have five contacts with other potential adopters of whom perhaps three would see the superiority of the new technique and adopt it themselves. T h e n these three would each "infect" with the new idea three out of five of their potential adopter contacts, and so on. T h r o u g h this interpretation, innovation diffusion came to be considered primarily a function of access to information. In light of this successful application of the logistic curve to cases of innovation diffusion, it is appropriate to ask whether the S-shaped curve of Iranian conversion reflects a phenomenon similar to that studied by investigators of modern technological change. T o begin with, it may readily be established that the S-shaped curve of Iranian conversion follows with great exactness the logistic curve as defined by mathematical equation. Graph 4 shows the comparison between the conversion curve plotted on graph 3 and the logistic curve calculated from parameters derived from two points on the conversion curve. B u t does the nearly exact match really reveal anything other than the fact that one S-curve is much like another? T h i s question cannot be answered by mathematics. All that can be positively stated is that the conversion to Islam of 469 Iranian families followed a pattern that has repeatedly emerged in the study of innovation diffusion in the twentieth century. Beyond that, one must look to what has been deduced from the curve by students of innovation diffusion and see if some of their insights can be transferred to the question of religious conversion in medieval times. In doing this it is important to specify in exactly what sense the curve is felt to be a useful abstract description of the conversion process. Whatever meaning the curve may have obviously does not come from simple 29
Conversion to Islam in the Medieval Period
G r a p h 4. Comparison of S-shaped conversion curve with logistic curve.
reproduction as in the case of the fruit flies. Although reproduction is naturally an important factor in determining the size of a religious population at any particular time, assuming—and in the case of Islam this is a fairly safe assumption—that apostasy was rare and that all of a convert's descendants adhered to the new religion, reproduction can be ignored in the absence of concrete evidence for different birth rates for members of different religious communities. Religious conversion is more closely analogous to technological diffusion than it is to the population growth of fruit flies in a bottle. Perhaps, therefore, the contagion model would be more helpful in explaining the curve's apparent applicability to Iranian conversion. Yet here, too, the matter is not clear-cut. An iron plow may be superior to a wooden plow in an absolute sense that would make its adoption an obvious act for anyone who had the means to obtain it and to whom its superiority was clearly demonstrated. One religion is not in the same sense 30
The Curve of Conversion in Iran
clearly superior to another. Without prejudgement of the matter of revealed truth or moral worth, it may safely be said that whatever superiority one religion may have to another, it is not clearly demonstrable in a practical or mechanical sense. Inevitably, some individuals reject the arguments that are put forward by their converted neighbors and cling to their ancestral faith. In many instances a practical superiority of one religion over another may be created, often for brief lengths of time but sometimes permanently. Such phenomena as persecution, differential taxation based upon religion, and direct or indirect financial rewards for converts obviously have an effect upon the rate of conversion. Yet these selective factors are frequently unrelated or only peripheral to the doctrines of the religion involved; they may change from one period to another. 12 Indeed such factors may be as much the result of stages in the conversion process as actual causes of conversion. Still, even while admitting that the adoption of a new religion is more complicated than simply perceiving a new technique or implement to be superior to an old one, it would be wise to keep in mind that access to information is a prerequisite for conversion. A second useful insight from the field of innovation diffusion relates to the bell curve (graph 2) of which the S-curve is a summation. 13 This bell curve represents standard distribution and is used in a vast variety of situations. Essentially it says here that most families in the sample converted to Islam in the middle of the curve and that the probability of someone's converting markedly decreases as one moves forward or backward in time from the middle of the curve. Statistically it is possible to divide the entire body of converts into categories according to how improbable it was for them to convert at the time that they did. The greatest improbability attaches to the earliest converts, who are pioneers in the new religion, and to the latest converts, who are lagging behind the bulk of the population. T h e set of terms used by specialists on innovation diffusion for describing these different groups would be the following: the first 2.5 percent are "innovators," the next 13.5 percent are "early adopters," the next 34 percent are the "early majority," 31
Conversion to Islam in the Medieval Period the next 34 percent are the "late majority," the final 16 percent are "laggards." 1 4 W h a t is useful about this division i n t o categories is that it suggests that people who converted at different times had very different motives a n d experiences. T h e two things to consider, then, as clues to the conversion process gleaned f r o m the study of innovation diffusion are the idea of looking at the adoption of something new as a f u n c t i o n of access to information about that thing and the notion of dividing the entire body of converts into groups according to the probability of their converting at the time they did. W i t h these in m i n d we can proceed to a discussion of the actual process of Islamic conversion.
32
4. Conversion as a Social Process Conversion to Islam in the Middle Ages was somewhat different from conversion to other religions. There was, for example, no rite involved comparable to baptism.1 The verb aslama, meaning "he submitted [to God]," is used to describe the procedure of becoming a Muslim, but where the verb occurs, there is no elaboration to indicate the real content of the act. Religious treatises speak of the simple utterance of the confession of faith, the shahäda, as the defining characteristic of adherence to Islam; but there is disagreement as to the precise nature of this utterance, whether it must be made with the heart or by the lips only. Nevertheless, since we lack any concrete statement to the contrary, it may be hazarded that the formal process of conversion to Islam consisted primarily of speaking eight words. More certain is the fact that conversion did not depend upon a priestly individual. For one thing, Islam did not have priestly individuals in an institutionalized fashion; for another, there is frequent mention of people who converted at the hands of Muslims who had no noteworthy position of any kind as well as at the hands of those who held governmental but not religious office. For present purposes, therefore, formal conversion, in the sense of utterance of the confession of faith, is not as significant as what might be termed social conversion, that is, conversion involving movement from one religiously defined social community to another. It is reported, for example, that various North African Berber tribes converted to Islam and subsequently fell away. This type of conversion, although the words of the confession of faith might actually have been mouthed by
33
Conversion to Islam in the Medieval Period the tribesmen, would not qualify as social conversion. T h e tribesmen were not leaving their non-Muslim tribe to join a different, Muslim tribe. Apostasy probably had little significance because formal conversion alone meant very little in any social sense. T h e same could be said at a later period for the conversion of various T u r k i s h tribes and possibly for nontribal groups as well, some of which are reported to have apostatized several times as groups. What is implied by the term social conversion is individual rather than communal action. Having performed the act of conversion, the convert henceforth saw his identity in terms of the new religious community of which he had become a member. T h i s possibility, in turn, implies or presupposes a society in which social identity was normally defined in religious terms as opposed, say, to tribal or national terms. Such a society prevailed over much of the area that converted to Islam in the Middle Ages, but it was the product of relatively recent historical development and was by no means universal. One of the most important social changes of imperial R o m a n times was the gradual development of religion as the focus of social identity. Why this development occurred is not altogether clear, although it unquestionably was related to the changes in religious orientation associated with the rise of Christianity and to the vitiation of the meaning of citizenship within the context of the universal R o m a n Empire. 2 Yet if the origin of this broad social change remains cloudy, its practical dimensions are easier to delineate. Different religious groups were organized somewhat differently, and organization changed over time; but the following features are of common occurrence: 1. A formal authority structure, often hierarchical. Within the various sects of Christianity the priestly hierarchy is well known. Sassanid Zoroastrianism was likewise structured in a hierarchical fashion, as was Manichaeism. 3 With J u d a i s m the situation is more complex since the destruction of the second temple left the religion without a center, but authoritative bodies still were recognized in Babylonia (Iraq) and Jerusalem. 4 2. Some degree of legal autonomy. With Christianity and J u d a i s m the jurisdiction of religious courts grew steadily under 34
Conversion as a Social Process the late Roman Empire, and bodies of religious law were compiled and written down. T h e areas in which the church had the greatest legal autonomy were those of religion and personal status, but some administrative and taxation functions fell to the religious authorities as well. 5 T h e Zoroastrian legal system is little known, but what is known suggests a closer identification of church and state than was the case with the other religions.® Nevertheless, Zoroastrian legal decisions continued to be rendered after the Sassanid Empire had fallen to the Arabs. 7 3. Fusion of language with religion. Although some religions, such as Manichaeism, utilized whatever languages were encountered, the more common pattern was for a given religion to utilize as a religious language the ancestral tongue of the bulk of its adherents. For Egyptian Monophysitism this meant Coptic, for Syrian Monophysitism and Nestorianism, Syriac, for Zoroastrianism, Avestan, and for Judaism, Hebrew. For purposes of instruction or exegesis, of course, vernacular languages were used wherever and whenever the liturgical language had fallen out of common use. 4. Recognition of religious authorities as spokesmen for the religious community and as exemplars of religious life. Here the role of the religious authority as the successor of the civic official is quite marked. 8 Christian bishops and Zoroastrian priests sometimes appear as the primary authorities of sizable communities. Given the preeminence of religiously defined social identification in the immediately pre-Islamic period in the Middle East, the notion of social conversion is both significant and quite specific. It may be proposed as an axiom of religious conversion that the convert's expectations of his new religion will parallel his expectations of his old religion. In the case of an ecstatic convert, the old religion may have failed to satisfy spiritual expectations which seem to have greater promise of fulfillment in the new religion. Such a convert might appear as a religious malcontent before conversion and very likely as a zealot or spiritual athlete after conversion. Most converts are nonecstatic, however. People who are more or less satisfied with their previous religious life and who convert more for mundane 35
Conversion to Islam in the Medieval Period than for spiritual reasons find life in the new religion more attractive the closer it approximates life in the old. In other words, a socially stodgy and conservative Zoroastrian would be more likely to become, upon conversion, a socially stodgy and conservative Muslim than a wild-eyed fanatic; and the attraction of Islam would be related to his perception of it as a religion that had room for stodgy conservatism.9 Naturally, no hard and fast determination of the individual reasons for and reactions to conversion can be derived from such an axiom, but the overall implication of the axiom is that as conversion progresses, the new religion becomes, in its social dimension, increasingly like the old. This is not to say that different religious doctrines or institutions inevitably have continuities or even close parallels, but simply that the social functions that were served by the previous religion are likely to be served by the new religion to an increasing degree as the membership of the new religion becomes dominated by converts. Since for Middle Eastern society in the pre-Islamic period, as has already been remarked, religion prevailed as the source of the individual's social identity, conversion to Islam normally resulted in a major change in the convert's social identification; but in the long run, conversion gave rise to strong pressures that affected the course of development of the new religion. For the individual, conversion meant that he was no longer a part of his old community. He may actually have been regarded as legally dead in the eyes of his former coreligionists. 10 Consequently, for defining his social life he looked to new exemplars to replace the bishop or rabbi or mobad (Zoroastrian priest) who had previously filled that role. He came to think of a new language, Arabic, as a sacred language, and he normally gave his children names from the Arabic onomasticon. If there were too few Muslims in his home community for him to feel that he could live a good Muslim life—a concept that was inevitably affected by how he had conceived a good Christian, Jewish, or Zoroastrian life—he was likely to emigrate to a community with a larger Muslim population. This last tendency might well be enhanced by economics insofar as a convert might find himself frozen out of his customary participation in 36
Conversion as a Social Process the economic life of his erstwhile coreligionists and consequently seek out a large Muslim community in whose economic life he could participate more or less in his accustomed fashion. 11 T o gether these changes constituted what has been termed social conversion. A given individual could obviously utter the profession of faith and go through none of these social changes, but such an individual would scarcely be noticeable as a member of the Muslim community. What built the Muslim community as a distinct and historically visible social entity was social conversion. It was also social rather than formal conversion that created pressures for change which affected the course of Islamic religious development. In the earliest stages of Islam, when the Muslim community was still confined to the Arabian peninsula, social conversion was as yet an uncommon phenomenon. It is not that the bedouin, of whose conversion the Quran speaks disparagingly, were merely mouthing words without understanding or appreciating their meaning; it is rather that the religious expectations of the bedouin were colored by their pre-Islamic religious experience. Outside of Mecca and the Yemen, religion in tribal Arabia impinged only slightly upon the social lives of the Arabs. 12 Even among nominally Christian Arabs religion appears to have had scant social impact. Sacred shrines and mantic seers known as kähins played only an occasional part in tribal life. T h e rules governing the normal turning points in personal life—birth, puberty, marriage, death—were determined by tribal custom more than by religious doctrine and required no supervision by priestly persons. 1:1 As a consequence of this tangential relationship of religion to Arab tribal life, the religious development of Islam was affected primarily by the original Meccan community where religion meant more than it did elsewhere in the peninsula. T h e impact of the conversion of the Arab tribesmen was most of all a steady dilution of the early sense of Islam as a social community. An Islam dominated numerically by converts of tribal nomadic origin was an Islam threatened with relegation to a subordinate social role. T h e Umayyads did not deliberately 37
Conversion to Islam in the Medieval Period promote an Arab kingdom at the expense of an Islamic caliphate; the tribal preponderance of the Muslim population dictated a course to be followed which they could not ignore and still continue in power. In part, indeed in large part, Islam survived this threat through efforts made in Mecca, Medina, Kufa, and Basra during its first century and a half by an Arab Muslim society that was tribal to a steadily decreasing degree and not nomadic at all. T h e pious circles of these cities raised and debated religious questions that were of little concern to most Muslims of that time. But socially speaking, the Muslim societies of these four cities were still dominated by Arab tribal identity to a strong degree. 14 T h e questions that were raised and debated were more concerned with the relation of the individual believer to God than the relation of the individual Muslim to his fellows. T h e socialization of Islamic thought was largely the product of pressures brought by non-Arab converts from the Christian, Jewish, or Zoroastrian communities. Insofar as Islamic social life was oriented toward Arab tribes to which all Muslims belonged by birth, or by fictive adoption in the case of converts, Islam did not provide the social structure afforded by the religions whence converts were to be made. T h i s lack was a barrier to conversion which could be broken down only by developing such a social structure on the basis of the Quran and the experience of the early community in Mecca and Medina. Certainly there was no conscious effort to create a society that would "compete" with the various non-Muslim societies in attractiveness. Yet it is hardly surprising that the interests of converts frequently centered around social questions as they searched in their new religion for the social structure they had abandoned by converting and that in their search they adumbrated the shape of such a structure. Looking back at the four characteristics mentioned earlier as hallmarks of pre-Islamic religious social structure, the first of them, a formal authority structure, was most strongly promoted in Islam by converts. Ibn al-Muqaffa', a convert from Zoroastrianism, is the best known early advocate of caesaropapism as a model for the caliphate, and the development of the cen38
Conversion as a Social Process tralized judicial system in early Abbasid times was a concrete step in that direction. 15 T h e r e may also be something of the same impulse in the attraction of many converts to Shiism, with its rigid authority structure based upon the theory of the Imamate. T h e truest expression of this drive, however, was the later emergence of the Muslim religious scholars, the ulama, as a quasi-priestly class. T h e second point, legal autonomy, is exemplified in the development of Muslim religious law, the sharia. Although civil law has always existed in Islam alongside religious law, it has been comparatively insignificant. Yet Islamic religious law, for all its importance, did not take shape until some two centuries after the origin of the religion. During the Umayyad period, when Arabs constituted the vast majority of all Muslims, the ideal of judgment was the sage determination of the caliph acting more as a tribal arbitrator than as a religious figure. T h e numerous stories of the hilm or judicial forbearance of Mu'äwiya bespeak this ideal. 16 Only when the convert population began to become numerically significant did religion come to the fore in the legal field. T h e third point, fusion of language with religion, needs some comment. Obviously, Arabic was as important when the Arabs were numerically predominant as it was after the convert population began to make itself felt. Formal study of the Arabic language for purposes of elucidating religious texts was a later development, however: only then did Arabic begin to become a sacred tongue of the type familiar in the other religious communities. It is noteworthy that many of the pioneering grammarians of the Arabic language were themselves non-Arabs. T h e fourth point is possibly the most important. It deals with the recognition of religious authorities as the spokesmen for religious communities. Since medieval Islam did not develop a clerical hierarchy or, indeed, a formal priestly office of any kind, no direct parallel to the bishop, rabbi, or mobad can be put forward. T h e converts to Islam did desire guidance, however, in how to live a good Muslim life. Since the role of exemplar was not filled by a formally invested religious functionary, certain obviously pious individuals came to be in39
Conversion to Islam in the Medieval Period formally recognized as exemplars. Recognition m i g h t derive f r o m m e m b e r s h i p in or association with the P r o p h e t ' s family or the families of other saintly early figures, b u t m o r e o f t e n it went to individuals w h o simply appeared to be good Muslims, notably to those w h o were studious and ascetic. Indeed, to some extent the Muslim p o p u l a t i o n of convert origin p r o b a b l y recognized as exemplars of good Muslim life individuals w h o m they j u d g e d to be so qualified not by still-uncertain Islamic social standards b u t by standards they were accustomed to associate with formally defined religious leaders in the non-Islamic communities. T h e M u s l i m ascetic or zähid, for example, may have taken the place in the m i n d s of converts f r o m Christianity previously filled by the m o n k . N o t only did the pious, the ascetic, a n d the studious achieve notice in this fashion, b u t they had the role of spokesman t h r u s t u p o n them. T h e convert p o p u l a t i o n used t h e m as their life models as well as their c o m m u n a l voices in the way M u s l i m A r a b tribesmen had previously used their tribal shaikhs. 1 7 Lacking a formally constituted priesthood, the c o m m u n i t y looked to t h e m as surrogate priests. Ultimately, the p r o d u c t of this evolution was the a c c u m u l a t i o n of power a n d a u t h o r i t y in the hands of this u n a p p o i n t e d clergy. In this entire process, it must be realized, the role of the Muslim p o p u l a t i o n of non-Arab convert origin was primarily that of applying pressure for social-religious d e v e l o p m e n t a n d p r o v i d i n g a receptive audience. T h e convert p o p u l a t i o n was proportionately still small w h e n this pressure began to be felt, and the individuals w h o responded were m o r e o f t e n t h a n n o t Arabs, at least u p to the t e n t h c e n t u r y w h e n the convert population began to overwhelm the A r a b p o p u l a t i o n numerically. T o be sure, specific non-Arabs m a d e significant c o n t r i b u t i o n s as the pressure began to b r i n g responses, b u t they were m u c h m o r e visible on a local level as social exemplars or proto-ulama t h a n they were at the top level of intellectual achievement. T h i s was u n d o u b t e d l y owing to m u c h of the convert p o p u l a t i o n ' s feeling that the lifestyle of someone of their own linguistic a n d social b a c k g r o u n d provided a m o r e m e a n i n g f u l model t h a n that provided by an A r a b w h o was ethnically a n d linguistically 40
Conversion as a Social Process a foreigner. How the fuller accomplishment of the conversion process affected this situation must be left for a later discussion, however. W h a t is necessary at this point is to introduce a second axiom of conversion, which states that leaving aside ecstatic converts, no one willingly converts from one religion to another if by virtue of conversion he markedly lowers his social status. More starkly put, if an emperor converts to a religion of slaves, he does not become a slave: the religion becomes a religion of emperors. T h e import of this axiom for the history of Islamic conversion in the medieval Middle East is great indeed. T h e conquests of Islam are just as properly viewed as the conquests of the Arabs. Being an Arab and being a Muslim were so much the same thing that people in the conquered territories were sometimes unaware of the specific religious character of the invaders. T h i s identity of Arab and Muslim continued to be the predominant view for over a century. As already mentioned, non-Arabs who converted to Islam were obliged to become mawäll, that is, fictive members of Arab tribes. 18 Only thus could they obtain any social identity as Muslims. Yet being a mawlä of an Arab tribe was fraught with disadvantages. Mawäll were regarded as racially inferior by many Arabs because they did not truly share the pure blood lineage that was the focus of tribal honor and loyalty. T h e y were discriminated against in marriage, denied inclusion on the military payroll, and made to suffer revilement and social slights. 19 Applying the second axiom of conversion, one must conclude that those who converted to Islam during the period when the mawäll were so heavily stigmatized must have been people for whom being second-class Arabs was superior to any other options. I n effect, for the Umayyad period, this statement implies that the major sources of non-Arab converts were two: prisoners of war who might seek through conversion to escape slavery and people, such as poor farm laborers, from the very lowest classes. Civil servants working for the new Arab rulers were not obliged to convert to keep their jobs. Certain others who felt economic pressure under the new regime preferred to convert to Christian41
Conversion to Islam in the Medieval Period ity and become clerics, thus evading taxation, rather than convert to Islam and become lackeys of the Arabs. 20 This highly discriminatory situation eventually changed, but while it obtained it shaped the character of the first several generations of converts. T h e product was a mixture of a relatively few upper class converts, largely former prisoners of war, and a larger number of riffraff. T h e former found a better reception in the new religious community than the latter, who were on occasion herded about like cattle; but all suffered from their subordinate status. It is scarcely surprising that some Arabs, such as Ibn Qutaiba, formed the opinion that non-Arab converts to Islam were, in general, the dregs of society. 21 It was this mixed group, however, that initially, by their very existence, posed questions about the social character of Islam which had such farreaching results in later centuries. T o examine stage by stage how this process occurred, we must return to a consideration of Iran and make use of the curve of conversion derived in the last chapter.
42
5. The Development of Islamic Society in Iran It has already been indicated that Iran was the first major segment of the vast territory conquered by the Arabs to accept their religion. It has further been argued that the acceptance of Islam by the population followed a specific pattern, namely, an S-shaped curve. T o appreciate the likelihood and potential importance of these two assertions, one must look at the political and religious history of Iran during the first four centuries of Muslim rule, the proposed duration of the conversion process, and see whether it is illuminated and made more comprehensible by them. T h e 5-curve in graph 3 has been reproduced in graph 5 with a number of political events and trends plotted above it and a number of religious developments plotted below it. If this graph is reasonably accurate, it should depict in summary form the political preoccupations and religious outlook of the Iranian population as it went through the dramatic change from being overwhelmingly non-Muslim to being overwhelmingly Muslim. However, it must be kept in mind that the completion of the conversion process described by the 5-curve does not represent total conversion of the Iranian population since a certain percentage remained non-Muslim well beyond A.H. 400. As a guess, a figure of 20 percent might be hazarded for this population of adamant non-Muslims, whose number was reduced only very slowly after the completion of the basic conversion process. If we look first at the arena of political activity, three noteworthy points spring immediately into view. First, the graph supports the conclusions of recent investigators of the Abbasid revolution who have seen it as primarily an Arab movement, despite its origin in eastern Iran. 1 T h e earlier idea that it rep43
Conversion to Islam in the Medieval Period
. Innovators
Early adopters
Α Λ Λ Early majority Late majority
Laggards
Graph 5. Correlation of historical developments with Iranian conversion curve.
resents a massive intrusion of Iranian influence into Muslim politics is impossible to sustain when it is realized that the population of Iran was only about 8 percent Muslim in 750 when the Abbasids came to power. That is, the conversion process represented by the S-curve was 10 percent completed, implying the conversion of 8 percent of the total population if the completion of the process is considered to leave as many as 20 percent unconverted. T h e second obvious observation is that non-Muslim revolts die out as the middle point in the conversion process is reached. From the assassination in 754 of Abu Muslim, the great Iranian leader of the Abbasid revolution, until the capture and execution of Bäbak, the last leader of a major non-Muslim revolt, in 838, Iran experienced an almost unbroken series of insurrec44
The Development of Islamic Society in Iran tions which were characterized by several common features: (1) they utilized the name of Abu Muslim as a rallying point, often through a doctrine of metempsychosis by virtue of which the leader of the revolt was viewed as a reincarnation of Abu Muslim; (2) while being non-Muslim, they were equally divorced from the official Zoroastrian establishment; (3) they shared a complex of terms and epithets—Mazdakite, Khurramdiniya, Rhurrâmï—which suggests a common appeal if not actual doctrinal similarity; (4) as time progressed, they receded to areas increasingly remote from centers of Muslim government and power. 2 Looking at these revolts in relation to the conversion curve, it is easy to visualize them as drawing upon a steadily shrinking base among the non-Muslim population. T o be sure, over half the population was still unconverted when the revolts died out; but since the percentage of the non-Muslim population that was ever inclined to participate in these desperate uprisings against massive Arab military power could never have been very large, it is quite conceivable that by the time the conversion process was half completed, the pool of potential recruits to such movements had dwindled to nothing. In this respect, it is of great interest that Dailam, a remote part of the Elburz mountain range which had been involved to some extent in the revolt of Bäbak, the last in the series, became a center for the development of revolutionary Zaidi Shiism. Indeed, direct connections can be made between the last of the Khurramis and the early adherents to the doctrines of Zaidi Shii preachers. 3 Here, quite tangibly, one can see the turnover point between revolution against Islam and revolution within Islam. T h e geographical recession of the revolts into ever more remote districts is also noteworthy from the point of view of the conversion process. T h e first revolts took place around major cities in eastern Iran and the last ones in the mountainous and sparsely settled region of Azerbaijan, yet there was no marked increase in the availability of military forces to crush the revolts, a factor that could possibly explain the recession. An alternative reason for this recession can be found in the exposition in chapter 3 of the S-shaped conversion curve. T h e r e an 45
Conversion to Islam in the Medieval Period analogy was drawn to the process of innovation diffusion in the twentieth century, which has been analyzed as primarily a result of dissemination of information. Innumerable refinements in the conceptualization of the information dissemination process have been proposed to take into account such things as mass media, relative degrees of respect given to various sources of information, and so forth. 4 Some of these refinements may be applicable to the study of Islamic conversion, but for most of them the data needed for verification and quantification are totally absent. One refinement, however, seems to be clearly applicable. T h i s is the idea of a retardation in innovation waves as the areas involved become more and more remote from centers of dispersal. In other words, information about a new idea filters slowly into remote areas, and this has the effect of stretching out the 5-shaped curve of innovation adoption. If there is a stage in the process of Islamic conversion, therefore, that conduces toward the development of anti-Muslim revolts, that stage should be attained, within a general conversion area, at ever later dates in increasingly remote localities. T h e effect of this sort of attenuation of the 5-shaped curve would thus be a series of ideologically or socially similar revolts taking place in increasingly remote areas as time progresses. Moreover, the revolts should die out as the conversion process gets beyond this stage in the most remote parts of the general conversion area. This is exactly what appears to happen in Iran, and it will be important to observe whether analogous happenings take place in other conversion areas. T h e third political phenomenon clearly shown on graph 5 is the appearance of independent dynasties of Iranian Muslim rulers beginning right around the midpoint on the curve. These dynasties become increasingly independent as the conversion process progresses and the control of the Abbasid caliphate correspondingly diminishes. What underlies this phenomenon is the basic fact that an independent Iranian Muslim dynasty presupposes a largeenough population of Iranian Muslims to furnish an army. T h e only other alternative would be for an Iranian dynasty to have 46
The Development of Islamic Society in Iran arisen based upon an Arab soldiery; and this, under the conditions of strong ethnic animosity that obtained in the first four centuries of Islam, was very unlikely. In this respect, it is of particular interest that the first of the notable Iranian dynasties, the Tahirid, which enjoyed de facto independence but retained its formal allegiance to the caliphate, was established by the general Tähir who commanded the army of the caliph alMa'mun in his civil war against his brother al-Amïn (809-813). This was the first genuinely powerful army in Islamic history composed predominantly of Iranians, and it was this military command that provided the continuing support for Tahirid autonomy. If the Tahirids became independent when Iran was approximately 40 percent Muslim (halfway through the conversion process), the Buyids and Samanids attained their apogees when Iran was rapidly approaching full conversion at a level we have guessed to be about 80 percent of the total population. Since this span of time encompasses an enormous change in the balance between Muslim and non-Muslim, it would be surprising if no sign of it appeared in the actions of the independent Iranian rulers. T h e question is, how might this change in the proportion of Muslims in the population be manifested? There may be several answers to this question, but one seems quite clear. At the beginning of the period of the independent Iranian dynasties, non-Muslims were still in the majority, and the possibility of an ultimate reversal of the process of Islamization was very real in people's minds. Whatever could be interpreted as playing upon political sympathies harking back to the pre-Islamic era was suspected of being anti-Muslim. Iranian independence was very suspect since it might conceivably lead to the expulsion of Islam from Iran. T h e trial and execution of the powerful Iranian general al-Afshin in 841 on charges of being a crypto-Zoroastrian, or at least an insincere Muslim, is a straw in the wind. 5 Given this sentiment, the Tahirids had to be extremely cautious in their assertion of independence. On the one hand, this meant refraining from a formal rupture with the caliphate, but more important, it meant avoiding any overt assertion of Iranian identity. Some Persian poetry in Arabic 47
Conversion to Islam in the Medieval Period script was probably composed at the T a h i r i d court, but it was certainly not strongly encouraged; nor did the Tahirids use any of the regalia or imperial terminology of the Sassanid Empire. 6 Indeed, they even went beyond stressing their origin as mawâlï of an Arab tribe and let it be believed that they were genuinely Arab. 7 Needless to say, these things were not done for the benefit of their Iranian soldiery but were simply efforts to avoid any possible charge of being anti-Muslim in a religious atmosphere in which Islam was still a minority religion. T h e Saffarid dynasty arose in southern Iran at a somewhat later stage in the conversion process, and their long history in their area of origin shows that they had substantial roots among the local population, roots that the Tahirids seem to have lacked. 8 T h e Saffarids made no great effort to play upon an Arab background or connection, and they certainly did not accept the authority of the caliphate, although they eventually reached a modus vivendi with it. What is most indicative of the changing religious climate, however, is the first appearance with the Saffarids of Persian panegyric poetry containing imagery drawn from the imperial Iranian past. 9 T h e regalia of the Sassanids did not come into play at this time, but symbols from the legendary earliest Iranian dynasty, the Kayanids, were used in an obvious attempt at reinforcing the legitimacy of the upstart Saffarids, who were far from being noble in their origins. Although this type of imagery does not appear to have been extensively used at this time, that it was used at all indicates that some people were beginning to imagine the possibility of a Muslim Iranian empire. A few decades earlier, such an empire would have seemed to involve an inherent contradiction. In the first half of the tenth century, the Samanids inaugurated a further stage in the gradual legitimation of the Iranian imperial past in Muslim eyes.10 Dari, the court Persian of the Sassanid Empire, became the court language, albeit charged with Arabic vocabulary and written in Arabic script. Persian poetry flourished, especially poetry recounting the epic exploits of the pre-Islamic Iranian shahs and heroes. Iranian Islam had become such an accepted concept that at-Tabarï's great chronicle, composed in Arabic by a historian of Iranian origin, was 48
The Development oí Islamic Society in Iran
abridged and translated into Persian by the Samanid vizir Bal'amï. Iranian Islam deserved a Persian-language Islamic history. T h e climax of the development being traced here came with the Bnyids in the second half of the tenth century. 11 Persian language and pre-Islamic Iranian culture were by then commonplace and required no further legitimation. The innovation of the Buyids was the use of the imperial regalia of the Sassanid Empire. A royal crown, the title Shähanshäh, imperial iconography on medals—all these and more were utilized by the rude Buyid princes to legitimize their regime in the eyes of an almost entirely Muslim Iranian population. There was no serious consideration that these acts might be construed as a threat to the existence of Islam in Iran because the progress of conversion had rendered unimaginable the idea of a non-Muslim Iran. From that time onward Iran's imperial past became wedded to her Islamic present to such a degree that the two titles adopted simultaneously by the Seljuq Turks when they conquered Iran were Sultan and Shähanshäh. 12 Turning from the arena of political history to that of religious and social history, we can make out a similar correspondence with the S-shaped curve of conversion, but the relationship, though more illuminating, is much more complicated. T o see it we must recall the two axioms of conversion proposed in the last chapter, which state, first, that the nonecstatic convert's social expectations of his new religion parallel his social expectations of his old religion and, second, that no one willingly converts from one religion to another if, by -virtue of conversion, he significantly lowers his social status. We must also consider once again an insight drawn from the study of modern innovation diffusion mentioned in chapter 3, namely, that the total body of adopters, in this case converts, can be divided into social categories according to the probability of their converting when they did. The point made by investigators of innovation diffusion is essentially this: the adoption of a new technique or device follows a standard distribution pattern, that is, the bell-shaped curve of which the S-shaped curve is a summation. The total 49
Conversion to Islam in the Medieval Period area enclosed beneath the bell-shaped curve is equal to 100 percent of all adopters. Now, it is a property of this curve that if it is cut by vertical lines drawn at regular intervals that are statistically determined, intervals known as standard deviations, the areas enclosed under the separate segments of the curve thus created will constitute regular percentages of the total area underneath the curve. As can be seen in graph 6, the area cut off by a vertical line drawn one standard deviation on either side of the mean, which is the center line of the bell curve, includes 34 percent of the total area under the entire curve. T h e area between the line marking one standard deviation and two standard deviations encloses 13.5 percent, and so forth. T h e importance of these divisions is that they are indicators
deviation
Graph 6. Adopter categories on bell- and S-shaped curves.
50
The Development of Islamic Society in Iran of the probability of adoption. Anyone who adopts the innovation beyond the line marking two standard deviations from the mean is performing an exceedingly improbable act in terms of the total population of eventual adopters. Anyone who adopts within the next category is acting improbably, but significantly less so than the adopters in the first category. T h e labels that have been assigned to these various groups have been mentioned already: the first 2.5 percent are "innovators," the next 13.5 percent are "early adopters," the next 34 percent are the "early majority," the next 34 percent are the "late majority," and the final 16 percent are "laggards." In this way, the entire body of adopters can be divided into artificial, but statistically meaningful, categories whose behavior may be analyzed separately. Studies of technological innovation pursued by this technique have yielded results that seem largely to reinforce the obvious: the better educated, wealthier, more cosmopolitan and adventurous segments of the population are generally either innovators or early adopters. 1 3 Also in the early adopter category may be found highly respected social leaders whose adoption paves the way for a flood of popularity of adoption represented by the early and late majorities. However, conversion to a new religion is not fully equatable with adoption of hybrid corn or a new weed killer. T h e benefits are normally less tangible and less immediately realized. Indeed, there may be no worldly benefits whatsoever. Consequently, it should not be expected that the social profile of the statistically defined categories of religious converts will be the same as the social profile of analogous categories in cases of technological diffusion. T o apply this methodological approach to the concrete case of the conversion of Iran to Islam, one first constructs categories. Reading dates from the S-shaped conversion curve in graph 5, we note that the first category, the innovators, includes those individuals who converted before the year 695, who equal approximately 2.5 percent of all converts. T h e second category, the early adopters, includes everyone who converted between 695 and 762, or about 13.5 percent of all converts. T h e early majority is composed of the 34 percent who converted between 762 and 820, the late majority of the 34 percent who converted be51
Conversion to Islam in the Medieval Period
tween 820 and 875, and the laggards are the remaining 16 percent whose conversion was essentially completed by 1009. In defining groups of converts according to these categories, of course, one is implicitly assuming that the groups differ in their social complexion ,in ways at least broadly analogous to the types of group differences found in instances of technological diffusion. This assumption, in turn, has a profound effect on a question that is invariably raised with regard to religious conversion, the question "Why?" What is implied by these categories is that the reasons for conversion differ according to the time of conversion. That is, the question must be answered very differently according to the period or stage of conversion of which it is asked. This assumption was already made in the last chapter when an axiom of conversion was applied to the earliest period of conversion. T h e question was asked: who was willing to convert at a time when conversion entailed highly discriminatory mawâlï status? The answer was: those people for whom becoming mawâlï did not result in a marked lowering of social status, namely, prisoners of war, who might thereby escape slavery, and people of very low social status who had no social status to lose. Unquestionably, one can find individuals who do fit into either category; but as groups, these are the most likely converts at this improbably early date, and there is evidence of substantial conversion for both of them. 14 It should be noted, of course, that these mawâlï are the innovators, the first 2.5 percent on the conversion scale, and that they are very different indeed from the type of innovator found in cases of technological diffusion. Moving to the early adopter group, the 13.5 percent who converted between 695 and 762, we find a substantially different situation. During this period the stigma attached to the word mawlä disappeared, and by the end of the period the word has ceased to signify a non-Arab convert altogether. As a result, it became increasingly possible to convert without being legally stigmatized within the community to which one was converting. But there was still a great stigma attached to the convert by members of the community from which he was converting. This 52
The Development of Islamic Society in Iran stigma can sometimes be seen in legal terms, as in the Zoroastrian religious laws according to which conversion to Islam was tantamount to legal death within the Zoroastrian community. It was probably more strongly felt, however, in social and commercial life. T h e marriageability of a convert's children was greatly diminished, and his participation in community affairs was severely limited. 1 5 T h e r e is also evidence that non-Muslims were encouraged, when possible, to do business with coreligionists in preference to Muslims. 1 6 T h e effect of this ostracism of converts by the numerically dominant non-Muslim communities was to delimit the social character of the early adopters. People who converted during this early period were those who could hope to find within the Muslim community to which they were converting a substitute for the social life they were abandoning by leaving their old religion. Since there were as yet very few Iranian converts, and the Arab Muslim population in Iran was concentrated in and around certain cities that had been selected as provincial administrative centers, it was extremely difficult to convert to Islam and lead a normal life without moving to one of the Arab governing centers. Consequently, the early adopters were largely people who were able to change residence upon conversion. W i t h i n this group could be found artisans, merchants, and religious and state functionaries; but what is most important is the general exclusion from the group of rural landowners. T h e s e were the dihqäns, a kind of landed gentry who constituted the backbone of the Iranian upper class in pre-Islamic times. Some of them converted, of course, and took up urban residence while retaining their estates; but for those who did not live close to an Arab governing center, conversion was not an attractive proposition. T h e innovators had moved to the Arab cities upon conversion for the same reasons that the early adopters did, but they had frequently done little more in the urban milieu than swell the ranks of the restive poor. W i t h the urban migration of the early adopters an important new development began. T h e relatively small towns on the Iranian plateau that the Arabs had chosen as administrative centers began to increase rapidly in size. A
53
Conversion to Islam in the Medieval Period period of massive urbanization commenced, a period that was characterized by the cities being Muslim in religion, the new population being predominantly Iranian, and the growth being unplanned and more or less unexpected by the Arab rulers. This urbanization, which started around the time of the Abbasid revolution, was carried to unprecedented levels by the continuing migration of new Muslims to the cities during the period of the early and later majorities, after which urban growth quickly tapered off. T h e people who gained the greatest advantage from the urbanization process, however, were the families of the early adopters whose timely conversion allowed them to get in on the ground floor of an urban land boom. It was their names that became affixed to the newly developed streets and quarters. 17 However, the new Muslim cities of the Iranian plateau were not important solely because of the wealth and grandeur they gradually acquired. They were important also because they represented the first extensive milieu in which a mass society of non-Arab converts to Islam exerted pressure for the elaboration of specifically Islamic social institutions. T h e assertion has frequently been made that Islam is, in its essence, an urban religion, and that a fully realized Muslim life can only be lived in an urban setting. 18 Against this commonplace observation must be weighed the fact that it is quite unlikely that at any given time prior to the twentieth century (excluding the primitive community in Mecca and Medina) more than 20 percent of the world's Muslims were living in cities. From this juxtaposition of statements one of two conclusions can be drawn: either Islam was not a characteristically urban religion, or most Muslims in premodern times did not live full Muslim lives. Since the second of these two conclusions involves a virtual contradiction in terms, one must consider very seriously the first conclusion. Is Islam an urban religion, or is it merely an illusion that it is so? One the one hand, there is an obvious snare in this type of question in that it suggests Islam to be a coherent unity with only one correct or full type of observance, an obviously absurd proposition. Yet on the other hand, leaving this cavil aside, the proponents of urban Islam as true Islam are able to muster persuasive arguments. Islam prior to the twelfth century 54
The Development of Islamic Society in Iran does appear to have a distinctly urban cast, although this is much less pronounced in the post-Mongol centuries. T h e resolution of this dilemma can best be found in examining the process of conversion. T h e reason that Islam has an urban cast before the twelfth century is that the migration of converts to the expanding Muslim cities made those cities the cradles of Muslim social and institutional development. Rural Islam, encompassing no less valid forms of Islamic religious observance, developed substantially later, partly because the remoteness of the countryside slowed the pace of conversion. It is characterized by rural Sufi organizations, local pilgrimages, veneration of saints, and the like. Furthermore, rural Islam being essentially illiterate, the Islam of the literate tradition is predominantly urban. Islamic urbanization began with the cantonments established by the Arabs in the aftermath of the initial conquests, primarily in Iraq; but since these cantonments were organized according to Arab tribal social patterns, they did not immediately pose a challenge to the Islamic religion. T h e migration to these incipient cities of non-Arab converts in the innovator period planted the seeds of discontent but did not markedly change the social situation. T h e immigrants were few, lacking in social status, and legally affiliated with the Arab tribes. W i t h the coming of the period of early adopters the situation changed appreciably. T h e immigrants were richer, more highly respected, and much more numerous. T h e i r conversion was directly connected with their urban residence. Moreover, they were no longer mawäli; the Arab tribal social system steadily faded away for Arab and non-Arab alike. W h a t the new converts sought, and what detribalized Arabs sought as well, was an expression of Islam, an Islamic way of life, that would satisfy their urge to have their new religion fulfill the social role of their old religion (or tribe, in the case of the Arabs) but within a specifically urban setting. T h e result was the elaboration of a variety of legal, educational, and communal institutions that were focused upon the city. T h e qadi court and the witness system in law, oral transmission of hadith in education, and the jämi' mosque as a com55
Conversion to Islam in the Medieval Period munity focus are the peculiarly urban religious institutions that attained prominence during the period of the early adopters. In the long run, none of these institutions, with the exception of the qadi court, retained its exclusive preeminence in the Islamic religion; but because of their early appearance they took on the aspect of being more basic to Islam than certain later phenomena. These institutions played important roles in the lives of the newly urbanized early adopters. Whereas many of the innovators had been lowly and desperate types who were easily induced to join heterodox and revolutionary Islamic sects (and probably non-Islamic sects) that offered hope to the hopeless, the early adopters tended to be people of greater substance and higher status who strove in every way to be good Muslims. After all, they had taken the risk of converting at a time when few of their countrymen were doing so, and if they failed to obtain acceptance as good Muslims, they could not hope to rejoin their previous non-Muslim community. As a result, the converts of this period and their descendants tended overall to be staunch supporters of the caliphal government, adherents of the theological and legal viewpoints most in favor with the government or with the local Arab community—in Iran this meant mostly the Mu'tazili theological school and the Hanafi legal school— and believers in the principle that the early bird catches the worm. T h e risk that they took in converting at an improbably early date, in adopting the religion of the alien ruling minority, seemed to entitle them to benefit from being part of the ruling elite. This is not to say that they converted for gain. They may have converted for any number of personal reasons. It is simply that certain benefits appeared to flow automatically and rightly from the act of conversion. T h e early adopters and the more respectable of the innovators joined with the Arabs to form the "establishment" in the nascent Muslim (as opposed to Arab) urban society. And whenever possible, they passed on this status and the attitude that went with it to their descendants. They formed the core of a potential hereditary religious aristocracy, but the full development of this potential was thwarted by the actions and attitudes of later groups of converts.
56
The Development of Islamic Society in Iran T h e next category, the early majority, comprises the 34 percent of all converts who converted between 762 and 820. T h i s period marks the beginning of the bandwagon phenomenon that resulted in over 60 percent of all converts making their move into the Muslim community in the century 772-869. Whatever bars to conversion had previously existed must have dropped very rapidly during this period. In particular, the rural landowners of pre-Islamic Iran became increasingly absorbed into the Muslim community along, one may believe, with the villagers who farmed their lands and looked to them for social and religious leadership. Basically, there are no grounds for distinguishing between the early and late majorities. With the removal or lessening of bars to conversion, such as ostracism or persecution by non-Muslims, the primary factor in the determination of the conversion process during this period was probably the dissemination of information. Increasingly, Islam was regarded as a permanent and irreversible aspect of Iranian life, and the idea spread that adherence to Islam was an absolute good in a manner analogous to the absolute advantage of a superior technological process or device. T h i s bandwagon process differs in one important respect from the process of technological diffusion, however. T h e study of technological diffusion has led to the conclusion that innovators and early adopters reap proportionally greater benefits from the new technique than members of the early and late majorities who adopt it after its market advantage has been diluted. 1 9 Insofar as the innovators and early adopters of Islam became part of the establishment in the emerging Islamic social system, this generalization holds true for the conversion process. What is different is the fact that in cases of technological diffusion, the innovators and early adopters tend to be comparatively wealthy, well educated, cosmopolitan, and adventurous while in conversion to Islam the analogous categories of converts tended to have less status and wealth and to be outside of the traditional elite represented by the rural landowners. Furthermore, the advantages of a new technique are, by and large, distributed equally over the entire population of adopters once the temporary advantages of early adoption become diluted; but in the 57
Conversion to Islam in the Medieval Period case of Islam, it was possible to pass on the advantages of conversion to one's offspring. In other words, everyone w h o drills a deep well has a deep well, b u t everyone who converted to Islam did not have an equal share in or access to positions of religious h o n o r and authority. W h a t this means historically is that the descendants of the innovators and early adopters who became part of the early Islamic establishment in Iran tried with substantial success to pass on their positions in the establishment to their children and grandchildren. T h e y quite reasonably desired to retain the advantages they had gained through early conversion. At the same time, however, the rural landowners, the traditional elite of pre-Islamic I r a n i a n society, were steadily m a k i n g their way into the Islamic fold as part of the early and late majorities. T h e s e people, because of their traditional position at the top of the social ladder, were not prepared to accept a socially subordinate status within their new religious community, particularly when the people to w h o m they would be subordinate were the descendants of people of comparatively low social status. This, then, was the dilemma of the dihqäns who climbed aboard the Islamic bandwagon: if, according to axiom, they would not willingly convert if, by virtue of conversion, they would significantly lower their social status, and if the descendants of the innovators and the early adopters had a tight grip on the positions of social eminence that dihqäns regarded as appropriate to their own social class, then how could they convert and live satisfactory lives as Muslims? Needless to say, there was more than o n e way o u t of this dilemma. Some landowners made mutually advantageous marriage alliances with establishment families. 20 T h e landowning family gained religious prestige, which was all-important in the developing Muslim society, and the establishment family gained the landed wealth that their ancestors had never had. In other cases, politics provided a means of securing the desired position of social eminence. I m p o r t a n t d i h q ä n families were f o u n d among the backers of the i n d e p e n d e n t Iranian dynasties, although they had to be careful in backing political adventurers not to lapse into heresy and risk losing everything. 2 1 A third 58
The Development of Islamic Society in Iran way out of the dilemma was more important than the first two, however, from the institutional point of view. T h i s was the route of patronage of what may be termed an alternative orthodoxy. Orthodoxy is a word that must be used gingerly in discussions of the Islamic religion because within Sunni Islam no person or discrete group has ever succeeded in gaining general acceptance as the authoritative arbiter of what is orthodox and what heterodox. On several occasions attempts have been made to assert such authority, but each time the attempt has proved unsuccessful within a relatively short period of time. Yet there remains a clear overtone of what may cautiously be termed orthodoxy in medieval Islam. T h i s took the form of setting boundaries and declaring that everything on one side of the boundary was orthodox and everything on the other side heterodox. T h e set boundaries varied from sect to sect, of course, and these variations were duly noted by heresiographers. 22 What is salient to the present discussion, however, is not what the boundaries were, but the fact that they provided for substantial leeway within the bounds of orthodoxy or, more accurately, nonheterodoxy. T h e r e could be several different orthodoxies that would be mutually acceptable insofar as they did not deem each other to be heterodox but that might otherwise be violently antagonistic. T h i s situation occurred in the fields of law, theology, and personal religious behavior. For some Iranians converting with the early and late majorities, therefore, an attractive way of avoiding the domination of the descendants of the earlier converts was to support and promote religious doctrines and forms that were within the orthodox category but not particularly in favor with the people whose domination they wished to avoid. Essentially, this route amounted to supporting religious doctrines and forms that developed relatively late, specifically, the Shafii law school, Ashari theology, and Sufism. These three became the cornerstones of the alternative orthodoxy that came to be opposed to the orthodoxy of the early adopters: Hanafi law, Mu'tazili theology, and asceticism (zuhd). 2 3 Although the specific doctrines and practices involved in these 59
Conversion to Islam in the Medieval Period opposing religious orientations did not inherently favor the political and social objectives of their backers, they had the capacity to be bent to this purpose. Hanafi law, in the hands of individuals whose vision of Islam was elitist and who were trying to protect their vested interest, developed a narrower and more exclusive definition of who was legally a Muslim than did Shafii law, whose partisans included comparatively recent converts who saw personal opportunity in the universal spread of Islam. 2 4 Similarly, Hanafi law mandated a single congregational mosque per city, while some Shafiis, no doubt recognizing that the existing, older jämi' mosques were normally dominated by the Hanafis, supported the idea of several such mosques in a city. 25 In theology a similar development occurred. Mu'tazili theology and Ashari theology both rest upon rationalistic argumentative techniques emanating from Greek philosophy. However, the former school of thought developed in such a way that a poorly educated believer had a difficult time understanding it. N o allowances were made for the tastes of the masses because the supporters of the Mu'tazili position did not consider their religion to be a religion of the masses. T h e i r view was essentially elitist, and their theology came increasingly to stress the concept of God's justice and the consequent responsibility of the individual believer for his own actions. It was not a formula .designed to comfort and sustain the common man; it was an intellectually rigorous and personally demanding doctrine tailored to and by the psychology of converts of the innovator and early adopter periods. 2 8 Ashari theology, on the other hand, made rational argumentation serve the needs of simple faith. Although it was rigorously argued, the simple believer did not need to understand the logic behind the system to appreciate the doctrine because the result of the logical argumentation was to support common sense readings of the Q u r a n and the general notion of God's close guidance of human affairs. T h u s , Ashari theology developed into a support for Islam as a religion of the masses. T h i s purpose was not inherent in the thinking of the eponymous originator of the doctrine; it was the work of later generations of disciples work-
60
The Development of Islamic Society in Iran ing primarily in Iran and stemming from families that converted with the early and late majorities. 2 7 Again, in the area of personal religious observance, there is a clear distinction between the elitism of the early-converting families and the populism of the later converts. Asceticism, the ideal of the earlier converts, was also an accepted ideal within Sufism, but Sufism combined it with several important additions. Sufism opened the way for direct contact between man and God and, equally important, contact between the ordinary believer and the saint who was himself capable of direct contact with God; it fostered the use of vernacular languages as religious vehicles and thus opened up Islam to people who were not educated in classical Arabic; and it introduced the notion that divine truth could be transmitted to the mystic adept directly without reliance upon the ulama establishment. Although different interpretations of Sufism included these ideas to varying degrees, the overall impact of Sufism was to bring Islam to the masses in a meaningful fashion. Without going into great historical detail it is difficult to describe the exact working out of the competition outlined in the preceding paragraphs. Several points must be stressed, however. 28 First, membership in a legal school ( m a d h h a b ) was normally hereditary, and it was almost unknown for someone to depart from his family's traditional orientation or to marry someone from a different legal school. T h i s situation alone would have produced a cleavage between early and late converts simply by virtue of the fact that the Shafii law school did not come into existence until after the early adopter period. Second, although the word for ascetic (zähid) and traditional ascetic practices, such as night vigils and supererogatory fasting, were at home among Shafiis, it is exceedingly rare to find a Hanafi Sufi in Iran prior to the twelfth century. Among Shafiis, Sufism paralleled and then engulfed asceticism; among Hanafis it was taboo. T h i r d , although in the realm of education there was no apparent rivalry between the descendants of early converts and
61
Conversion to Islam in the Medieval Period the descendants of late converts—this because each side accepted the other as orthodox and hence as legitimate teachers—in the political sphere the two groups were consistently and often violently opposed. From the middle of the tenth century onward rioting and internecine strife marked the relations between the political-religious factions. Fourth, the people involved in this deep rivalry did not perceive themselves as being early converts or late converts; conversion was frequently too far in the past to be of apparent relevance. What they did perceive was that they and like-minded families, from whom they selected suitable marriage partners, shared a common view of Islam, either a conservative elitist view or what can be termed a populist view; and this view was important enough to be worth fighting for. As an analogy, the Yankees and the Irish in Boston, to take a simplistic example, have very different views of society and politics, and it is commonplace to relate these views to ethnic and religious differences. Yet of even greater importance is the date of immigration of the two communities to North America. Nevertheless, if a historian in the distant future knew nothing about the history of immigration, he would have a difficult time perceiving the importance of the latter factor since it is not commonly referred to in newspaper accounts of current Boston politics. This concealed polarization of the Islamic community in Iran along lines determined in large part by the timetable of conversion was immensely important for the development of Islamic society, for out of the rivalry of the two opposed views of Islam as a social religion there emerged a complex of institutions that eventually assumed preeminence in all Muslim territories. Conversion brought about the need to develop social institutions that would be peculiarly Islamic, and the conversion process dictated that these institutions would be born out of conflict instead of through a peaceful and orderly evolution. T h e conversion process also dictated that the arena for this conflict would be an urban one and occur earlier in Iran than in any other area. Other areas experienced analogous changes in the course of their conversion to Islam, however, and the final elaboration and spread of social institutions involved a far broader horizon than 62
The Development of Islamic Society in Iran Iran alone. Therefore, further discussion of social evolution will be postponed until after the conversion process in these other areas has been examined. But before this examination can begin, it will be necessary to consider once again the question of methodology.
63
6. The Curve of Muslim Names T h e analysis of the conversion process in Iran began with a sound mathematical curve derived from concrete and reasonably unbiased data and interpreted according to the techniques used in numerous arguably analogous situations. T h e study of conversion in other areas depends upon the soundness of that analysis. Because of limitations in the nature of the available quantifiable data, however, the study of conversion in these other areas will be substantially more tentative and speculative. It is important to point this out because the abundance of graphs contained in this chapter might otherwise give the impression that one is now moving into a field of more, rather than less, rigorous quantitative analysis. T h i s chapter is essentially a step-by-step argument designed to show that techniques used to analyze conversion in Iran may be legitimately transferred to other areas despite the fact that one cannot demonstrate their soundness from equivalent data in each of those other areas. Clearly, it is important to examine the logic and consistency of this argument, but ultimately the value of the approach taken here lies in its utility in forwarding our understanding of the history of the medieval Islamic world. Although it is important not to allow methodology to substitute for substance, if one is in touch with the reality of the historical past, ideally the two should support each other. Therefore, the plausibility of the methodological argument presented in this chapter must be determined in part by the utility of its application to regional conversion history in succeeding chapters. T h e 5-shaped conversion curve for Iran was produced from a subgroup of 469 genealogies. One would like to be able to tie this result to the other 5465 biographies contained in the 64
The Curve of Muslim Names sources from Nishapur and Isfahan. This can be achieved by observing the names that the individuals identified as converts to Islam in the 469 cases already studied gave to their sons and comparing the results with the patterns of naming appearing in the entire group of biographies. If one looks generally at the names given by Muslim converts to their sons, names that are all Arabic because of the way in which the subgroup of genealogies was selected, it is immediately evident that only a small portion of the vast Arabic onomasticon was utilized. Over two-thirds of the sons of converts were given names that fall into two categories: obviously Muslim names with strong religious significance, to wit, Muhamhad, Ahmad, 'Ali, al-Hasan, and al-Husain, or names that occur in biblical as well as Quranic tradition, such as Ibrahim (Abraham), Ismä'il (Ishmael), Yüsuf (Joseph), and so forth. When the frequencies of these two categories of names are tabulated for the entire group of biographies, it becomes clear that this concentration on certain names is neither fortuitous nor unimportant. Curve A on graph 7 shows the frequency of occurrence of the names Muhammad, Ahmad, 'Ali, al-Hasan, and al-Husain in a body of 6726 biographies pertaining to Iran— the total group of biographies from Nishapur and Isfahan combined with those contained in adh-Dhahabi's Kitäb al-'Ibar fi al-khabar man ghabar; curve Β shows the frequency of biblical/ Quranic names among the 5934 biographies from Isfahan and Nishapur. What is most striking, of course, is the dramatic shape of curve A, a shape that resembles the bell-shaped curve used earlier to describe the process of conversion. Is it possible that there is some kind of relation, direct or indirect, between the curve of conversion already discussed and this curve of the frequency of five distinctively Muslim names? It is not easy to investigate this possibility. For one thing, the subjects of biographies whose names are represented on graph 7 were patricians and not normally the sons of new converts. Therefore, despite the shape of the curve of popularity of the five names, it can by no means be assumed that it is the direct result of the movement of converts immediately into the patrician class. Indeed, specific biographical data clearly indi65
Conversion to Islam in the Medieval Period
Graph 7. Frequency of name types in Iranian biographies. A, the names Muhammad, Ahmad, 'Ali, al-Hasan, al-Husain; B, biblical/Quranic names.
cate that in most cases there was a lag of several generations before a convert family became accepted into this class. Furthermore, the fact that the biographies are of city dwellers raises other questions of representativeness. If the S-shaped curve of conversion were to be plotted beside the curve of popularity of the five Muslim names, however, it would immediately be seen that the peak of the popularity curve coincides with the point of transition from the category of the late majority to that of the laggards. After this point, the popularity of the five names plummets sharply. T h i s relation does not appear to be mere coincidence. T h e grounds for believing that this popularity peak is meaningfully related to the curve of conversion are essentially psychological. Looking more closely at the names given to the sons of converts within the 469 genealogy subgroup defined earlier, one can identify three specific groups. T h e relative proportion 66
The Curve of Muslim Names
Graph 8. Frequency of name types among sons of converts. A, the names Muhammad, Ahmad, 'Ali, al-Hasan, al-Husain; B, biblical/Quranic names; C, names from pre-Islamic Arabic onomasticon.
of names in each category is depicted in graph 8. Curve A shows the popularity of the five distinctively Muslim names: Muhammad, Ahmad, Ά ΐ ϊ , al-Hasan, and al-Husain. Curve Β indicates the popularity of biblical/Quranic names. Curve C represents all other names with the exception of those construed with 'Abd and a few others, such as al-Mahdï, that are obviously Muslim but not as easily classified as the five distinctive Muslim names represented by curve A. 1 T h e graph also shows the chronological categories of conversion established earlier from the S-shaped conversion curve. These categories show a division of naming customs into three distinct stages. In stage one, which covers the period of the innovators, the first 2.5 percent on the conversion curve, virtually all sons of converts were given names from the pre-Islamic Arab onomasticon. Instead of choosing distinctively Muslim names, these 67
Conversion to Islam in the Medieval Period converts selected names with no marked religious connotation whatsoever (such as Habib, Sahl, al-Fadl) beyond the clear understanding that any Iranian bearing an Arabic name was almost certainly a Muslim. T h e reason for this pattern is presumably the obligation put upon converts to become mawäll, that is, clients or fictive members of Arab tribes. Since it is certain that many mawäll families adopted the tribal nisba as a last name and assimilated as much as possible to their adoptive tribe, it is scarcely surprising that the names they chose to give their sons were typical of the Arabs in the tribe and hence drawn from the general Arabic onomasticon inherited from pre-Islamic times. Stage two covers the period of the early adopters. It is marked by a steadily decreasing popularity of nonreligious Arabic names and a steadily increasing popularity of distinctively Muslim names. It is also marked by a sharp rise and then leveling off in the popularity of biblical / Quranic names. Particularly noteworthy is the fact that this last group increases in popularity more rapidly than the distinctively Muslim names only to be overtaken by them at the end of the period. T h i s rather shortlived fad of using biblical/Quranic names appears to represent a desire on the part of some converts of the early adopter period to cloak to some degree their children's religious affiliation. It should be noted that at the end of the first Islamic century, when the curve of biblical/Quranic names peaks, the process of Iranian conversion was only about 5 percent completed. Since it is otherwise known that Arab tribal identification in Iran was beginning to lose its significance by that time, it is quite likely that some Iranian converts were living or working without meaningful Arab tribal protection in overwhelmingly non-Muslim Iranian environments. 2 T h i s fact, it has been argued, prompted many converts to migrate to Arab urban administrative centers. T h a t it should also have prompted some of these individuals to choose for their sons names that might be understood to be Christian or Jewish, as well as Muslim, as a form of social protective coloration is hardly surprising. Stage three emerges gradually from stage two. It is marked by the fantastic rise in poplarity of the five distinctively Muslim 68
The Curve of Muslim Names names. T h e explanation for this remarkable popularity can once again be read from the conversion curve. Whereas in the earliest stages of conversion identification as a friend of the Arabs was desirable, and somewhat later religiously ambiguous (though not threateningly Zoroastrian) names came briefly into vogue for protective coloration, once Islam was unquestionably there to stay and rapidly growing stronger with a bandwagon effect, as indicated by the steep ascent of the middle of the conversion curve, it became increasingly desirable to display publicly one's conversion and give ostentatiously Muslim names to one's sons. Names that might be mistaken for Christian or Jewish names steadily lost popularity, and nonreligious Arabic names were even less desirable in what was increasingly an Iranian Muslim society. T o relate the naming practices of new converts to Islam to the naming practices of the Iranian patriciate, one simply compares the popularity curves shown on graph 7 with those on graph 8. T h i s comparison is shown in graph 9. T h e solid lines represent the popularity of the five distinctive Muslim names (A) and of biblical/Quranic names (B) within the total body of biographies. T h e broken lines A' and B ' represent the popularity of the same groups of names among sons of converts. It should be recalled in studying graph 9 that the group yielding lines A' and B ' does not at all overlap the membership of the larger group. A' and B ' are popularity curves for ancestors of a subgroup of patricians extracted from the larger body and not for that subgroup of patricians themselves. Aside from the greater regularity of curves A and B, which is presumably nothing more than a function of having a much larger sample, there are three particularly noteworthy comparisons to be observed. First, the general trend of the two pairs of curves is roughly comparable, with certain important exceptions. What this indicates is that the naming practices of new converts were not entirely dissimilar from the naming practices of the social and religious elite. T h i s is an important observation since it suggests that the naming practices of the elite may be taken, with due caution, as reflections of the general society of converts even though the elite constitute in many other ways 69
Conversion to Islam in the Medieval Period
laggards
Graph 9. Comparison of graphs 7 and 8. A, Muslim names from graph 7; A', Muslim names from graph 8; B, b i b l i c a l / Q u r a n i c names from graph 7; B', b i b l i c a l / Q u r a n i c names from graph 8. Vertical line indicates division between late majority and laggards.
a distinctly biased sample from which to make such generalizations. Second, the popularity of b i b l i c a l / Q u r a n i c names is regularly much higher among convert sons than among the patriciate at large even though the general shape of the popularity curves is roughly similar. T h a t this should be the case is quite understandable on the psychological grounds mentioned previously. Ostracism by non-Muslims, and hence the desire to give one's sons protective coloration through religiously ambiguous naming, would obviously be much more strongly felt by new converts than by well-established Muslim families. T h i r d , the popularity of the five distinctively Muslim names is strikingly similar among the two groups up to the peak in
70
The Curve of Muslim Names curve A, and then the two curves diverge entirely. Again, this peculiar pattern can be reasonably explained on psychological grounds. As the bandwagon effect indicated by the sharp rise in the middle of the S-shaped conversion curve set in, it became quite generally perceived that being demonstrably a Muslim was a desirable objective. Whether new convert or established Muslim, one did not want one's children to be mistaken for non-Muslims or to have one's own piety challenged. T h e period under discussion, it may be recalled, was one of incipient factional conflict associated with conversion and of growing social identity and political power among the patrician class. T h e appearance of piety among the leaders of the nascent Muslim society was as important as it was among new converts. What is most significant, however, is the sharp divergence of the two curves that begins, as already pointed out, at the start of the laggard period on the conversion curve. As new conversions became rare because of the diminished pool of nonMuslim potential converts, the desire of new converts to display their conversion by giving their sons distinctive Muslim names only intensified. Non-Muslims were few in number, mere remnants of once dominant communities. If one wished to break with such a minority community, one wished more than ever to make the break total. Among the patrician establishment, on the other hand, naming patterns loosened up considerably. Since everyone of importance had converted by then, there was no one to compete with in piety of naming. Competition between old and new convert families remained and took on new and more important chracteristics leading to open political and social conflict, but in the field of name giving, there was simply no one left to impress. By the year 1000 it even became acceptable once again for a patrician to give his son a Persian instead of an Arabic name. 3 What is implied by this line of reasoning, of course, is that it is not at all fortuitous that the popularity curve of the five Muslim names should peak at the beginning of the laggard period on the conversion curve. T h e psychological impact on naming of the conversion process seems to require such a development. Indeed, the divergence of curves A and A' would 71
Conversion to Islam in the Medieval Period be difficult to account for in any other way. Nevertheless, a margin for misestimation in the various assumptions and guesses used to relate the peak in the Muslim name curve to a specific point on the conversion curve obviously exists. Yet despite this possibility, it is of utmost importance to make an attempt at this sort of correlation if the conversion curves of other lands are to be estimated. For it is only in a territory such as Iran where there is a language change in individuals' genealogies that a conversion curve can be drawn directly from primary data. T h e same procedure will be attempted for Spain in chapter 10, but for other regions historical circumstances prevent its application. In Semitic-language areas that eventually became Arabic speaking, such as Iraq and Syria, the distinction between Arabic and non-Arabic names in genealogies is indistinct, and the interpénétration of Arab and non-Arab naming customs may well have antedated Islam. T h e progressive adoption of the Arabic language in Egypt and Tunisia presents a similar problem. In all these areas, however, a sufficient volume of biographical material is available to calculate the curves of popularity of the five distinctively Muslim names used for Iran. Graphs 10-13 show the results of these calculations. Unlike Iran, for which virtually identical curves representing large samples taken from three separate sources permitted the drawing of a single, smooth average curve, these countries' graphs show lines that are uneven and sometimes internally conflicting. T h i s is largely due to the samples' being a few hundreds in size instead of a few thousands. Nevertheless, there is sufficient regularity to permit certain pertinent generalizations. T o anticipate the most crucial of these generalizations, the regional graphs all support the notion of there being two waves of popularity in the use of the five names. T h e first wave is approximately synchronous in all areas and can be attributed to the increase in popularity of the names among ethnic Arab populations that for the most part entered the land d u r i n g the conquest period. T h e second wave of popularity is that proper to the convert population and accordingly sets in much later. T h e peak of this second wave, following the preceding line of argument, should coincide approximately with the onset of the 72
The Curve of Muslim Names laggard period on the hypothetical conversion curve of each region. Looking now at each graph separately, we see that Iraq (graph 10) presents the most reliable picture. T w o curves are shown that virtually coincide. Curve A is derived f r o m the Shadharât adhdhahab of Ibn al-'Imäd and curve Β f r o m the Kitäb al-'Ibar of adh-Dhahabï. In many instances the f o r m e r a u t h o r simply repeats the work of the latter, b u t in the case of Iraq before the Muslim year 225, the fact that adh-Dhahabï consistently includes 50 percent more Iraqi biographies than I b n al-'Imäd warrants his inclusion as a separate source. T h e average n u m b e r of biographies per twenty-five years in I b n al-'Imäd is fifty-two and in adh-Dhahabï seventy. W h a t is most striking about these two curves is that although they show a slight peak or bulge a r o u n d the Muslim year 260, which is when the Muslim n a m e curve of Iran peaks, after that year the Iraqi curve rises to a
Graph 10. Popularity of Muslim names in Iraq. A, names from Ibn al-'Imäd; Β, names from adh-Dhahabï.
73
Conversion to Islam in the Medieval Period
Graph 11. Popularity of Muslim names in Egypt.
much higher peak in 365. At the same time, the Iranian curve in graph 2 is steadily descending. For Egypt (graph 11) there is a problem of scanty sources. The curve shown is taken from Ibn al-'Imäd who essentially repeats the information in adh-Dhahabi. T h e average number of biographies per twenty-five years here is only 7.5. T h e curve produced is marked by two dramatic peaks, the first around 240 and the second in approximately 365. Although there is no easy way to check the representative character of this curve short of finding a large independent body of biographical data, a good general impression of its reliability can be gained by observing the popularity of Muslim names among the fathers and grandfathers of the subjects of Ibn al-'Imäd's biographies. T h e samples for this comparison are obviously even smaller than those for the curve shown, but the figures are indicative nonetheless. The earlier peak on the curve shows up in a similar peak among fathers' names in the following twenty-five-year period and in 74
The Curve of Muslim Names grandfathers' names some fifty years after that. This interpretation is obviously somewhat imprecise since twenty-five years is too short for a complete generation, but the data unquestionably tend to corroborate what is shown in the graph. As for the later peak, there is a fifty-year time lag before it appears among fathers' names, and twenty-five years after that it appears among grandfathers' names. In other words, three comparatively small name samples taken from the same source tend toward the same conclusion that there was an earlier and a later wave of popularity. For Syria (graph 12) there is sufficient divergence in the coverage given by Ibn al-'Imäd and adh-Dhahabl to justify showing both curves on the graph. T h e average number of names per twenty-five-year period is fifteen for the former compiler and seventeen for the latter, so one is again dealing with small samples, although double the size of those for Egypt. As
Graph
12. Popularity of Muslim names in Syria. A, names from Ibn al'Imäd; Β, names from adh-Dhahabï.
75
Conversion to I s l a m in the Medieval P e r i o d
in the case of Egypt, however, the tabulation of fathers' names in Ibn al-'Imäd clearly reflects the curves shown on the graph. T h e first peak around 265 is reflected in an upsurge in the number of fathers with distinctive Muslim names fifty years later. T h e second peak around the year 380 is likewise reflected in the the fathers' names of the next twenty-five-year period. T h e same tendencies can be seen in grandfathers' names as well. It seems safe to conclude, therefore, that two waves of popularity of Muslim naming hit Syria just as they did Egypt; and this conclusion reinforces the similar indication of the Egyptian name curve, making up somewhat for the very small data base of the latter. Finally, graph 13 shows the name curve for Tunisia. Here the data were less plentiful still than they were for Egypt and Syria, and as a consequence it was necessary to aggregate the names in fifty- instead of twenty-five-year groups. Furthermore, names from two entirely separate sources were combined to produce
Graph 13. Popularity of Muslim names in Tunisia.
76
The Curve of Muslim Names the curve. O n e source was I b n al-'Imäd. T h e other was a compilation of biographies of noted members of the Maliki law school made by I b n Farhün. 4 Virtually all the biographies in I b n Farhün's collection pertain to N o r t h Africa or Spain, b u t only those specifically relating to the province of Ifrïqiya (modern Tunisia) have been used. It is possible to combine these two sources without concern for distortion arising f r o m duplicate entries in both books because I b n al-'Imäd preserves very few biographies of people b o r n before A.H. 300 (A.D. 913) a n d I b n F a r h ü n very few of people born after that date. T h e resultant curve, partly because of the fifty-year time period used, shows very clearly the two popularity peaks already remarked in Egypt and Syria. T h e main difference between these latter two graphs and the one for T u n i s i a is the earlier appearance of the trough between the two peaks in the T u n i s i a n case. T h e popularity curve for Muslim names in Spain is more complex than the ones just presented and will be discussed separately in chapter 10. If we look simply at the f o u r graphs presented here, however, a clear pattern stands out despite their various inconsistencies a n d weaknesses. All b u t the T u n i s i a n graph support the idea that in the period A.D. 864-888 the five Muslim names became simultaneously popular t h r o u g h o u t Islamic territories. In T u n i s i a the peak is somewhat earlier, b u t the percentage of Muslim names is still almost at peak level in 864. It is at roughly this same time, as has already been pointed out, that the Muslim name curve in Iran reaches its significant peak. Can it be concluded f r o m this correspondence that to the extent that the popularity of these names indirectly reflects conversion, all territories converted at the same time? Obviously not, for then there would be n o accounting for the subsequent peaks in territories other than Iran. Instead, it appears likely that the individuals c o n t r i b u t i n g to this first wave of popularity were the descendants of the Arabs who settled in the conquered lands and were feeling the effects of gradual detribalization, a p h e n o m e n o n that may have been felt slightly earlier in isolated T u n i s i a . Chronologically extensive bodies of purely Arab biographies are hard to find, b u t short-term samples f r o m I b n al' I m ä d a n d an early biographical dictionary of Kufa and Basra 77
Conversion to Islam in the Medieval Period by Khalifa b. al-Khayyät clearly show that distinctively Muslim names did grow greatly in popularity among Arab tribesmen in urban centers. 5 In Iran this tendency coincided with the conversion of the Iranian population; in other areas it resulted in a separate peak of naming popularity, or, in the case of Iraq, a separate bulge. If the first, chonologically coincident surge of popularity indicated by the four graphs is to be explained by reference to the Arab immigrant population then it is to the second peak that one must look for the wave of popularity indicating the conversion of the local populations to Islam. T h e detribalization phenomenon that presumably stands behind the popularity of Muslim names among Arabs is quite separable from the psychological factors influencing naming among converts; thus, there is nothing particularly surprising in the two elements in the Muslim population's displaying different naming patterns. In Iraq the peak representing the popularity of Muslim names among converts comes in the year 976, in Egypt in the same year, in Syria perhaps a few years later, and in Tunisia around the year 1010. It should be noted, however, that since the conquest of Tunisia did not become firm until some thirty-five years after similar consolidation was achieved further east, the Tunisian naming peak occurs at relatively the same point in that land's Muslim history as do the peaks in Egypt, Syria, and Iraq. If the various surmises and approximations contained in the preceding analysis are sound to any appreciable degree, it should now be possible to pinpoint at the peak of the second wave of popularity the transition from late majority to laggards on the S-shaped conversion curve. With this point marked and the date of conquest known, it becomes possible to construct with little additional information a hypothetical conversion curve for each of the four territories modeled after the Iranian curve. One can then test the plausibility and utility of each curve by comparing it with the political and religious history of each region. This will be the program of the next several chapters. But already three interesting tentative conclusions may be drawn. First, it appears that the heavily Christian areas of Iraq, 78
The Curve of Muslim Names Egypt, Syria, and T u n i s i a converted at approximately the same rate, taking into account the later date of the conquest of T u nisia. Second, the conversion of Iran, a predominantly Zoroastrian land, significantly preceded the conversion of Iraq, Syria, and Egypt. And third, the hypothesis suggested at the beginning of chapter 3 that Iran looms extraordinarily large in the intellectual history of the third, fourth, and fifth Islamic centuries partly because of relatively early conversion seems to be confirmed.
79
7. Iraq
T h e fruits of the analytical procedure described in the last chapter must now be tasted to see if they are palatable. However, this chapter and the three following it should not be taken for comprehensive histories of the conversion process in the areas covered. Only one approach to the history of conversion is being tested here, and the data adduced in support of this approach are intended to be more suggestive then exhaustive. If the heuristic value of the approach can be demonstrated, it may be hoped that future scholarship will produce a more detailed history of conversion based upon the curves and interpretations suggested here. Iraq affords both a good and a bad starting point. It is a good starting point because of the comparatively large samples of names available for analysis and because its history is quite well known. It is a bad starting point because it was the capital province of the Abbasid caliphate and before that the seat of the two largest and most active Arab urban centers in the conquered lands, Basra and Kufa. T h e history of Basra, Kufa, and Baghdad, the Abbasid capital, has often been taken as representative of the history of Iraq as a whole, so that the history of the Arabs in Iraq tends to overshadow the history of the much more numerous indigenous population. T h e question of the size of the population living in the Tigris-Euphrates valley is a complex one, of course, as medieval population estimates invariably are. In all likelihood, all that will ever be possible to achieve is an estimate of order of magnitude, and for this the work of J . C. Russell provides a usable figure. He estimates, primarily by assuming a similarity in population density between Syria and Iraq, a population of 9.1 mil80
Iraq
lion for the Tigris-Euphrates valley in the early Islamic period. 1 Although great reliance cannot be put upon the precision of this figure, it at least gives a plausible number against which to compare the estimate of the Umayyad governor 'Ubaid Allah b. Ziyäd that in the year 64 of the hijra there were one hundred and fifty thousand mawâlï or Muslim converts in Basra.2 I argued earlier that in the early stages of Islamic conversion converts usually migrated to Arab-dominated urban centers. This pattern seems to be as true of Iraq as of Iran. T h e great preponderance of the convert population at this early date was situated either in Basra or Kufa. Therefore, allowing a comparable number of mawâlï for Kufa and Basra, it can be hazarded that the convert population of Iraq in A.H. 64 was roughly three hundred thousand. This figure is no more precise than that for the total population of Iraq, but it does indicate a plausible order of magnitude, and it further suggests that by that year some 3 percent of the Iraqi population had converted to the new religion. Given a figure of 3 percent for the amount of the population converted by A.H. 64 and 84 percent for those converted by A.H. 365 when the peak in the Muslim name curve indicates the transition from late majority to laggards, it becomes possible to derive the parameters for a logistic or S-shaped conversion curve. T h e resultant curve is shown plotted as a solid line on graph 14. For comparison, the S-shaped curve of Iranian conversion is shown by a broken line. Graph 15 shows the proposed S-shaped curve of Iraqi conversion with a number of political events and institutional and religious developments marked on it. Also marked are the categories of converts that were established earlier in the discussion of conversion in Iran. T h e dividing lines between these categories are all later than those proper to Iranian conversion, but they are later by different margins. The innovator periods for the two regions end within twenty years of each other. Using Christian dates, the innovator period in Iraq ends in 675 as opposed to 695 in Iran. This span is less than that of a generation, and given the looseness of the estimate of 3 percent for the population Converted by the year 64 of the hijra, it is not un81
Conversion to Islam in the Medieval Period
Graph 14. Cumulative S-curve of Iraqi conversion compared with Iranian conversion curve.
reasonable to suppose that the innovator periods actually faded out at pretty much the same time. For the early adopters, however, the difference is more striking. In A.D. 762 the early adopter period in Iran terminates, while in Iraq it goes on until 791, or roughly the beginning of the caliphate of H ä r ü n ar-Rashïd, whose reign may have been perceived as a period of classic grandeur in part because it marked the beginning of the great surge of Islamicization. At that point the bandwagon effect sets in, but it is less explosive than in Iran. Between 791 and 975, 68 percent of the conversion process is accomplished, b u t this is a space of nearly two centuries; in Iran the same amount of conversion took place in just a little over one century between 762 and 875. Finally, as mentioned earlier, the laggard period commences in 975 in Iraq as compared with 875 in Iran, a difference of a full century. Chronology aside, there are other differences between the 82
Iraq
Graph 15. Correlation of historical developments with Iraqi conversion curve.
conversion process in the Tigris-Euphrates valley and in Iran. Although there is no way of quantifying the magnitude of these differences, three seem particularly noteworthy. First, there is a larger prisoner-of-war element in Iraq in the period of innovators because of the concentration of the Sassanid army there at the time of the conquest and the sending back to Basra and Kufa of captives taken during campaigns further east.3 Second, government administrative personnel, most of them converting in the early adopter period, loom larger in Iraq as an important body of converts, both because of the heavy concentration of administrators in the capital province and because of the bias of the historical literature toward recording the personalities and events connected with the operation of the caliphal establishment. Third, Iranians, being an upper class minority of foreign origin before the Arab conquest, had less resistance to conversion than they had in Iran proper, presumably because join83
Conversion to Islam in the Medieval Period ing the Arabs was a way of continuing a degree of social superiority over the Aramaic-speaking majority of the population. T h e non-Muslim Iranian presence in Iraq diminishes greatly in the early adopter period indicating extensive early conversion of this group. 4 When one turns to the arena of political history, these various chronological and social differences between conversion in Iraq and Iran reveal themselves in different ways. In Iran, nonMuslim religious revolts flourish prior to the attainment of approximately 50 percent on the conversion scale, although the last revolts were in remote areas, which suggests that such resistance movements actually tended to fade away in more central areas well before the 50 percent point was reached. After that point, Iran became dominated by independent Muslim Iranian dynasties that progressively adopted the symbols and regalia of the pre-Islamic Iranian empires. In Iraq the proximity of large Arab armies and the concentration of the population in an easily traversed river valley of relatively limited size made non-Muslim religious revolts virtually impossible from the moment of conquest on. Instead, Iraq was visited by a steady succession of Khariji revolts that almost exactly coincides with the period of the early adopters. T h e Kharijis came into existence as a political movement in the 670s, and uprisings were still reported as late as 796 in more remote northern Iraq. Kharijism, of course, is a variety of Islam, and Arabs unquestionably predominated in these revolts. Nevertheless, mawâlî were also drawn to Kharijism either because of the egalitarian character of Khariji ideology or, perhaps more important, because for the humbler stratum of converts, which formed one of the mainstreams of conversion in the innovator and early adopter periods, violent rebellion offered hope of the material betterment that had not been achieved by simple conversion. 5 Converts of prisoner-of-war or administrator origin do not appear to have been involved with the Kharijis. In short, leaving the Arabs aside and taking the point of view of the mawâlï converts, one can see the Khariji revolts of Iraq as the equivalents of the non-Muslim religious revolts in Iran.® T h a t they did not continue as long is a function of Iraq's easy and 84
Iraq uniform geography, which did not provide mountainous refuges and isolated pockets of population. As for Iraq's political history following the point of 50 percent conversion in 882, there are again parallels with Iran. Instead of independent local dynasties, one encounters a progressive deterioration in the central government's ability to control the countryside. Earlier, during the Samarra period (836-893), there had been a similar breakdown caused by the infighting among different components of the army, but the vigorous regency of al-Muwaffaq and the reigns of his son and grandson had demonstrated unexpected resource and governing ability in the Abbasid family. Still, the recovery during those reigns could not prevent the decline of state power that set in as a result, in part, of the conversion process. On the one hand, rebels became able to cut off and hold, sometimes for appreciable periods, large chunks of territory. T h e Zanj rebellion in southern Iraq and the virtually independent tribal principality of the Hamdanids in the north were harbingers of future political dissolution. Finally, in 945, the advent of Buyid power in Iraq brought what was essentially a permanent end to the secular power of the Abbasid caliphate. After that the ability of small dynasts to set up independent principalities in Iraq was limited only by the power of the caliph's foreign captors. T h e theory of unified central government had lost its hold on the population. T h e way in which conversion appears to have played a causative role in this process of political decay can be seen, again, from what happened at a parallel point in Iran. T h e rise of the independent Muslim Iranian dynasties was made possible by the realization that Islam in Iran was a permanent fact and could not be harmed by a fractioning of political power. T h i s realization could only be produced by conversion, for that alone could give the individual Muslim the assurance that he and his fellow Muslims were dominant. T h a t prime desideratum satisfied, would-be dynasts were free to recruit supporters for a variety of reasons. In Iraq the caliphate was a much more central issue. T h e caliphate's existence was justified, in the first instance, by the need to hold the Muslim community together in the face of ad85
Conversion to Islam in the Medieval Period versity, the need to protect it and guarantee its survival. In the long run, this was the only permanently valid justification for the existence of the caliphate. Consequently, whenever Islam's survival came to be taken for granted, the caliph's underlying source of authority evaporated leaving only a brittle superstructure. More than any other single element, the accomplishment of majority conversion" among an area's population brought about the feeling of Islam's permanence and of the caliph's superfluousness. If the rebels that surfaced on every hand toward the end of the ninth century had seriously threatened Islam in Iraq, they would have been resisted by military contingents from other lands hurrying to the aid of the caliph; but they did not pose such a threat, and thus they were met only by local political and military forces. 7 Similarly, if the Buyid occupation of Baghdad and seizure of the caliph in 945 had had any significant effect upon the local populace, Iraq would have been ungovernable; but the caliphate was already an outworn institution in its capital province, and the Buyids did not even bother to replace the Sunni incumbent with a puppet who shared their Shii views. Yet half a century later the Abbasid caliph was still an important symbol of Islamic unity and permanence on the eastern marches where Mahmud of Ghazna was making conquests in India. Although it is possible to see the increasing seclusion of the caliph and ritualization of his functions during the two centuries of the early and late majorities as responses to the expectations that the non-Arab convert population had of their ruler, it is only conjecture since the tastes of the court were substantially isolated from those of the population at large and since many court functionaries came to Baghdad from other parts of the caliphate. A different phenomenon of the same period seems to correlate well with the notion of bandwagon conversion, however: the phenomenon of urbanization. T h e large cities of Kufa, Basra, and Wasit which dominated Iraqi politics throughout the first century after the conquest were, in origin, Arab military encampments, and they retained their predominantly Arab character despite a growing non-Arab mawâlî element. With the foundation of Baghdad in 762 a 86
Iraq second period of urbanization began, however, a period characterized by the influx of non-Arab converts into new cities. Baghdad was the first. T h e army it was built to accommodate was largely Arab, b u t the extensive u n p l a n n e d civilian quarters that grew on the outskirts of the R o u n d city of al-Mansür were heavily populated by non-Arabs. T h e growth of Baghdad was phenomenal, b u t even more p h e n o m e n a l was the creation of the city of Samarra only a h u n d r e d miles u p the Tigris f r o m Baghdad starting in 836. Samarra, like Baghdad, began as a caliphal capital and c a n t o n m e n t for troops, b u t it also had a very substantial population of civilians who were predominantly of non-Arab origin. 8 Since there is little evidence that the entire population of Baghdad a b a n d o n e d the older capital for the newer one, it must be concluded that much of the civilian population of Samarra represented a continuation of immigration f r o m the countryside, immigration that had a reflection in the reduced density of population in the Diyala River basin northeast of Baghdad. 9 Although the lure of the grand new cities may have had somet h i n g to do with this r u r a l - u r b a n immigration, a more plausible reason for it can be f o u n d in the desire of converts to migrate to purely or heavily Muslim communities, as was the case in Iran. Indeed, the fate of the two capitals reflects the d w i n d l i n g of convert migration as the rural areas rapidly became overwhelmingly Muslim. Samarra was a b a n d o n e d as a capital in 893, the m i d p o i n t on the conversion scale, and Baghdad seems to have suffered severe population loss d u r i n g the century of Buyid mastery f r o m 945 to 1055. T o be sure, political and economic reasons can be adduced to account for this termination of what had been a period of rapid urbanization, b u t economic crisis a n d political upheaval had also occurred earlier w i t h o u t impeding the trend. It seems much more likely that urbanization was b r o u g h t to an end largely by the fact that new converts no longer felt the need to migrate since they were already living in heavily Muslim communities at the time of conversion. It is noteworthy in this connection that as Baghdad and Samarra declined, a n u m b e r of secondary Iraqi cities took on increasing political a n d cultural importance. 1 0 87
Conversion to Islam in the Medieval Period I n the area of social and religious development, Iraq presents a somewhat different picture from Iran. In Iran it was possible to construe the ubiquitous factional rivalry of the tenth and eleventh centuries as a logical outcome of the conversion process. Conversion produced a partial inversion of the social hierarchy of the pre-Islamic period, and this in turn engendered a rivalry between the families of early converts and the families of later converts for control of the Muslim establishment and direction of the future course of Muslim social development. T h e early convert families were generally linked to the Hanafi law school, Mu'tazili theology, and asceticism as a form of extreme religious observance; the late convert families were similarly linked to the Shafii law school, Ashari theology, and Sufism. As the capital province and the cockpit of early Muslim theological controversy, Iraq does not afford an easy analogy with Iran. Indeed, there appears to be one complex of conversionlinked factional rivalries associated with the city of Baghdad, and another rather different complex for the province as a whole. T h e history of factional rivalry in Baghdad is well recorded. T h e early dominance of the Mu'tazili theological doctrine, which was closely associated in Iraq with certain tendencies in Shiism, was confirmed during the period of the mihna, the government-sponsored persecution of non-Mu'tazilis that lasted from 827 until 847. T h e Mu'tazilis were resisted, within Baghdad in particular, by what eventually crystallized into the Hanbali school of law and theology. T h e r e is substantial evidence that the social support for this resistance movement came from the descendants of individuals who had converted around the middle of the early adopter period, and their eventual attainment of a dominant position in the capital confirms that their goal was the seizure of control of the Muslim establishment from the backers of the Mu'tazilis. 11 As for the Mu'tazilis themselves and their supporters, some of them were very early converts to Islam and many others were of Arab descent. Factional rivalry does not stop with the Mu'tazilis and the Hanbalis, however. In the tenth and eleventh centuries the most visible rivalry in Baghdad was between the Hanbalis and the
88
Iraq Shiis. Although some Shiis were closely associated with the early Mu'tazili movement, this later Shiism was much more popular and its appeal more emotional. T o all appearances it represents a quite late stage in· the conversion process, a more extreme form of the populism characteristic of the Shafii-Ashari-Sufi faction in Iran. Its supporters were evidently part of the late majority who converted to Islam quite late and who had no implicit claim to social leadership based on prior social eminence. In fact, the Shiis never achieved the degree of power or recognition that the Hanbalis had achieved before them. If we look at the countryside at large, the Mu'tazili-Hanbali rivalry does not seem to be particularly important. This is doubtless because most converts migrated to the large Muslim cities during the period of this rivalry leaving the countryside and the smaller communities predominantly non-Muslim in character. Factional rivalry only reaches the countryside in a visible manner right around the 50 percent point on the conversion curve. It takes the form of the Qarmati movement, a rather mysterious form of Ismaili Shiism that sprang up at roughly the same time in eastern Arabia, southern Iraq, northern Iraq, and Syria. 12 As with the earlier Khariji revolts, a problem is posed by the mixture of Arab tribesmen and non-Arab peasants in the same movement, but the latter element was particularly in evidence in southern Iraq. T h e Qarmati movement in Iraq also resembles the earlier Khariji revolts in its Utopian, revolutionary message and ultimate futility. Quite clearly, the supporters of the Qarmatis did not have a strong stake in the existing society. In comparison with Iran, then, the political cleavages engendered by conversion in Iraq are more complicated. T h e social elite of the pre-Islamic period, represented by government administrators and the Iranian segment of the population, tended to convert at an earlier date in Iraq than did the dihqäns of Iran. Although this prevented a clear-cut inversion of the preIslamic class hierarchy from taking place, it did not prevent the development of a rivalry between early and late convert families. This rivalry was played out almost exclusively in the great
89
Conversion to Islam in the Medieval Period cities, however, instead of being dispersed among a larger number of smaller cities as it was in Iran, and the players were not as evenly matched socially as they were in Iran. T h e Hanbalis eventually attained the degree of social respectability that their Iranian counterparts, the Shafiis, had from the beginning; but for a long time they represented a distinctly lower social element. Finally, the converts of the late majority did not follow the guide of their social betters and become spear carriers in the factional disputes of the major Sunni factions as they did in Iran. Instead, they supported the respectable but déclassé doctrines of moderate Shiism in the capital city and, at least to some extent, the Utopian revolutionary doctrines of Qarmati Shiism in the countryside. An important consequence of this more complicated socialreligious situation in Iraq was a lack of clear direction in institutional development. Instead of acting as the pioneers of a popularized Islam as the Shafiis did in Iran, the Hanbalis found themselves confined between the elitist Mu'tazili camp and the even more popular Shiites who cut them off from their natural constituency of late converts. Like the Shafiis, the Hanbalis supported common sense intepretations of the Koran, and some of them favored Sufism; but they did not foster fraternal forms of social organization. When the Hanbalis eventually encountered in the late eleventh century the new Shafii-Ashari form of Islam that had been developing independently in Iran, they looked upon it as an opponent to be resisted as fiercely as they resisted the Shiis of Baghdad. 13 They may have quite realistically seen it as a rival for the support of the same social groups that had previously been Hanbali backers. From the earliest conquests, Iraq had been the center of intellectual and religious development in the conquered lands. After the Abbasid revolution made it the capital province in 750, there was every reason to expect it to retain its preeminence. T h a t it did not, that it was surpassed in these avenues of development by Iran, which was conquered later and was much less heavily colonized by Arabs, is only partly the product of the more rapid rate of conversion in the latter territory. It was also due, in part, to a more complicated social profile of conversion 90
Iraq in Iraq which produced fruitless factional infighting instead of more creative, if equally vicious, bipartisan rivalry, and to the cleavage between capital and countryside which curtailed the tendency to develop religious institutions on a mass basis and made the various factions more dependent upon court favor than upon their own resources.
91
8. Egypt and Tunisia
Since graphs 10, 11, and 13 show the popularity of the group of five distinctive Muslim first names peaking in Egypt and Tunisia at about the same time it does in Iraq, the line of analysis followed in this book would seem to suggest that the preceding chapter might be simply repeated here, or else this one skipped altogether. After all, similar peaks tend to yield similar graphs that are subject to similar interpretations. T h e historical development of Egypt, however, leaving Tunisia aside for the moment, has never precisely duplicated the historical development of any other part of the Middle East, and there is good reason to question whether it really follows Iraq in this particular case. T h e shape of the popularity curve is the first apparent indication of a different line of development. According to graph 16, which compares the curves of graphs 10 and 11, the popularity in Iraq of Muslim names among the indigenous convert population and among ethnic Arabs shows not so much two clearly separate peaks as one great rise with the Arabs' earlier attainment of a maximum appearing as a shoulder rather than a completely separate prominence, while the curve for Egypt, in striking contrast, shows a deep valley between the two peaks of popularity. Unfortunately, the importance of this contrast is clouded by the fact that the very low totals of biographies per twenty-five years in Egypt (average 7.5) by themselves create great fluctuations in the shape of the curve with a change of only one or two names from one group to the next. That two peaks exist in the Egyptian case is certain, but their height and the depth of the trough between them may well be exaggerated by the smallness 92
Egypt and Tunisia
Graph 16. Comparison of graphs 10 and II. A, popularity of Muslim names in Iraq (curve from Ibn al-'Imäd only); B, popularity of Muslim names in Egypt.
of the samples. Nevertheless, the apparent contrast with Iraq is too striking to dismiss so simply, particularly since the curve for Egypt is very closely paralleled by that for Syria shown on graph 12. Assuming, then, that the contrast is significant, one possible source of the disparity is that somehow the much larger population of Arabs that entered Iraq during and after the conquests may have generated a pattern of adoption of Arabic names markedly different from that caused by a smaller number of Arabs entering Egypt. If the overall number of Arabs were a significant factor, however, then it might be expected to have had an effect upon the total conversion process as well. T h e innovation diffusion model, after all, rests ultimately upon access to information on the part of potential adopters. In a largely nonliterate society in which information was disseminated 93
Conversion to Islam in the Medieval Period primarily by word of mouth, a smaller number of Arabs could be expected to have contacted and possibly influenced a smaller number of possible converts. However, even without concrete or reliable figures regarding the number of Arabs who migrated to Egypt as compared with Iraq, this complication can be largely discounted. Quite imprecise but not unreasonable published estimates of the overall population of the two regions indicate that the exposure to Arab Muslim contact was probably roughly comparable. T h e figure of 9.1 million was used to indicate the order of magnitude of the Iraqi population at the time of the Arab conquest, and estimates of equivalent generality for Egypt at that time indicate a population on the order of 2.7 million. 1 In other words, even if the number of Arab Muslims migrating to Egypt was only a third to a quarter as great as the n u m b e r migrating to Iraq, the potential for spreading information about the new faith would have been roughly equivalent. A more promising route than demography for investigating the difference between the Egyptian and Iraqi data lies in Egyptian political history. It is striking that between the great depression in the curve around the year 950 and its highest peak around 975 one of the most important political events in Islamic history occurred: Egypt was conquered by the Fatimid dynasty and made into the capital province of an Ismaili Shii caliphate. T h e r e are two obvious ways in which this event might have affected the curve of Muslim names. T h e religious doctrines of the new rulers could have led people to believe that an obviously Muslim name would be a prudent one to bestow upon a child. Or else the new dynasty could have given a segment of the population with predominantly Muslim names an opportunity to rise to sufficient prominence to have their biographies recorded. T h e former possibility has little to recommend it. T h e early Fatimid regime was not known for its intolerance of non-Muslims, and there were several Fatimid political figures of note who had religiously neutral names, notably al-Jawhar, the conqueror of Egypt, and Buluggln b. Ziri, the Berber tribal chief who assumed control in T u n i s i a as a Fatimid governor after the shift of the capital to Egypt. Moreover, the popularity of the five
94
Egypt and Tunisia Muslim names drops sharply in the mid-eleventh century while the Fatimids were still ruling. T h e other possibility is far more attractive. Prior to the Fatimid invasion, Egypt had been a part of the Abbasid caliphate ruling from Baghdad, albeit at times a quasi-independent part. Not only were its governors usually appointed from Baghdad, but many lower level administrative personnel were as well. 2 As for the favor shown by this essentially foreign administration to the Egyptian populace, it is more likely to have fallen upon the kindred Arabs than upon the native converts to Islam. With the advent of a Fatimid regime, this tendency must certainly have been reversed. T h e Arabs who had been prominent in the military in Egypt during much of the period since the original Arab invasion were supplanted along with other troops from the east by Berber and other troops entering in the conquering army from Tunisia. These earlier military forces represent elements in the local population that lost favor under the new order. In the same way the pro-Abbasid Arab element in other branches of government service was gradually purged and supplanted. 3 One does not hear of prominent administrators continuing for long in office under the Fatimids as happened in Iran when the Seljuqs swept away the native Iranian dynasties. As for the people who supplanted these administrators, it is easier to believe that they were native Egyptians than that they were immigrants from Tunisia since Tunisia was still a Fatimid territory and could not be simply stripped of its educated population to staff the much larger Egyptian bureaucracy. In short, the sharp and sudden rise in frequency of characteristically Muslim names is quite plausibly explained by a change in government favoritism incident upon the installation of the Fatimid regime. By comparison, the convert population in Iraq would appear to have eased quite gradually into positions of prominence. Explicit confirmation of this hypothesis would require a detailed examination of the family backgrounds of large numbers of early Fatimid officials and their immediate predecessors in office. In the absence of such a study, a strong indication of the soundness of the hypothesis can be gleaned from the graph of
95
Conversion to Islam in the Medieval Period popularity of Muslim names in Tunisia (graph 13). As pointed out previously, the first popularity peak in Tunisia comes a bit earlier than the analogous peaks in other areas, but what is of particular interest in comparison with Egypt is the fact that the deep trough between the two peaks is earlier than the similar trough in Egypt and Syria. As a result, the recovery in the popularity of Muslim names in Tunisia takes place between 913 and 961 while in Egypt it takes place between 950 and 975. Since the Fatimids took power in Tunisia in 909, this finding means that in both countries there was a dramatic increase in the percentage of children given distinctive Muslim names immediately following the advent of Fatimid rule. T h e Tunisian data thus reinforce the equation suggested by the Egyptian data between Fatimid rule and the popularity of Muslim names, an equation that is best accounted for in both cases by a change in governmental orientation from the Arab segment of the population to the indigenous convert segment. Just as important as the earlier occurrence in Tunisia than in Egypt of the trough in the name curve is the fact that the Egyptian curve peaks in the period immediately after the trough, around 976-1000, while the Tunisian curve rises more gradually from its low point to crest at roughly the same time in the year 1010. What this indicates is that the Fatimids came to power in Tunisia well before that peak in the naming curve that applies specifically to converts had been reached, while in Egypt Fatimid rule barely preceded the peak. T o demonstrate this difference between the two areas more fully, a graph of a type already familiar will prove helpful. Graph 17 compares major stages in Egyptian and Tunisian political history with the putative curve of conversion common to them both, that curve being taken directly from the one calculated earlier for Iraq. T h e correlation between events in Egypt and Tunisia and events in Iran discussed earlier is quite striking. In all three cases the outbreak of non-Muslim rebellions occurs at the same point on the respective curves of conversion. In all three cases the later revolts took place in less accessible parts of the country, the last one in Egypt being in Upper Egypt and late rebellions in Tunisia centering in the 96
Egypt and Tunisia
Graph 17. Correlation of historical developments with Egyptian and Tunisian conversion curve.
mountainous Zab region. 4 T h e fact that such native revolts continued longer in Iran than in Egypt and Tunisia is probably due to the abundance of mountain refuges and remote locales in the former region, making it much less amenable to control than the Nile valley and the Tunisian coastal plain. As for the immediate causes of the revolts in these widely separated provinces of the caliphate, there is no detectable similarity. However, the underlying cause in all three cases would seem to have been the growing realization that Islamic dominion was a threat to the survival of the native non-Muslim communities. T h e next important stage in the political history of Iran and Egypt is characterized by the appearance of semiindependent dynasties and the development of significant urbanization through migration of converts. In Iran the semiindependent Tahirid dynasty, which appears at almost exactly the same point on the Iranian conversion scale as the Tulunid dynasty does on 97
Conversion to Islam in the Medieval Period the Egyptian conversion scale, was ethnically Iranian but clearly depended to an important degree upon Arab support. In Egypt the Tulunid governors were of Turkish origin and did not utilize the local populace for military purposes. This is possibly an important reason for their being unable to resist the reabsorption of Egypt into the sphere of caliphal control in 905. Nevertheless, the point of the comparison between the Tahirids and the Tulunids, who were followed after a thirty-year hiatus of direct caliphal rule by the similarly independent Ikhshidids, is that the Muslim community felt sufficiently entrenched in Egypt to be willing to cut the umbilical cord uniting them to caliphal authority in Baghdad, just as the Iranian Muslim community had at a parallel stage. This degree of communal confidence was in both areas the result of attaining the conversion of over a third of the native population. It is unreasonable to suppose that Ahmad b. T u l u n was the first governor of Egypt to dream of becoming independent and keeping the province's tax revenues for himself or that he was the first who had sufficient personal ability and will to carry out such a scheme. What is much more easily imagined is that he was the first governor to have the local political backing to put the scheme into operation. This political backing was in part the product of Muslim communal confidence based upon conversion. Similar support had become available to the Tahirids almost a half century earlier because of the earlier timetable of conversion in Iran. T h e best substantiation of this interpretation of events during the T u l u n i d period is the notable urban efflorescence that accompanied it. Just downstream from the original Arab cantonment city of Fustât a new capital was built named al-Qatâ'ï. 5 What is recorded about al-Qatâ'ï, as about the simultaneous and analogous construction of Samarra in Iraq, is how grand it was and how expensive. What is more important from the point of view of conversion, however, is how it was populated. There is no indication that Fustät was deserted when al-Qatä'I was built or that any great migration took place from outside Egypt to the new capital. As with Samarra, the new urban population must have come from within the country; and since, as has already been demonstrated, migration to Muslim governing centers was 98
Egypt and Tunisia frequently associated with conversion in its early stages, the building of al-Qatâ'ï strongly reinforces the notion that Tulunid independence was associated with substantial and rapid conversion as indicated by the conversion curve. 6 T h e case of Tunisia diverges somewhat from that of Iran and Egypt, but the difference is illuminating. Egypt and Iran accepted the pretensions of semiindependent local rulers because the Muslims there had attained a sufficient feeling of security as a religious community and resented the taxation of their wealth by and for a distant capital. In Tunisia the remoteness of the province from the Abbasid capital and the rebelliousness of the only nominally Muslim inhabitants of Algeria had led quite early to the formal investiture by the caliph of the Aghlabid family with the position of hereditary governor in return for a set amount of annual tribute. As a result, resentment over alienation of wealth was scarcely felt by the Tunisian Muslims, and what resentment there was could be controlled by enrolling the malcontents in the military forces that gradually carried out the conquest of Sicily. T h e Aghlabids are sometimes portrayed as the first semiindependent dynasty setting a pattern that was subsequently followed by the Tahirids, the Tulunids, the Hamdanids, and others. As can be seen quite clearly from the conversion curve, however, the Aghlabids are in fact unique, coming as they, do before any local Muslim demand for autonomy could have appeared. A more valid comparison is that between the Tahirids, the Tulunids, and the Fatimids in Tunisia. T h e fact that Tunisia goes directly to full independence around the midpoint on the conversion scale is necessitated not only by the nature of Fatimid ideology but by the unique sort of semiindependence previously enjoyed by the Aghlabids. It is particularly noteworthy that after their triumph in 909 the Fatimids built their new capital on the coast at al-Mahdiya, while the Aghlabids had earlier constructed for themselves an inland capital at al-'Abbäsiya only three miles from the Arab cantonment city of Qairawän. T h e move from the interior, where the Arab invaders concentrated at the time of the conquest, to the coast, which had a much larger indigenous popula99
Conversion to Islam in the Medieval Period tion, is emblematic of the shift in orientation f r o m A r a b to nonArab convert that coincided with the rise of the Fatimids. Furthermore, the city b u i l d i n g that took place u n d e r b o t h dynasties was u n d o u b t e d l y fueled by the u r b a n migration of converts to Islam. T h e final m a j o r stage of political development in Iran and Egypt is characterized by full independence and hostility toward the caliphate in Baghdad, a stage that T u n i s i a had reached substantially earlier. In Iran this stage began with the Saffarid dynasty just after 870 and was later extended by the Ziyarids and Buyids. In Egypt full independence did not occur until the successful invasion of the Fatimids in 969, a full century later. In terms of the different conversion scales, however, the two stages are reached at almost exactly the same time. W i t h the culmination of the period of the late majority comes the drive for total independence. W h a t had been a tentative willingness to cut ties with the central authority when a third of the population had converted becomes a positive desire when conversion has affected over two-thirds of the population. Subordination to foreign authority became intolerable to a thriving and self-confident local Muslim community. Some sort of ideological justification for total independence was a normal concomitant of this development, b u t it made little difference whether that justification was Khariji (the Saffarids), I m a m i Shii (the Buyids and Ziyarids), or Ismaili (the Fatimids). N o n e of these sectarian theologies sank deep roots into the population of the lands their proponents led into independence of the caliphate; they were merely convenient justifications for actions that were deeply desired by the local populations whose acquiescence was thereby easily obtained. T h i s point brings us back to the sudden peak in the curve of Muslim names following the establishment of the Fatimid regime in Egypt compared with the more gradual rise to a simultaneous peak in T u n i s i a . It has previously been argued that the peak in Muslim n a m i n g occurs at the beginning of the laggard period in conversion for quite understandable reasons. Now it has f u r t h e r been argued that the Fatimid dynasty conquered Egypt, with surprising ease after sixty years of futile attempts, 100
Egypt and Tunisia at just about the same point on the conversion scale because that was the precise time when the Muslim Egyptian population became interested in total independence from the Baghdad caliphate. Therefore, the personnel change from predominantly Arab to predominantly non-Arab that marked the early stage of Fatamid rule in Egypt is quite reasonably reflected in a sudden peak in the curve of Muslim names followed by a rapid decline. T h e synchronization of Fatimid rule with Muslim name popularity is not mere coincidence, then, but the mark of a significant stage of attainment on the scale of conversion, a stage that is quite independently confirmed by analogous events in Iran. In T u n i s i a , with its unusual political history, the Fatimids took power much earlier, in terms of the conversion scale, than they did in Egypt. T h e reasons for Tunisia's greater vulnerability might be sought in its isolation at the western tip of the caliphate or in the anomalous form of government represented by the Aghlabid dynasty; but whatever the reason, the fact that the Fatimids took over at such a comparatively early point and redirected the attentions of the government from the Arab Muslims to the non-Arab converts explains the more gradual rise in popularity of Muslim names to a peak that generally coincides with that reached in Egypt. T h e peak in both cases represents the beginning of the laggard period of the conversion process, but the political events that correlate with the stages of conversion are somewhat peculiar to T u n i s i a and do not exactly coincide with what happened at parallel stages in other regions. In response to this lengthy and extremely speculative set of arguments relating Egyptian, T u n i s i a n , and Iranian history, the question might well be raised as to whether the basic argument is not actually a circular one. Since the peak in Muslim naming was hypothesized at the end of chapter 6 to represent a certain stage in the achievement of conversion, is it not unsound to seek to support this hypothesized relationship with an argument synchronizing political events and stages of conversion that itself depends upon the correctness of the hypothesis? T h e answer to this, of course, must be yes. On a purely logical basis the foregoing argument is difficult to defend. Its defense must be found in the comparison with the course of events in other 101
Conversion to Islam in the Medieval Period provincial areas, such as Syria and Spain, and in its reasonableness in view of the scattered known details pertaining to Egyptian and T u n i s i a n conversion to Islam. Fortunately, two recent investigations have been m a d e into the chronology of Egyptian conversion using types of data entirely different f r o m those used in this study. In one, Ira Lapidus has concluded the following: " T h e n i n t h century seems to be the crucial epoch in the conversion of Christians to Islam . . . Not the A r a b conquest itself, b u t the n i n t h century was the great watershed. By no means, however, was the conversion of the Copts complete. Although the majority of the populace became Muslim, we cannot say that the minority of Christians was anything less than very substantial in the n i n t h and tenth centuries. T h e Coptic u r b a n populations seem to have remained in vigor in the eighth, n i n t h , and tenth centuries." 7 A r a t h e r similar conclusion was reached by Michael Brett f r o m a review of materials similar to Lapidus's: "Christians remained a majority at least until the tenth century." 8 Brett f u r t h e r remarks that " f r o m the n i n t h century . . . the conversion of the b u l k of the population of Egypt a n d N o r t h Africa is largely assumed." 9 C o m p a r i n g these sober b u t impressionistic conclusions with the proposed curve of conversion for Egypt, one finds complete agreement. T h e conversion process passes the halfway point just before the beginning of the tenth century. T h e n i n t h century is, indeed, the watershed period with over a third of all conversions taking place between 800 and 900, b u t Brett's estimate that the Christians were in the majority as late as the year 900 may well be accurate since the conversion curve makes no allowance for the u n k n o w n percentage of Christians who never converted at all. T h i s i n d e p e n d e n t corroboration of the chronology set by the conversion curve is particularly i m p o r t a n t since the likelihood that a curve derived f r o m name frequencies would correspond exactly with such i n d e p e n d e n t assessments is very remote unless one assumes that n a m e frequencies are, in fact, indirect reflections of conversion. It would now be desirable to proceed to a consideration of the social consequences of the conversion process, b u t the social and religious history of Egypt and T u n i s i a prior to the Fatimid 102
Egypt and Tunisia period is not known with enough clarity at present to attempt this with any confidence. As in other areas, the conversion process brought about a period of urban growth, but did it bring about the gradual emergence of a dominant urban social class parallel to the patricians of Iran? Or did such a class emerge as rapidly? T h e indications that native converts in Egypt and Tunisia were assimilated much more slowly into the Muslim establishment than they were in Iran and Iraq raise doubts. Even within the strictures of the proposed axioms of conversion, the structure of provincial Byzantine society may well have caused the social profiles of different chronological categories of converts to be quite unlike those we encountered earlier. Or even if the social profiles were analogous, the attitude of the Arabs toward the Egyptian and Tunisian native populations may have been so negative as to stifle or retard the development of a body of converts honored for their religious eminence. A separate study would be necessary to examine these possibilities. Does this uncertainty mean, then, that the rivalry that is in evidence elsewhere between early convert families and late convert families does not apply to Egypt and Tunisia? Chronology alone would suggest that the element in the Egyptian and possibly the Tunisian population that felt most receptive toward Fatimid rule, and perhaps Fatimid theology, may have been composed of relatively late converts. This would be the case if the Fatimids served as the functional equivalent of the ShafiiAshari-Sufi tendency in Iran. As counter-orthodoxies, both were characterized, although in different ways, by mysticism, populism, and opposition to the entrenched legal establishment. But without knowing which and how many Egyptians and Tunisians welcomed Fatimid rule, one can do no more than suggest this comparison. T h e only sure datum is that Sunni factional rivalry along law school lines never approached the importance in Egypt and Tunisia that it had in Iran, Iraq, and Syria. In short, the social and religious impact of the conversion process upon Egypt and Tunisia is a subject that remains to be investigated.
103
9. Syria
T h e basic features of the popularity curve of Muslim names in Syria after the year 840 are so similar to those of the Egyptian curve (graph 18) that the two areas may reasonably be supposed to have a common naming history in that period, or a naming history determined by common influences. Syria, therefore, can be considered a test area for some of the arguments adduced to explain the changing naming patterns of Egypt.
Graph 18. Comparison of graphs 11 and 12. A, popularity of Muslim names in Syria (curve from Ihn al-'Imäd only); B, popularity of Muslim names in Egypt.
104
Syria Before discussing the two great peaks of coincident popularity in 864-888 and 961-985, however, the earlier portion of the Syrian curve must be explained. That Syria should have a complicated curve in the earlier period while Egypt has none at all is not surprising. From 661 to 750 caliphs of the Umayyad dynasty ruled the entire Muslim empire from the capital province of Syria, and this political prominence understandably led to a more salient role for Syria than for Egypt even for a generation or two into the Abbasid period, although, to be sure, the two provinces together did not even approach Iraq in overall importance (see graph 1). More problematic than the mere existence of an earlier curve in Syria is the sharp dip it takes between 816 and 840. Although this data may be distorted by that being a period yielding exceptionally low numbers of Syrian biographies, the presence of a noticeable dip in the Iraqi name curve at exactly the same time at least suggests a substantive reason for the dip. However, following the analysis of the preceding chapters, whatever that reason might be, it presumably affected the Arab portion of the population only, seeing as it had but a minor impact in Iraq where converts achieved representation in the biographical compilations gradually but a major impact in Syria where early biographies are mostly of Arabs. This analysis depends, of course, upon acceptance of the proposition that the first great peak in the name curve in Syria and Egypt is caused almost exclusively by the naming practices of the Arab population rather than those of the indigenous convert population. Little concrete substantiation of this proposition has hitherto been provided because suitable evidence is available only for Syria. Verifying the identity of the subjects of biographies as ethnic Arabs or as indigenous converts is difficult and uncertain, which means that determining the actual proportion of Arab to non-Arab in any sample is a futile undertaking. In Syria, however, a plausible quantitative indicator is afforded by the likelihood that almost the entire Arab tribal population, whether resident in Syria before Islam or immigrant to Syria during and after the Arab conquest, lived to the east of the Jordan-Biqä'-Orontes valley that separates coastal Syria from the inland areas on the fringe of the desert. 1 By noting the spe105
Conversion to Islam in the Medieval Period cific place of origin of individuals in biographical dictionaries as being f r o m coastal Syria or inland Syria, therefore, it is possible to reach some conclusion a b o u t A r a b versus non-Arab representation. T a b l e 3 gives, by fifty-year periods according to birthdates, the n u m b e r of biographies in the Shadharät adh-dhahab f r o m inland and f r o m coastal Syria and the percentage of the total represented by the latter group. Q u i t e obviously, the quantities involved are vanishingly small in the early periods, b u t the percentages suggest a natural division into chronological periods. Prior to the year 791 the percentage never rises above 13 although there is every reason to believe that the population of the coastal area, as here defined to include the central valley, constituted a far greater percentage of the Syrian population as a whole. After 791 the percentage f r o m the coastal region quickly doubles, and for two centuries roughly a q u a r t e r of the biogra-
Table 3. Relative number of biographies from inland and coastal Syria. Chronological period
Approximate Coastal birthdates Syria
Inland Syria
Total
Percentage from coastal Syria
' 646-694 695-742 . 743-790
2 2 5
23 31 33
25 33 38
8 6 13
II
' 791-839 840-887 888-936 937-984
5 9 8 9
16 26 24 28
21 35 32 37
24 26 25 24
III
' 985-1033 1034-1081 .1082-1130
4 7 9
15 28 51
19 35 60
21 20 15
IV
1131-1179 1180-1227 1228-1276 1277-1324 1325-1373 1374-1421 1422-1470 .1471-1519
36 62 43 62 61 26 61 21
92 163 91 130 129 25 142 64
98 225 134 192 190 61 203 85
37 28 32 32 32 43 30 25
I
where Ρ is the proportion of the eventual total and Τ the time at some point in the process. T h e 100 in the numerator represents 100 percent or the upper limit of the process. T h e letter e represents the constant upon which natural logarithms are based. T h e parameters a and b change from one application of the curve to another. For further discussion see Peter R. Gould, Spatial Diffusion, Association of American Geographers Resource Paper no. 4, Washington, D.C., 1969, pp. 19-21. 9. Raymond Pearl, Studies in Human Biology (Baltimore: Williams 8c Wilkins, 1924), pp. 558-583. 10. T h e curve is discussed at some length in Margaret J . Hagood, Statistics for Sociologists (New York: Henry Holt, 1947), pp. 272-289, and in F. E. Croxton and D. J . Cowden, Applied General Statistics (New York: Prentice-Hall, 1941), pp. 452-^61. But by 1973 Taro Yamane could write, "today it is hardly ever used, but will be presented here mainly for historical interest," Statistics: An Introductory Analysis (New York: Harper Sc Row, 1973), p. 1057. 11. Everett M. Rogers, Diffusion of Innovations (New York: Free Press of Glencoe, 1962), pp. 152-158. Rogers draws conclusions about the applicability of S-shaped curves based on a large number of separate studies. An example of such a study may be found in Zvi Griliches, "Hybrid Corn: An Exploration in the Economics of Technological Change," Econometrica, 25 (1957), 501-522. T h e most extensive claims for the general validity of analysis by means of logistic curves are made by Robert L. Hamblin, R. Brooke Jacobsen, and Jerry L. L. Miller in A Mathematical Theory of Social Change (New York: John Wiley & Sons, 1973) chaps. 3-4. They also include a useful discussion of the history and theoretical foundations of the curve's application. 12. T h e irregularity and changing character of Islamic taxation policy vis-à-vis non-Muslims is well studied in Daniel C. Dennett, Conversion and the Poll Tax in Early Islam (Cambridge, Mass.: 143
Notes to Pages 31-35 Harvard University Press, 1950). T h a t there was at least occasionally an economic component to conversion in later periods when taxation policy had become stabilized is attested to by the poem on the Banü Säsän scoundrel underworld by Abu Dulaf preserved in athTha'âlibî, Yatïma ad-dahr, ed. M. M. 'Abd al-Hamïd (Beirut: Dar al-Fikr, 2d ed., 1973), I I I , 359. According to him confidence men would feign conversion to Islam apparently in order to be showered with gifts as rewards for their enlightenment. 13. T h e logistic curve is not precisely the same as the integrated normal distribution curve, but there is so little difference between the two that the logistic curve is usually used because it is easier to apply. For a comparison of the two curves see C. P. Winsor, "A Comparison of Certain Symmetrical Growth Curves," Journal of the Washington Academy of Sciences, 22 (1932), 73-84. 14. Rogers, Diffusion of Innovations, pp. 159-171. 4. Conversion as a Social Process 1. Circumcision was unquestionably performed in at least some instances, but it is rarely mentioned. For evidence of it see Jean De Menasce, "Problèmes des Mazdéens dans l'Iran musulman," Festschrift für Wilhelm Eilers (Wiesbaden, 1967), p. 225. For Zoroastrians the act of conversion to Islam would include removal of the sacred thread worn on the body, but this was naturally of little interest to Muslim chroniclers. 2. A good survey of social and religious change in the Hellenistic period is to be found in F. E. Peters, The Harvest of Hellenism (New York: Simon and Schuster, 1970). 3. T h e perfecti of Manichaeism constituted an elite group of religiously pure individuals. For the Zoroastrian ecclesiastical structure see, for example, Richard N. Frye, The Golden Age of Persia (New York: Barnes & Noble, 1975), p. 17. 4. S. D. Goitein, A Mediterranean Society, II, The Community (Berkeley: University of California Press, 1971), pp. 5-40. 5. N. Pigulevskaya, Les villes de l'état iranien aux époques parthe et sassanide (Paris: Mouton 1963), pp. 106-111. 6. Pigulevskaya, Les villes de l'état iranien, pp. 102-111. 7. De Menasce, "Problèmes des Mazdéens," pp. 220-230; B. T . Anklesaria, The Pahlavi Riväyat of Äturfarnbag and Farnbag-srös, 2 vols. (Bombay: Industrial Press, 1969). 8. Steven Runciman, Byzantine Civilization (Cleveland: World, 144
Notes to Pages 36-41 1956), pp. 89-91; Richard N. Frye, The Heritage of Persia (Cleveland: World, 1963), pp. 209-211. 9. T h e notion that new converts to a religion are usually zealots arises from the practice of studying only the early stages of a conversion process. More conservative later converts pass into the new religion without giving much evidence of fanaticism or zealous observance. 10. De Menasce, "Problèmes des Mazdéens," pp. 224-225, gives a Zoroastrian legal text confirming that the property of a convert to Islam could legally be seized by other Zoroastrians. 11. Anklesaria, Pahlavi Riväyat, II. Zoroastrians were constrained by their law from purchasing goods in Muslim markets (pp. 137-138) and from consorting with non-Zoroastrians in caravanserais (p. 129). They could sell livestock to Muslims only if there was no other way of earning a living; De Menasce, "Problèmes des Mazdéens," p. 223. Constraints of this sort must eventually have broken down as the Zoroastrian community became a minority, but initially they must have amounted to economic ostracism for many converts to Islam. 12. Edouard Farès, L'honneur chez les arabes avant l'Islam (Paris: Adrien-Maisonneuve, 1932), chap. 4. 13. Farès, L'honneur chez les arabes. 14. Martin Hinds, "Kufan Political Alignments and their Background in the Mid-Seventh Century A . D . , " International Journal of Middle East Studies, 2 (1971), 346-367; M. A. Shaban, Islamic History, A.D. 600-750 (A.H. 132) (Cambridge: Cambridge University Press, 1971). 15. Ibn al-Muqaffa', "ar-Risäla fï as-sahäba," Äthär Ihn alMuqaQa' (Beirut: al-Hayät, 1966), pp. 345-361; "Kädi," Encyclopaedia of Islam, new ed., IV, 373-374. 16. Henri Lammens, Etudes sur le règne du calife omaiyade Mo'äwia Ier (Paris: Paul Geuthner, 1908), pp. 66-108. 17. In one surprising instance in the mid-ninth century, a group of ascetics forming a brotherhood (ikhwän) may actually have used the word monk (rähib) to describe themselves; Abü Nu'aim, Akhbär Isbahän, II, 54, 57, 140. 18. T h e term mawlä is used here in its generally accepted sense. Recently completed, Ph.D. dissertations by Patricia Crone (Oxford University, 1974) and Daniel Pipes (Harvard University, 1978) indicate that the actual meaning of the word during the period in question is much more complex, although it does carry the sense of 145
Notes to Pages 41-48 "convert Actively linked to an Arab tribe" in many instances. T h e retention of the older understanding of the word is not intended to deny the newly revealed nuances but rather to indicate in shorthand fashion that body of early converts who were subject to discriminatory actions on the part of Arab Muslims. 19. A striking example of Arab derision of mawäli is provided by al-Jähiz in Kitäb al-qawl ft al-bighäl, ed. C. Pellat (Cairo: Mustafa al-Bâbï al-Halabï, 1955), p. 87, where a poem likens miscegenation of Arab and non-Arab to Arab women fornicating with donkeys and mules. 20. Michael G. Morony, " T h e Effects of the Muslim Conquest on the Persian Population of Iraq," Iran, 14 (1976), 54. 21. Gérard Lecomte, Ibn Qutayba (mort en 276/889), l'homme, son oeuvre, ses idées (Damascus: Institut Français de Damas, 1965), p. 345. 5. The Development of Islamic Society in Iran 1. M. A. Shaban, The 'Abbäsid Revolution (Cambridge: Cambridge University Press, 1970), pp. 155-158. 2. Gholam H. Sadighi, Les mouvements religieux iraniens au Ile et au lile siècle de l'hégire (Paris: Les Presses Modernes, 1938); M. Rekaya, "La place des provinces sud-caspiennes dans l'histoire de l'Iran de la conquête arabe à l'avènement des zaydites (16-250 H./ 637-864 J.C.): particularisme régional ou rôle 'national'?" Rivista degli Studi Orientali, 48 (1973-1974), 137-151. 3. Rekaya, "La place des provinces," pp. 144-148. 4. Lawrence Brown, Diffusion Dynamics: A Review and Revision of the Quantitative Theory of the Spatial Diffusion of Innovation. Lund Studies in Geography, series B, no. 29 (Lund, Sweden, 1969). 5. at-Tabari, Abü Ja'far Muhammad, The Reign of al-Mu'tasim (833-842), tr. E. Marin (New Haven: American Oriental Society, 1951), pp. 111-119, 123. 6. C. E. Bosworth, " T h e Tähirids and Saffärids," in Cambridge History of Iran, ed. Richard N. Frye (Cambridge: Cambridge University Press, 1975), IV, 104-106; Gilbert Lazard, Les premiers poètes persans (Paris: Adrien-Maisonneuve, 1964), I, 12, 17-18. 7. C. E. Bosworth, " T h e Heritage of Rulership in Early Islamic Iran and the Search for Dynastic Connections with the Past," Iran, 11 (1973), 53-54. 146
Notes to Pages 48-58 8. Bosworth, "Tähirids and Saffarids," pp. 106-135; E. de Zambaur, Manuel de généalogie et de chronologie pour l'histoire de l'Islam (Bad Pyrmont: Heinz Lafaire, 1955), pp. 200-201. 9. S. M. Stern, "Ya'qüb the Coppersmith and Persian National Sentiment," in Iran and Islam, ed. C. E. Bosworth (Edinburgh: Edinburgh University Press, 1971), pp. 535-555. 10. Richard N. Frye, "The Sämänids," in Cambridge History of Iran, IV, pp. 136-161. 11. H. Busse, "Iran under the Büyids," in Cambridge ^History of Iran, IV, 250-304. 12. The earliest coin I have seen bearing the titles as-sultän ala'zam and shähanshäh was dated 437/1045-1046. It was privately owned. 13. Rogers, Diffusion of Innovations, pp. 169-170. 14. It is inconceivable that the Umayyad governor al-Hajjäj b. Yüsuf could have maltreated mawâlî as he did, driving them back to their villages and branding them with their village names, if they had been people of wealth and social standing; William Muir, The Caliphate: Its Rise, Decline, and Fall (Edinburgh: John Grant, 1924), pp. 335-336. On lower class converts see Morony, "The Effects of Conversion," p. 51; on prisoners of war, pp. 49-51. 15. See chap. 4, note 10. A woman whose only male relations were Muslim had to have a Zoroastrian guardian appointed for marriage purposes, Anklesaria, Pahlavi Riväyat, II, 48^49. 16. De Menasce, "Problèmes des Mazdéens," p. 223. 17. Five of the seven streets mentioned by name in Abü Nu'aim's biographical dictionary of Isfahan are quite obviously named for the individuals who developed them. Fully half of the better than fifty streets known for the city of Nishapur are named for individuals who were not particularly famous otherwise. 18. The classic statement of this theory is in G. E. von Grunebaum, "The Structure of the Muslim Town," in Islam: Essays on the Nature and Growth of a Cultural Tradition, ed. G. E. von Grunebaum (London: Routledge and Kegan Paul, 1964), pp. 141158. 19. The precise role profitability plays in the adoption of technological innovations is debatable (Griliches, "Hybrid Corn," pp. 516-521; Rogers, Diffusion of Innovations, pp. 136-143), but it is not questioned that the sense of deriving material benefit plays a crucial role. 20. For examples see Bulliet, Patricians, pp. 89,201-202. 147
Notes to Pages 58-67 21. As an example, the Mïkâlï family, which held high rank in pre-Islamic Iran, first rose to prominence as Muslims through their support of the Tahirids; Sa'ïd Nafïsï, ed., Tärtkh-i Baihaqï (Tehran: University of Tehran Press, 1332 [1953]), III (notes), 973ff. 22. For examples see al-Baghdädi, 'Abd al-Qähir b. Tähir, Moslem Schisms and Sects, tr. K. C. Seelye (New York: Columbia University Press, 1919), I, 27-39. 23. For a description of these two religious-political positions see Bulliet, Patricians, chap. 3. 24. Ira Lapidus, " T h e Conversion of Egypt to Islam," Israel Oriental Studies, 2 (1972), 255-256. 25. al-Mâwardï, Abü al-Hasan 'Ali, Les statuts gouvernementaux, tr. E. Fagnan (Algiers: Adolphe Jourdan, 1915), pp. 217218.
26. Of course, most early Mu'tazilis were Arabs, but the doctrine was formed against a background of religious debate between Muslims and leaders of other religions. T h e points of superiority claimed for Islam by the Mu'tazilis provided a defense for converts against the polemic of their erstwhile coreligionists; H. A. R. Gibb, Mohammedanism, 2d ed. (New York: Oxford University Press, 1962), pp. 112-115. 27. T h e best account of AsharT development is in Michel Allard, Le problème des attributs divins dans la doctrine d'al-Ash'ari et de ses premiers grands disciples (Beirut: Imprimerie Catholique, 1965). For specific Ashari family histories see Bulliet, Patricians, chaps. 9, 11. 28. These points are discussed at length from a different vantage point in Bulliet, Patricians, chaps. 3-4. 6. The Curve of Muslim Names 1. A large number of names are construed with 'Abd, meaning slave or servant [of God]. These theophoric names, a few dating back to the pre-Islamic period, include some that are quite common and some that are quite rare. Since an a priori judgment as to which might be expected to be common and which rare seemed unwise, the entire category was left out of consideration in this particular set of calculations involving rather small quantities. In the Mamluk period and later, 'Abd Allah becomes a standard name for Muslim converts, but it is not commonly used this way in the earlier period. 148
Notes to Pages 68-87 2. By the middle of the tenth century the occurrence of nisbas of Arab tribal origin in Nishapur had fallen permanently below 10 percent from almost 30 percent a century earlier. This indicates the reduced salience of tribal identification; Richard W. Bulliet, " T h e Social History of Nishapur in the Eleventh Century," (Ph.D. dissertation, Harvard University, 1967), graph 5. 3. For example, among the fifteen hundred or so patricians dying after 425/1033 listed in al-Färisi's biographical dictionary of Nishapur, there are five named Shah and two named Shähfür, both from the Iranian imperial past. Among the approximately three thousand earlier biographies from Nishapur these names never occur. 4. Ibn Farhün, Kitäb ad-dibäj al-mudhahhab ft ma'rifa a'yän 'ulama' al-madhhab (Cairo: 'Abbäs b. 'Abd as-Saläm b. Shaqrün, 1351 [1932]). 5. Khalifa b. al-Khayyät, Abü 'Amr, Kitäb at-tabaqät, ed. S. Zakkär (Damascus: Ministry of Culture, 1966). 7. Iraq 1. J. C. Russell, Late Ancient and Medieval Population, Transactions of the American Philosophical Society, new series vol. 48 (Philadelphia, 1958), p. 89. 2. Julius Wellhausen, The Arab Kingdom and Its Fall, tr. M. Weir (Beirut: Khayat, 1963), p. 402. 3. Morony, " T h e Effects of the Muslim Conquest," pp. 49-51, 56-58. 4. Morony, " T h e Effects of the Muslim Conquest," p. 57. 5. Wellhausen, The Arab Kingdom, p. 72. 6. There were contemporary Khariji revolts in Iran, as well, that surely had a similar mawäl! component, but they were overshadowed by the non-Muslim revolts. 7. On occasion is was explicitly stated by local ulama that mere dynastic struggles were not the concern of the ordinary Muslim; W. Barthold, Turkestan Down to the Mongol Invasion, 3d. ed. (London: Luzac, 1968), pp. 267-268. 8. Gaston Wiet, Baghdad: Metropolis of the Abbasid Caliphate (Norman: University of Oklahoma Press, 1971), pp. 40-43. 9. Robert M. Adams, in Land behind Baghdad (Chicago: University of Chicago Press, 1965), has demonstrated that Baghdad and Samarra were each some ten times larger than the preceding Sassanid capital of Ctesiphon (p. 98). At the same time, the density of settle149
Notes to Pages 87-95 ment in the immediate hinterland of the capital was much reduced from Sassanid levels (p. 99). He remarks that "terminal Sassanian and early Islamic settlements in these areas often neatly alternated with one another along the same canal branches. Since in most cases the early Islamic sites were newly settled after Sassanian times, this suggests that the Sassanian abandonment was associated with a social upheaval sufficient to break off the tradition of residence at most of the Sassanian sites" (pp. 81-82). Morony ("The Effects of the Muslim Conquest," p. 47) concurs with Adams's suggestion that the indicated social disruption stemmed from warfare and natural calamity in the late Sassanid period. However, since the dating was established by pottery styles and there is no reason to assume that Sassanid pottery styles were immediately abandoned after the Arab conquest, I would suggest that what is actually indicated by this pattern is a migration of Muslim converts to new Muslim villages where Muslim ways were followed—including in pottery design—and a concomitant abandonment of non-Muslim villages where Sassanid styles persisted. This type of convert migration would complement the rural-urban pattern described elsewhere. 10. T h e increase in the twelfth century of the importance of the Jazïra area (northern Mesopotamia) relative to lower Iraq as indicated on graph 1 is a token of the distribution of Islamic prominence among smaller cities. 11. Ira Lapidus, "Separation of State and Religion in Early Islamic Society," International Journal of Middle East Studies, 6 (1975), 379-382. 12. S. M. Stern, "Ismâ'ïlis and Qarmatians," in L'Elaboration de l'Islam. (Paris: Presses Universitaires de France, 1961), pp. 99-108; E. Ashtor, A Social and Economic History of the Near East in the Middle Ages (Berkeley: University of California Press, 1976), pp. 160-167. 13. George Makdisi, Ibn 'Aqil et la résurgence de l'Islam tradionnaliste au XIe siècle (Damascus: Institut Français de Damas, 1963), pp. 340-375. 8. Egypt and Tunisia 1. Russell, Late Ancient and Medieval Population, pp. 89-90. 2. A careful study of a family of Iraqi officials serving in Egypt is contained in Hans Gottschalk, Die Mädarä'iyyün: Ein Beitrag zur Geschichte Ägyptens unter dem Islam (Berlin: De Gruyter, 1931).
150
Notes to Pages 95-111 3. M. Talbi, "Droit et économie en Ifrïqiya au IlIe/IXe siècle," paper presented to Princeton University conference on the economic history of the Middle East in 1974. 4. For information on native rebellions in Egypt, see Ira Lapidus, " T h e Conversion of Egypt," pp. 256-257; for Tunisia see Charles-André Julien, Histoire de l'Afrique du nord, 2 vols. (Paris: Payot, 1961), II, 30-33. 5. On the urban expansion of Fustât-Qatâ'ï-Cairo, see Encyclopaedia of Islam, new ed., articles by J . Jomier, "Fustät," II, 957-958, and J . M. Rogers, "al-Kähira," IV, 424-441; and also J . Abu-Lughod, Cairo (Princeton: Princeton University Press, 1971), pp. 13-36. 6. For the period prior to A.D. 750 there is substantial evidence connecting conversion with rural migration; Dennett, Conversion and the Poll Tax, pp. 110-114. Dennett also points out that conversion was numerically not very significant during this period (p. 115) and that ostracism of converts by the Christian community was a heavy burden (p. 87). 7. Lapidus, " T h e Conversion of Egypt," p. 260. 8. Michael Brett, " T h e Spread of Islam in Egypt and North Africa," in Northern Africa: Islam and Modernization, ed. Michael Brett (London: Frank Cass, 1973), p. 4. 9. Brett, " T h e Spread of Islam," p. 9. 9. Syria 1. Shaban, Islamic History, A.D. 600-750, pp. 40-41; Wellhausen, The Arab Kingdom, pp. 131-134. 2. An introduction to the chaos of Syrian political history may be found in Marius Canard, Histoire de la dynastie des H'amdanides de Jazîra et de Syrie (Paris: Presses Universitaires de France, 1953), I, 598-618. 3. Stern, "Ismâ'ïlîs and Qarmatians"; Canard, Histoire de la dynastie des H'amdanides. 4. al-Maqdisï, Ahsan at-taqâsïm ft ma'rifa al-aqâlïm (Leiden: Brill, 1885), compare pp. 179-180 on Syria with pp. 323, 336 on eastern Iran. 5. M. G. S. Hodgson, "Durüz," Encyclopaedia of Islam, new ed., II, 631-634; Η. Ζ. Hirschberg, " T h e Druzes," in A. J . Arberry, ed., Religion in the Middle East (Cambridge: Cambridge University Press, 1969), II, 330-348; Philip K. Hitti, The Origins of the Druze People and Religion (New York: Columbia University Press, 1928); 151
Notes to Pages 111-118 Bernard Lewis, The Assassins (New York: Basic Books, 1968); L. Massignon, "Nusairi," Encyclopaedia of Islam, old ed., III, 963-967. 6. Xavier de Planhol, Les fondements géographiques de l'histoire de l'Islam (Paris: Flammarion, 1968), pp. 95-100. 7. Encyclopaedia of Islam, new ed., artirles on "Bektâshiyya," I, 1161-1165; "Barghwäta," I, 1043-1045; "Hä-Mim," III, 134-135. 8. Non-Muslim elements are not clear-cut, but the Druze sometimes practice baptism (Hodgson, "Durûz," p. 364), and Nusairi food taboos, specifically against camels, suggest a Jewish or Christian origin (Massignon, "Nusairi," p. 966). For various other theories on the ethnic derivation of the Druze, see Hitti, The Origins of the Druze People, chaps. 3-4. Hitti's own theory of an Iranian origin (chap. 4) rests on scant and insubstantial evidence. 9. Hitti, The Origins of the Druze People, p. 14; Carleton S. Coon, Caravan: The Story of the Middle East (New York: Holt, Rinehart and Winston, 1966), p. 158. 10. Spain 1. T h e five works were published in Madrid between 1882 and 1890 in the series Bibliotheca Arabico-Hispana edited by Francisco Codera et al.: I-II, Ibn Bashkuwäl, Khalaf b. 'Abd al-Malik, Kitäb as-sila fi ta'rlkh a'imma al-andalus wa 'ulamä'ihim wa muhaddithlhim wa fuqahâ'ihim wa udabä'ihim; III, ad-Dabbï, Ahmad b. Yahyä, Bughya al-multamas fi ta'rlkh rijàl ahi al-andalus; IV, Ibn al-Abbär, Muhammad b. 'Abd Allah, al-Mu'jam fi ashäb al-Qâdï al-Imäm Abi 'Ali as-Sadafi; V-VI, Ibn al-Abbär, Muhammad b. 'Abd Allah, at-Takmila li-kitäb as-sila; VII, Ibn al-Faradï, 'Abd Alläh b. Muhammad, Kitäb ta'rlkh 'ulamä' al-andalus. T h e method of determining the approximate date of conversion in each genealogy was the same as that described for Iran in chapter 3. 2. It could be argued that the graph suggests a slightly earlier date for the peak of the curve, but 350/961 was selected because of the likelihood that changes from Berber to Arabic names may have caused the curve to peak somewhat earlier than it would have if only indigenous converts were involved. 3. Volumes I-III and V-VII of the Bibliotheca Arabico-Hispana were used (see footnote 1 of this chap.) along with the Spanish entries in Ibn al-'Imäd, Shadharät adh-dhahab, and Ibn Farhün, ad-Dibäj al-mudhahhab. 4. Obviously, the pooling of biographies from different sources 152
Notes to Pages 120-136 runs the risk of duplicating data, but the compilations give the impression of being quite independent, indicating that the distortion of the curve that might arise from duplicated data should be minimal. 5. The works used were volumes III and VII of the Bibliotheca Arabico-Hispana by ad-Dabbï and Ibn al-Faradi. 6. T h e curve as reconstructed seems to have too high a percentage in the earliest period. This may be because of the inclusion of an unknown number of Berbers in the group of 154 genealogies. 7. Ibn Hafsün's rebellion was supported largely by Muslim converts and Christians. He himself was from a convert family and reconverted to Christianity during the course of the rebellion; E. Lévi-Provençal, Histoire de Espagne musulmane, 3 vols. (Paris: G.-P. Maisonneuve, 1950-1953), I, 300-310, 368-380. 8. Al-Ghazâlï's works were declared heretical and burned early in the twelfth century; W. M. Watt and P. Cachia, A History of Islamic Spain (Garden City, N.Y.: Doubleday, 1967), pp. 116-117. 9. Watt and Cachia, A History of Islamic Spain, pp. 123-125. 10. The revolt in western Spain led by the Sufi Ibn Qasï from 1142 to 1151 shows the political potential of Sufism as well as a special appeal to Christians. One of the few things reported about the religious foundation of the revolt is that the khutba (Friday sermon) was delivered with three people in the minbar (pulpit) representing Muhammad, Jesus, and John the Baptist. David R. Goodrich, "A Süft Revolt in Portugal: Ibn Qasï and His Kitäb Khal' alna'layn," Ph.D. dissertation (Columbia University, 1978), chap. IV-A. 11. The Consequences of Conversion 1. T h e accepted "orthodox" theory of the caliphate began to crystallize in the al-Ahkäm as-sultäniya of al-Mâwardï (d. 1058); E. I. J . Rosenthal, Political Thought in Medieval Islam (Cambridge: Cambridge University Press, 1962), pp. 21-37. 2- Bulliet, Patricians, chap. 4. 3. T h e best study covering the relationship between the government and the religious elite in this area is Ira Lapidus, Muslim Cities in the Later Middle Ages (Cambridge, Mass.: Harvard University Press, 1967). 4. Those data are taken from the biographical dictionaries of al-Häkim an-Naisäbüri and 'Abd al-Ghâfir al-Fârisï cited in chap. 3, note 1. 153
Notes to Pages 136-138 5. Data from Abu Nu'aim al-Isbahäni's biographical dictionary cited in chap. 3, note 1. Similar observations could be made from other biographical compilations as well. For example, tabulation of the geographical nisbas in the table of contents of eight volumes of the biographical dictionary of Baghdad, by al-Khatîb al-Baghdâdï, Ta'rlkh Baghdäd (Beirut: Dar Al-Kitab Al-Arabi, [n.d.]) indicates that the total number of scholars reaching Baghdad from Cairo and Damascus combined in the ninth and tenth centuries was less than half as great as the number arriving from Nishapur. 6. For a description of how this system worked, see Lapidus, Muslim Cities. 7. The absorption of the religious establishment into the state bureaucracy after the thirteenth century is a major problem of Islamic history that has received relatively little attention. Some observations on the problem are to be found in Richard W. Bulliet, " T h e Shaikh al-Isläm and the Evolution of Islamic Society," Studia Islamica, 35 (1972), 53-67.
154
Index
Abbasids, 54,80,90,95,105,107,109, 110,132,137; Arab basis of their movement, 43-44; decline of, 85-86 al-'Abbäsiya, 99 Abu Muslim, 44,45 Afghanistan, 11 al-Afshln, 47 Aghlabids, 99, 101
Baghdad, 12,80, 86,87, 88,90,100,101, 109,132,135,136 Bal'ami, 49 Barghwäta, 111 Basra, 38, 77,80,81,83, 86,136,137 Bektashis, 111 Berbers, 33,94, 111, 116-121 passim, 123,131 Biographical dictionaries, 5-6, 106; indicators of mobility of scholars, 136; professions indicated in, 132-133; regional variations in, 7-9 Brett, Michael, 102 Buluggïn b. Ziri, 94 Buyids, 13,49,85, 86, 87,100,110 Byzantium, Byzantine Empire, 1,11, 103,114
Ahmad b. Tulün, 98. See also Tulunids Algeria, 9, 11,99, 112, 114 Almohads, 9, 111, 119,120,129 Almoravids, 119, 129 al-Amin, 47 'Änah, 11 Anatolia, 9,11, 111 al-Andalus (Muslim Spain), 11. See also Spain Antioch, 136 Arabia, 11, 16, 37, 89,114; effect of tribal migrations on, 8 Arabic language, 13, 18, 36, 61; and conversion to Islam, 39; in Spain, 115-116 Arabs, 1,17,18, 3 5 , 3 7 , 5 4 , 5 5 , 5 6 , 6 8 , 7 2 , 77,90,115,118,119,131,132; ¡nearly Islamic Egypt, 92-95; role in Syrian conversions, 105-108 Aramaic language, 84 al-'Arish, 11 Asceticism, 59, 61; in early Islam, 40 Asharis, 59, 60, 88, 89,103, 126 Assassins, 111, 112 Avestan language, 35 Ayyubids, 9, 137 Azerbaijan, 11, 45 Bäbak, 44, 45 Babylonia, 34
Caliphate, 38, 48, 85, 100, 101,129,132 Christianity, Christians, 34, 36, 41, 68, 78,102,114,120,121,123 Conversion: effect on central authority, 125,128-129,132,137-138; formal rite of, 33; generational differentiation in, 21-22; social aspect of, 33-34; theory of, 128; two axioms of, 35,41 Converts: social identity of, 36-37 Coptic language, 35 Copts, 102 Cordoba, 125 Crusaders, crusades, 107,110 Çukurova, the, 11 Dailam, 45 Damascus, 136 Darï language, 48 Demography, 15; and conversion in Egypt, 93-94; religious, 16-17
155
Index adh-Dhahabï, 13, 65,73, 74, 75 Dihqân, 53, 58, 89 Diyala River, 87 Druze, 111, 112 Egypt, 9, 11,13, 14, 16, 74-75, 78, 79, 92-113 passim, 115,118, 124,131, 132, 134, 136, 137; conversion of Christians in, 102 Elburz Mountains, 45 Euphrates River, 11 Factionalism: conversion as stimulus to, 129-130; in Egypt and Tunisia, 103; in Iran, 61-62; in Iraq, 88-91; and rate of conversion, 71; in Spain, 125-126; in Syria, 110-111 Fatimids, 13, 14, 94,95,96,99, 100,101, 102, 103, 109, 110, 111, 112, 132, 137 Fustät, 98 Genealogies, 121; in Persian convert names, 18-19 Hadith, 55, 132 Hamdanids, 85,99, 108 Hâ-Mïm, 111 Hanafis, 5 9 , 6 0 , 6 1 , 8 8 Hanbalis, 13, 88,89,90 H ä r ü n al-Rashïd, 82 Hebrew language, 35 Hims, 136 Ibn al-'Arabl, 126 Ibn Farhun, 77 Ibn Hafsûn, 125 I b n al-'Imäd, 9, 12, 13, 73-77, 133 Ibn al-Muqaffa', 38 Ibn Qutaiba, 42 Ifrjqiya, 77. See also Tunisia Ikhshidids, 98, 108 India, 86 Indonesia, 4 Innovation diffusion, 57; applied to religious conversion, 28-32; and conversion in Iran, 46, 49-51 Iran, 7 , 1 1 , 1 3 , 1 6 , 1 8 , 4 3 , 4 5 , 6 2 , 63,64, 77, 87, 88, 96, 97, 98, 100, 101,103, 111, 112, 131,132, 135, 136, 137, 138; categories of converts in, 51-53;
conversion of, compared with Iraq, 81-85; conversion of, compared with Spain, 121-124; independent Muslim dynasties in, 46-49 Iraq, 11,13,16, 73-74, 78, 79, 80, 89,91, 93,94,95,96,107,108,112,113,115, 124, 131, 134, 135, 136, 137; conversion of, compared with Iran, 81-85 Isfahan, 18,65, 136 Islam: orthodoxy of, explained, 59; in rural setting, 55 Ismailis, Ismailism, see Shiis, Shiism al-Jawhar, 94 al-Jazira, 11. See also Mesopotamia Jerusalem, 34, 136 Jordan,11 Jordan-Biqä'-Orontes valley, I I , 105 Judaism, Jews, 34, 35, 36, 68, 115 Karkh, 12 Kayanids, 48 Khalifa b. al-Khayyät, 78 Kharijis, Kharijism, 84, 89, 100 Khurrämdiniya, 45 Khurramis, 45 Kitäb al-'ibar ft khabar man ghabar (adh-Dhahabï), 13,65,73 Koran, see Q u r a n Kufa, 38, 77, 80, 81, 83, 86,136,137 Kurdistan, 11 Lapidus, Ira, 102 Law schools (madhähib), 61, 110. See also individual law schools Libya, 11 Logistic curve, 27-28; and innovation diffusion, 29-32 Maghrib, 9. See also North Africa al-Mahdiya, 99 M a h m u d of Ghazna, 86 Malikis, 77, 126 Mamluks, 9, 137 al-Ma'mün, 47 Manichaeism, 34, 35 al-Mansür, 87 Manzikert, Battle of (1071), 11 al-Maqdisï, 111 Mawlâ (pl. mawält), 41, 55, 68, 81, 84
156
Index Mazdakites, 45 Mecca, 37, 38,54 Medina, 38, 54 Mesopotamia, 108 Middle East, 2. See also individual countries Mihna, 88 Mobad, 36, 39 Mongols, 7, 8,130,131, 137 Monophysitism, 35 Morocco, 9, 11, 111, 114 Mu'âwiya, 39
Qarmatians, 89, 90,110; a n d local autonomy in Syria, 108-109 al-Qatâ'ï, 98, 99 Quantitative history, 4-5 Quran, 38, 60, 90
M u h a m m a d (the Prophet), 13, 40 Mulük at-tawä'if, 125, 138 Mu'tazilis, 59, 60, 88, 89 al-Muwaffaq, 85 Names, naming, 4; curves of, compared, 72; in Egypt, 74-75, 76, 78, 79, 93-95; and Fatimid conquest of Egypt, 100-101; importance oí nisba in, 10-11, 12, 68; in Iraq, 73-74, 78, 79; Muslim, 65,68-71; Persian, 71; pre-Islamic, 67-68; in Spain, 77, 116-119; structure of medieval Arabic, 10; in Syria, 75-76, 78, 79,104-106; in Tunisia, 76-77, 78, 79, 95-96; use of biblical, 65, 67, 68 Nestorianism, 35 Nile River, 97 Nishapur, 18, 19, 21, 65,134, 135, 136 North Africa, 2 , 9 , 7 7 , 1 1 4 . See also Algeria, Morocco, Tunisia Nusairis, 111, 112 Ottoman Empire, Ottomans, 9, 14 Palmyra, 11 "Party kings," see Mulük at-tawä'if Patricians, 18; in Egypt a n d Tunisia, 103; emergence of, 133-135, 136-138; and rates of conversion, 65-66. See also Ulama Persian Empire, see Sassanids Persian poetry, 47-48 Philosophy, Greek, 60 Prophet, the, see M u h a m m a d (¿adi, 55,56 Qairawän, 99
Raqqa, 11 Rasulids, 8 Regionalism: in Islamic history, 7; as a factor in conversion, 17 Religion: in pre-Islamic Arabia, 37 Religious authority: role in conversion, 35, 39-41 Revolts, non-Muslim, 108; in Egypt and Tunisia, 96-97; in Iran, 44^16; in Spain, 125 R o m a n Empire, the, 34, 35 R o u n d city, see Baghdad R u m , see Anatolia Russell, J. C„ 80 Saffarids, 48, 100 Samanids, 48-49 Samarra, 85, 87, 98 Sassanids, 1, 34, 35, 49, 83, 114 Seljuqs, 7-8, 49 Shadharät adh-dhahab ft akhbär man dhahab (Ibn al-'Imäd), 9, 10, 12-13, 73, 106 Shafiis, 13, 59, 60, 88, 89, 90, 103 Sharia, 39 Sharifs (of Mecca), 8 Shiis, Shiism, 13, 39, 86, 90,130; Imami, 100; Ismailis, 89,94, 100, 109. See also Fatimids Sicily, 99 Social history, Islamic, 3 Society, Islamic: formation of, 1-3 Society, Middle Eastern: religion as focus of, 34-37 Spain, 9, 77, 102, 131, 132, 134,135, 136, 137; Arab conquest a n d conversion of, 124-125; Christian reconquest of, 119, 129, 130; independent Muslim dynasties in, 125, 138; indigenous population of, 114, 116,123; isolated character of, 114-115 Sufism, Sufis, 55, 59, 61, 88, 89, 103, 125, 126,137 Sunnis, 15, 86,90, 125, 130
157
Index Syria, 8 - 9 , 1 1 , 1 3 , 1 4 , 1 6 , 75-76,78,79, 80,89,96,102,103,113,115,118,124, 131,132,135,136,137; politics a n d conversion in, 108-110; religious sects in, 110-112 Syriac language, 35
at-Tabari, 48 Tähir, Tahirids, 4 7 ^ 8 , 9 7 , 9 8 , 9 9 T a k r i t , 11 Tarsus, 11 T a u r u s Mountains, 11 T a x a t i o n , Islamic, 35, 42 Theology, Islamic: and conversion in Iran, 59-61 Tigris River, 11, 83; population in valley of, 80-81 Timur, 7 T r a d i t i o n s of the Prophet, see Hadith Transoxania, 11 Tribalism: in early Islam, 37-38 T u l u n i d s , 97-99,108 Tunisia, 11,76-77, 7 8 , 7 9 , 9 4 , 9 5 , 9 6 , 9 7 , 99-103,107, 108,110,115,118,124
Turco-Mongolian steppe, 114 Turkey, 11 'Ubaid Allah b. Ziyäd, 81 Ulama, 14, 39,40,117 Umayyads, 37,41,81,105,110,137; in Spain, 125 Urbanism, Islamic, 68, 134; a n d conversion in Iraq, 86-87; converts migrate to centers of, 81; in Egypt and Tunisia, 98-100; institutions of, 55-56; in Iran, 53-54; n a t u r e of, explained, 54-55 Wasit, 86 West Africa, 4 Yemen, 8, 37 Zab, the, 97 Zaidis, 111 Zanj rebellion, 85 Ziyarids, 100 Zoroastrianism, Zoroastrians, 34, 35, 36, 38, 45,47, 53, 69, 79
158