declining subcaste endogamy in India: the Hyderabad Kayasths, 1900-75

The subcaste, iati, is the endogamous unit in the Hindu caste system, the unit within which marriages must be arranged. The belief that living beings belong to separate species, dif-ferentiated by biogenetic substance and code, is fundamental to the caste system; one’s dharma, “duty or prescribed behavior,” is encoded in one’s physical substance (Inden and Marriott 1974). The preservation of the social and moral order depends upon marriage and procreation within the subcaste; members of a subcaste share particular attributes and behavioral patterns which set them apart from others. The conceptual categories of caste and varna (Sanskrit for “category” or “order”) have been utilized to classify and rank the numerous subcastes of the Indian subcontinent (Cohn 1968,1971; Srinivas 1962). For example, there are said to be 12 subcastes in the Kayasth caste, and the Kayasth caste has been placed variously in the Brahman (priest), Kshatriya (ruler] or Sudra (worker) varna, depending upon different regional caste systems and interpretations of Anglo-Indian law. The sub caste, however, is the endogamous unit, the most meaningful group for the everyday social relations of its members. level of caste intersubcaste marriages Some caste associations have such inter-marriages the in marriage patterns whose degree This paper explores the declining endogamy of the Kayasth subcastes of Hyderabad, india, by analyzing patterns an marriage patterns analyzed this of each pair of subcastes each The a in the three time periods. The Kayasths from a is

The Kayasth caste in Hyderabad city, in the Indian state of Andhra Pradesh, originally came from northern India. The Kayasths are known throughout India as a "writing caste"; their traditional occupation has been government service since their appearance in the 16th century. A few Kayasths migrated to Hyderabad as part of the administrative entourage of Nizam ul Mulk, governor of the Deccan province for the Mughal empire in the early 18th century. Kayasth kin groups developed in Hyderabad city, capital of the autonomous state established by the first Nizam's successors in the late 18th century. The Kayasths in Hyderabad city belong to seven different subcastes: Asthana, Bhatnagar, Gaur, Mathur, Nigam, Saksena and Srivastava. Leonard (1978b3 has described the historical development of the Kayasth subcastes and kin groups; some 320 khandans (Urdu for "patrilineage" or "family") are presently established in Hyderabad city.
Historical and contemporary data on members of the caste come from fieldwork conducted by Leonard in 1965Leonard in -66, 1970Leonard in -71 and 1976. Genealogies and family histories were reconstructed through interviews, with a particular emphasis on marriage connections. The information was cross-checked whenever possible by reference to written materials, primarily the administrative records of the former Hyderabad state in which Kayasths appear not only as subjects of the records but as record keepers. While many of the genealogies collected go back to the 18th and 19th centuries, the information is most complete and reliable for the 20th century.
The marriage patterns recorded for Hyderabad Kayasths in the 19th century confirmed subcaste endogamy. But in the 20th century, members of the Hyderabad Kayasth caste began arranging marriages beyond the subcaste boundaries, and this trend has increased.
However, the great majority of marriages beyond the subcaste have remained within the Kayasth caste. Such marriages, outside the subcaste but within the caste, are termed "exogamous" for purposes of this analysis.

analyals
All recorded marriages from 1900-75 were tabulated by date of marriage, sex and subcaste of spouse. Kayasths marrying outside the Kayasth caste, even though shown in Tables 1-3, were omitted from further analysis to avoid the problem of trying to estimate the number of endogamous marriages in the "other" category. When discrepancies occurred among genealogies from different subcastes as to the number of marriages, the larger number was used.' The marriages to Kayasths from outside the city of Hyderabad were included. The original data for each time period appear as Tables 1.2, 3. The columns repre sent females by subcaste; rows represent males by subcaste; and the frequencies within the tables are the numbers of marriages recorded. Reading down the columns, we can see how many males were taken in marriage by the female members of the subcastes; reading across, we can see how many females were taken in marriage by the male members of the subcastes. For example, in the third time period (Table 31, there were four marriages b e tween Asthana males and Bhatnagar females, while there were three marriages between Bhatnagar males and Asthana females. Although it is true that a sizable proportion of the marriages were exogamous, the actual numbers do not necessarily reflect patterns of preference between the groups because of the unequal population sizes. Unequal population sizes and differential sex ratios disguise the underlying preference patterns and make analysis of the original data misleading. The subcastes range in size from the Nigams (28 individuals recorded as marrying from 1951 to 1975) to the Mathurs (556 individuals marrying in that same period). Indices of preference based on the original data correlate more with the relative size of the subgroups than with preference. Before meaningful indices of preference can be calculated, the probabilities must be adjusted to reflect "true" patterns rather than size biases.
To correct the problem of unequal group sizes, and to discover the pattern within a table, we first normalize the data so that group sizes are equal (Mosteller 1968;Romney 1971). Romney describes a process by which subgroup sizes can be "equalized." The procedure is simply an iterative one of first setting the column totals to 100 and adjusting the cell values to percentages. Then, setting the row totals to 100, the cell values are readjusted to percentages. The process is continued until the row and column totals agree.3

698.9
Note: See note 3 underlying preference patterns, to the right of each example, are identical. The difficulty in ascertaining the "true" pattern of association present in a table of this sort is greatly increased the greater the discrepancy among group sizes.
In order to ascertain the underlying marriage patterns and compare them over time, we have normalized the Hyderabad marriage data in Tables 4, 5 and 6. The entries within the tables represent preferences for intermarriage among the groups. In the normalized tables we can see the marriage preference patterns that were present in the first three tables. Visually we can compare the relative amounts of endogamy within a given table, as well as the relative amount of endogamy of a given group across time. For example, the Asthanas have 100 percent endogamous marriages in the first time period, 64.5 percent in the second time period and 28.5 percent in the third time period. We can quantify the degree of endogamy for each pair of groups for each time period  to make more accurate comparisons between groups and across time. In Table 7, we used Mosteller's cross-products ratio (alpha) to describe each relationship because we desired a measure of association that would reflect the patterns within the tables without being sensitive to the unequal group sizes.' Alpha has a value of one when neither an endogamous nor an exogamous preference is present. When alpha is less than one and approaches zero, an exogamous preference is present. When alpha is greater than one and approaches infinity, an endogamous preference is present. Hence, the larger alpha is, the higher the preference for endogamous marriage. Table 7 shows the degree of endogamy for each pair of subcastes across the three time periods. The columns represent the three time periods and the rows represent all the possible combinations of the subcastes. The degree of endogamy i s expressed as the square root of alpha, so that we can talk in terms of the "odds of endogamous marriage."5 We can also calculate the overall measure of endogamy by using Romney's odds ratio (personal communication, 19791.' In time period one, the odds are 76:l in favor of marrying endogamously. In time period two, the odds become 16:l. Finally, in time period three, the overall odds have dropped to 5:l in favor of endogamous marriage. Exogamy has changed from 1 chance in 76 to 1 chance in 5 over the time periods studied. Clearly, a substantial change has taken place, and it is even more evident between the pairs of subcastes for each time period. The results appear in Table 7. As i s evident by looking across the columns in Table 7, some groups have changed dramatically. The most dramatic change has taken place among the Asthanas and the Saksenas. These groups did not intermarry prior to 1925, but in the third time period they are almost as likely to marry each other as to marry endogamousl y.

discussion
Historical developments in Hyderabad city and state, and their effect upon the Kayasths, help explain these shifts in marriage preferences. While much of the Indian subcontinent passed under the rule of the British East India Company in the 18th and 19th centuries, Hyderabad state stayed independent. The administration, modeled on that of the Mughal decllnlng subcaste endogamy In lndla SO9 empire, was presided over by the Nitam (governor or ruler) and a predominantly Muslim ruling class; Persian was the official state language. Hindus in crucial military or recordkeeping positions, including some Kayasths, ranked among the nobility as well, but Hindus predominated in middle and lower-level managerial and clerical positions. By the mid-19th century, most Kayasths in Hyderabad held hereditary positions in the Nizam's Mughlai (Mughalderived) administration. The administration was decentralized. Control of positions was delegated to kin groups which followed bilateral inheritance practices. Positions acquired by families were allocated by the head of the family to all male relatives of appropriate age, whether consanguineal or affinal. The earliest Kayasths attained noble status, properties and positions, and their descendants dominated endogamous groups in the city. Within some of the subcastes, several separate kin groups had developed, dependent upon particular resources and residing in particular localities (see Figure 1). In some cases, a separate kin group was the consequence of a later migration; in others, it was the consequence of allegedly impure ancestry. But the kin groups were basically property groups, transmitting their resources with minimal interference from the state.
By the mid-19th century, however, the state was under severe economic pressure. Nearbankruptcy and political pressure from the British Resident in Hyderabad' forced the gradual introduction of a new administrative system, modeled on the Anglo-lndian bureaucracy in British India. The prime minister recruited a new class of English-educated officials from British India, and for some 30 years Mughlai and Anglo-Indian administrations coexisted in Hyderabad. From 1883, following the death of the prime minister who had maintained the dual administrative system, the modern administrators seized political power from the Mughlai nobles and administrators (Leonard 1978al with the establishment of modern schools in the old city, did young Kayasths embark upon new educational and occupational paths. These young men formed Kayasth voluntary associations in the next few decades, associations based on particular neighborhoods and led by members of particular subcastes, but at times bridging these divisions. The associational activities, particularly a controversial all-India Kayasth Conference hosted in 1938, drew together in social reform efforts some educated young men in modern occupations; these efforts included exogamous marriages. But these "reform" marriages, based on an ideology of caste unification, were not among the first exogamous marriages for the Hyderabad Kayasths. The first marriages across subcaste lines, in the period 1900-25, occurred between members of the poorer kin groups of four subcastes in the old city. The Saksenas, Bhatnagars and Gaurs had developed two kin groups each by 1900. The Srivastavas were comprised of at least eight kin groups by 1900, and these varied widely in socioeconomicstatus (Leonard in press). In each of these subcastes, the higher-status kin group was led by families which had arrived earlier and risen higher in terms of noble rank and positions. The economic impact of the late 19th-century administrative and judicial changes in Hyderbad state first affected Kayasths in the poorer, lower-status kin groups. They were the first to change their marriage patterns, replacing more expensive long-distance endogamous marriages with local exogamous ones.
Urban residential patterns helped shape the shifting marriage patterns. In the first of the three time periods, exogamous marriages characteristically occurred between residents of the same residential locality in the old city (the Bhatnagars, Saksenas and Srivastavas in Husaini Alam, and the Gaurs, Saksenas and Srivastavas in Shahalibanda). Not until the second period, when "reform" marriages began among the higher-status subcaste kin groups, decllnlng rubcarto endogamy In Indla did residents of the new city marry residents of the old, or did residents of noncontiguous localities in the old city intermarry.
The three subcastes which did not begin exogamous marriages in the first period shared one important feature: structurally, each was a single kin group. In size, residence and characteristic occupation they differed greatly. The smallest group, 5 or 6 Nigam families, lived in the old city adjacent to the largest group, the 28 Mathur families. The Nigams were dominated by one family which held hereditary positions in the Nizam's household estab lishment. The Mathurs were dominated by the wealthy and noble "Malwala" family, hereditary keepers of half the land revenue records for Hyderabad state. The Asthana Kayasths, falling numerically between these two groups, all lived across the river in the new suburbs of Hyderabad city in 1900. They had come only in the late 19th century as members of the modern administrative class, and the founding family still headed the kin group. These three subcastes, each an endogamous group led and controlled by a single family, did not change their marriage patterns until the second period.
In the 1930s and 1940s, exogamous marriages within the Kayasth caste began to be advocated and practiced by "modernizing" men among the Kayasths of higher socioeconomic status. These leaders were not from the families of highest traditional status and income; they were the upwardly mobile, educated young men involved in the caste associations. Occupationally, they were moving beyond the declining hereditary stipends held by some families to take salaried positions in the modern administrations and to enter modern professions. Saksena subcaste members were leaders, and the exogamous marriages which they arranged clearly reflected common educational and professional experiences. Those who became kinsmen in some of these exogamous marriages included fellow students of St. George's Grammar School, colleagues in the Office of the Accountant General, and colleagues in district police and judicial service (Leonard 1978b:268-270).
The second period, then, saw the initiation of exogamous marriages by upwardly mobile, middle-class Kayasths. Such marriages were now occurring at two socioeconomic levels, and they remained separate; that is, marriages were being made across subcaste lines at both levels, but not across class lines. The data show sharply increasing rates of participation in exogamous marriages by most subcastes; the reality was more complex, since these marriages occurred at two different class levels and for different reasons.
A closer analysis of the Nigams and Asthanas, groups which joined the trend towards exogamous marriages in the second period, illustrates this complexity. A split occurred in the Nigarn kin group, and the poorer families began intermarrying with those lower-class Kayasths who had been arranging exogamous marriages locally. The leading family continued to make endogamous marriages with North Indian Nigams; it also began participating in local exogamous marriages, but with the Kayasths of higher socioeconomic status in Hyderabad. The Asthana kin group also split, and a few members moved to the old city and intermarried with Kayasths there. The Asthanas in the new city also began exogamous marriages, a few with local higher-status Kayasths, but most with an entirely different set of partners: Kayasths of other subcastes in North India. They also continued to marry endogamously outside of Hyderabad. The Mathur Kayasths remained highly endogamous until the third period, although a few individuals joined the "reform" associational activities and married exogamously. The unusual historical constitution of the subcaste, and i t s continuing dependence upon one dominant family, account for its slow change. When the Malwala family had secured hereditary claim to the record-keeper position in the second half of the 18th century, it o b tained fixed and permanent landholdings and stipends, and its members rose in the Hyderabad nobility. The family then recruited Mathurs from northern and western India, performing exchange marriages with the new families and establishing them in Hyderabad.
Written marriage contracts specified that the Malwalas would give each new family a residence, certain provisions (basic food items and cloth distributed from a central store house), and regular allowances. The Mathur families clustered in the locality around the Malwala Palace and shared some of the palace facilities, like the wedding hall. Kinship ties among these families were many and close, since they practiced exchange marriages and restricted their marriage network to the 28 families in the city by 1900.
A multidimensional scaling diagram, a spatial representation of Table 7, brings out some of the relationships described above, and points to 0thers.O The degree of exogamy b e tween groups can be thought of in terms of distance: groups more likely to intermarry would be closer together than groups less likely to intermarry. (A group's location is dependent upon its intermarriage with all other groups.) Figure 2 is a spatial representation of the degree of exogamy among subcastes for time period three. For example, in Table 7 we see that the two groups most likely to intermarry, for the third time period, are the Asthanas and the Saksenas. In Figure 2, the Asthanas and the Saksenas are located close together; the Caurs and Mathurs, the least likely to intermarry, are the farthest apart.
The spatial representation allows us to examine patterns of preference in exogamous marriage and avoidances. We have presented only the results from time period three because the earlier time periods were too endogamous to have discernible patterns of exogamy. For the third time period, we can see that an exogamous network exists, even though the overall preference is for endogamy. Most striking i s the relationship between the Asthanas and Saksenas (marked with a double line in Figure 2). They are as likely to intermarry as they are to marry endogamously. The solid lines connect subcastes that are about twice as likely to marry endogamously as exogamously. In this pattern, we see that the subcastes most likely to intermarry form a triangle: the Saksenas, Asthanas and N i g a m~.~ The Saksenas, in the center of the network, are the most involved in intermarriage and appear to be preferred by all. Subcastes connected by dotted lines are about three times as likely to marry endogamously as exogamously. All other relationships, with decllnlng subcaste endogamy In lndla stronger endogamy preferences, are not shown. However, we can see that avoidances are strongest among the Mathurs, Gaurs and Nigams.
Important changes have also occurred with respect to the geographic range of marriage patterns. Figure 3 adds the variable of the spouse's childhood residence, either inside or outside Hyderabad city. The vertical axis shows the percentage of individuals taking a spouse from outside the city;" the horizontal axis is the proportion of exogamous marriages by subcaste, using the normalized figures obtained in Tables 4, 5 and 6. Each sub caste is represented by a line; the three dots on each line move from left to right to show the subcaste's position for time periods one, two and three. At the beginning of the 20th century, five of the subcastes had substantial proportions of individuals marrying outside the city." The direction of movement is across the graph, from high endogamy in time period one to low endogamy in time period three, and down, towards increasingly localized marriages.
The localization of marriages clearly depends upon the size of the endogamous group residing locally. The two largest groups, the Mathurs and Saksenas, tended to restrict marriages to the city population;" for them, the trend towards localization evidenced by the other subcastes is simply a continuation of past marriage practices. For the others, changing the definition of the endogamous group from the subcaste to the caste enlarged the pool of potential local marriage partners. The Asthanas and Srivastavas increased local exogamous marriages at the expense of extralocal endogamous marriages; the Bhatnagars, Gaurs and Nigams have made local exogamous marriages at the expense of local endogamous marriages.' Figure 3 shows that the proportion of individuals having spouses from outside Hyderabad in the first and second time periods is associated with the size of the subcaste population: the smaller the local group, the more marriages i t s members make outside the city. The Asthanas are the only exception to this: although they are the third largest group, they marry most frequently outside the city. Socioeconomic status helps explain the Asthanas' different pattern of geographic preferences. The Asthanas were approximately two generations ahead of the other Hyderabad Kayasths in acquisition of modern educations and jobs in the modern administration and professions, and their per capita income was higher than that of members of the old city subcastes." They had the resources to continue marriages outside the city, and their credentials placed them in an all-India arena for purposes of marriage, whether endogamous or exogamous. Another factor explaining the constriction of marriage patterns to Hyderabad city, the recent change in the city's political culture, is more difficult to assess. In 1948, Hyderabad state was incorporated by military action into newly independent India. In 1956, the state was dismembered and the city became the capital of a new state, Andhra Pradesh, with a hinterland largely drawn from a former British Indian province. Territorial boundaries and governmental institutions were restructured, and administrative personnel were displaced. Andhra Pradesh was the first state formed along linguistic lines in India's linguistic reorganization of states; Telugu is the official language and Telugu-speakers dominate politics and government service today. Many Kayasths felt adversely affected by these political changes, and the cultural changes were equally far-reaching. The Hyderabad court had retained a Mughal cultural heritage which had died out in British-ruled India. The Hyderabad Kayasths, long-standing participants in Mughlai administration and culture, were still well educated in Urdu and Persian. Younger generations were learning English and adapting to the more modern administrative and professional occupations, but their historical legacy set them apart from the Telugu-speaking majority in the new state. The exogamous marriages now being made locally by Hyderabad Kayasths are a reaction to the new political and cultural forces; those who were part of the old city Mughlai administration and culture are preserving its memory through marriages with local families who participated in it."

concluslons
The Hyderabad Kayasths are moving from a highly endogamous system towards a moderately endogamous one. To put it another way, the trend is from the subcaste to the caste as the unit of endogamy." Another trend is towards the localization of marriages-marriages arranged within Hyderabad city rather than outside of it. It is apparent that these trends are not occurring at a uniform rate among the different subcastes-some have changed their marriage patterns dramatically. Several factors proved important in explaining the subcaste variations. The number and socioeconomic status of kin groups in each subcaste were significant factors, based on historical developments in the 18th and 19th centuries. The changing nature of economic resources-the "property" of the kin groups under the Mughlai and modern administrations-had consequences for marriage patterns. The variable timing of occupational change ("modernization") was also important in producing the class differences which underlie the new marriage patterns. Comdecllnlng subcaste endogamy In lndla bined with class, size of a subcaste population was a factor crucial to the increased localization of marriages. Despite the strong influences of subcaste and class on marriage arrangements, the seven Kayasth subcastes in Hyderabad are more similar today in their marriage preferences and patterns than at any time in the past. notes