Where new caste names were accepted by the 1931 census to record the strength of discrete jatis or schismatic groups (which become, in effect, new jatis), they are reflected on our maps. On the other hand, when the names returned were such as to mask differences among groups that, in fact, remained socially distinct, as in the case of the aforementioned "Adi-" groups of South India, they are, if feasible, not reflected on our maps. Where such returns were numerically significant we have computed the percentages of the jatis involved in the 1921 district populations as given in the census for that year and assumed that those percentages were still applicable for the year 1931. Similarly, where data for Muslim jatis were suppressed in the 1931 census, we have calculated their per- centages as of 1921 or, in Sind and the Central Provinces, as of 1911 and utilized those figures on our maps wherever relevant.

For the area of Ceylon no twentieth-century census data on caste are available. The castes we have depicted there and their estimated proportion of the total population are derived from a variety of sources whose accuracy cannot be readily specified.

The spelling of jati names may differ from one census to another and, for a given census, from one region to another, in the latter case usually reflecting the forms of various lan- guages and dialects. On our maps we have normally opted for the most widely utilized spelling of the 1931 census. But occa- sionally, where two distinct forms enjoy widespread currency (e.g. "Goala" and "Ahir" in Bihar), both are specified.

Before concluding these introductory remarks for our maps of castes, tribes, and comparable groups, it is in order that we consider the relevance of the 1931 data for an understanding of the present social composition of the several countries and regions of South Asia. From what has already been said it is obvious that the caste system is not, as is often supposed, static. On a priori grounds, given the general acceleration of social change in the post-independence period and the official and unofficial opposition to a caste-rooted social order, we may assume that group-specific flux in the system, which was so evident in the reports accompanying the tabulations of the 1931 data, is substantially greater today than it was then. Nevertheless, the weight of evidence in the literature we have examined and our own firsthand field experience lead us to propose that where there is no clear evidence to the contrary the broad regional distributions of castes, tribes, and so forth, presented in plates X.C.1-3 remain valid for the present day. But, as is documented elsewhere in this atlas, a number of local factors have been at work to alter those distributions. Partic- ularly noteworthy are those changes occasioned by the mass movements of refugees across the Indo-Pakistani borders, both in the east and in the west, at the time of partition and after- ward; the conversion of millions of formerly untouchable Mahars to Buddhism in Maharashtra and Madhya Pradesh and the consequent diminution in the remaining strength of that caste; and the more gradual changes resulting from per- sistent immigration of both Hindus and Muslims into the Brahmaputra Valley of Assam, of nontribal peoples into pre- dominantly tribal regions, and of rural dwellers into the larger cities and towns. Another change of a qualitative rather than a quantitative sort is in the very significance of jati in the emerging social order. Its influence, we are led to believe, is very much on the wane in contemporary Pakistan and pre- sumably in Bangladesh and Sri Lanka as well. In India too its importance is being eroded, gradually in the countryside and more rapidly in the cities. Just how much longer it will con- tinue to be a primary social category, however, is a matter on which we shall not venture a judgment.

X.C.1. Most Numerous Caste, Tribe, or Other Ethnic Group

Over much of South Asia, the regional life-style and the pat- tern of social relationships of the rural population are greatly influenced by what anthropologists have termed "the dominant caste." The term "dominant caste" normally relates to the jati that holds a preponderant share of the land, that wields the greatest power in secular matters, and that typically, though not always, is the most numerous in a given region. The cate- gory, however, is subjective; in some areas differences of opin- ion may exist as to which caste is dominant, while in others (e.g., central Uttar Pradesh) it may be doubted whether any single jati merits being characterized as dominant. Further, it must be noted that while a given caste may be recognized as generally dominant in a particular region, many individual villages therein may be dominated by different groups.

A comprehensive map of South Asia showing dominant castes wherever they are believed to occur has yet to be pre- pared. But in the absence of such an undertaking, the map presented on plate X.C.1 should advance our comprehension of the regional situation. What the map actually depicts is the most numerous single caste, tribe, or comparable ethnic group in each district (or small princely state) of South Asia, irre- spective of its status. The groups are colored according to socioeconomic functional categories; however, the fact that several of the color tones used on the map apply to more than a single functional group should not be taken to mean that such commonly colored groups (B and C, D and E, and G and H) are equivalent, either in function or status. In fact, no ascription of rank is at all intended for groups E, G, H, and I. The remaining five groups, A-D and F, include the whole of the Hindu population, exclusive of Hinduized tribal peoples, and are hierarchically ordered by ritual status, with Brahmans in the highest position and subgroup F.3, Depressed Castes, in the lowest. Within each district, the proportion of the specific caste, or such, depicted to the total population is indicated, in four classes, by line patterns or, where the proportion is less than one-eighth, by the absence of a pattern.

It is readily apparent that most of the area mapped falls into one or another of three categories: D, "Cultivator and/or Pas- toral Castes," over most of what are today India and Sri Lanka; G, "Muslim Groups," over virtually all of what are today Paki- stan and Bangladesh; and I, "Tribes," over most of the hilly regions of central and northeast India and Burma. Usually, but not always, the specific jatis named in categories D and G will function locally as dominant caste groups, especially where they constitute a fourth or more of the total population. But, where the proportion is less than one-eighth, the case for assuming such a status may well be weak (e.g., as with the Goalas in much of Bihar). As for the tribes depicted under category I, their local dominance will be largely a function of the degree to which their hilly homelands have been pene- trated by immigrant lowland groups. Of the remaining map categories, B, "Rajputs," and E, "Ethnic Burmans," can be presumed, by and large, to be dominant in those areas where they appear on our map. Categories A, "Brahmans," C, "Vaish- yas," and H, "Syrian Christians" are all quite limited in their extent as most-numerous caste or other social group (the Vaishyas showing up in only a single district) and, given the coincidental fact that each coexists in its respective area(s) of numerical ascendancy with other numerous and powerful groups, we do not feel that a priori judgments on their domi- nance can reasonably be advanced. Of the one group remain- ing, F, "Low Status Castes," we can, however, state fairly safely that, whatever their numbers may be (it may be noted here that no single such caste ever constitutes as much as a fourth of the total population), they never have enough power to be in the position of a dominant caste at the district level of analysis.

Two tables accompany the map on plate X.C.1. The first, "Brahmans and Exterior Castes," calls for little comment. One may wish to consult the data on Brahmans, however, in com- bination with map (b) of plate X.C.2 and those on exterior castes ("exterior" being a census term roughly equivalent to "depressed" or the modern "scheduled") in combination with map (d) of plate X.C.3 and various maps of plate X.A.8. The second table, "Three Most Numerous Tribes, Castes, or Other Ethnic Groups," will convey some impression of which jatis are likely to be dominant or at least influential at the provin- cial or state level. Almost all the groups named in that table will also be found somewhere on the map to the right, usually, but not necessarily, in the state under which they are tabu- lated. Failing that, all but a few (listed for Sikkim and Coorg only) are to be found on some map on plate X.C.2 or 3. (The atlas index will help in finding them.) Thus one will virtually always be able to identify the functional group in which the named jatis might best be placed. Typically, the jati named as the "Most Numerous Group" in the table will also be the domi- nant caste in the administrative unit, taken as a whole. But there are a few obvious exceptions, the tribal Gonds in the Central Provinces and Berar and the depressed Chamars in the United Provinces being most worthy of note. In such cases one will have to consider the data under "2nd" and "3rd Most Numerous Group" to find other jatis of highly influential, if not clearly dominant, status.

Sources

For India

India, Census (1911), (1921), (1931).

For Ceylon

S. Casey Chetty (1834, listed under General References); R. Pieris (1956); B. Ryan (1953); A. Sievers (1964).

Other Useful Works on Caste in General (in addition to works cited in the General Bibliography)

E. A. H. Blunt (1931); Y. B. Damle (1961, listed under Bib- liographies); L. Dumont (1970); G. S. Ghurye (1961); J. H. Hutton (1963); I. Karve (1961); E. R. Leach (1960); M. Marriot (1965); J. E. Schwartzberg (1965), (1968); M. N. Srinivas (1962).

X.C.2. Selected High-Status Castes or Other Ethnic Groups, 1931

Each of the seven maps on atlas plate X.C.2 deals with a specific high-status ethnic group for which the term "caste" may or may not be deemed applicable. The groups in question generally fall into categories A, B, C, and G of plate X.C.1. (A single exception will be noted below.) The three maps on the left of the page relate to Muslim groups; those to the right to predominantly Hindu groups. On two of the latter maps Hindu castes are aggregated with their functionally similar Sikh and Jain counterparts. (Intermarriage across religious lines is not uncommon, so long as intracaste endogamy is maintained, in the cases of castes with adherents in two or more kindred communities. Intermarriage of Hindus with the more alien Muslim or Christian communities, however, is rare.) The maps are arranged vertically according to the rela- tive status of the groups considered. On all seven maps a com- mon legend is employed to show, by districts, what portion of the total population the group makes up. The groups are rep- resented only when that portion is more than 1 percent.

The map entitled Saiyads shows the distribution of those Muslims who claim descent from the tribe of the Prophet Mu- hammad and who, by virtue of that claim, are accorded par- ticular respect within their religious community. Their total number in India, though not precisely obtainable, may be es- timated at somewhat less than two million in 1931. The con- centration of this group in what is today Pakistan is hardly surprising. Elsewhere they are relatively prominent in and ad- jacent to the former Muslim-ruled princely states of Hyder- abad, Bhopal, Junagadh, and Rampur, and in scattered dis- tricts elsewhere, a number of which contained former Muslim administrative centers (Delhi, Lucknow, Allahabad, Patna, and Ajmer). In only three thinly populated districts, all in the far northwest, did the proportion of Saiyads in the total popu- lation exceed 5%.

From Vedic times, the brahmanic theorists have divided Hindu society into four great classes or estates, called varnas. These four are the Brahmans, priests and learned men; Ksha- triyas, rulers and warriors; Vaishyas, traditionally household- ers, currently mainly a mercantile group; and Sudras, tradi- tionally servants, today mainly cultivators and artisans. The varna system, though of theoretical interest, means relatively little in day-to-day life. One's jati is what really matters, and the assignment of specific jatis to particular varnas is a matter allowing for great variations of interpretation. The depressed (now "scheduled") castes as a whole are not considered to fall in any of the four varnas, whence their occasional designation as "exterior castes" or "out-castes."

Brahmans (map (b)) constitute the highest status group within the Hindu fold. Though enumerated by the census as a single caste, on a par with other jatis, Brahmans in fact com- prise a bewildering array of endogamous subgroups, each of which might be considered a jati in its own right. Collectively these constitute the priestly Brahman varna. The subgroups may for the most part be aggregated into five northern and five southern divisions, the "Pancha Gauda" and the "Pancha Dravida," respectively, lying on either side of the broad dashed lines on our map. (Note the Pancha Gauda outlier on the Malabar Coast.) The principal region of identification of each of these divisions is suggested by the placement of their names, printed in capital letters. Specific Brahman jatis, only the more important of which are shown, are indicated in their region(s) of relative concentration by lower-case letters, followed by key letters in parentheses to show the division to which they belong, where such information could be obtained. Excluded from our map are a number of castes commonly designated