Urban is as Urban Does:

The Tirunelveli Kattabomman District c. 1823[1]

David Ludden

University of Pennsylvania



          Though arguments about India's de-industrialization under colonial rule have a long history, and though arguments about premodern urbanism and urban  trends in India are prominent in historiography, it is difficult to reach firm conclusions about the non-agricultural economy and the urban sector in India before 1850.  The biggest body of data before 1750 concerns taxation in Mughal north India.  It does not permit clear separations of non-agricultural and urban economic activity, and it contains very little information on the composition or distribution of population.  To describe Mughal urbanism, therefore, Shireen Moosvi identifies pargannas containing known urban centers and compares their taxation with neighboring pargannas.  She concludes that arid and semi-arid regions were the most urbanized (particularly Rajasthan).  But after all, "urban centers" here were virtual oases -- fortress and market towns surrounded by arid plains.[2]  Her data and method reveal no urban centers in Bengal, where we would find non-agricultural economic activity spread more evenly throughout the countryside, and where locational advantages were less pronounced for specific centers.  But does this mean that Bengal was less "urbane" in the composition of its economic life than Rajasthan?  Hardly. 


Measuring Urbanism


          Around 1800, data appear that permit a closer look at the composition of economic activity and population in South India.  In Tamil Nadu (and adjacent areas in Andhra and Karnataka) we have very detailed economic data on the Company's jaghir (Chinglepet) in 1765-1795, the Baramahal (Coimbatore and Salem) in 1790-1804, the Ceded Districts (Bellary, Anantapur, Kurnool, and Cuddapah) in 1810-1825, and Tinnevelly District in 1823.  These data (at the Tamil Nadu Archives and the National Archives) contain more detail than we have for any part of South Asia before 1947.  The 1823 Tinnevelly census has more detail on the sources of taxation and on caste groups than any later source, and it allows a reconstruction of the urban and non-agricultural economy that is more fine-grained than anything before.       


          The Dehazada and Census of the Province of Tirunelveli, 1823 covers 645 census villages and records up to 300 items on the economy and population of each village.  I began using this data by constructing a large sample including 185 items that appear in 159 census villages in four contiguous taluks: Shermadevi, Brahmadesam, Tenkasi, and Shankarankoil (now in the Tirunelveli Kattabomman District).  Dehazada items are items for revenue enumeration and the 122 standard dehazada items that appear in all four of these taluks are as good a count of economic assets as we can find for any area in South Asia before 1947.[3] 


          Data in the Tinnevelly dehazada permit us to assess the scale, composition, and spatial organization of the non-agricultural economy, and to describe the character of pre-modern urbanism in southern Tamil Nadu.  Toward this end, I compiled indexes that measure degree to which forty‑three dehazada items indicating non-agricultural assets and activities are concentrated in census sites compared to population and farm acreage. An index value of 1.00 for total looms, for instance, means that the percentage of the sample total of looms in a census village is the same as that village's percentage of total farmland (Index A) or total population  (Index B) in the sample.  For each census village, the simple average of index item values yields a non‑agricultural activity index.  High index values, well  above the sample average, indicate that a census village contains a  high proportion of artisan and/or commercial activity; for simplicity, we can say that this means it is more "urban" in economic terms.  Low index values for a village render it more rural: this means in practice that it contained few taxable assets other than land counted in the 1823 Census. 


          It is traditional to consider larger centers of population to be more urban, but for the premodern period this is not viable, because the area within which these populations are contained is not specified, so their economic activity could thus be primarily agricultural and spread over a large territory.  To designate a site an urban center, we would want it to have relatively high population density, high proportions of non-agricultural economic activity, and high total population.  In the 1823 Tinnevelly census (as in other similar sources) the size of a census site (called a "village" but often composed of many settlements) was determined by the state revenue politics.  Larger places with larger populations were often simply larger areas of revenue authority, under the command of higher ranking officials. 


          Smaller sites, indeed, tend overall to be slightly more urban as measured by indexes of economic activity.  Huge villages average lower than the sample average for indexes A (0.94) and B (0.76); small villages average slightly higher (A=1.05; B=1.38).  Smaller sites often also have much higher population densities.  After all, when a small place became sufficiently wealthy, it could become a separate revenue unit, and such units proliferated over time in fertile agrarian regions like Bengal and riverine Tamil Nadu.  Ambasamudram, for instance, had sixteen subordinate ur in 1477, but only had three subordinate villages in 1823 and a population of only  3,952.  In the interim, it had become surrounded by more and more independent sites of local political authority.


          Similarly, many tiny and relatively isolated villages in foothills and plains may have been break‑away settlements in which  farmers claimed highly productive land and upstream irrigation waters.   Thus though in the dry plains tiny villages were often very poorly  endowed with taxed assets other than land, some of the smallest census sites had high concentrations of looms, mat frames, gunny frames, toddy shops, arrack shops, and other commercial assets, in addition to artisan and merchant castes.


          Table 1 combines population size with Index A and gives a reasonable sense of urbanity in our sample.  A quarter of the sample population lived in places with relatively high proportions of non‑agricultural assets: urban centers, towns, and suburbs.  If we omit all these more urban areas from the sample, we still see a gradient from higher to lower village concentrations of non‑agricultural assets among the remaining 115 census villages.  The resulting  picture is thus not one of sharp disjuncture between urban and rural  locations, but rather of a gradual slope from higher to lower index values, from more urban to more rural census villages, with some extreme cases at both ends.



Table 1: Economic Differentiation in Four Tirunelveli Taluks, 1823[4] 

(Percentages are of Sample Total)


                     Census                  Economic  Population Per

Locality Types       Villages   Population   Index     100 Farm Acres


                       N   %    Average  %   Average   Total    Wet


I. Urban Areas        36  22     2,253  34     5.87     116     229              

   Large Centers       6   4     4,813  17     4.28     101     280

   Major Towns        15   9     1,492  13     4.52     103     193

   Urban Suburbs      15   9       454   4     8.81     145     214

II.  Small Towns       8   5       427   2     2.32      63     158

III. Rural Areas     115  72     1,492  62      .73      24     121

        Large         16  10     2,633  25      .78      23     152

        Midsize       22  14     1,446  19      .69      20     123

        Small         77  48       398  18      .72      30      57




          Irrigation highlights the long history of agricultural development that underlay urbanization.   On average, urban sites altogether have 3.54 times more of the sample area's irrigated land than of its total farmland.  Small towns have a similar ratio (3.18), but rural sites have ratios down from 2.42 to 0.04.  The low proportion (3%) of uncultivated wet land (tarisu) in urban centers indicates the intense utilization of wet land in urban areas.  Wells also concentrated in urban settlements.  So it is clear that urban sites of manufacturing and trade grew in proportion to local water, paddy, and food.


          Statistically, however, irrigated arable acreage is not a good predictor of economic urbanism.  Large urban centers have lower proportions of wet land in their borders than big towns.  Many trade and manufacturing assets lie in localities with little irrigation.  Many small villages contain large irrigated tracts but few trade and manufacturing assets. Irrigated places in 1823 were rural and urban; they could be either small agricultural villages or major trade and manufacturing centers.  Spreading water on the land did not suffice to make it a center of manufacturing and trade.


          Economic urbanization depended on extensive economic relations with many localities in networks of trade and power.  Five of the six large urban centers in our sample lay in the medieval core political territory of Tirunelveli, in tracts watered by the Tambraparni and Chittar rivers, at the heart of the regional transportation system.  One, however, Shankarankoil, lay in a tract askew of central corridors with poor soil and wretched irrigation: its population -- like that of most major urban centers, I think -- could not have fed itself with local cultivation alone.  Altogether, large urban centers have 17% of the sample population but only 9% of cultivated wet land.  They have only 18 cultivated wet acres for every one hundred people, whereas 40 is the average for rural areas, 55 for small rural areas.  However intense their local rice farming might have been, urban centers must have fed themselves by buying food from their hinterlands.  Agricultural villages would have sent grain to urban centers.  This food, like the investments that built local irrigation, came into urban centers because of their position in networks of accumulation and exchange.


          The political partitioning of south India was an important determinant of any locality's economic position, because the segments of states constituted central place hierarchies.  Four of six urban centers were taluk headquarters, with long histories of regional political significance.  Major urban centers and big towns together had 30% of the sample population, only 21% of its cultivated wet land, but 80% of government headquarters (katcheries), 44% of town custom stations, 63% of road custom stations, and 53% of state granaries.  Government grew in market centers and markets grew in states.  So urban centers developed as concentrations of consumption, public activity, trade, and redistribution.  Urban centers and big towns, with 30% of the sample population, had 43% of principal temples, 48% of mandapams, 76% of chattrams, and 56% of the major annual festivals; they had 58% of all two‑story houses, 51% of terraced houses, and 53% of tile roof houses in the sample region.  


          The urban growth of trade and manufacturing affected many aspects of local life.  Urban sites held about one third of the sample population.  They had much higher population densities and quite different caste compositions than small towns and rural census villages.  Table 2 indicates, however, that only one set of manufacturing assets concentrated very dramatically in urban centers, gunny frames.  This reflects the fact that major urban centers contained a high proportion of high‑cost production for long‑distance trade.  They held 70% of all silk looms and 78% of all Kaikkolar looms.  94% of all silk looms were in major urban centers and big towns together. 


          By contrast, major urban centers and big towns, with 30% of the sample population, only had 64% of all taxed looms, 46% of the silver shops and 61% of the brazier shops.  Urban sites had significant concentrations of trade and manufacturing assets, but the production system -- and textile production in particular -- involved the combination of activities and assets that were dispersed throughout networks in which urban centers were nodes of accumulation and exchange.  It is reasonable to assume that assets accounted for in the 1823 census were commercially active and therefore taxed -- they constitute a set of assets registered at the intersection of the commercial economy and the state taxation system.  Some of these assets (and their associated production activities) were distinctly rural: only 38% of all carts and Muslim looms in the sample, and only 11% of Shanar looms and 20% of Pallar looms were registered in urban sites.  


          The activities of textile production -- cotton growing, cleaning, spinning, weaving, dyeing, packing, transport, and finance -- were spread throughout the central place hierarchy.   Weaving was concentrated in the 21 of 159 census villages which had a higher share of sample's total number of taxed looms than of its population.  These twenty-one weaving centers include all six major urban centers, three big towns and two small suburbs, but also one small town and five small villages.  Even small rural localities sustained commercial weaving, in part from local demand and in part from trade along the urban‑rural continuum that connected villages to long‑distance trade.  Concentrations of Muslim and Kaikkolar weavers of high-quality cloth appear in five small villages.[32]  Moreover, tiny hamlets -- which altogether account for a little more of the sample population (18%) than major urban centers (17%) -- contain many more of some types of taxed commercial assets: oil mills (28 to 12), betel gardens (45 to 4), betel bazaars (26 to 5), carpenter's yards (67 to 32), potters works (66 to 25), periodic festivals (105 to 68) and bazaars (141 to 118), not to mention milk cows and buffaloes, plows, bullocks, and commercial trees (coconut, supari, jack, mango, tamarind, and palmyra).    


          Because commercial and manufacturing activities were spread across the urban‑rural continuum, urban and rural localities were not distinguished by the presence of taxable commercial and manufacturing assets, but rather by the scale of accumulation and by the preponderance of farming assets in their boundaries.  No part of the region -- even the most rural areas -- lay outside the commercial economy. Even the smallest rural villages were differentiated from one another in their economic character by their position in regional networks of economic specialization.


          Compare Pattamadai, with 552 residents in the Tambraparni valley, two miles from Seranmahadevi (taluk headquarters), and Karisattan, with 450 residents in the dry land seven miles from Shankarankoil.  At first sight, Pattamadai would seem to be the more commercial­ly active: it has 98 water works, 47 religious buildings, a major annual festival, 14 periodic festivals, a customs station, 17 terraced houses, and 41 merchant houses, all sustained by trade along on the active road between Tirunelveli town and Shermadevi, and by the 21% of its total arable acreage irrigated from the Tambraparni.  Yet this village and its 81 Mirasidars seem to have lived by working the transit trade, farms, and trees alone; for virtually no manufacturing assets appear in its census record, the exception being a sugar mill; and in addition to irrigated land, it boasted 22,732 taxable trees, mostly palmyras. 



Table 2: Ranked Average Indexes of Concentration for Village Economic

         Index Items, by Locality Groups

                                         Average Index Value

             Rank   Item              Urban  Small  Rural

                                      Areas  Towns  Areas

               1   Gunny Frames        24.74   .00   .04 

               2   Toddy Shops         14.72   .34   .18 

               3   Arrack Shops        14.06   .91   .22 

               4   Brazier Shop         9.67  3.68   .29 

               5   Carts                8.60  3.43   .77 

               6   Asses                8.32  1.81   .72 

               7   Annual Festivals     7.60  3.34   .45 

               8   Terraced Houses      7.52  2.41   .91 

               9   Muslim Looms         7.39   .00   .49 

              10   Kaikkolar Looms      7.16  3.67   .26 

              11   Total Looms          6.83  2.32   .40 

              12   Sugar Mills          6.60  2.07   .62 

              13   Beetle Bazaars       6.51  5.65   .80 

              14   Goats                6.35   .41   .80 

              15   Silversmith Shop     6.30  2.30   .62 

              16   Bleaching Places     6.23   .55   .86 

              17   Pallar Looms         6.19  2.19   .63 

              18   Jack Trees           6.06  4.53   .62 

              19   Shanar Looms         5.76  3.62   .77 

              20   Public Buildings     5.57  4.17   .79

              21   Silk Looms           5.54   .00   .14

              22   Potter Works         5.30  4.21   .80 

              23   Horses               4.92   .98   .80

              24   Coconut Trees        4.89  2.78   .86 

              25   Lime Kilns           4.79   .00   .88 

              26   Razor Cases          4.76  3.10    76 

              27   Total Population     4.65  2.32  1.01 

              28   Bazaars              4.63  1.54  2.16 

              29   Houses               4.58  2.45   .97 

              30   Periodic Festivals   4.51  3.86  2.13 

              31   Religious Buildings  4.12  2.86   .99 




          By contrast, only 8% of arable in Karisattan was irrigated. It had only 556 taxable trees and was clearly not a market center, though it did have three periodic festivals, one bazaar, one betel bazaar, one arrack shop, and two merchant houses.  But  Karisattan had one iron forge, thirteen Kaikkolar looms, one carpenter's yard, two potters works, and two bleaching places for cloth; and it had one milk cow for every two inhabitants (compared one for every eight in Pattamadai) and 600 sheep and goats (compared to none in Pattamadai).  The people in Karisattan clearly produced a lot to sell.


          Who the people were in these two villages helps to explain their economic differences.  In Pattamadai, a Pandya Vellala stronghold, all but two mirasidars appear as non‑residents.  Pandya Vellalas were 21% of the population; Pallars and Pariahs together were 50%; both being about twice the average percentage for sample villages, which is typical for a Tambraparni village.  Pattamadai's abundant palmyras were surely tended by its large Shanar population (at 18%, about twice the census village average).  No Pandya Vellalas lived in Karisattan, but instead, a small group of Chola and Tulu Vellalas.  The untouchable castes amount to only 17% of its population, well under the sample census village average (24%).  Most resident Karisattan jatis had histories of migration ‑‑ Kaikkolars, Chetties, Reddiyars, Totiyars, and Maravas ‑‑ and along with the immigrant Vellalas, they constituted 60% of the village.  Karisattan's bleaching places were obviously worked by its large group of resident washermen (at 4%, four times the sample average).  Its population of Kaikkolar weavers (16%) is eight times the sample census village average.


Creating Urbanism


          In Pattamadai and Karisattan, we see two localities with different social and economic profiles, one at the ancient core of the region and the other on its periphery, built as production environments by distinctive caste clusters.  Pattamadai and Karisattan typify the pattern of production localities in Tirunelveli recorded in the 1823 census: local caste clusters reflect distinctive social formations of economic organization.  These arose historically because family decisions to move and settle in particular places distributed labour and capital among places within networks of human mobility, trade and power.  The historical combination of specialized caste migration and settlement strategies, the spatial clustering of castes endowed with complementary assets, the availability of ritual and social techniques for organizing power among caste groups, and ready communica­tion along the urban-rural continuum made the early-modern economy responsive to its world market environment. 


          Family strategies and the formation of caste clusters are clearly visible in census data, which describe jatis of many different kinds.  Important economic differences among castes appear in their spatial distributions, which can be taken (very roughly) to represent aggregate family preferences and the outcome of residential settlement strategies.  Some castes formed large populations, embraced many occupational distinctions, and spread across the entire region, all along the urban‑rural continuum. 


          Other populous jatis concentrated only in one region or at one end of the urban-rural spectrum.  For instance, Vellalas and Brahmans -- who together dominated the political economy of the Tambraparni valley, where they had a much bigger demographic share than elsewhere -- demonstrate different residential preferences.  Brahmans lived much more often in urban sites; and among Brahmans, Smarta Telugus (who migrated into the region in post‑medieval centuries) had a much stronger tendency to live in more urban sites than Smarta Tamils or Sri Vaishnavas.  On the other hand, (native) Pandya Vellalas spread over the entire region in exactly the same pattern as the population generally; as did the immigrant Tondaimandala Mudalis; whereas immigrant Karaikkattu Vellalas were split between very rural and strong urban preferences, and Chola and Tulu Vellalas much preferred hamlets like Karisattan. 


          Other large jatis that migrated into the region were split on similar lines.   Among Maravas, the Kondaiyankottais and Kotalis favored rural and urban settings, respectively, which in effect meant that relatively few members of the much larger Kondaiyankottai group lived in Zamindar (formerly Poligar) fort towns: Uttumalai, Surandai, and Chokkampatti.  Shanars had very strong rural preferences.  Pallars (the largest landless labour caste) settled in the same spatial pattern as did the population as a whole, but eschewed urban centers, which suggests a distinct caste profile for urban labour.  The smaller landless labouring jati, Pariahs, show a stronger tendency to live in urban sites, which often have high Pariah percentages -- for instance,  Seranmahadevi (16%) and Ambasamudram (8%) -- suggesting that Pariahs played more active roles than Pallars in manufacturing.


          Small, specialized caste groups, which comprise about half the total sample population, show the most marked urban and rural preferences.  This is critical for understanding the role of caste in the early-modern economy.  For it seems that there were social mechanisms for fissioning and inventing castes that produced countless small, specialized groups.  Most very small castes in the census do not appear on all four taluks census tables that comprise our sample, and they can thus be presumed to have been significant only in small areas or localities, at least in official perceptions.


          The number of castes in major urban centers is much smaller, proportionately, than the total number of people counted in urban centers; and a significantly higher percentage of all castes than of population appear to have lived in small rural villages.  This indicates that a great many very small jati groups lived only in small rural settlements.  A few tiny castes that demonstrate strong rural preferences do appear on our list: Vyravers, Chola and Tulu Vellalas, and Chucklers.  But most small jatis in my listing are in urban settlements: non‑Brahman temple specialists ("the religious establishment"), Karaikkattu Vellalas, Chetties, Patnulkaran and Pattashalaiyan silk weavers, Kaikkolars, Chalupans, Cudashelkarans, Panans, Dyers, Pattulu Muslims, Pariah Christians, and Cunniyans.  Small castes were much more likely to be enumerated if they were seen by state officials as significant for urban society, the political core of the state and economy.  Large urban centers were composed disproportionately of small, specialized caste groups involved in state politics, trade and manufacturing.


          The census figures give a glimpse of the complex and moving social world of residential decisions that defined locality units of production.  In rural areas, the major dominant castes (Vellalas, Maravas, and Brahmans) and the major landless labourer caste (Pallars) usually comprise the majority of the population.  But concentrations of artisans (weavers, potters, Panchalars), merchants (Chetties, Villay Jedians), and subordinate farming castes (Shanar, Pariah, and omitted jatis) are common.  If we analyze the caste composition of census villages, starting at the rural end of the economic continuum, and moving toward major urban centers, we see more and more frequent local concentrations of jatis that we would expect to relocate readily into locations of market opportunity; and we also see higher numbers of these castes in census village populations.  But this is a gradual progression, like the rising slope of manufacturing assets up the urban-rural continuum.  For opportunities in trade and manufacture were also concentrated in small rural settlements, like Karisattan, with its cluster of Kaikkolars and washermen, working in the textile industry. 


          A dry urban suburb, Vagaikulam (population 837), just south of Brahmadesam, had the highest village percentage of Panchalar artisans (17%) in our sample, a high concentra­tion of brass works, and many Shanar looms, lime kilns, and betel bazaars.  A small rural village, Pottalpudur (population 504), six miles northwest of Brahmadesam, had the next highest percent Panchalar population (12%), with concentrations of mat and gunny frames and oil mills.  The highest village proportion of Chalupens also appears in two urban suburbs: in one, near Shermadevi (Vellankoil, population 294) Chalupens were 16% of the population; in the other, six miles east (Kilachevel, population 656) they were 12%; and both had huge numbers of gunny frames.        


          Decisions by specialized caste groups to move into specialized settlements activated the manufacturing economy, most prominently, the weaving industry.  The demographic outcomes of weaver mobility can be startling.  Kaikkolars were a majority of the population (56%) in Turtikulam Devasthanam (population 312), an urban suburb of Uttumalai town, and prominent in the major urban centers of Ambasamu­dram (33%), Shankarankoil (20%), and Tenkasi (12%).  Tenkasi's many Muslim looms and 29% Muslim population suggest that a third of its population engaged in cloth trades.  A quarter of Viravanallur's 4,833 urban population were silk weavers (Patnulkarans and Pattashalaiyans), who worked a local concentration of silk looms.       


          The weaving industry also involved regular movement among settlements.  Of course, cotton, silk, dyes, looms, bleaches, thread, gunny sacks, and cloth circulated among sites, each produced and used in specialized places.  Weavers and loom owners also moved for work across village lines.  In Tenkasi and some other Muslim weaving centers, like Ravanasamudram (near Brahmadesam, population 489) and Kallidaikurichi (population 4,651), the resident Muslim population would have been sufficient to work the looms counted in those places.  But where Muslim looms were most concentrated in proportion to population, in the urban suburbs of Vellankoil and Minakshipuram (population 996, near Tirunelveli town) there were no Muslims resident in the census village at all.


Caste Territory


          Caste settlement strategies produced patterns of jati residential association that defined the social composition of production localities.  In using 1823 census data to deduce the logic behind the formation of jati clusters, I assume that if jati populations were randomly distributed, there would be no significant correlations between any two castes across the villages in the sample; and if they were evenly distributed, all village caste populations would correlate with one another and also with total population.  Then I compute correlations with Pearson's "r" measures of linear relationship, so that a maximum correlation of 1.0 between two jatis means that an increase in the population of one caste would be matched exactly by proportionate increase in the other for 117 villages in my sample with populations greater than zero. 


          Looking at the correlation data, we see that all jatis correlate significantly with total population, so that normally all jati populations in a village increase in some proportion to village population.  Significant correlations among jatis thus do not necessarily suggest caste residential associations that need explaining: they may simply derive from normal associations between all jati populations and total population.  To find indications of caste decision‑making about jati residence that created clusters of jatis and patterns of association, we look for non‑significant or negative correlations and also for high positive correlations. 


          Looking at Brahmans for instance, shows that they are not all alike, not even Smartas.  Only Smarta Tamils correlate with population density, village economic types, and wet cultivation.  The small population of Mudivar Tadvatis were very prone to settle in drier areas, as reflected by their correlation with total farm acres, a figure dominated by dry acreage; but they also settled in dry urban areas.  Brahmans correlate pretty well with one another, but not in every case: Smarta Tamils and Smarta Telugus emphatically do not.  This would be the result of post‑medieval partitioning of Smarta territories along the Tambraparni River, which seems to parallels the partitioning by Telugu Nayakas and Maravas of the dry periphery to the north.   The medieval alliance of Vellalas and Brahmans that dominated the political economy of the Tambraparni valley is clearly visible in 1823 residential patterns, in which Smarta Tamils cluster with Pandya Vellalas and Pallars.  But migrations in the post‑medie­val period brought Smarta Telugus into the region in political alliance and residential association with immigrant Vellalas --Tondaimandalam Mudaliars -- whose leader, The Medai Delavoy, encouraged allies among Vellalas and Telugu Brahmans to migrate south.  In this migration, it seems that Smarta Telugus were granted land in villages outside Smarta Tamil control, where they remained until 1823: for none of the Vellalas are as segregated from one another as are these two Brahman groups.  In addition, Smarta Telugus tend slightly more than Smarta Tamils to live in urban centers and are much more positively associated with weavers and merchants, indicating that they played a role in the rise of commercial centers, all across the urban‑rural continuum, during post‑medieval centuries.


          Non‑correlations between Brahmans and Maravas, Shanars, Christians, and Muslims are typical of these latter jatis, which always tend to be relatively isolated other jatis in their own areas of residential concentration.  Marava groups do not even correlate with one another.  The only castes that correlate with Christians are the castes that produced most of the Christian converts: Shanars, Paravas, and Pariahs.  Maravas, Shanars, Christians, and Muslims preferred very strongly to live in their own settlements, outside any strong local association with virtually any other caste. 


          On this evidence, several conclusions can be drawn that pertain to the social construction of economic locations.  A medieval Tambraparni cluster of castes (Tamil Brahmans, Pandya Vellalas, and Pallars) persisted until in irrigated farming areas until 1823, though like other castes, Brahmans and Pandya Vellalas did settle in proportion to total population everywhere.  Post‑medieval immigrant groups still remain clustered together in 1823, in part because they jointly settled open land and in part because dominant castes partitioned the landscape politically by building up their own jati clusters, in regions under their control, as in Pudukottai.  As a result, Smarta Telugus are much more closely associated with Tondaimandala Mudalis than with Pandya Vellalas, and not at all with Smarta Tamils, who cluster more with Pandya Vellalas than with other Vellalas. 


          Commercially and occupationally specialized jatis are often immigrants into the region and into localities; in a sense, their mobility makes them perpetual migrants, over the long term.  All these migrant castes are more closely associated with one another than with dominant castes or the medieval Tambraparni cluster of castes.  Although exceptions to this rule occur when a small group like Chetties concentrated in an area dominated by Pandya Vellalas and Smarta Tamils.  Immigrant clusters populated whole regions of migratory settlement, for clusters of castes opened whole regions together, as the Telugus did in northern Tirunelveli.  Also, since dominant political powers ‑‑ the Medai Delavoy, Madurai Nayakas, Nawab of Arcot, and English East India Company ‑‑ migrated into the region with natural allies and powerful patronage, we expect to see residential associations among their allied groups, and we do.  Urbanization brought immigrant and commercially mobile jatis together into clusters.  Telugu castes, Muslims, and Parava Christians thus have high correlation with artisan and merchant groups, which have strong correlations with one another.  Urban and rural jati populations differed significantly in the proportion of Brahmans, non‑Brahman temple specialists, merchant and artisan groups, and even untouchables, for Pariahs are more prominent than Pallars in urban areas. 


          Numbers in the 1823 census that represent local clusters of caste populations are remnants of colonial accounting and traces of the social history that formed local units of production.  A small portion of manufacturing output from these localities entered world markets, but the internal dynamics of caste society that organized localities for production gave the entire eastern peninsula the powers to respond to overseas demand that made south India a world manufacturing center in the early-modern period.  Locally, caste society organized labour and capital for production.  The mobility of caste groups and political struggles over centuries had carved up the landscape for coherent clusters of castes organized around the exploitation of opportunities in particular locations.  The hierarchy of caste dominance in each location was part of its organization as a segment of the state and allowed political assets to be translated into productive purposes. 


          At the pinnacle of local society, patriarchs within dominant caste groups, like the Medai Delavoy Mudaliars (though most men like him worked on a much smaller scale) served a hinge role connecting localities to wider networks of economic and political power.  In this context, specialized caste families moved over the landscape in response to opportunities for investment and employment during the rise of demand for regional manufactures after 1700.  A politically powerful local financier would invest to attract these specialists -- weavers being prominent among them -- who were then integrated into local society by established patterns of patronage and ritual participation.  Localities that thrived as units of manufacturing became stable residential settings for requisite combinations of workers, financiers, artisans, traders, and service castes, whose output combined with that of others in regional networks of trade.  The 1823 census records a moment in that history.  Soon after, economic conditions that sustained most of the local textile manufacturing units of south India disappeared, and the economic adaptability of caste society faced the domination of industrial capitalism.


Urban Tirunelveli


          Table 4 covers the whole district as depicted in the 1823 Census.  It shows the percentage of the population in urban centers with population over 5,000.  


          I have added together Tirunelveli, Palayamkottai, Melapalayam, and other smaller sites that comprise the Tirunelveli urban complex, which forms a massive urban center at the heart of the region. This central urban complex comprise six percent of the total census population and was four times bigger than the next two urban complexes, surrounding Sivakasi and Srivilliputtur.   The remainder of the list includes Srivaikuntam, Alvar Tirunagari, Tenkasi, Varttiraya Iruppu, Shankaranayanarkoil, and three Zamindari headquarter towns, Shattur, Sivagiri, and Shapatur.  This list would clearly not comprise the list of major urban centers in the later nineteenth century.  The reason for this difference is primarily that boundaries around political territories changed.  This list represents a set of centers of population and political influence that was of particular importance at the end of the eighteenth century.



          We can learn a lot about what was important in the early modern economy from the Census portrait of these centers. 


          1. As a group, they were not "urban" in a modern sense at all, and in fact most of them contained such large populations primarily because they contained large areas of farmland within their borders.  They included within them centers of dense population, but as territories only the Tirunelveli Urban Complex, Alvar Tirunagari, and Tenkasi look from their population densities like "cities."


          2.  Though irrigated land was exceptionally important for the creation of wealth and for state revenue in this period, it is not so important in the definition of major centers of population and economic power.  Most of these centers are indeed in the drier parts of the region.  This is consistent with other survey data from the same period and earlier, which shows that drier regions of South India -- for instance in Rayalaseema, northern Mysore, and in Coimbatore -- were becoming important centers of economic activity, not only but significantly because of the expansion of cotton, tobacco, and other lucrative forms of cultivation, in addition to the expansion of weaving and other manufacturing activities.


          3.  These urban centers did have distinctive agricultural characteristics, however.  The most densely populated centers also had the highest proportion of irrigated land under cultivation.  These were places where agricultural land was used intensively.   These were also places with large endowments of agricultural assets other than land, most importantly, animals and trees. 


          4.  What most distinguished these centers economically was that they were centers for the accumulation of revenues of various kinds from their hinterlands and thus places where assets other than those directly used in agricultural could accumulate.  The Tirunelveli Urban complex benefited more from the flow of shares of agrarian product than other centers, but they all show significant concentration of non-agricultural assets of various kinds -- including religious buildings, bazaars, shops, looms, gunny and mat frames, iron forges and blacksmiths shops, and fancy houses -- that is, tile roof, two-storey, and terraced houses, which are counted in the Census.


          5. The non-agricultural economic activity that receives the most attention in the history of this period is weaving, and cloth manufacturing was important in all these centers.  It is interesting to note, however, that in proportion to other non-agricultural assets, looms are not as numerically preponderant as we might expect, and they are most numerous proportionately oddly enough in Shankaranayanarkoil.  In the more intensely urbanized towns of Tirunleveli, Sivakasi, and Tenkasi, building fancy houses seems to have been a more prominent kind of work than weaving, though there were many looms in these places; and in other towns, building and other activities related to religious buildings of various kinds stands out.  That the wealth which financed the construction of these houses and religious buildings was primarily agricultural, however, is well indicated by the proportion of total listed agricultural to non-agricultural assets in these centers.  Only the Tirunelveli Urban Complex, Sivakasi, and Tenkasi stand out as places in which non-agricultural assets are of disproportional importance in the economic profile. The cloth industry was most prominent in the export trade, but in the profile of employment, it was much less prominent in this region.


          6. Even so, because the weaving industry was expansive and dynamic in this period, it is interesting to look at the composition of looms in the Census to get an idea of how the industry worked.  The most outstanding feature in the count of looms in the Census is that each center had a concentration of a different kind of looms.


          Even in The Tirunelveli Urban Complex, weavers of different types were concentrated in different places with all the Silk Looms in Tirunelveli and Kaikkolar Looms in Palayamkottai.


          7.  This spread, concentration, diversity, and spatial segregation of weavers  that is visible in the Census raise the final points that I want to make about this Early-Modern economic landscape. 


          a. The composition of the population in each center was its most distinctive characteristic.  Every one of these centers is distinctive as a social space.  To some extent, this reflects the fact that migration and settlement patterns had partitioned the Tirunelveli region among different combinations of social groups -- which were formed within distinctive political territories -- during the early modern period; and these centers developed at the center of different population subregions. 


          So for example, Brahmans formed as much as 23% of the population in Alvar Tirunagari but only 1% in Sivakasi and less than 1% in all the Zamindar headquarter towns.  Even within the Tirunelveli Urban complex, virtually all the Brahmans were settled in Tirunelveli, whereas the weavers were concentrated in Palayamkottai and all the Muslims in Melapalayam.  Maravas are concentrated in centers in the Marava tract.  Such distinctions could be recounted in many more details, because the number of social groups counted in the census is so large.  The number of named social groups goes down from 75 in Srivilliputtur to 14 in Melapalayam with the average population per group ascending from about 100 to about 300.  So that in these centers, on average, groups distinguished by name had about 200 members, and the specific set of groups that comprise the population of each subregion surrounding each major center consisted of about 50 groups in different combinations. 


          As we can see immediately when we compare this enumeration of the population to later Censuses --more modern ones -- the social composition of the population is much more complex and highly differentiated than established views of "caste society" suggest.  Indeed, when the first really modern account of Tinnevelly District was published in the 1870s, the author -- the Collector, Mr.Pate -- took great pains to describe what he called "the typical caste composition of a Tinnevelly village.  Clearly, in 1823, there was no such typical village and the social distinctions that were made among groups were much more finally attuned to local conditions.


          b.  In addition to reflecting the diversity of the populations of their subregions, urban centers were distinctive as urban social spaces.  As a group they had a much lower percentage of Maravas and much higher proportions of Weavers and Panchalars.  The urban spaces in the irrigated tracts had much higher percentages of Brahmans in their populations than surrounding rural areas; and the same can be said of Paraiyars, while the inverse is true for Shanars, who tended to live in less urbanized settings.  But the group that set these urban spaces apart most was the Muslims.  Here as in most of India, the Muslim population was concentrated in urban centers and their presence came in the early modern period to become a reliable social and cultural marker of urbanism.  Muslims, like other groups, lived disproportionately in their own settlements, which are sometimes visible in the 1823 Census, which shows that Melapalayam was almost entirely occupied by its 5,000 Muslims, who lived almost nowhere else in the Tirunelveli Urban Complex.




(1)  Migration brought a great variety of different social groups into specific places where they concentrated as groups.  


          a. The social landscape was thus very fluid and people were moving around from place to place with considerable frequency in search of economic opportunities. 


          b. But they strongly tended to settle in clusters of social groups that developed effective control over subregions and localities within them. 


          c. This social concentration of people with distinctive skills and employment profiles produced an economic landscape that was highly differentiated, with specific regions and localities concentrating on a specific range of productive activities in landscapes that supported them and became marked by them. 


(2) This produced an urban-rural continuum in which only a few rather large urban centers stood out from a number of smaller ones and only a few urban centers that could be said to be centers of specifically "urban" economic activity, because most urban centers were simply centers of accumulation and power in their hinterlands, with concentrations of the same forms of labor that could be found in the smaller settlements.  This is the spatial pattern typical of the pre-railway era, but the consolidation of an urban hierarchy in which major centers stand out permanently became typical only in the 17-18th centuries.


          a.  Urban economies were characterized by concentrations of weaving, but the construction industry seems to have been much more important, as it could be sustained by a wider range of income streams.  Religious institutions, trade in high-value consumer goods, and the government service sector (including finance) were also prominent non-agricultural economic activities


          b.  This was predominantly an agricultural economy that supported a wide range of small investments in non-agricultural production (as in weaving, mining, and other crafts) rather than an industrializing economy with increasing concentrations of proletarian labour.  The great dispersion of non-agricultural economic activities enabled the majority of the population to be involved in some way in what would become characteristically of  "urban" occupations. In consumer goods trade, processing, and finance, the textile sector, construction, religious institutions, and government, many wage-paying activities were available in the everyday agrarian work environment.  (These would later be concentrated in industrial centers, lowering non-agricultural income options in farming communities, and also income earning options in the domestic economy.)


(3)  It was an economy that was widely and deeply commercialized and filled with people who were willing and able to take advantage of economic opportunities within the social groups that organized their family and their work lives simultaneously.


(4)  Social space was intricately segregated and partitioned among social groups, with their own places and forms of worship, leaders, and sources of livelihood.



[1] MIDS Conference paper, WINWORD version; MS-WORKS embedded tables omitted; they are available on request from author) This is a draft for the conference on Tamil Nadu at the Madras Institute for Development Studies, APril 2, 1996.  Please do not cite or quote without author's prior permision.  Work for this paper was supported by a grant from the American Institute for Indian Studies.

[2] The Economy of Mughal India, c.1595: A Statistical Study, Delhi, Oxford Univesity Press, 1987, 299-321, amd Map10.

[3] For an introduction to these data see David Ludden, "Agrarian Commercialism in Eighteenth Century South India: Evidence from the 1823 Tirunelveli Census," Indian Economic and Social History Review, 25, 4, 1988, 493‑519. Rpt. Merchants, Markets and the State in Early Modern India. Editor: Sanjay Subrahmanyam, Oxford University Press, Delhi, 1990, 215‑241.

[4]Note: Taluks are Tenkasi, Shankarankoil (without Zamindaris), Brahmadesam, and Shermadevi.  Index values measure concentrations of 44 dehazada items (Table 2) in each village compared to farm acres.  Farm acres include all revenue classifications, including waste. "Wet" means "classified as irrigated."  Source: Tamil Nadu Archives, Revenue Department Sundries No.39, "Census and Dehazada of the Province of Tirunelvelie."

 [See Ludden 1988, for details.]