This essay was completed while the author held a Fellowship from the National Endowment for the Humanities, for which he is grateful. He appreciates also earlier help from the American Institute of Indian Studies.

¹ The only important work on the subject is by Kingsley Davis, who provides a fine synthesis of statistical material to be found in Indian census reports. His estimates of population growth in the period before the first national census in 1871, however, are open to question. Davis did not try as a primary task to determine the reasons for population growth, and his occasional comments about the causes of epidemics and high death rates must be taken with great caution, for reasons discussed below in this essay. See Davis, Kingsley, The Population of India and Pakistan (Princeton, 1951), p. 42 and passim.Google Scholar

² Knowles, L. C., The Economic Development of the British Overseas Empire, 3 vols (London, 1924–1936), I, pp. 351–2.Google Scholar

³ Anstey, Vera, The Economic Development of India (London, 1929), p. 41.Google Scholar

⁴ Woodruff, Philip, The Men Who Ruled India, 2 vols (London, 1953), II, p. 109.Google Scholar

⁵ Coupland, Reginald, India: A Restatement (London, 1945), pp. 54–66.Google Scholar

⁶ Davis, , op. cit., p. 27.Google Scholar

⁷ See, in particular, Morris, Morris D., ‘Towards a Reinterpretation of Nineteenthcentury Indian Economic History’, Journal of Economic History, XXIII (1963), pp. 608–18.Google Scholar

⁸ Report on the Census of the North West Provinces …1853 (Calcutta, 1854), pp. 409–13.Google Scholar

⁹ Census of India, 1921, XVI, Pt 2, p. 6. The census of 1871 actually was taken in several years—in northern India it was done in 1872.Google Scholar

¹⁰ Report of the Sanitary Commissioner [hereafter RSC], United Provinces, 1906, p. 2; and 1918, p. 4.Google Scholar

¹¹ This essay was originally sent to Modern Asian Studies before the appearance of Elizabeth Whitcombe's able book on Agrarian Relations in Northern India (Berkeley, 1972).Google Scholar Her analysis of the problems of agrarian development, particularly pages 61–119, helps make more readily intelligible the patterns of population change and disease, the focal points of this essay. One of Dr Whitcombe's themes is that canals led to the salinization of the soil, the decay of well irrigation, and to excessive reliance on commercial crops for export; railways and canals sometimes disrupted drainage and interfered with older patterns of cultivation. In these and other ways there was a high ‘cost of innovation.’ Many of Dr Whitcombe's conclusions agree with earlier criticisms of British policy, but she enlightens them through an admirable technical expertise, and by finding the environmental and sociological roots to notions which previously had been only rather weakly supported. It can readily be seen that environmental disruptions and economic distress produced by modernization would have a negative impact on public health, as will be discussed below, and as Dr Whitcombe briefly notes in her work, page 81.

¹² This relationship, which will be discussed below, was not limited, of course, to the areas of Agra and Oudh. For a documented example of the link in the Punjab, see Christophers, S. R., ‘Malaria in the Punjab’, Scientific Memoris by Officers of the Medical and Sanitary Departments, No. 46 (Calcutta, 1911), pp. 12, 72;Google Scholar for Bengal, see Bentley, C. A., Report on Malaria in Bengal (Calcutta, 1916), pp. 53–62.Google Scholar

¹³ Census of NWP, 1853, pp. 409–13.

¹⁴ RSC, NWP, 1876, pp. 46–7; RSC, UP, 1919, p. 30; District Gazetteers of the United Provinces, III, pp. 20–1, 45; RSC, Bengal, 1868, pp. 163–5, 177–9, 252–3, 343, 370, 424–9.Google Scholar

¹⁵ Census of India, 1911, XV, Pt I, pp. 12–15;Google ScholarDist. Gaz. UP, IV, p. 35.Google ScholarThe condition of northern India at the British accession, and British difficulties in creating a sound agrarian settlement are revealed more fully in State Archives of Uttar Pradesh, Sadr Board of Revenue, XIV, Progs 12 July, 26 July, and 24 December 1805.Google Scholar

¹⁶ Census of India, 1911, XV, Pt I, pp. 12–15.Google Scholar

¹⁷ Report on the Settlement of Banda (Allahabad, 1881), pp. 107–9, 127;Google Scholarsee also State Archives of Uttar Pradesh, Sadr Board of Revenue, XI, Progs 23 October 1832, Minutes by R. M. Bird and W. Fane.Google Scholar

¹⁸ Dist. Gaz. UP, XIX, p. 32.Google Scholar

¹⁹ Ibid., XVIII, pp. 1012–15, 41.

²⁰ A process was employed of counterchecking against two separate lists; if discrepancies were found new returns were obtained. See Census of NWP 1853, pp. 12–12–22 and passim.Google Scholar

²¹ Dist. Gaz. UP, XVIII, pp. 10–15, 41.Google Scholar

²² Sadr Board of Revenue, CCL, Progs 19 April 1844.Google ScholarDist. Gaz. UP, II, pp. 40–1; IV, pp. 35–6; VIII, pp. 27–8; XII, p. 24; XIII, pp. 30–1; XVII, p. 30; XIX, pp. 32–3; XXII, pp. 27–8; XXIII, p. 31; XXVI, pp. 31–2; XXX, pp. 29–30.Google Scholar

²³ Ibid. See also, Dist. Gaz. UP, V, p. 32; XI, p. 29.

²⁴ Sadr Board of Revenue, CCLXIX, Progs 31 December 1844;Google ScholarDist. Gaz. UP, V, p. 32.Google Scholar

²⁵ Gazetteers of the North West Provinces, II, p. 39.Google Scholar

²⁶ The degree of accuracy or error in census returns requires comment, since the main arguments of this essay are developed through the use of Indian vital statistics. Among historians of South Asia the inaccuracies and quagmires of census data, particularly material from the earliest census reports, have been broadcast and accepted virtually as articles of faith. There are many weaknesses in Indian population and other statistics; but broad criticism appears to have failed thus far to distinguish the considerable differences in reliability by locale, or to focus on the means by which compensation may be made for distortions. Even for India as an entirety there perhaps has been some exaggeration of the unreliability of census returns. The census compilers and the current authority, Kingsley Davis, found the returns of 1871 rather inaccurate, off by about 8 per cent. But for the next decennial returns Davis considered errors to be under 3 per cent and to be less than 1 per cent thereafter Further, it is necessary to consider the impact of any statistical inaccuraies. There has been universal agreement that any mistakes in early census returns were those of undercounting, which were corrected gradually in each following decennial census. Any serious undercounting in the first national census, then, would simply mean that population was higher and crowding more severe in the early 1870s than had been thought. The phenomena to which this article points—the subsequent stagnation of population despite a high birth rate, the severe crowding, the fatal consequences to a poor and land-hungry people of epidemic and endemic diseases— would only be made more striking by the upward revision of the census of 1871. Most important, the vital statistics of northern India appear to have been the most accurate in the country. Among the reasons were the techniques developed in, and the public exposure to, the four censuses compiled in Agra and Oudh before 1871, a unique undertaking. While there were some inaccuraies, particularly in the border districts of Oudh, mainly the first national census simply improved on the relative reliability of some of the inquiries of the 1860s. By 1891, the Sanitary Commissioner of the North West provinces believed that the annual registration of births and deaths was ‘Undoubtedly effected with considerable accuracy.’ The Army Sanitary Commission concurred that there was ‘no doubt’ that ‘considerable accuracy has been attained’ —a ‘wonderful accuracy,’ given the problems of compilation. (Causes of death, however, still were not reliably reported.) For the whole decade 1881–91 there was only an ‘inappreciable difference’ of less than two-tenths of a per-cent between the sum of annual registration figures and the tabulations of the census. This conformance of statistics validated registration figures, according to the Army Sanitary Commission, but tacitly corroborated the accuracy of the census as well. Most likely, census errors considerably exceeded an ‘inappreciable’ two-tenths of a per cent. Mistakes in individual districts definitely were greater. There is no reason to believe, however, that returns for northern India were sufficiently unreliable to cast disbelief on the broad trends in population which they recorded. See India Office Library, Sanitary Proceedings of the North West Provinces, 02 1894, Vol. 4506, pp. 3512–6; Davis, op. cit., p. 27.

²⁷ Report of the Board of Revenue on Revenue Administration [hereafter RBRRA], North West Provinces, 1877–78, pp. 1–6; 187812–79, p. 2.Google Scholar

²⁸ Ibid., 187912–1980, p. 5.

²⁹ Census of India, 1921, XVI, Pt 2, p. 26.Google Scholar

³⁰ RBRRA, 189312–1894, p. 3.Google Scholar

³¹ RSC, UP, 1908, p. 2.Google Scholar

³² Ibid., 1896, pp. 2–3; 1897, Orders of Government, p. 2; RBRRA, 1896–1897, pp. 112–8.

³³ While the rise in the recorded death rate may reflect to a limited extent a greater precision in vital statistics, there cannot have been so drastic an improvement over the ‘considerable accuracy’ of returns in the 1880s and 1890s as to explain the extraordinarily high death rates in some years in the next two decades. There was, moreover, a balancing factor minimizing the degree to which an improvement in data could explain an increased death rate. The idea that mortality rates rose because of improved tabulations reflects the erroneous notion that these rates were simply the result of counting heads. Mortality rates were obtained by measuring the annual registration of deaths against the decennial census returns. If there was any significant amount of inaccuracy in vital statistics in the 1870s, for example, undercounting in the census would have compensated to some extent for under-registration of deaths, by providing a lower base of population to which annual death tolls were compared, and, hence, artificially raising mortality rates. Any later improvement in census enumerations would have provided some balance against increased accuracy in annual death tallies, by expanding the population base to which yearly mortality figures were compared, and, consequently, holding down death rates. Any belief that the first national census was highly inaccurate, then, strengthens the view that there was a real, not a statistically induced, rise in death rates in the late nineteenth and early twentieth centuries. For causes underlying years of high mortality see RBRRA, 1907–1908, pp. 1–8, and Orders of Government, p. 1; 1908–1909, pp. 1–5.Google Scholar

³⁴ Ibid., 1904–1905, Orders of Government, pp. 1–2; 1905–1906, Orders of Government, p. 1.

³⁵ India Office Library, Indian Sanitary Proceedings, 1908, Vol. 7887, p. 1757.Google Scholar

³⁶ RSC, Bengal 1868, p. 343; RSC, NWP, 1875, pp. 47–9.Google Scholar

³⁷ RSC, UP, 1907, pp. 1–2; 1908, pp. 2–5; RBRRA, 1906–07, pp. 1–3; 1907–08, pp. 1–7; 1908–09, pp. 1–5.Google Scholar

³⁸ RSC, UP, 1910, p. 4; 1911, pp. 2–4; 1919, pp. 2–7, 13–14; 1920, pp. 5, 8, 13; Administrative Reports of the United Provinces, 1915–16, pp. 1, 27, 44; 1917–18, pp. 28, 45–6.Google Scholar

³⁹ Dist. Gaz. UP, XXXII, p. 35;Google ScholarCensus of India, 1911, XV, Pt 1, p. 16.Google Scholar

⁴⁰ Dist. Gaz. UP, I, pp. 4, 16, 44–5, 85; XXXIV, pp. 60–3; XXXV, pp. 43–5, 52–3.Google Scholar

⁴¹ Census of India, 1911, XV, Pt I, pp. 11–16;Google ScholarRBRRA, 1877–1878, p. 6; 1879–1880, pp. 5–6; 1893–1894, p. 3; 1896–1897, pp. 6–8; 1904–1905, Orders of Government, p. 1.Google Scholar

⁴² RSC, NWP, 1884, p. 29; RSC, UP, 1906, p. 8; 1920, p. 15.Google Scholar

⁴³ Census of India, 1921, XVI, Pt 2, p. 26.Google Scholar

⁴⁴ Admin. Reports, UP, 1917–1918, pp. 45–6.Google Scholar

⁴⁵ See my essay on ‘Malaria and Mortality in Bengal,’ Indian Economic and Social History Review, IX (07 1972), pp. 132–60.Google Scholar

⁴⁶ Clemesha, W. W., ‘Influence of Railway Construction on Malaria’ Records of the Malaria Survey of India, I, No. 2, pp. 163–4;Google ScholarChristophers, op. cit., passim; Bentley op.cit., passim;Google Scholar see also, Bentley's, Malaria and Agriculture in Bengal (Calcutta, 1925),Google Scholar and his essay on Some Economic Aspects of Malaria in Bengal (Calcutta, 1923).Google Scholar

⁴⁷ SirHehir, Patrick, Malaria in India (London, 1927), pp. 12–13, 46;Google ScholarBentley, and Christophers, , ‘The Causes of Blackwater Fever in the Duars’, Scientjfic Memoirs, No. 35 (Simla, 1908), pp. 22–3.Google Scholar

⁴⁸ Bentley, , The Relation Between Obstructed Rivers and Malaria in Lower Bengal (Calcutta, 1923), pp. 1–12;Google ScholarMajumdar, S. J., Rivers of the Bengal Delta (Calcutta, 1942), pp. 76–93;Google Scholar see also India Office Library, Indian Medical Proceedings, 1873–75, Vol. 526, pp. 501–30.Google Scholar

⁴⁹ Hehir, op. cit., pp. 12–15.Google Scholar

⁵⁰ Dist. Gaz. UP, III, p. 29.Google Scholar

⁵¹ Dist. Gaz. UP, III, pp. 43–6.Google Scholar

⁵² It is difficult to obtain an entirely satisfactory criterion of crowding. The absolute density of population serves as a rough guide; it does not take into account the proportions of rural and urban population in individual districts, but figures of absolute density are not totally useless, since less than 10 per cent of the population of northern India was ‘urban’ in 1872, according to British classification, and only about 11 per cent by 1921. A more serious problem is that the categories of ‘urban’ and ‘rural’ are themselves unsatisfactory. At what size does a village or town cease to be ‘rural? When does a ‘town’ no longer contain an agrarian population? In northern India, dairy farming and other ‘rural’ functions were carried on within the largest cities. A more useful guide to crowding is that of the actual agricultural population, those who earned their living from farming or pasturage. These figures provide a slightly, although not startlingly, different ratio of comparative crowding than is shown by absolute density, as can be seen below for some representative districts in 1872.

Even the classification of agricultural population, and the further adjustment for unculturable land, do not provide an entirely trustworthy picture. It does not tell us how many people had been forced out of agrarian occupations by scarcity and competition for land, swelling the lists of the urban poor. Nor does it explain another issue germane to crowding on which complete information for northern India is not readily available: size of individual holdings and contractual relations between landlord and peasant. Clearly, crowding was more severe where extensive estates limited the acreage of the ‘average’ cultivator-owner. Nevertheless, density, particularly the density of agricultural population, provides a useful general guide to which districts were most crowded; whether or not landlordism was rife, where population was heavy and increasing there was pressure on the land in some form. Agra and Shahjahanpur probably were the most crowded districts on the western plain because they had the greatest absolute densities, were among the least ‘urbanized,’ and had the highest densities of agricultural population. Moreover, there were clear signs of danger in Agra, in that there were trends toward deurbanization and toward an increased proportion of agriculturalists, although good land was all under cultivation.

⁵³ RBRRA, 1879–1880, App. I, Ia; 1880–81, Ia; 1917–18, p. 2; 1918–19, p. 3; Review of the Revenue Administration of Oudh, 1870, App. II, p. 6; 1884, App. 7a; Census of NWP, 1853, pp. 409–13.Google Scholar

⁵⁴ Prices and Wages in India, XXVII (Calcutta, 1923), pp. 8–11, 74–5, 96–7, 154–5, 213–56, and passim.Google Scholar