12 | Making of the 1891 Census

By Gaurav Kalyani

Published on: 6 June 2026

Introduction

On the night of 26 February 1891, the second general census of India was conducted across the vast expanse of the subcontinent. This enumeration, which encompassed approximately 287 million persons, roughly one-fifth of the world's estimated population at the time, represented what its Commissioner, Jervoise Athelstane Baines, rightly described as "the heaviest enterprise of the kind that has ever been undertaken."

The 1891 census was built upon foundations laid a decade earlier by the 1881 census. Its success can be attributed to a longer and more meticulous period of preparation than had been afforded to its predecessor. Acting on advice from the 1881 Commissioner, Sir William Plowden, the Government of India appointed Jervoise Athelstane Baines as Census Commissioner nearly 15 months in advance of the enumeration date.

The Magnitude and Scope of the Census

The scale of the 1891 census challenges comprehension even by contemporary standards. The final returns enumerated 287,223,431 persons. 

The territorial coverage of the 1891 census was expanded from its predecessor. The new census brought within its scope Kashmir, Sikkim, Upper Burma, and a portion of the frontier tribes in Assam and the far east, the areas which were excluded from the 1881 exercise. These additions substantially added to the population under enumeration (Baines, 1893). 

However, certain tracts remained excluded: Nepal and Bhutan were not included, nor were the border tribes of Balochistan or the coastal settlements dependent on Aden. The small state of Sikkim was enumerated only informally, with its population excluded from the general returns.

The complete enumeration encompassed both British territory and feudatory states. British territory comprised 62 percent of the area but 77 percent of the population, while the feudatory states contained 38 percent of the area but only 23 percent of the population. This disparity was due to the inclusion within the states of extensive wild tracts, desert portions of the Indus valley, and forest-covered hills of the central belt: regions where population density remained low (Baines, 1893).

Methodological Refinements

The 1891 census introduced several small but important methodological changes, reflecting the evolving needs of Indian administration. The most significant innovation was in the treatment of occupational statistics. The 1881 census had attempted to enumerate only workers, those who actually performed labour contributing to family income (Baines, 1893). 

This approach was deemed to be ‘worthless’ for obtaining a full return, as Baines acknowledged. For 1891, the instructions were altered to now enumerate the supporting power of each means of livelihood, including the dependents. The General report noted that, “general distinction between worker and non-worker is more or less obliterated, and the results sought for are made into the number of persons supported by each occupation (Bains, 1893).”

This approach ensured that the occupation column would not be left blank for any individual, even an infant. It resulted in a comprehensive picture of how each occupation supported the population, rather than merely a count of workers.

The schedule also reflected an attempt at a more nuanced understanding of Indian social organization. 

The 1881 census had combined religious and caste information in a single column, leading to considerable confusion. The 1891 schedule separated these elements into four distinct columns: 1) religion, 2) sect of religion, 3) caste, tribe, or race and 4) subdivision of caste, tribe, or nationality. 

The separation of these categories was intended to avoid the confusion found in the previous enumeration and to allow for more precise analysis of the relationship between social organisation and other variables.

Another significant methodological change we can see is literacy. The 1891 census introduced a distinction between those learning (under instruction) and those literate (able to read and write but no longer under instruction), while specifically excluding those who could only sign their names from the literate category.

Additionally, a column was added to record knowledge of English, reflecting the growing importance of English-language education in Indian society. However, Baines noted that the distinction between learners and literates proved problematic in practice as the literate status was considered superior to that of a mere pupil (Baines, 1893).

The treatment of age was also refined, though Baines notes the persistent difficulties in it, "In India the return is peculiarly liable to be defective," he wrote, adding, "when once launched into the busy world, age is to the masses a matter of no importance." 

Additions were also made to the general tables regarding infirmities (insanity, deaf-mutism, blindness, and leprosy). For the first time, these conditions were tabulated in connection with caste and race, allowing the administration to identify the localization of certain diseases within specific social groups.

The Logistics of Enumeration

The central pillar of the administrative preparations for the census was the Agra Conference of December 1889. This assembly brought together Baines and the Provincial Superintendents who had managed the 1881 census, including figures such as Denzil Ibbetson and J.A. Bourdillon. Over ten days, these experienced officers debated the rules for enumerators, the subjects to be included in the schedule, and the tabular forms for exhibiting results (Baines, 1893; Natarajan, 1971).

The conference ensured that the onus of planning fell on those who were more closely aware of social complexities of Indian populations, rather than relying on Standard European models that were proven unsuitable during the 1881 census. 

Following the conference, Baines personally visited every Provincial headquarters to coordinate preliminaries with the newly appointed local Superintendents, ensuring that instructions were uniform and explicit across the diverse administrative units.

After these initial preparations, fundamental to the operation was the selection of the census date. The night of 26 February was chosen as it fell three days after the full moon, close enough to benefit from moonlight for the enumerators' rounds, but sufficiently removed from the full moon itself to avoid the major religious gatherings that drew pilgrims away from their homes. 

The enumeration procedure itself was adapted to Indian conditions. With only about six persons in every hundred able to read and write, the European model of having householders prepare their own schedules was impossible. Instead, the census followed the two-stage process that was employed in the two previous censuses. 

First, enumerators filled out schedules for all habitual residents before the census night. Then, on the appointed night, they made a second round, updating the record by striking out those who had died or were absent and entering those found on the premises who had not been present during the preliminary visit: travellers, guests, and the newborn.

Nearly 70 million schedules were printed in 17 different characters. The scale of paper consumption was staggering. About 290 tons of paper were used for the schedules alone. The linguistic diversity of India necessitated translations into multiple languages, though Baines noted with satisfaction that only 17 languages proved necessary "apart from a small issue of schedules in dialects specially confined to a few Hill tracts (Baines, 1893)."

The organization of the census followed a hierarchical structure. At the imperial level, Baines served as Census Commissioner, working from the India Office in London after the enumeration was complete. Provincial Superintendents directed operations in each major province and large state, working through district officers who supervised the actual enumeration. 

Baines particularly praised the district officers, noting that their co-operation and energy enabled the preliminary totals to be furnished for publication within five weeks of the census date.

The enumeration of special populations required additional arrangements. The homeless, travellers, and those on railways or boats required separate treatment. Military populations, both British and Indian, were enumerated under special rules. The census even accounted for the possibility of multiple counting. Baines noted that "one distinguished military officer on his way from the east to the north of India found that he had been returned three times in the course of about 15 hours (Baines, 1893)."

The Cost of the Census

Despite its unprecedented scale and complexity, the 1891 census was remarkably cost-effective. The total cost of the imperial government was Rs. 2,597,161 (approximately £173,000 at contemporary exchange rates), or approximately Rs. 10.81 per 1,000 persons enumerated in the provinces. This represented a slight reduction from the Rs. 11.49 per 1,000 spent in 1881. By comparison, the 1881 census in England cost nearly five times as much per capita as the Indian operation.

This efficiency was made possible by the reliance on unpaid or nominally paid local agents and the administrative experience of District Officers, for whom the census was a heavy addition to their current duties. 

However, variations between provinces were noted. In Burma, where native enumerators were not habitual to clerical labour were paid accordingly, at the ‘fancy rates’, incurring the highest cost at Rs. 19.37 per thousand. The Panjab, which did not undertake the more elaborate caste tables, showed the lowest at Rs. 9.28 per thousand (Baines, 1983). 

Bengal, despite its difficult enumeration conditions, the general absence of village staff, and the lack of information available as to boundaries, managed to reduce expenditure by more than 17 percent while dealing with a population nearly 7 percent larger than in 1881.

Comparison with the 1881 Census

The 1881 census had established the template upon which the 1891 enumeration would build. The decade between the two censuses witnessed a population increase of 27,821,420 persons in the tracts enumerated on both occasions. The British provinces accounted for 19,294,509 of this increase, while the feudatory states contributed 8,526,911. The annual rate of increase stood at 9.3 per mille, which, while substantial, placed India twentieth among twenty-eight countries for which comparable data were available.

One significant difference between the two enumerations lay in the treatment of previously omitted populations. The 1891 census included territories that had not been enumerated in 1881, making direct unadjusted comparison across the full scope misleading, because the increase in the population was not entirely due to natural growth. 

The report attributes it mostly to the increased accuracy from the previous census and people being more open to enumeration during the 1891, as compared to the 1881 census. The census authorities therefore carefully distinguished between tracts enumerated in both years and those added to the enumeration in 1891, presenting separate tables to account for this variation (Baines, 1983).

The growing confidence of the population about the census process is another notable difference. Baines observed that there was "a certain growth of confidence, as well as of population," leading to the appearance in the 1891 returns of persons, particularly among forest tribes and young women, who had likely been omitted from the 1881 enumeration. This improved coverage was particularly evident in the returns of girls between the ages of ten and fifteen, though Baines acknowledged that "reticence about this portion of the household is by no means uncommon, more particularly in the north of India (Baines, 1983)."

Conclusion

The 1891 census was a notable achievement in the statistical history of India. Yet the census also revealed the limitations of statistical knowledge in the face of profound cultural complexity. The persistent challenges of systematic under-enumeration of women, the preference for round numbers in age returns, the confusion between caste and occupation, and the difficulty of distinguishing between religious sects all testify to the difficulties of imposing Western statistical categories onto Indian social reality. 

Baines acknowledged these limitations throughout his Census 1891 report, noting that "uncertainty as to the true meaning of the statistics prevails to but an insignificant extent" only in the simpler returns, while more complex subjects like occupation remained "grievously handicapped (Baines, 1983)."

The value of the 1891 census, however, lies in its comprehensiveness. As Baines concluded, "it is equally indisputable that for many years to come each census will be more accurate than its predecessor." The 1891 enumeration provided a baseline against which future changes could be measured, a stock-taking of British India at the height of the imperial era that continues to inform historical and demographic scholarship to this day (Baines, 1983).

F. J. Mouat (1891), states in his note to the Royal Statistical Society before the census begins that,

This underscores the magnitude of both the process of enumeration and the volume of data it generated (Mouat, 1891). 

References:

Alborn, T. (1999). Age and Empire in Indian Census 1871-1931. The Journal of Interdisciplinary History, 30(1), 61–89.

Baines, J. A. (1893). General Report on the Census of India, 1891. London.

Mouat, F. J. (1890). Note on the Imperial Census of India 1891. Journal of Royal Statistical Society, 53(4), 689–694.

Natarajan, D. (1971). Indian Census Through A Hundred Years—Part I. Census of India.