32 | Counting amidst communalism: Making of the 1931 Census

By Gaurav Kalyani and Shivakumar Jolad

Published on: 22 Aug 2026

Introduction

On February 26, 1931, approximately two million individuals were engaged in conducting a population count across British India and the princely states. While most regions were counted simultaneously, Burma was enumerated two nights earlier, on February 24, because a local festival prevented synchronization with the rest of the country.

The final calculations recorded a total population of approximately 352.8 million people living across 1.8 million square miles. These figures indicate an average population density of roughly 195 persons per square mile and a population increase of nearly 34 million compared to the 1921 census data.

The census of 1931 was directed by J. H. Hutton, described in the discussion of his own paper before the Royal Society of Arts as an anthropologist of acknowledged standing. That background shaped the final product of this exercise. The 1931 report devoted unusual attention to caste, tribe, race and social custom, and commissioned original anthropometric fieldwork. This essay is a comprehensive summary derived from the extensive census reports written by Hutton.

The area this census covered was comprised of 15 major provincial administrations, including Assam, Bengal, Bihar and Orissa, Bombay, Burma, Central Provinces and Berar, Madras, Punjab, the United Provinces of Agra and Oudh, and the North-West Frontier Province (NWFP). Apart from them, 718,562 square miles of area under princely states was also covered. 

It excluded tracts lying beyond the administrative frontiers in the North-West Frontier Province and Burma. It also did not cover any portion of the peninsula that was politically outside the Indian Empire, such as Nepal, Bhutan, and the French and Portuguese coastal settlements. The Settlement of Aden and Perim Island were included under the Bombay Presidency's operations, but the wider Aden Protectorate was excluded. The Andaman and Nicobar Islands were covered, though certain highly isolated populations (such as the Shompen of the Nicobar interior) were excluded from detailed tables.

Population wise, it covered one sixth of the world’s population at the time, resulting in the publication of 28 volumes of reports and tables. The total expenditure was approximately Rs. 4,876,000, an increase from the Rs. 4,000,000 spent in 1921; however, this change was primarily due to accounting adjustments that incorporated administrative charges previously excluded from the census budget (Shirras, 1935).

When measured by the cost per thousand people counted, the 1931 operation cost came to Rs. 12.8, as compared to Rs. 14 for the previous one. Comparatively, the costs of conducting the census were much higher in Great Britain. Based on these metrics, the 1931 census was the largest operation of its kind at the time and was conducted at a lower relative cost than contemporary European examples (Shirras, 1935).

Census Machinery and Enumeration Process

The Census Act (Act X of 1929) was passed before a Census Commissioner had even been appointed, so by the time Hutton took charge, the legal framework was fixed and he had little chance to ask for the powers he needed. The Act provided limited authority to force municipalities to conduct a proper census. While its penalties were sufficient for dealing with occasional uncooperative individuals, they were not intended to handle large-scale organized resistance. It would matter in 1931.

Under the Census Commissioner was a Superintendent of Census Operations for each province and a census officer for each Indian state (Shirras, 1935). These were not permanent posts. Their departments were created from scratch every decade and wound up as soon as the tables were printed. Hutton thought that this was wasteful, as experience was continually getting lost and had to be relearned.

Staffing was precarious in practice as well as in principle. The Bombay Superintendent, who faced the worst obstruction of anyone, carried on until compilation was nearly finished and then had to hand his material to a successor. The Bihar and Orissa Superintendent broke down early, leaving his replacement to pull a badly delayed preliminary enumeration back on schedule and the Baluchistan officer was made Wazir of Kalat State and had to write his report while running a princely administration.

One organisational innovation was a conference of all census superintendents at Delhi in January 1931, where among other things the approach to listing ‘depressed castes’ was agreed. Hutton's verdict was that it had come far too late. According to him, if this decision had come a year or two early, the draft codes could still be revised and it might have produced a single common code instead of separate versions for each province.

The country was divided into districts, then charges, circles and blocks. A block was normally thirty to forty houses and the responsibility of a single enumerator (Shirras, 1935). Above him sat a supervisor, a charge superintendent, a district census officer and the district magistrate. 

Nearly all did census work on top of their ordinary jobs. Revenue officials, village accountants and schoolmasters formed the backbone, with police, magistrates, railway staff, forest and port officers and municipal employees drawn in as required (Shirras, 1935).

The enumerators themselves were unpaid volunteers. Hutton opened his report by thanking that body of some two million people, noting that many were out of pocket and that most received nothing but a printed certificate. The report claims that the reliance on them was unavoidable, as paying two million people would have made the census unaffordable. However it was also a weak point. Volunteers were getting harder to find, and a voluntary agency was exceptionally vulnerable to political influence.

The count itself came in two stages. Some weeks before census night, houses were numbered and a preliminary enumeration was carried out. Enumerators visited every household in their block and wrote out the schedules in full. This two-stage method, combined with the general immobility of the population, made it possible to get a reasonably accurate count on a single night across a subcontinent (Shirras, 1935).

Even so, Hutton acknowledged that the census was never fully synchronous. In some Himalayan districts the snow arrived long before February, so those populations were counted the previous autumn; in remote tracts with poor communications a non-synchronised count simply gave better results (Hutton, 1932). Special arrangements covered people in hospitals, asylums, jails, camps and in transit (Shirras, 1935), and rules were issued for a census of Indians on the high seas.

The Census Schedules and Data compilations

The general schedule was a single form covering each individual, with columns for block number and serial number; religion; sex; civil condition; age; race, tribe or caste; birthplace; occupation and subsidiary occupation; whether earner or dependant; mother tongue and subsidiary language; literacy; and infirmities.

Several details were changed. The old column asking the "occupation on which dependent" became a simpler "earner or dependant" question. Hutton objected to infirmity questions, although they were included. He wanted the leprosy return dropped, since the census figures fell so far short of medical estimates as to misrepresent the position, but the Department of Public Health used them as a basis for extrapolating true numbers and the return stayed.

A separate schedule sought a return of the "educated unemployed". Large numbers went out and very few came back, so the return was omitted from the tables as a misrepresentation of the facts. The results were varied. Madras recipients asked why they should fill in a form from a government that had no job to give them. In Burma, the mostly Indian unemployed of Rangoon feared repatriation. In Bengal, the unemployed bhadralok suspected they were being listed as political suspects. 

The industrial statistics collected in 1921 were not attempted at all, partly because Hutton thought a tally of horsepower, handlooms and mechanical shuttles did not belong in a population census.

Tabulation

After the census schedules were received, the data for each individual was manually transcribed onto separate slips of paper. During this process, personal names and specific administrative identifiers, such as block and circle numbers, were excluded. To facilitate categorization, different colors were used to represent various religions, while specific symbols indicated the sex of each person. These slips were then organized into labeled compartments and counted to produce the final statistical tables. 

The procedure for recording sex and marital status on census slips was modified to improve efficiency. In previous years, copyists used six different types of pre-marked slips to capture these details. The updated 1931 method utilized slips with only two printed sex symbols, which copyists then manually adjusted with a tick or cross to indicate marital status.

Additionally, the use of various regional vernaculars in the records and the high volume of complex caste-related categories made centralizing a mechanical punching operation difficult.

Two experimental methods were tested to improve the efficiency of census data collection. In Baroda State, the process of slip-copying was replaced by having enumerators record information directly onto perforated tickets, which were color-coded by sex to reduce the time and cost involved. Hutton recommended wider adoption of this method wherever enumerators could be properly trained.

In the Andaman and Nicobar Islands, a manual system using steel skewers to sort cards was found to be effective for smaller population groups. Hutton concluded that while mechanization was a likely future development, its successful implementation would require the establishment of a permanent statistical department rather than the temporary administrative structure used for each decennial census.

What 1931 counted differently

Four things set the 1931 census apart from its predecessors, beyond the methods.

The exterior castes

The Government of India's instructions asked that attention be paid to information conducive to better knowledge of the backward and depressed classes. Framing a single all-India definition proved impossible, so each province was asked to draw up its own list. It was to be determined using local enquiry into which castes suffered disability such as being barred from temples, schools or wells, or whose contact required purification on the part of high-caste Hindus. Muslims and Christians were excluded, as were hill and forest tribes who had not become Hindu. 

Earlier figures had ranged widely with the Hartog Committee's approximately 30 million, the Franchise Committee of 1919's roughly 42 million, the 1921 Census Commissioner's minimum of 52.68 million, and estimates as high as 70 million. 1931 was the first attempt to replace estimate with enumeration.

Fertility

For the first time, a serious attempt was made to collect data on family size and reproduction. Returns were obtained for about 900,000 families drawn from all grades of society and from every part of India except the United Provinces. Educational or medical agencies were generally used, and wherever possible figures were collected through women. 

The material was uneven. Many returns had to be rejected for particular tables, in some cases more than half. However Hutton regarded them as well-mixed, even if they were numerically inadequate samples. Baroda and Travancore made particularly careful enquiries.

Age and the actuary 

The age returns were treated by a new method of grouping developed by H. G. W. Meikle after his actuarial examination of the 1921 figures, and adopted for the first time in 1931. A full actuarial report and set of life tables was prepared by L. S. Vaidyanathan and printed with Hutton's own report. Hutton considered the 1931 age returns probably more accurate than ever before.

Race

Herbert Risley's 1901 census report had been accepted virtually as the official textbook on Indian ethnography. Hutton judged that enough work had been done since to require a restatement of the problem. To test the data underlying Risley's conclusions, the Census Department obtained the co-operation of the Zoological Survey of India, and B. S. Guha carried out anthropometric measurements concentrated on Brahman castes. It was chosen because of their comparative strictness and therefore stability of type. Other castes were also surveyed.

Two smaller changes were notable. Burma's figures were tabulated by race rather than religion, on the ground that religious tabulation had never suited that province. At the cost of some inconsistency in the all-India tables, data was restructured into races. Second, in the tables for literacy and civil condition by selected castes, the arrangement was changed from a purely provincial basis to show corresponding units from different parts of India together (Hutton, 1933a).

Impact of contemporary politics

The 1931 census, like that of 1921, ran straight into a mass political movement. Gandhi's march at the salt-pans of Dharasana coincided with the opening of operations. In 1921 he had blessed the census at the last minute, however in 1931 that blessing did not arrive. Although he was not known to have advised a boycott himself, other Congress leaders did. 11 January 1931 was notified by the Congress Committee as Census Boycott Sunday.

The boycott was not taken up with much enthusiasm, but where it took hold it was effective. Ahmadabad was a serious case. The city was built largely of pols — blind alleys, each with an outer gate, tenanted by a specific caste community. On the census night the pols barred their doors and defied the enumerators. Police forced an entry into many, but obstruction continued. Most inhabitants stayed out in the streets, and the bands of small boys (vanar sena) made question and answer inaudible with catcalls, jeers and songs (Shirras, 1935). 

The census was carried out effectively in six of twenty-four wards; the other eighteen were incomplete and the city's total had to be estimated. Broach and Surat were probably defective too. Hutton put the resulting deficiency for India at about a tenth of one per cent — a small error nationally. 

In Delhi, house numbers were obliterated and enumerators were obstructed. One was refused access to the house where Gandhi was staying, though the Deputy Commissioner had already arranged for Gandhi to be censused at the city gate on his way in.

In other places, enumerators ran into the opposite problem. In the Punjab, where political rivalries ran high and a communal franchise based on census returns seemed possible, each community wished to poll its full strength. 

Exterior castes, badgered by one party to return themselves as Hindus, and as Sikhs or Muslims by others, labelled themselves Ad Dharmi, adherents of the original religion. Disputes over whether a man was Ad Dharmi or Sikh led to several affrays and at least one homicide. 

On the Orissa borders, pan-Oriya propaganda and the counter-propaganda it provoked were both detrimental to census taking. There the stakes were explicit as the Orissa Boundary Committee had already requisitioned census figures.

Meanwhile, caste associations continued a tendency observed at every census since 1901. They treated the Census Department as a sort of herald's college and put forward claims of all sorts to improve their standing. What Hutton found new was that, where castes had once split into ever finer divisions, sub-castes and castes were now combining under a new name. 

Grazier castes were combined as "Yadava", artisan groups as Vishwakarma or Jangid and so on. It was a move to strengthen their numerical position politically.

Impact of legislative developments were also observed. The Sarda Act (Child Marriage Restraint Act), 1929 produced a definite fall in the accuracy of the record of civil condition. Hutton estimated the error at not less than 0.5 per cent and probably higher. It was largely because girls married in contravention of the Act were deliberately returned as unmarried. 

Matters were made worse by attempts to produce census schedules as evidence in criminal courts, including in Sarda Act prosecutions. The schedules had been destroyed in the ordinary course before the question was settled, but Hutton argued that the next Census Act should protect the information obtained, since any other course was bound to be seen as a breach of faith with those enumerated.

Afterwards

The most consequential thing the 1931 census produced was the count of about fifty million people classified as ‘exterior castes’. Questions of religion and caste assumed particular importance because of the impending political reforms and the likelihood that figures would be needed to determine franchise and constituencies. 

Figures had in fact already been requisitioned by both the Franchise Committee and the Orissa Boundary Committee. Both for social and political reasons, Hutton wrote, it was necessary to know these numbers, with reference to representation in the body politic and to social work aimed at improving the position of those counted. The census was no longer only a description of Indian society; it had become an input into the machinery that would help determine political rights.

Hutton's own conclusions were mostly about procedure. He wanted the census made de jure rather than de facto, so enumeration could be spread over a longer period and carried out by local revenue staff instead of volunteers. 

He wanted a permanent statistical department instead of what he called a “pitiable ephemeron”, which was unable to put into practice the knowledge acquired from its too brief experience. He also suggested that in a federated India each unit might be responsible for its own enumeration, with results still handled centrally. This proposal drew immediate objection from statisticians who thought a census had to be highly centralised to be sound. 

Since the 1931 census proved to be the last census to conduct caste enumerations, its findings continue to impact contemporary politics and governance in post-independence India.

(Authors: Gaurav Kalyani works as Research Associate at the Center for Legislative Education and Research, FLAME University, Pune; 

Shivakumar Jolad works as Associate Professor (Public Policy), and is the Chair of Center for Legislative Education and Research and Director India State Stories, FLAME University, Pune

Gaurav contributed to conceptualization and did research and primary writing;  Shivakumar contributed to conceptualization, research and editing)

Bibliography

Hutton, J. H. (1932). Census of India, 1931. Journal of the Royal Society of Arts, 80(4154), 783–805. https://www.jstor.org/stable/41362598 

Hutton, J. H. (1933a). Census of India, 1931. Vol. I, India, Part I: Report. Manager of Publications.

Hutton, J. H. (1933b). Census of India, 1931. Vol. I, India, Part V: Administration report. Government of India Press.

Shirras, G. F. (1935). The census of India, 1931. Geographical Review, 25(3), 434–448. https://www.jstor.org/stable/209312