38 | The 1941 Census of India: Counting a Population in Wartime
Updated: 2 days ago
By Siddharth Ramkumar and Gaurav Kalyani
Published on: 22 October 2026
"The war has laid its hand on the Indian census as on every other activity of the Indian Government and people... This census saw more changes in methods than had previously taken place in the whole 70 years since the census began."
-M. W. M. Yeatts, Census Commissioner for India
Introduction
The eighth census of India was taken in February and March 1941, about eighteen months after the Second World War began. It counted 388,997,955 people in British India and the princely states (Yeatts, 1943a) , and it was the last census of the whole subcontinent before the Independence and Partition in 1947.
The census count itself was carried out in full, with several new methods that later became standard practice. But the Government of India, citing the cost of the war, soon ordered that only a small part of the collected information be sorted for British India (Yeatts, 1942, 1943a) . The result was a census that was ambitious yet limited.
This essay synthesizes three main works by M. W. M. Yeatts, the Census Commissioner for India - the Administration Report (1942), the Tables volume (1943a), and a paper read at the Royal Society of Arts in London in January (1943b). Along with them, the reports of provinces and states, are also utilised for this work. Yeatts directed the operation, so his account is also a defence of his own decisions. Some provincial officers disagreed with him, and their views are noted.
The Planning
Census work normally begins years before the count. In 1941 it did not. Provincial superintendents were meant to take charge in December 1939, but the war had started three months earlier. The Government of India did not decide whether to hold a census at all until February 1940. That left only about a year for planning. A second decision, on how much of the data would be tabulated, was not made until the enumeration was over (Yeatts, 1942).
This delay had profound effects. Sorting offices that were normally booked months ahead were not available. Bombay lost the buildings it had reserved, and Bengal lost a suitable one (Yeatts, 1942) . In Madras, the first instruction circular to the districts went out only on 30 May 1940 (Elwin, n.d.).
A conference in Delhi between 20–22 February 1940 gave the work a start. It brought together the eleven provincial superintendents and the officers of several agencies and large states, twenty-two in all. Yeatts explained the new methods and then let each superintendent adapt them within a few common rules (Yeatts, 1942).
The Census Act of 1939 let provincial governments charge census costs to municipal funds, but local government was a provincial subject, so the section was not valid. Yeatts chose not to press the point, and most municipalities paid as before (Yeatts, 1942). The Act also could not compel anyone to serve as an enumerator. This nearly caused a crisis in Delhi, where the head of a large firm disputed the power to appoint his staff. Yeatts settled it by persuasion. He recommended a permanent census act, which Hyderabad adopted (Mazhar Husain, 1945). Other states passed their own acts. In Rewa, a Durbar order gave tahsildars power to fine anyone who refused census duty (Dube, 1944).
Effects on the Census staff
The census rested on a small group of officers. Of the provincial superintendents, only Dracup in Bombay had worked on the previous census (Yeatts, 1943a). The states had more continuity. Baroda's Mukerjea was the most experienced of all (Yeatts, 1943a), Hyderabad's superintendent could call on staff from 1931 (Mazhar Husain, 1945), and the officer in Rewa was on his third census (Dube, 1944).
The war had pulled most officers away. From July 1941 Yeatts worked on the census part-time, because he was also a Joint Secretary handling American purchase and procurement (Yeatts, 1942). Scott in the North-West Frontier Province left his post that month for other duties (Yeatts, 1942). The Census Superintendent for Central India left for Persia on war service just after the count (Dube, 1944). In Mysore, Krishna Rao (1944) was made Air Raid Precautions Controller after Pearl Harbour. His draft report went into storage until he was relieved on 16 August 1942, and the popular edition appeared on Christmas Eve that year. The work also took a toll on their health. Five of Yeatts's provincial team fell ill, Yeatts himself directed key central operations from a hospital bed following major surgery, receiving urgent census files within 36 hours of his operation. He spent December 1940 recovering in a nursing home (Yeatts, 1942, 1943a).
Below the superintendents, the work fell on district officers, revenue staff, village officials and teachers, who received no extra pay (Yeatts, 1942). The war added to their workload. In Madras, Elwin (n.d.) recorded that revenue officers were also busy with war propaganda, civic guards and the Governor's War Fund. Elwin dropped a plan for special housing enquiries so as not to ask more of them.
The war also affected public attitudes. In Sind, the superintendent, Lambrick (n.d.), addressed public meetings partly to remove fears that the census was linked to army conscription. Yeatts noted the same rumour from East Bengal to Rajputana, though not in the South (Yeatts, 1942). In Hyderabad, the superintendent reported that some young men hid their age for fear of recruitment (Mazhar Husain, 1945).
Superintendents tried to overcome these problems by touring. Lambrick (n.d.) travelled about 1,000 miles a month in his own car, and the Hyderabad superintendent visited all but five tahsils (Mazhar Husain, 1945). In Cochin, some supervisors who claimed to have no doubts did poorly when asked to enumerate in front of the class (Menon, n.d.), and in Kashmir some officials had not read the instructions (Wreford, 1943).
How the Census was conducted
Until 1931, the census had two stages. Enumerators first prepared a record of each household over several weeks. Then, on a single night, they went around again to correct it for visitors, births and deaths. Yeatts abandoned this in 1941 (Yeatts, 1942).
His case was that the careful record built over weeks was checked and tested, while the final night's round was hardly checked at all. In a year of communal tension, he feared that the final round would invite false returns. The one-night method also needed about two million enumerators, which meant drawing in outsiders workforce. In wartime, railway staff were already stretched, and on the Frontier few enumerators would go out after dark (Yeatts, 1942).
The new rule was to record each person at the place of normal residence, with sunrise on 1 March 1941 as the reference time. Enumeration was spread over 10-15 days, followed by a short round to adjust for births and deaths (Yeatts, 1942).
The change removed some old problems, though not all. There was no count of travellers on platforms and trains, although Baroda still made each of its 200-odd railway stations a separate circle (Mukerjea, 1941). Weddings caused trouble, since astrologers named 27 and 28 February as good days, and a Muslim gathering at Unava, which drew about 20,000 people, fell close to the date (Mukerjea, 1941). Baroda ruled that guests and pilgrims from within the state be counted at home (Mukerjea, 1941), and Patiala counted its wedding parties before they left (Suchdev Singh, n.d.).
In autumn 1940, the house-numbering exercise was widened into a first count. Officials recorded the people in each house by sex and broad age. Yeatts argued that it would help estimate the number of slips needed, it would give an approximate record if war made the real count impossible, and it would also check against a corrupted count (Yeatts, 1942). It served that third purpose. In the United Provinces, the house-list and final totals differed by more than two percent in only one district (Yeatts, 1942). In Baluchistan, officers traced the larger gaps to nomads moving between districts (Gastrell, n.d.).
Several units used the house list for more. Mysore's form also recorded water supply, schools, medical institutions, cultivated area and cattle (Krishna Rao, 1944). Assam added village notes on flooding, cholera and water supply (Marar, 1940), and Hyderabad painted house numbers in durable colour so that they would last until the next census (Mazhar Husain, 1945). Quality was uneven. Krishna Rao (1944) wrote that hardly one house list in Mysore could bear careful scrutiny.
Previous censuses relied on large, unwieldy paper household schedules (20'' × 31'') containing multi-column entries for entire families. The large household schedule was abolished and replaced by individual paper slips bound into pads of 100. Enumerators wrote answers straight onto individual slips of about four by six inches, bound in pads of 100. This removed the copying stage in the sorting office, where errors could creep in (Yeatts, 1942). Baroda had tried the idea in 1931 (Mukerjea, 1941).
Figure 1 - Specification of the individual paper slips bound into pads of 100 (p.32);
Figure 2 - Text Diagram/Layout Sketch in the Administration Report Volume I, Part II: Administration Report (1942), Chapter II ("Pad System", p. 33)
The slips carried no regional script. They showed only question numbers and symbols, so one design could serve the whole country (Yeatts, 1942). A tick meant yes and a cross meant no, and a few letters stood for common answers. A code number at the top identified the area, and the press could print all slips for one code at a stretch. In Patiala the code gave the state, tahsil, charge and circle (Suchdev Singh, n.d.). Questions on infirmities such as leprosy, deafness and insanity were removed, since Yeatts thought a census could not measure them (Yeatts, 1942).
Nearly 400 million slips were printed at the Government of India Press in Calcutta and sent to every tehsil. Yeatts reported a saving of more than a lakh of rupees, even though paper prices had risen by about 30 percent (Yeatts, 1943b).
Local officers found some weaknesses. Krishna Rao (1944) thought the old schedules were safer. He noted that the symbols caused most errors, a lost slip could not be replaced, and doubtful entries could not be checked. Sorting for sex, religion and civil condition, once done during slip-copying, now took his office nearly three months. In Baroda, one enumerator counted the god Mahadev in a village, recorded him as a married Hindu, and filled the other boxes with crosses (Mukerjea, 1941).
By abandoning the one-night count and expanding block sizes from 30 houses to 120–250 houses, census authorities halved the total volunteer enumerator army. The nationwide enumerator force was reduced from 2,023,516 in 1931 to 1,056,813 in 1941. This drastic reduction allowed census superintendents to replace untrained private volunteers with official revenue officers, village patwaris, and village headmen. The following table shows the number of enumerators in 1941 as compared to 1931.
Table: Number of enumerators, 1931 and 1941
Unit | 1931 | 1941 | Change |
India | 2,023,516 | 1,056,813 | −48% |
Madras | over 370,000 | 82,000 | about −78% |
Punjab | 141,152 | about 71,000 | about −50% |
Patiala | 10,664 | 4,329 | −59% |
Cochin | 6,305 | 3,681 | −42% |
Hyderabad | 78,066 | 53,951 | −31% |
Baroda | 9,409 | 8,466 | −10% |
Source: Yeatts, Administration Report; Cochin, Hyderabad, Patiala and Baroda reports.
Local conditions also shaped the choice. Village officers suited the work in Madras and Punjab. But Bihar had almost none, so the census used the police stations, which Yeatts disliked because the public did not always speak freely to the police. Bengal used the presidents of Union Boards (Yeatts, 1942). In Patiala, teachers and literate pensioners did most of the work, and about 1,200 rural blocks had no enumerator of their own, so the circle supervisor counted them (Suchdev Singh, n.d.). Baroda used 5,403 teachers and 867 village accountants, and only 235 private persons against 521 in 1931 (Mukerjea, 1941). Yeatts noted that quality improved where the reduction went furthest (Yeatts, 1942).
Special Populations and Difficult Territories
The census had to deal with people who did not fit a house-to-house model.
Railways and the military - Railway staff were treated like other household populations. The armed forces counted their own members through their units, and Yeatts found this simpler than civilian enumeration. For security reasons, troops were not shown separately in the published figures (Yeatts, 1942, 1943a).
The frontier - Parts of the North-West Frontier could not be enumerated in the usual way. The frontier statement records 2,377,599 persons, of whom 753,261 were enumerated and 1,624,338 were estimated (Yeatts, 1943a). The enumerated areas were new to the census. In the three pricely states the count was carried out entirely by state servants. Only six questions were asked, and a woman could be entered as the wife or daughter of a named man (Scott, 1942).
The Powindahs - Each autumn, nomadic families from Afghanistan crossed into India with their camels and flocks, to graze and to find work. Little was known about this movement, and Yeatts decided the census could help. The count was built on the posts where grazing fees were already collected, so that the tribesmen met a process they knew. The count found 47,427 people, with 413 women for every 1,000 men. The figure was probably low, since some families used routes that avoided the posts (Powindah Census, 1942).
Mountain Kashmir - In Jammu and Kashmir, snow made a February count impossible in high villages. A total of 425 villages, with the whole of Ladakh, Astore and Karnah, were counted from 1 December 1940. The records then had to travel. Ladakh's papers were carried by coolies who were held up by snow on the Zoji La until 11 April, and the Srinagar city record reached the census office, a mile away, only on 27 March (Wreford, 1943).
Assam and Travancore - Assam's tea gardens were made into separate charges, as were towns and hill areas, and each doorway of the coolie lines counted as a house (Marar, 1940). In Travancore, the Sabarimala pilgrimage overlapped with the preliminary count, and the floods of July 1940 slowed house-listing (Narayanan Tampi, 1942).
Administration, Cost and Problems
The Administrative Report complained that the census was a central subject in law but was carried out almost wholly by provincial staff. The Government of India does not pay enumerators, so it depends on provinces, municipalities and village officials, over whom the Census Commissioner has no direct authority. Yeatts proposed a clearer division, with the centre directing and tabulating and the provinces responsible for enumeration. He also suggested that superintendents be made deputy secretaries in their provincial governments. Only the United Provinces did so, and he credited the quality of its count partly to this link (Yeatts, 1942).
Some central offices did not co-operate. Yeatts criticised the Survey of India in the United Provinces, some Calcutta offices and the Reserve Bank, and said the Collector of Customs in Bombay had sought an exemption from census duty. At the same time, he called the system of unpaid citizens counting their own country creditable and advanced, and said other countries impoverished by the war might envy it. It had also arisen by accident, he wrote, and now needed careful nursing (Yeatts, 1942).
Money was also tight due to the ongoing war. The appendices recorded Rs 1,465,802 for enumeration and Rs 645,049 for tabulation, with recoveries of Rs 360,607, which left a net sum of Rs 1,750,244 (Yeatts, 1942). In his paper to the Royal Society of Arts, Yeatts (1943b) put the cost at Rs 4-3 per 1,000 persons, against Rs 4-8 in 1931, even though about 15 percent more people were counted and prices had risen by 20 to 30 percent. The figures left out unpaid labour, and tabulation was far smaller than before, so they are not directly comparable with 1931. At unit level, the budgets were small. Cochin's commissioner started with a clerk, a typist and two peons (Menon, n.d.). In Kashmir, the commissioner worked from his own house for ten days before he found an office, and his budget was cut from Rs 42,627 to Rs 31,054 (Wreford, 1943).
Officers also reported small frictions. Mysore and Patiala froze transfers of census staff (Krishna Rao, 1944; Suchdev Singh, n.d.). Mysore found that some officers were paid allowances two or three times and others without a receipt, and that late pads from charge superintendents delayed the whole abstraction by over a fortnight (Krishna Rao, 1944).
Some of Yeatts's points concerned statistics. Local governments had used very different definitions of a town. For example, Bhopal counted places of 2,000 people, while Madras ignored places over 5,000 (Yeatts, 1942). He proposed a minimum of 5,000 for the common tables, and argued that totals should not be quoted to the last digit, since births and deaths make them out of date within minutes of the count (Yeatts, 1942).
What Was Tabulated and What Was Not
Every census began with a first sort of the slips by sex and community, and for British India this became the whole programme. Population by sex and community for districts, tehsils and towns. The slips holding age, marital status, livelihood, literacy, birthplace and fertility were put into storage in district treasuries without being sorted. Yeatts regretted most the loss of the full age and civil condition table, which he thought were the most valuable of all. In 1943 he wrote that the slips were kept ready, but that his staff would be dispersed and he doubted that anyone would be available to resume the work (Yeatts, 1942, 1943a).
Table: Tabulation done in 1941
Subject | British India | States that paid for full tabulation |
Population by district, tehsil and town | Published | Published |
Community, by district | Published | Published |
Age and civil condition | Not tabulated | Tabulated |
Means of livelihood | Not tabulated | Tabulated |
Literacy | Only where a province paid | Tabulated |
Birthplace | Not tabulated | Tabulated |
Mother tongue and script | Not tabulated | Varies |
Caste | Bengal, Bombay and Central Provinces only | Varies |
Source: Part II - Tables; Provincial reports.
Many princely states paid for complete work, such as Mysore, Travancore, Cochin, Gwalior, Baroda and Hyderabad, along with four of the Punjab states (Yeatts, 1943a). Some states had never done their own sorting. In Rewa, none of the local staff knew anything of it, and a tahsildar was put in charge of the new office (Dube, 1944). Hyderabad took on nearly a thousand temporary workers in two large government buildings (Mazhar Husain, 1945). Yeatts noted that due to this uneven work, parts of the country such as the regions of Kerala and Mysore are better represented in the figures than the figures in the regions of Indus and Ganges plains, which held nearly half the population (Yeatts, 1943a).
The states also tested the new methods. Yeatts saw sampling as a way to save labour when the slips already covered the whole population, and every fiftieth slip of each sex was marked for it (Census Commissioner for India, 1940–1941; Yeatts, 1942). A test in Punjab found that the sample's community distribution differed from the full sort by less than one percent (Yeatts, 1942).
Delhi became the first census unit to be tabulated by machine. Hiring punched-card machinery for all of India was impossible in wartime, so Yeatts used the spare time of machines the government already owned. With the Central Board of Revenue’s permission, census cards were run through its Hollerith 80-column equipment in New Delhi during off-peak weekend hours and night shifts.
The Delhi experiment, he wrote, “conducted against every kind of difficulty, has shown that this is quite possible” (Yeatts, 1943a). Tight coding was employed to fit 3 persons instead of one per card. This cut the cost of cards by about two-thirds. Leaving aside machines hired at the end, the Rs 7,000 estimate was missed by only Rs 228. The whole of Delhi province was processed for Rs 8,780, or roughly Rs 10 per 1,000 records. Rare categories, such as unusual birthplaces or minor mother tongues, would have used too many columns, so coders sorted them by hand before punching (Yeatts, 1943a).
In Tonk, the records were put on cards sorted by needle, and this also worked well (Yeatts, 1943a). Yeatts did not overstate the case for machines stating, “Sorting will always be the weak point of a mechanical census so far as time is concerned. For all the impressive outturns of a machine there remains the hard fact that it can rattle out broadly only one thing at a time, whereas a hand sorter can sort for an entire table at once” (Yeatts, 1942).
Census Findings
Population and Growth
The operations spanned a vast territory of 1,581,410 square miles, encompassing 655,892 rural villages and 2,703 urban towns. The total count of occupied residential dwellings reached 76,035,345 houses. The final tables recorded 388,997,955 persons, of which 201,025,726 were males and 187,972,229 were females, or about 935 women for every 1,000 men. British India held 295,808,722 people and the states and agencies 93,189,233. Against the 1931 total of 338,119,154, adjusted for boundary changes, this was an increase of 50,878,801, or 15 percent. Yeatts noted that this gain was larger than the population of any European country except Germany and Russia (Yeatts, 1943a).
Table: Population variation between 1941 and 1931 censuses
Region / Administrative Level | 1931 Population | 1941 Population | Decennial Growth (Persons) | Decennial Growth (%) | Density (Persons / Sq. Mile) |
All-India Total | 338,119,154 | 388,997,955 | +50,878,801 | +15.0% | 246 |
British Provinces | 256,757,818 | 295,808,722 | +39,050,904 | +15.2% | 341 |
Bengal Province | 50,115,548 | 60,306,525 | +10,190,977 | +20.3% | 779 |
United Provinces (U.P.) | 48,408,482 | 55,020,617 | +6,612,135 | +13.7% | 518 |
Madras Province | 44,205,243 | 49,341,810 | +5,136,567 | +11.6% | 391 |
Bihar Province | 32,367,909 | 36,340,151 | +3,972,242 | +12.3% | 521 |
Punjab Province | 23,580,864 | 28,418,819 | +4,837,955 | +20.5% | 287 |
Bombay Province | 17,992,053 | 20,849,840 | +2,857,787 | +15.9% | 272 |
Central Provinces & Berar | 15,323,058 | 16,813,584 | +1,490,526 | +9.7% | 170 |
Assam Province | 8,622,791 | 10,204,733 | +1,581,942 | +18.3% | 186 |
Orissa Province | 8,025,671 | 8,728,544 | +702,873 | +8.8% | 271 |
Sind Province | 3,887,070 | 4,535,008 | +647,938 | +16.7% | 94 |
N.W.F. Province | 2,425,076 | 3,038,067 | +612,991 | +25.2% | 213 |
Princely States & Agencies | 81,361,336 | 93,189,233 | +11,827,897 | +14.3% | 130 |
Hyderabad State | 14,436,148 | 16,338,534 | +1,902,386 | +13.2% | 198 |
Rajputana Agency | 11,570,583 | 13,670,208 | +2,099,625 | +18.1% | 103 |
Central India Agency | 6,643,761 | 7,506,427 | +862,666 | +13.0% | 144 |
Mysore State | 6,557,302 | 7,329,140 | +771,838 | +11.8% | 249 |
Travancore State | 5,095,973 | 6,070,018 | +974,045 | +19.1% | 792 |
Kashmir & Feudatories | 3,646,243 | 4,021,616 | +375,373 | +10.3% | 49 |
Growth was not even, The North-West Frontier Province, Punjab and Bengal grew fastest, by 20 to 25 percent, while Madras, Bihar and the Central Provinces grew by 10 to 12 percent (Yeatts, 1943a). Yeatts explained the rapid growth in two areas as the filling of new land. In the Punjab canal colonies and in states such as Bikaner and Bahawalpur, irrigation had opened thinly settled land, and the population moved in. Bikaner grew by about 40 percent as Sikh settlers followed the water. East Bengal was a younger, wetter land where people could still multiply on new soil. He also credited falling death rates, and, for some northern places, better counting than in 1931. These explanations were his own and were not backed by a breakdown of births, deaths and migration (Yeatts, 1943a, 1943b). Hyderabad's superintendent said reliable vital statistics were not available to test such causes (Mazhar Husain, 1945).
Towns and Cities
A defining demographic feature of the 1931–1941 decade was the dramatic acceleration of urban growth, which was mostly by industrialization, rural-to-urban migration, and municipal expansion. The urban population was 49,696,053, or 12.8 percent of the total. Cities of 100,000 or more rose from 35 in 1931 to 58 in 1941, and their combined population from 9.14 million to 16.53 million. Part of the rise came from places crossing the 100,000 mark, so it was not a growth rate for a fixed set of cities. Five of the new cities were in the United Provinces and four in the Punjab (Yeatts, 1943a).
Yeatts (1943b) described the growth of large cities in his London paper as something one could see in the streets. Cawnpore nearly doubled. Lahore, Amritsar, Rawalpindi and Jullundur each grew by about 50 percent, and Jamshedpur by 77 percent to about 150,000. Hyderabad City became the fourth largest in India. Ahmedabad, which could not be counted in 1931, now had about 590,000 people.
Sex Ratio
The 1941 Census recorded an All-India sex ratio of 935 females per 1,000 males, reflecting a slight decline from 1931 and continuing a long-standing national deficit of women. This gap followed a geographical pattern, with severe female deficits in northern areas like Baluchistan (703) and Punjab (847) contrasting with female surpluses in southern territories such as Orissa (1,069) and Cochin State (1,042). Furthermore, male labor migration to industrial hubs created acute urban distortions, reducing female ratios to 452 per 1,000 in Calcutta and 581 in Bombay. Yeatts connected these figures to deep-seated health and social issues, including early marriage, poor healthcare access, and high maternal mortality (Yeatts, 1943b).
Literacy and other measures
Literacy figures for British India were provisional, taken from the first count made by enumerators and not from a checked sort. Yeatts arranged this step so that some literacy data would exist even if tabulation was cut. The figures give about 47 million literate persons for India, or 12 percent. Where the sex breakdown was available, 19 percent of men and 5 percent of women could read and write. Among the provinces, Bombay was leading, Bengal was well placed, and the United Provinces and Bihar were far behind (Yeatts, 1943a).
The strongest results came from states with full tabulation. In Travancore nearly half the population was literate, including more than half of the men and over a third of the women (Yeatts, 1943a). Cochin showed about a third, and Baroda about a quarter (Mazhar Husain, 1945). The Travancore superintendent warned that the 1931 count had used a stricter test, so that an apparent rise of 97 percent fell to 49 percent once the definitions were matched (Narayanan Tampi, 1942; Yeatts, 1943a).
The census also asked fertility questions, including the mother's age at the birth of her first child. Yeatts said his main purpose was to draw attention to the poor state of vital registration (Yeatts, 1943a). Age returns remained weak everywhere. Baroda recorded 375 centenarians against 311 in 1931 (Mukerjea, 1941), and Rewa's officer wrote that tribal people and villagers seldom knew their ages (Dube, 1944).
Community, Caste and Tribe
One of the most important changes in the 1941 census was how Yeatts decided to count religion and caste. The central change was to classify people by community and not by religion. Yeatts argued that religion returns had never worked for tribal people, because an enumerator faced with a tribesman who had no idea what the question meant tended to record him as Hindu. In 1941, persons of tribal origin were shown as Tribes whatever religion they returned. The effect was significant. The tribal share rose from about 2.4 percent in 1931 to about 6.6 percent, and the Hindu share fell from about 68.2 to about 65.9 percent. Yeatts insisted that no one had been moved from one group to another, just a different thing was being measured (Yeatts, 1943a).
His summary was that, of every 100 Indians, 66 were Hindus, 24 Muslims and 6 of tribal origin (Yeatts, 1943a). Scheduled Castes were about one-eighth of the whole population, Indian Christians about 1.5 per cent, and persons of European origin numbered about 135,000. Muslim, Sikh and Christian figures remain broadly comparable with 1931, but Hindu and tribal figures do not. Table XIII covers 386,666,623 persons, not the full total, because about 2.3 million people in frontier areas were not classified by community (Yeatts, 1943a).
Yeatts dropped the central all-India caste tabulations, due to the constrained finances. He wrote that even in 1931 the caste table was “severely limited for financial reasons”. He thought the time had passed for “this enormous and costly table” as a central charge. He further stated that due to the ongoing war, money was also tight, so there was “no justification for spending lakhs on this detail”. He thought that even a full tabulation of British India would not have produced an all-India caste table (Yeatts, 1943a).
Instead he argued and advocated that the center should carry only what it needed for central purposes: age, means of livelihood, community, civil condition, literacy, birthplace and mother tongue. Anything extra, caste included, should be paid for by whoever wanted it (Yeatts, 1943a).
In the Administrative Report (Yeatts, 1942) he notes that many Hindus in Bengal, and to a lesser extent elsewhere, refused to return any caste. He thought this refusal sprang from a mistaken fear that admitting divisions would weaken Hindu unity. Because of this, he expected caste returns to become less and less useful, and said that if the trend continued the census might as well “come down to the plain community question”.
Legacy and Limitations
The 1941 census holds a double character. As a record, it is scant, as compared to previous censuses. However, as an operation, it introduced more changes than any census before it, and its one fully tabulated subject, community by district, turned out to matter most. The Boundary Commissions of 1947 used its district community figures in Punjab and Bengal.
Many of its methods lasted. Yeatts argued for continuity between censuses such as a house list kept up to date, permanent house numbering, proper registration of births and deaths, and sampling. He described the old arrangement as a ‘Phoenix System’, an ad-hoc administrative structure assembled from scratch every decade and completely disbanded within twenty-four months (Yeatts, 1942, 1943a). He condemned this temporary cycle as both a financial mistake and an intellectual crime that repeatedly destroyed operational momentum and continuity. To modernize the process under rigid wartime budget constraints, Yeatts introduced mechanical tabulation as an experimental step toward institutional continuity.
The Census Act of 1948 gave the census a permanent legal footing, and the 1951 census kept the extended counting period in place of the single night.
Sampling also had an afterlife. The marked slips, one in every fifty of each sex and so a 2 percent sample (Yeatts, 1942) that came to be called the “Y-Sample”, stayed untouched in district treasuries until 1945, when the Population Data Committee recommended that they be recovered. The work was led by P. C. Mahalanobis at the Indian Statistical Institute in Calcutta, where the sample slips were transferred on to Hollerith punched cards.
From these cards, age, livelihood and fertility tables for undivided India were reconstructed between 1948 and 1956. The gap Yeatts regretted most was thus partly filled, and by the combination of sampling and machine tabulation that he had tested in 1941, at a fraction of the cost of a full count.
Yeatts directed the census and defended its choices, and his comments on communal activity, tribes and the public reflect the standpoint of a colonial official. The provincial and state reports add what he left out. The problems like the strain on unpaid staff, the errors the slips brought with them, and the labour of sorting are discussed and highlighted. Read together, they show a census taken under pressure, while accepting that the war would limit what could be published.
(Authors: Siddharth Ramkumar is an undergraduate student at FLAME University;
Gaurav Kalyani works as Research Associate at the Center for Legislative Education and Research, FLAME University, Pune
Siddharth did primary research and writing; Gaurav contributed to primary research, writing and editing)
Bibliography
Main sources
Yeatts, M. W. M. (1942). Census of India, 1941: Vol. 1. India. Part II: Administration report. Government of India Press.
Yeatts, M. W. M. (1943a). Census of India, 1941: Vol. 1. India. Part I: Tables. Manager of Publications.
Yeatts, M. W. M. (1943b). The Indian census of 1941. Journal of the Royal Society of Arts, 91(4634), 182–194. https://www.jstor.org/stable/41363062
Provincial reports
Census Commissioner for India. (1940–1941). Sampling: Census Commissioner’s office file notes [Unpublished file notes]. Census of India, 1941, Vol. 1.
Dube, N. K. (1944). Census of India, 1941: Central India States. Rewa State: Report and tables. Rewa State.
Elwin, (n.d.). Census of India, 1941: Vol. 2. Madras: Administration report [Typescript]. Government of India.
Gastrell, (n.d.). Census of India, 1941: Vol. 14. Baluchistan: Administration report [Typescript]. Government of India.
Krishna Rao, P. H. (1944). Census of India, 1941: Vol. 23. Mysore. Part III: Administrative. Government Press.
Lambrick, (n.d.). Census of India, 1941: Vol. 12. Sind: Administration report [Typescript]. Government of India.
Marar, K. W. P. (1940). Census of India, 1941: Assam census handbook. Part I: Preliminary. Assam Government Press.
Mazhar Husain. (1945). Census of India, 1941: Vol. 21. H.E.H. the Nizam’s Dominions (Hyderabad State). Part I: Report. Government Central Press.
Menon, (n.d.). Census of India, 1941: Vol. 19. Cochin State: Administrative volume. Government of Cochin.
Mukerjea, S. V. (1941). Census of India, 1941: Vol. 17. Baroda: Some results of the census of 1941 with tables and subsidiary tables. Baroda State Press.
Narayanan Tampi, A. (1942). Census of India, 1941: Vol. 25. Travancore. Part IV: Administration report. Government Press.
Powindah census: North-West Frontier Province and Baluchistan. (1942). Manager of Publications.
Scott, I. D. (1942). Census of India, 1941: Vol. 10. North-West Frontier Province. Appendix: Trans-border areas, report and tables. Manager of Publications.
Suchdev Singh. (n.d.). Patiala State census report, 1941. Bhupindra State Press.
Superintendent of Census Operations, Central Provinces and Berar. (1940–1941). Census of India, 1941: Vol. 8. Central Provinces and Berar: Collection of census circulars. Government of Central Provinces and Berar.
Wreford, R. G. (1943). Census of India, 1941: Vol. 22. Jammu and Kashmir. Part IV: Administration report. Hallir Government Press.
Comments