Censusnama

Shivakumar Jolad (opens in a new tab) has been posting some crazy historical datasets lately. I haven’t gotten around to playing with them yet, but boy, oh boy, am I excited. These are really important datasets that help us understand the history of India better.

India is gearing up for Census 2027, the largest administrative exercise in its history, an attempt to count more than 1.4 billion people. It will also count caste, which makes the last time India did that suddenly very relevant again.

That was 1931. The 1931 Census was the last census of undivided India, counting about 352.8 million people across British provinces, princely states and agencies, down to districts, towns, cities and castes. It was also the last to publish full caste data: the 1941 Census, taken in wartime, only tabulated population and religion. So for almost a century, 1931 has been the number people reach for whenever caste is debated. The Mandal Commission leaned on it in 1980, and Census 2027 will be measured against it.

A team at FLAME University’s India State Stories project has turned all this into a treasure trove. Censusnama is a long-form series of more than 40 essays tracing the Indian census from its pre-colonial roots and the first count of 1871–72 through to 2027, on how categories like caste, language and religion were invented and what they did to Indian society, with interactive explorers along the way. They’ve also extracted the tables from every colonial census report from 1871–72 to 1941 and made them downloadable.

The census datasets on Harvard Dataverse are the jewels. The first (opens in a new tab) digitizes the Imperial Tables of the 1931 Census, grouped by theme: population and settlement, demographics, migration and birthplace, education and literacy, language, religion, caste, tribe and race, occupation and livelihood, and health and infirmities. The second (opens in a new tab) links historical and contemporary caste lists: the 1931 caste and religion tables, the 1931 list of “exterior castes,” the Mandal Commission’s 1980 lists, the current central list of Other Backward Classes, and state-wise lists of Scheduled Castes and Scheduled Tribes. Put together, you can trace how a community was classified under the British and how it’s classified for reservations today.

The same team has also mapped the other half of colonial India: the princely states. Their database (opens in a new tab), built from the government’s Memoranda on Indian States, 1939, covers everything from giants like Hyderabad, Mysore and Jammu and Kashmir down to hundreds of tiny estates and talukas in Kathiawar, with each one’s area, population as of 1931, annual revenue, and its ruler’s title, name and caste. My favorite detail from the notes: Hyderabad had its own currency, and 116 Hali Sicca rupees were worth 100 British Indian rupees. There’s also an interactive visualization (opens in a new tab) if you’d rather explore than download.

Even from the descriptions alone, two things stand out about how they’re built. The figures and category labels are kept exactly as printed in 1931, not modernized, and known printing errors are documented rather than quietly corrected, which is exactly how historical data should be handled. Every workbook comes with a codebook, and all three datasets are released under a CC0 licence, free for anyone to use for anything. They’re also a reminder of the point I made about Object 05: datasets that once took years of drudgery to build are suddenly within reach. B.