Census official count of population of a country. Medieval governments, both eastern and western, had conducted occasional censuses for taxation and military call-ups. Dooms Day Book or Domesday Book, a survey of land and landed people conducted by King William I in 1086 AD, is considered to be the earliest recorded census in history.

The first census undertaken in India was Emperor akbar's survey and settlement of 1582 AD which had made a detailed enumeration of mouzawari rent-rolls. The survey known in history as Todar Mal's Bandubast recorded rent-payers, land under cultivation, categories of land according to productivity, rent structure of various categories of land, particulars of landholders, and so on. Though Bengal was not yet wholly under Akbar's kingdom, Todar Mal made a theoretical survey of Bengal sarkars as well. He enumerated the revenue rolls of seventeen Bengal sarkars or provinces.

Census began to obtain its modern form in the late eighteenth century. The first modern census was undertaken by the US government in 1790. The first British census on a decennial basis took place in 1801. Since then census science has developed into a unique exercise involving decennial assessment of socio-economic trends. In 1801, there was also a notional census of Bengal districts. Collectors, magistrates, and judges of Bengal districts were required to return estimated population of districts within their respective jurisdictions. The method of the census was to enumerate population of respective districts, taking five heads on an average for every household.

In the 1840s and '50s, there were several surveys and censuses enumerating revenue paying and rent-free estates as well as the population of Bengal.

The thakbast survey (1840s-1850s), which fixed the village boundaries of Bengal districts for the first time, prepared a sketch map for every mouza and made an enumeration of the village population, which was recorded in the map of the concerned mouza. The map contained a table giving the mouza's statistical details such as, households, population (based on average five heads per household), occupation, and the Hindu-Muslim break-up of village inhabitants. The chart also contained environmental enumeration such as livestock, cultivated land, wasteland, jungle-land, marshes, rivers and canals, bridges and culverts. These records provided basic material for w w hunter when he compiled his 20-volume Bengal gazetteer entitled Statistical Account of Bengal published in London in 1876.

The first decennial census in Bengal was undertaken in 1872. But the attempt failed to achieve its objectives. People, suspicious of the population count, did not cooperate with the enumerators. However, this census led to the startling revelation that Bengal was a Muslim majority province. The second and third censuses of 1881 and 1891 fared much better, though earlier defects could not be entirely eliminated. According to demographers the censuses of the first three decades of the 20th century were fairly reliable.

The censuses of 1931 and particularly of 1941 lost their credibility due to fabrication of census data. Hindus and Muslims, motivated by communal enthusiasm, tended to submit false returns in favour of their respective communities. The census of 1951, the first census for East Bengal (East Pakistan), suffered weaknesses of a different kind. Recent two-way migration trends made the population count for many districts useless. The census of 1961 is regarded as the most reliable count undertaken since 1901.

The decennial census is now losing its ground to modern information technology. computer technology is now capable of generating, updating, and retrieving data of all kinds with exactness and accuracy. No traditional census, how carefully it may be managed, can approach the exactness and variety of computer generated census data.

Table Variation in population since 1872

Source: Census of India, 1931, and 1941; Bangladesh Population Census, 1991, 2001  [Sirajul Islam]