It is not established fact. It is not, in any ordinary sense, a measurement at all. The purpose of this article is narrow and worth stating plainly at the outset. It is not to deny that famines killed millions of Indians under British rule, because they did. It is not to defend the conduct of Lord Lytton in 1877 or of Churchill’s war cabinet in 1943, both of which deserve the criticism they receive. It is to show that the specific figure of 100 million, and its inflated sibling of 165 million, is an artefact of an arbitrary statistical baseline and the wholesale reassignment of disease mortality to deliberate policy. Once you understand how the number is built, you cannot in good conscience keep citing it as a death toll.
Where the number actually comes from
Strip away the rhetoric and the method is simple arithmetic. You take India’s recorded death rate in a given period. You choose a lower rate and call it “normal.” You subtract, multiply by the population, and every death above your chosen line becomes an “excess death” laid at the door of the British Raj.
Everything therefore depends on the line you draw, and Sullivan and Hickel draw two of them. Using India’s own 1880s death rate as the baseline yields roughly 50 million excess deaths, which they call conservative. Assuming instead that pre-colonial India enjoyed the mortality of sixteenth and seventeenth century England, a rate of 27.18 per 1,000, yields 165 million.2
The headline 100 million is not the output of a third calculation. It is an assertion laid over the gap between the two. Their own words are that while the precise number is sensitive to the assumptions made about baseline mortality, it is nonetheless clear that somewhere in the vicinity of 100 million died prematurely.2 Read that twice. The sensitivity is conceded in the first half of the sentence and then overridden by the word “clear” in the second.
Pause on what that means. Their published range spans from 50 to 165 million, a factor of more than three, and where you land inside it is decided entirely by an assumption they cannot verify. They concede the point themselves, admitting that they do not know for certain what India’s pre-colonial mortality rate was.2 A quantity that triples depending on an unmeasured guess is not a finding. It is a sensitivity analysis wearing the costume of a body count, and the round hundred million is the number that reaches the headline because it sounds like history rather than like the midpoint of two speculations.
A counterfactual is not a corpse
The deeper problem is conceptual, and it survives even if you accept every figure in their spreadsheet. “Excess deaths relative to a chosen baseline” and “people killed by British policy” are not the same quantity, and sliding between them is the entire rhetorical engine of the claim.
Excess mortality is a descriptive statistic. It tells you how a period compared with a counterfactual you invented. To convert it into an accusation of policy-induced killing, you must show that the deaths would not have occurred under a plausible alternative government facing the same monsoon, the same pathogens and the same nineteenth century technology. Sullivan and Hickel never do this. They assume it, by holding a single low death rate flat across forty years and treating every deviation above it as a colonial crime.
That assumption is ahistorical. Death rates in the mid to high thirties and low forties per 1,000 were not an Indian anomaly manufactured in Whitehall. They were the ordinary condition of every pre-industrial society on earth, driven by epidemic disease, unclean water and the violent year to year swings of harvest and weather. Tudor and Stuart England, the very society Sullivan and Hickel hold up as India’s healthy alternative, buried a large share of its children before adulthood and saw its own mortality spike with each bad harvest and epidemic.3 A flat baseline mechanically transforms every failed monsoon and every epidemic into “excess” mortality attributable to policy, when much of it is simply the background noise of a world before germ theory, vaccination and mechanised food transport. The model cannot tell the difference between a grain policy and a bacterium, because it was never designed to.
The pandemic the model cannot see
Nowhere does this failure show more starkly than in the single largest mortality event inside their forty year window: the influenza pandemic of 1918 and 1919.
India was the worst-hit country on earth in that pandemic. The demographic historian David Arnold puts the toll at no fewer than twelve million lives.4 The most careful modern reconstruction, using panel data from the Census of India, concludes that mortality in the sampled districts was at most 13.88 million, a figure the authors present as a significant downward revision of the older estimates it replaces.5 The usual range cited for India as a whole is ten to twenty million.6 Those deaths fall squarely inside the window and push up the 1910s death rate that the whole calculation feeds upon. The 1918 influenza was a global viral catastrophe that killed somewhere between twenty and fifty million people across every continent, in independent nations and colonising powers alike, from the United States to Iran to Western Samoa.6 It was not a policy. It was a virus, and a novel one against which no government anywhere had a defence.
Confronted with this, Sullivan and Hickel do not remove the pandemic from their total. They absorb it, leaning on Mike Davis’s suggestion that unusually high Indian influenza mortality was itself a product of British grain requisitioning, and keeping the deaths in the ledger.7 This is the moment the exercise ceases to be demography and becomes advocacy. To fold the deadliest pandemic in modern history into a figure marketed as “policy-induced mortality,” on the strength of a single contested secondary claim, is not rigour. The same window contains the bubonic plague that entered Bombay in 1896 and killed some ten million people over the following two decades, alongside the endemic malaria and cholera that were the ordinary killers of the subcontinent.8 A method that cannot separate a flea-borne bacillus or an airborne influenza virus from a tariff schedule is not measuring the effect of British policy. It is measuring the disease environment of the tropics and sending Britain the invoice.
The romance of the pre-colonial baseline
The larger of the two figures, 165 million, rests on the assumption that India’s natural, pre-British death rate was that of Elizabethan and Stuart England. This is presented as generosity to the data. It is in fact the least defensible move of all.
There is no reliable measurement of pre-colonial Indian mortality, a point Tirthankar Roy has made forcefully: the pre-colonial record survives largely in court chronicles and travellers’ tales, not in the census data the British later collected, so the very baseline the argument needs simply does not exist in comparable form.9 Into that vacuum Sullivan and Hickel insert the mortality of a distant European society and treat the gap as a colonial death toll. This is not conservative. It quietly assumes that, left alone, India would have been healthier than Tudor England, and then charges Britain for the difference.
The assumption also requires a heavily romanticised picture of pre-colonial governance. Mughal India was no stranger to catastrophic famine: the Gujarat and Deccan famine of 1630 to 1632, under Shah Jahan, is thought to have killed around three million people.10 The much-cited contrast, in which benevolent Mughal emperors opened their granaries while callous Britons exported grain, leans on exactly the kind of ruler-praising court chronicle that Roy warns against, documents produced to flatter the sovereign who paid for them. It is worth noting that inflation of famine figures is not unique to the colonial ledger. The famous claim that the Bengal famine of 1770 killed ten million, a third of the province, is now regarded by demographers such as Tim Dyson as very probably an overestimate, resting on Company officials who rarely left Calcutta and on out-migration mistaken for death.11 Numbers assembled from thin data and strong feeling tend to run large, whoever is doing the assembling.
What happens to the conservative figure
The 165 million collapses on contact. The 50 million is the serious number, and it deserves a serious answer rather than a wave of the hand, because it does not depend on Tudor England at all. It uses India’s own recorded 1880s mortality as the baseline, and the recorded rise from 37.2 deaths per 1,000 in the 1880s to 44.2 in the 1910s is real data from the census.2
So take the 50 million and subtract. The influenza pandemic accounts for something in the region of fourteen million. The Bombay plague accounts for perhaps ten million across the two decades after 1896. The famines of 1896 to 1897 and 1899 to 1900 together account for around six million. That is roughly thirty million, and the accounting is necessarily rough, since these figures mix excess-mortality estimates with total-death estimates and were produced by different methods for different purposes. Still, on the most conservative baseline its own authors offer, something in the order of twenty million excess deaths remains after the great epidemics and the great famines have been removed.
Three things account for most of that residue, and none of them is a policy. The first is the baseline decade itself. The 1880s sat in a relatively benign interval between the catastrophe of 1876 to 1878 and the compound crisis of 1896 to 1900, so adopting it as “normal” builds a favourable decade into the denominator and guarantees that the following decades look bad by comparison. The second is endemic disease, above all malaria, which was the largest single killer in India across the whole period and which surged in epidemic years driven by rainfall and irrigation patterns rather than by Whitehall. The third is measurement: vital registration in British India was severely incomplete in the 1880s and improved over the following decades, so part of the recorded rise in the death rate is a rise in the recording, not in the dying.12
What remains after all of that is genuinely contested, and it is the part of the argument worth having. A real fall in living standards, real declines in nutritional status, and real failures of famine relief plausibly did leave India with higher mortality than better government would have produced. That is an argument about millions, conducted in the space where historians actually disagree. It is not an argument about a hundred million, and it cannot be settled by subtracting one assumed number from another.
What is actually true
None of this lets the British record off the hook, and an honest critique gains its force by conceding what is real.
Famine mortality under British rule was severe and repeated. Within the very window Sullivan and Hickel analyse, the famine of 1896 to 1897 and that of 1899 to 1900 together killed on the order of six million people.13 The Great Famine of 1876 to 1878, just before their window opens, killed somewhere between 5.6 and 8.2 million.14 Lord Lytton’s administration behaved with genuine callousness in that famine, allowing grain exports to swell to record levels while restricting relief to labour-camp inmates on rations that were often below subsistence.7 The Bengal famine of 1943, in which perhaps two to three million died, involved real and defensible charges of policy failure reaching to Churchill’s war cabinet.15 The debate over deindustrialisation and declining real wages, drawing on the work of Robert Allen and others, is a serious scholarly argument that deserves engagement rather than dismissal.16
Here is the decisive comparison. Inside the window, recorded famine deaths amount to roughly six million. That is a grave indictment of imperial famine policy. It is also somewhere between a tenth and a fifteenth of the figure now in circulation. The remaining tens of millions are supplied almost entirely by two devices: an invented baseline that reclassifies ordinary endemic mortality as “excess,” and the annexation of the 1918 influenza pandemic and the Bombay plague as acts of British policy. The distance between “several million died in famines that better government could have mitigated” and “a hundred million were killed” is the distance between history and propaganda, and it is manufactured by arithmetic, not discovered in graves.
The number should be retired
The 100 million figure should not be cited as a death toll, because it is not one. It sits in the middle of a threefold range, anchored at its upper end to the mortality of Tudor England, and inflated by counting a global influenza pandemic and India’s endemic disease burden as acts of British policy. Tellingly, when Roy challenged the claim, its authors did not defend the number by showing the deaths were caused; they simply restated that the excess-death arithmetic was undisputed, as though a counterfactual subtraction were the same thing as a grave.7
Cite the famines. Cite Lytton. Cite Bengal in 1943. Cite the real and contested literature on living standards. Those arguments can be defended in daylight. The hundred million cannot, and every time it is repeated as fact, the genuine history of famine under British rule is made easier to dismiss, because it has been chained to a figure that does not survive contact with its own footnotes.
Footnotes
-
Dylan Sullivan and Jason Hickel, “Capitalism and extreme poverty: A global analysis of real wages, human height, and mortality since the long sixteenth century,” World Development 161 (2023). The paper gives a range of 50 to 165 million excess deaths for India depending on the assumed baseline mortality rate, and states that the figure exceeds the combined famine deaths of the Soviet Union, Maoist China, North Korea, Pol Pot’s Cambodia and Mengistu’s Ethiopia.
-
Dylan Sullivan and Jason Hickel, “How British colonialism killed 100 million Indians in 40 years,” Al Jazeera, 2 December 2022. Source of the 100 million headline, the England baseline of 27.18 deaths per 1,000, the recorded rise in the Indian death rate from 37.2 per 1,000 in the 1880s to 44.2 in the 1910s, the authors’ own admission that the pre-colonial rate is unknown, and the formulation that the precise number is sensitive to baseline assumptions but that it is “clear” the toll is in the vicinity of 100 million. Note that the op-ed gives the conservative 50 million estimate for the period 1891 to 1920, while the 165 million figure and the headline claim are both stated for 1881 to 1920.
-
E. A. Wrigley and R. S. Schofield, The Population History of England 1541 to 1871: A Reconstruction (Cambridge University Press, 1981). On the high and volatile mortality of early modern England, the society Sullivan and Hickel adopt as India’s counterfactual baseline.
-
David Arnold, “Death and the Modern Empire: The 1918 to 1919 Influenza Epidemic in India,” Transactions of the Royal Historical Society 29 (2019), which states the epidemic cost at least twelve million lives, more than any other country.
-
Siddharth Chandra, Goran Kuljanin and Jennifer Wray, “Mortality From the Influenza Pandemic of 1918 to 1919: The Case of India,” Demography 49, no. 3 (2012): 857 to 865. Using panel data models and Census of India data, the authors find mortality of at most 13.88 million for the districts in their sample, against 17.21 million on the assumptions of Kingsley Davis (1951), and conclude that Davis’s influential estimate is overstated by at least 24 per cent.
-
Siddharth Chandra and Eva Kassens-Noor, “The evolution of pandemic influenza: evidence from India, 1918 to 1919,” BMC Infectious Diseases 14 (2014): 510, giving an estimated Indian death toll of between ten and twenty million against global mortality estimates of twenty to fifty million.
-
Jason Hickel and Dylan Sullivan, “On the mortality crises in India under British rule: A response to Tirthankar Roy,”
, 7 January 2023. Here the authors confirm that Roy did not dispute the scale of their excess-death estimate, incorporate the 1918 influenza deaths into their total via Mike Davis’s grain-requisitioning argument, and set out their reading of Lytton-era relief policy and Mughal famine relief.
-
On plague mortality, commonly estimated at around ten million deaths in India over the two decades following the 1896 Bombay outbreak. See I. Klein, “Plague, Policy and Popular Unrest in British India,” Modern Asian Studies 22, no. 4 (1988).
-
Tirthankar Roy, “Colonialism did not cause the Indian famines,” History Reclaimed, 18 January 2023. On the absence of comparable pre-colonial mortality data, the unreliability of famine-frequency comparisons drawn from hagiographic sources, and the monsoon ecology of the Deccan.
-
On the Gujarat and Deccan famine of 1630 to 1632, with deaths commonly estimated at around three million. See W. H. Moreland, From Akbar to Aurangzeb: A Study in Indian Economic History (1923), and the “Famine Tales” project, University of Exeter.
-
Tim Dyson, A Population History of India: From the First Modern People to the Present Day (Oxford University Press, 2018), pp. 80 to 81, on the likely overstatement of the traditional ten-million figure for the Bengal famine of 1770, and the conflation of out-migration with mortality.
-
On the incompleteness and gradual improvement of vital registration in British India, and on malaria as the largest single cause of recorded mortality in the period, see Dyson, A Population History of India (2018), chs. 4 and 5, and I. Klein, “Malaria and Mortality in Bengal, 1840 to 1921,” Indian Economic and Social History Review 9, no. 2 (1972).
-
Dyson, A Population History of India, and the demographic work of Pravin and Leela Visaria on the Indian famines of 1896 to 1897 and 1899 to 1900, whose combined excess mortality is estimated on the order of six million.
-
Tim Dyson, A Population History of India (2018), p. 137, citing official estimates of about 5.6 million excess deaths for British-administered provinces in the Great Famine of 1876 to 1878; more recent demographic estimates reach approximately 8.2 million.
-
On the Bengal famine of 1943, with mortality generally estimated at two to three million. See Dyson, A Population History of India (2018), and Amartya Sen, Poverty and Famines: An Essay on Entitlement and Deprivation (Oxford University Press, 1981).
-
Robert C. Allen, “Poverty and the Labor Market: Today and Yesterday,” Annual Review of Economics 12 (2020), on the estimated rise in the share of the Indian population in extreme poverty under British rule; and Allen’s earlier work on Indian real wages from the Mughal period onward.
