How many were killed in the 1971 war?

Those interested in this page, may also be interested in: 

- Tribunal judgements and the death toll in 1971

- 1971 war time death numbers, further thoughts

- Journalists reporting on the numbers of deaths

If anyone has any additional information or references that you think should be added to this post relating to information on the numbers of those who died/killed in 1971, please e-mail me or leave a comment at the end.

-----------------

This is my third of four posts looking at the Tribunal's 3 October 2011 order charge-framing Sayedee on 20 counts of international crimes involving genocide and crimes against humanity relating to the 1971 war of independence. The first post looking at the charges against Sayedee is here. The second post looking at some legal issues discussed in the tribunal is here.

The introduction to the 3 October order stated, in its historical introduction that 'As a result' of the actions of the Pakistan military, and the role of the collaborators,

'3 million (thirty lacs) people were killed more than 200,000 (two lacs) woman raped, about 10 million (one crore) people deported to India as refugees and million others were internally displaced...'.

This statement by the tribunal provides an opportunity to look at the question of how many people died as a result of the 1971 war - a controversial issue within Bangladesh. In certain nationalistic circles, to raise what I consider to be legitimate questions about the 3 million figure can draw strong emotions.

The tribunal in its order does not provide or refer to any evidence or material on record to support the figure of 3 million, treating it as a historical fact. (To read about what subsequent judgments of the Tribunal say on this click here)

Although this number is treated as though it is an official government figure, there is as far as I can see little evidence, if any, to support it.

It is not uncommon of course for there to be widely different estimates of the numbers who died as a result of a war. People involved in any conflict have reasons to either over or under estimate the numbers who died. Moreover, it is difficult, in any post conflict situation, to make accurate estimates of the numbers of dead, as even recent wars in Iraq, for example, testify. Sarmila Bose's criticism in her recent book (see below) of the failure of the Bangladesh government after the war to have undertaken proper studies that would have provided more reliable estimates of the number who died do seem rather unfair.

Although writing about this subject is like sailing in perilous waters, I think it is fair to make the following points on this question.

(a) Where did the 3 million figure come from?
It seems that the figure of 3 million may first have appeared in a daily newspaper, Purbodesh on 23 December 1971, when an editorial stated that 'enemy occupation' had resulted in the deaths of 'about 3 million innocent people.' Subsequently, the Soviet newspaper, Pravda, is reported to have published a news report quoting this figure, and then the figure was more widely distributed in the Bangladesh media, when the news agency ENA, picked up the Pravda piece and distributed an article stating that 'The communist party newspaper Pravda has reported that over 30 lakh person (3 million) were killed throughout Bangladesh by the Pakistan occupation forces during the last nine months.' This news agency article was apparently widely covered in Bangladesh newspapers, including for example, the Bangladesh observer, on 5 January 1972.

The figure however only became widely popularised when Sheikh Mujib on 10 January, the day he returned to Bangladesh, stated, "Three million people have been killed. I believe there is no parallel in the history of the world of such a colossal loss of life for the struggle of freedom.' He is also said to have made similar comments around this time in a number of international TV interviews.

Significantly, Sayyid A Karim, Bangladesh’s first foreign secretary, supports the contention that Mujib picked the figure of 3 million up from these sources, when he wrote in a footnote in his autobiography of Sheikh Mujib that, 'As for the number of Bengalis killed in the course of the liberation war, the figure of 3 million mentioned by Mujib to David Frost in January 1972 was a gross overstatement. This figure was picked up by him from an article in Pravda, the organ of the communist party of the Soviet Union.’

There is, of course, another story that is often recounted which suggests that Sheikh Mujib made a mistake when he said 3 million died, and that he really meant 3 lakh (300,000) were killed. However, in light of Karim's comment, I am more willing to support the contention that he picked it up from the newspaper reports of the time.

(b) Govt inquiry into numbers
In late January 1972, according to press reports at the time, the then government set up a committee to look into the number of war dead, under the chairmanship of the Deputy Inspector of Police. However no final report was published, and there have been suggestions that the decision not to publish the report was because the inquiry had not found anywhere near the number of 3 million deaths.

(c) Who died?
The question of the number of people who died in 1971 cannot be properly answered without considering what deaths form part of that number.

Sisson and Rose, in their 1990 book, 'War and Secession: Pakistan, India and the creation of Bangladesh' make the point that whatever the figure for the dead may be, '... it is still impossible to get anything like reliable estimates as to (1) how many of these were 'liberation fighters' killed in combat, (2) how many were Bihari Muslims and supporters of Pakistan killed by Bengali muslims, and (3) how many were killed by Pakistani, Indian or Mukti Bahini fire and bombing during the hostilities. One thing is clear - the atrocities did not just go one way, though Benglai Muslims and Hindus were certainly the main victims.'

The distinction between liberation war fighters killed in combat and civilian deaths is an important one. It is also important to recognise that amongst the civilian deaths were Biharis civilians killed by 'Liberation fighters' or their supporters.

In addition, of course, there are many deaths that are the result of war but are not the result of fighting and atrocities - but disease and malnutrition (see below the study by the cholera hospital on this very issue).

(d) Studies and estimates on number of dead

There have been a number of other estimates on the number of people who died in 1971.

(i) The Peace Research Institute in Norway along with Uppsala University in Sweden, have collected information on the numbers of 'battle deaths' in all wars since 1900. Apparently, on the basis of eye-witness and media reports as well as other data, they have estimated that about 58,000 people died in battle in 1971 in Bangladesh.

However, it has proven difficult to clarify the methodology upon which they came to this figure and relying on press reports (if this is what they have done) is clearly a far from accurate method of ascertaining the number of deaths. It should be noted that this figure does not seem to include the numbers of deaths of civilians.*

(ii) More recent research conducted by academics at the Institute for Health Metrics and Evaluation, at the University of Washington in Seattle, and published in 2008 in the British Medical Journal, analysed World Health Organisation population surveys, looking at sibling deaths, to estimate the number of 'direct violent deaths' in different wars in different countries. Their calculation suggest that the number of 'deaths from war injuries' in 1971 was closer to 269,000 – five times the number of the Scandinavian researchers which involved only 'battle deaths'. Their figures range from 125,000 to 505,000. It should be noted this does not include other war related deaths.

The reports also says that the 'estimates presented here should be viewed as conservative', pointing out that the surveys were unable to capture 'families with no survivors'. ****

(iii) Perhaps the most reliable estimate of number of deaths as a result of the war in 1971 is found in a little known study published in 1976 by the Cholera Hospital (now the ICDDRB) in a prestigious journal called 'Population Studies'. One of the article's three authors was Lincoln Chen who subsequently became a very noted public health specialist and is due, I understand, to be honored by the Bangladesh government for his contribution to independence in 1971.

The article looked at changes in population numbers in the rural area of Matlab Bazaar Thana. The Cholera hospital had collected detailed population data, including details of birth and deaths, on this area since 1963, and so was able to compare the population figures collected in 1972 with those collected during the war and prior to it to make an estimate of the number of 'excess' deaths.

It found that in 1971, in this particular area of the country there were 868 'excess' deaths.

Extrapolating, to the whole of Bangladesh it estimated that about 500,000 people had died because of the war. The article does concede, of course, that there are limitations to making such a generalisation - and the article does not suggest that 500,000 can in any way be said to be an exact figure. However, the detailed neutral objective analysis of one particular rural population using data that was collected systematically before, during and after the war makes it an extremely important piece of research.

It should be noted, in fact, that a considerable number of the 'excess' deaths that this study recorded in 1971 were the result of malnutrition and diseases caused by the war, and not deaths as the result of specific violent action by the Pakistan military, and its collaborators. (In fact, since the system of data collection was initially set up in peacetime, without anticipation of war, there was no direct category to put atrocity deaths, so they would have been put in the categories of 'other' or 'general' which both showed significant increase in death rates in 1971).

What this suggests is that, assuming for one moment that the ball park figure of 500,000 dead is about right, then the number of 'atrocity' deaths amongst the excess 1971 deaths will be, according to this study, less than 500,000.

Because of the importance of this study, I am extracting a number of paragraphs (emphasis is added):**

The war in Matlab Bazar followed the chronology of the national conflict. During the March 1971 disturbances, many labourers returned to their village homes in Matlab Bazar. The army made its first appearance in the area in April with an expedition to Chandpur, an inland port about 13 miles south of Matlab Bazar. In June, the thana headquarters were raided by the army, and in July the army instituted periodic counter-insurgency operations in the area. This precipitated out-migrations from the thana. As the conflict progressed, the insurgents grew in strength and the intensity of the hostilities increased. By November, insurgents controlled the entire thana. .....

The conflict undoubtedly had different impacts on different geographical areas within Bangladesh. Indi- genous food production and the distribution of imported foods varied between one locality and another. Moreover, those areas containing large numbers of Hindus, the religious minority, or those in which fighting actually took place were more severely affected. Matlab Bazar thana was not at the extreme on any of these factors. However, it is unique in that the population was served by a diarrhoea treatment unit; this unit provided care to about 2,500 in-patients a year. The base crude death rate in Matlab Bazar of 15-3 per 1,000 is thus probably lower than in the nation as a whole, commonly estimated at 17 0. Withdrawal of diarrhoea treatment during the conflict could have led to a further increase in the death rate. Thus, the impact of the war in Matlab Bazar, while not representative of the nation as a whole, illustrates and reflects in a qualitative sense the consequences of the civil war. Accepting these limitations, it appears useful to estimate the probable overall demographic impact of the war on Bangladesh. Assuming a national population of 70 million in 1971-72 and the Matlab Bazar base vital rates, there would have been about 3-20 million births and 1 07 million deaths in 1971-72 in the absence of the war. The decline in the crude birth rate by eight per cent in 1972-73 suggested that there were approximately 260,000 births either averted or postponed by the conflict. With increases in the death rate in 1971-72 and 1972-73, of 40 and six per cent respectively, this implies an overall excess number of deaths of nearly 500,000. By any standard this was a major disaster. 

The full article can be downloaded here

(iv) There do remain respected people who suggest that there is some evidential basis to the 3 million. So for example, Mofidul Huq, a publisher and respected war crimes activist points to the census data of June 1969 which, he says, estimated the population to have been 69.8 million with a growth rate of 2.8%. ‘At this rate the population was supposed to be 80.1 million in 1974 when the census was made. But the actual count in 1974 was 76.4 million. The great number of missing persons indicate the estimate of three million deaths in Bangladesh genocide was not an exaggerated figure,’ Huq told me.

More recently, this argument has been detailed further in a blog [this section was updated on 27 Feb 2013]

It is important to note, however, that the analysis in this blog depends on UN estimates of population - not census data. As one demographer states: these are 'interpolated (estimated averages) calculated by guessing at what the population might be based on the counts for the preceding and following census counts.'  There was a census done in 1974 (not mentioned in the blog in fact) which states that the population was nearly a million higher than the UN estimates (71.5 mil compared to 70.6 mil). Moreover this 1974 census is itself considered by demographers to have been a particularly poorly conducted survey leading to an under-count. It is notable that the adjusted figure for the 1971 census is nearly 5 million more at 76.4 mil. (See, for example: 'Estimates of Recent Trends in Fertility and Mortality in Bangladesh', written by the Committee on population and Demography, National Research Council (1981). This states at p.15: 'Looking at the population as a whole the post enumeration check for the 1974 census indicated an under enumeration of 19.3 per cent in the four major towns and 6.5 percent elsewhere, giving an adjusted population of 76,938,000 for 1 March 1974'

Moreover any gap in the figures could, in any case just as easily be explained by the large number of migrations to India along with depressed fertility.

The point here is really that using (a) estimates of population and (b) poor census data are not a good basis for making an assessment of the numbers of 1971 war deaths.

(v) Another view comes from Dr. MA Hasan, the Convenor of the War Crimes Facts Finding Committee who, since the fall of General Ershad in 1990, has with the aid of a groups of researchers, traveled around the country, 'village to village', to uncover accurate information on the numbers of dead.

Hasan however thinks that 3 million is an ‘exaggerated’ number – with the real figure being closer to 1.2 million. He said to me: ‘We have identified 946 killing fields or mass graves. ... Our research suggests that for every one grave that we have found there are four others which have been built upon or are not accessible. That makes a total of about 5000 graves.’

He goes onto explain that from their experience the average number of bodies found in each mass grave is about 100.

‘If there are 5000 graves with 100 bodies in each that is 500,000. In addition, from talking to villagers throughout the country, we think that the number of bodies buried represents only about 30 per cent of the total number of deaths, with the remainder disposed of into the rivers. Most bodies were washed away. If we add these in the total comes about 1.2 million.’ He however said that the figure of deaths may be as high as 1.8 million.

(The above information was given in a telephone interview undertaken in Nov 2011 when this post was written. It should be noted that in a paper given at Kean University, he is a little more definite about the 1.8 million figure: '[T]here is no doubt that nearly 1.8 million people were eliminated during the war of 1971.' (p.3) - though in this paper he does not set out the basis for his greater certainty.)*****

It is to Hasan's credit that he has tried to undertake a detailed estimate of war dead. However, his figures do remain pretty speculative.

(vi) One should also note the work of MJ Rummel***, who in 1997 published a book about the numbers of deaths in a series of conflicts each of which he calls a 'democide', collating estimates from books, and articles. To read about his work on the 1971 conflict, you can look at this, and the detailed table here.

Below I set out the key part of the table. The first column describes whether the author of the book/report refers to those who 'died' or those who were 'killed' (and includes some other descriptions as well). The next three columns, I think, describes how the author descrive the number (i.e 'at least', or 'as high as'). The next column describes where the number is reported - and the names refer to the authors of the book/report. And the fifth column, refers to how the author describes where he or she sourced the number.

Description in Source about what number is being measured	Estimate in book or report			Author	Specific source if known
	Low	Medium	High
Killed		50,000		Chauderi (from Time Magazine)	Statement by Bhutto in March 1972
Dead	300,000		500,000	Sisson and Rose	From an Indian official who ‘held a responsible position on the issue of Bangladesh’
Killed	500,000			Glasser & Possony	Source not stated
Killed	500,000		3 mil	Pike	Source not stated
Dead	1.25 mil		3 mil	Harff and Gurr	Bengali nationalist
Killed	1mil			Chalk and Jonassohn	Source not stated
Dead	1 mil			Chowhdury	Source not stated
Victims		1 mil		Chaudhuri	From The Times, Guardian and Le Monde
Genocide Killed			2 mil	Mascheranos	Source not stated
Dead	2 mil		3 mil	Stohl	From Robertan Cohen, John Simpson, Jana Bennett
Killed	2.2 mil		3 mil	Kuper	Source not stated
Killed	3 mil			Chaudhuri	From Pravda’s Dhaka Correspondent
Killed	3 mil			Chaudhuri	From Sheikh Mujib interview
Killed		3 mil		Chaudhuri	From Bangladesh government report
Killed			3 mil	Chaudhuri	From surveys of Bangladesh newspapers
Butchered		3 mil		Roy	Source not stated
Victims		3 mil		Sisson and Rose	India’s figures of Pakistan atrocities
Killed		3 mil		Kuper	Source not stated
Slaughtered		3 mil		Muhith	Source not stated
Killed		3 mil		Payne	Source not stated
killed		8 mil		Chandra	Source not stated

He concludes after averaging out all the different estimates given in different sources - but ignoring the first and the last, that:

'I give a final estimate of Pakistan's democide to be 300,000 to 3,000,000, or a prudent 1,500,000'

It should be noted that all Rommel does is to take the numbers set out in different books and newspapers and averages them out. He does not assess their value or integrity, which is of course  a serious weakness in his work. He also does not mention the Matlab research, the most authentic of the research that has been undertaken into this question.

One should perhaps also note that Rommel does the same thing with the 'Bengali massacres of non-Bengalis, primarily the Biharis' and concludes a 'a range of 50,000 to 500,000 [were] killed, more likely 150,000.' Again, there are similar weaknesses with these estimates.

(vii) More recently, the academic Sarmila Bose has, controversially entered into the fray. In her book, 'Dead Reckoning' where she states:

'From the available evidence discussed in this study, it appears possible to estimate with reasonable confidence that at least 50,000 - 100,000 people perished in the conflict in East Pakistan/Bangladesh in 1971, including combatants and non-combatants, Bengalis and non-Benglais, Hindus and Muslims, Indians and pakistanis. Casualty figures crossing one hundred thousand are within the realm of the possible, but beyond that one enters a world of meaningless speculation.'

Without getting into the many critiques that people have about this book, on this particular point it is really difficult to see how she can say anything about the numbers of dead 'with reasonable confidence.' Her book does raise some legitimate queries about how particular 1971 incidents are viewed, and it may well be the case (as she suggests) that popular notions about the numbers who died at for example Shankharipara (in March 1971) and at Chuknager (in May 1971) have been exaggerated, but at the same time none of the research that she did, really allows her to make any conclusions about the numbers who died.

(viii) In her book, Bose quotes an interesting paragraph from the book by Simmon and Roee (mentioned above). This states that:

"India set the number of victims of Pakistani atrocities at three million and this is still the figure usually cited. We interviewed two Indian officials who had held responsible postions on the issue of Bangladesh in 1971. When questioned about the actual numbers of deaths in Bangladesh in 1971 attributable to the ciil war, one replied 'about 300,000'. Then when he received a disaproving glance from his colleague, he changes this to '300,000 to 500,000'"

It is interesting perhaps to note that I have heard, from a number of different sources, that the United State's unofficial view is that 300,000 people died in the 1971 war.

(ix) One should of course finally mention the Hamoodur Rehman Commission which was set up after the war by the new Pakistan government of Zulfikar Ali Bhutto. It remained classified for a long time, but recently most of it has been published in Pakistan. Chapter 2 of the report stated that:

"According to the Bangladesh authorities the Pakistan army was responsible for killing three million Bengalis and raping 200,000 East Pakistan woman. It does not need any elaborate argument to see that these figures are obviously highly exagerated. So much damage could not have been caused by the entire strength of the Pakistan Army then stationed in East Pakistan, even if it had nothing else to do." (para 32)

Its own estimate was much lower. It stated:

".. the latest statement supplied to us by the GHQ shows aproximately 26,000 persons killed during action by the Pakistan Army. This figure is based on the situation reports submitted from time to time by the Eastern Command to General Headquarters." (para 33)

Make of this what you will. But I think this figure, for obvious reasons, has as much credibility as the 3 million figure: Not very much at all!

(e) Number of woman raped [this section was updated on 20 March 2013]
On the question of the number of woman who were raped in 1971, the figure of 200,000 given by the tribunal has also it seems come to be an 'official' figure. The figure appears to have come from an estimate given by the Australian doctor, Dr Geofrey Davis, who worked in Bangladesh from March 1972 for about six months. There is therefore a more legitimate source for this figure than perhaps the 3 million estimate for the number of people died.

Much less has been written about the issue of rapes in Bangladesh but I think it is fair to see there is doubt about the accuracy of this figure. The Australian based academic, Bina D'Costa, who is amongst the few who has researched this area in some detail, thinks that this number is ‘too high’. From interviews of those involved in the abortion and the adoption programme that took place in Bangladesh after the war, she considers that ’25,000 forced pregnancies is correct’, though she accepted this may be a ‘rather conservative estimate.’ Those 25,000 cases consist of rape and forced impregnation of women and girls kept in rape camps or elsewhere (like bunkers, or in some cases in houses), who later either approached or were taken to the women's welfare division. ICRC documents refer to these cases and note that at least 25,000 cases of pregnancy from abduction, rape and forced marriages occurred during the war.

This figure however does not include those who were raped in their houses or elsewhere; who were then taken overseas (in most cases to India) by families for abortion or adoption. It also, doesn't include rapes and resulting pregnancies of Bihari women either.

(f) Do numbers matter?
Does it really matter how many people actually died in 1971?

It is perhaps important to note from the beginning that the actual number of deaths is not relevant to whether the particular accused at the International Crimes Tribunal have committed the offences of genocide or crimes against humanity to which they have been accused.

Moreover, whether 3 million, 300,000 or indeed even 30,000 were killed, the number of deaths in 1971 was very very large. And no-one can really deny that. There is enough substantiated evidence to suggest that whatever the exact number of deaths, a very large number of civilians were killed.

Yet, at the same time, arguably it is important for the sake of accuracy that people do not claim that a particular number of people died - whether it is too high or too low - which has no basis at all in the evidence.

As a result, coming back to the tribunal's remark in its 3 October order about the number who died, it may well have been preferable for it not to have mentioned these particular figures. Maybe the prosecution will provide evidence to support this figure in the course of the trial but, as yet, it has not done so.

------
* 7 Nov 2014: Unpublished information has recently been brought to my attention that this research dealt with 'battle deaths', not 'war time deaths'. A number of clarifications has therefore been made to these two paragraphs. Thanks to Ishtiaq Rouf for this clarification. I would suggest that the people responsible for this research please put this material in the public domain, or provide links if it is already  in the public domain.

** 7 Nov 2014: This extract from the study was added, along with a link to the article itself.

*** 15 Nov 2014: This section on Rummel's research was added.

**** 17 Nov 2014: Clarification made to explain that the study involved only estimates of 'violent war deaths' and not all war-related deaths, and also that it does not include data on families where there were no survivors.

***** 18 Nov 2014: A link to a paper given by Hasan, which is undated. Thanks to Muhammad Nazmul Islam for this information.