Statistics of Democide
Chapter 1: Summary and Conclusions [Why Democide?...]
Other Democide Related Documents On This Site
After a well organized military buildup in East Pakistan the military launched its campaign. No more than 267 days later they had succeeded in killing perhaps 1,500,000 people, created 10,000,000 refugees who had fled to India, provoked a war with India, incited a counter-genocide of 150,000 non-Bengalis, and lost East Pakistan.
This democide is listed in Table 8.1 (lines 26 to 30), which gives an overview of Pakistan's war-dead and democide from 1958 to 1987, the period over which Pakistan has had authoritarian rule, usually military governments. There have been periods without martial law, constitutions have been drawn up, and elections have been held. But these were hardly open and fair, democratic rights and liberties were still absent, and the military still largely controlled major policy from behind the scenes. In Table 8.2 I detail the sources and calculations for the 1971 West Pakistan genocide.
Table 8.1 lists several estimates of other democide during the period of military rule (lines 26-30). It also gives the population figures and final democide rates (lines 82 to 92). I calculate the latter for Pakistan as a whole (lines 84 to 85), East Pakistan (line 88), and for the Awami League in East Pakistan (line 91). Although it would be useful to calculate how the proportion of the different ethnic or religious groups were killed, such as the Hindus or Biharis, there is not enough information in the sources for me to determine a reasonably credible figure.
Turning now to Table 8.2, it begins with estimates of war-dead (lines 1 to 21). While the estimates for the largely military war-dead in the Indo-Pakistan War are reasonable, given the size of the forces and the rapidity of the Indian advance (the war was over in two-weeks), some of those for the civil war or military dead must include democide as well. The Mukti Bahini guerrilla forces numbered about 100,000, the Pakistan army about the same. Some estimates give the civil war or overall military death toll as equal to or even two times the combined armed forces involved (lines 14, 15, 19, and 20). This is not credible, even considering that many civilians were caught up in the war and guerrillas were rapidly replaced by volunteers when they were killed. Accordingly, I have consolidated the civil war estimates at a much more sensible level (line 16) and summed this (line 21) with that for those killed in the Indo-Pakistan War, ignoring the two overly inflated military dead estimates (lines 19 and 20).
In the table I next list a variety of democide estimates (lines 23 to 158). Some of these have to be read carefully. There were two major democides in East Pakistan, one of the Hindu and Moslem Bengalis by Pakistan; the other of the non-Bengalis (largely Urdu speaking Biharis) by the Bengalis. Estimates often do not indicate whether they cover both democides, although the source and context of an estimate may suggest that it is only for that by the Pakistan army. Moreover, some overall estimates may also include combat deaths. With this in mind, I have used various subclassifications for the estimates, including putting those that may include combat deaths under a war and democide heading (lines 170 to 178).
The sources give a number of estimates covering only part of the democide period (lines 47 to 55). I have proportionally projected these to the whole period of nine months [(9 x estimate)/(months covered by estimate)], except for two estimates that are for two months (lines 53 and 53a). Their result would have been 4,500,000 killed, obviously much too high. In any case, these I simply and conservatively tripled to cover the whole period. Regardless, the resulting low and high values (line 56) do not depend on them. The mid-value, however, is the average of all the projected estimates.
Malnutrition, disease, and exposure deaths among the refugees constituted democide. These deaths resulted directly from these pitiful people, largely Hindus, fleeing for their lives before the murderous Pakistan Army. In the table (lines 59 to 62) I give some clearly incomplete estimates of these deaths. They are low enough that I can assume they are included in the estimates of the overall democide.
Turning now to the overall estimates of the Pakistan democide (lines 65 to 79), there are two that are clearly excessively low or high (lines 65 and 79) and that I ignore in the consolidation (line 80). While any leader's admission that his country killed 50,000 people is to confess to a terrible crime, some estimate this number were killed in the first two days of massacres in Dacca alone (line 31). Casting out the unique estimate of 8,000,000 dead hardly need be defended.
Beneath the consolidated overall toll I show my calculation from the partial estimates (line 81). These are rather close. Consolidating both ranges, I give a final estimate of Pakistan's democide to be 300,000 to 3,000,000, or a prudent 1,500,000 (line 82).
Then there is Bengali massacres of non-Bengalis, primarily the Biharis (lines 84 to 158). How much of this was democide (intentional killing by government or its agents) is a question. In this part of the world there is a history of ethnic communal violence and massacres between Hindu and Moslems, and Biharis and Bengalis. However, for the reasons given in Death By Government I will treat these massacres as democide.
The first set of estimates (line 86 to 93) cover only part of the period. And these cannot be projected to cover the whole period, since most of the killing took place in the first two months. Accordingly I simply consolidate them into a minimum of 50,000. Note that two of the lowest estimates are limited in place (lines 86 and 87) and to a body count (line 86).
Many of those who collaborated with the Pakistan Army were killed by the Awami League and its supporters during the civil war and after. Only one estimate is available of this number (line 97), which seems very low given the deep hatred on both sides and the pervasive killing. Accordingly, I give an estimated low of 5,000 murdered (line 99). This is probably very conservative, but I do not have enough information to estimate how much to increase it.
A number of estimates of specific massacres are listed by town, city, or district (lines 102 to 152). Most of these come directly from or are based on the reports of survivors. Some of these are from different sources apparently covering the same massacre (e.g., lines 105 and 106); the great difference in the estimated number of victims is a warning as to how seriously to take them. Since all are from two sources, I summed the estimates for each source (lines 153 and 154), and will use these sums below to derive an overall democide (lines 164 and 165).
There are two overall estimates (lines 157 and 158). One of 500,000 dead is an "impression" Aziz got from interviewing hundreds of repatriates who survived the massacres (line 157). This is my high for the two estimates (line 159).
I can now put together the various estimates of the Bengali--Awami League--democide (lines 162 to 166). Consolidating these, I get a range of 50,000 to 500,000 killed, more likely 150,000.
Finally we can turn to the overall results. First are those estimates in the sources that appear to be covering both war and democide dead (lines 171 to 178), which I consolidate (line 179). Then, we have the various subtotals arrived at previously, which I can now bring together (lines 182 to 184). From these I calculate the total democide (185) and the sum of this and the war-dead figures (186). Then for comparison I show the consolidated total previously determined from the estimates (line 187). The two are close enough such that I can take the sum total (line 186) as the final total (line 188) for this period. Note that its low is lower and its high higher than the estimated total (line 187). I cannot average the two mid-values or take the lower one from the estimate total, because then the subtotals (lines 182 to 184) would not add up to the final total. Were the mid-values radically different, I would have to readjust my previous consolidations and calculation (such as for lines 21 and 167), but the difference does not justify that here.
* From the pre-publisher edited manuscript of Chapter 8 in R.J. Rummel, Statistics of Democide, 1997. For full reference to Statistics of Democide, the list of its contents, figures, and tables, and the text of its preface, click book.
1. Rummel (1994, Chapter 13).