Statistics of Democide
Chapter 1: Summary and Conclusions [Why Democide?...]
Other Democide Related Documents On This Site
Table 3.1 presents the sources, estimates, and calculations on Japanese democide in World War II. There is one major omission, however. Democide in China during the Sino-Japanese War that begun in 1937, and merged with WWII in December 1941, is excluded. This democide has been separately calculated in Rummel (1994), and only the total derived there is given in the table (line 386) in calculating the overall democide.
The first part of the table (lines 2 to 42) calculates the number of Japanese that died in Japanese wars, 1937 to 1945. This amounted to 1,771,000 to 3,187,000 Japanese, most likely 2,521,000 (line 42). Of this number, 672,000 probably were civilians (line 32), virtually all killed in American air raids (including the two atomic bombs).
The first democide I consider is against prisoners of war and interned civilians (lines 45 to 93). Most of these figures are official, and were presented at the Tokyo War Crimes Trial.
The overall number of POWs and internees killed was about 138,000 (line 93). Since this is largely based on official figures released shortly after the war, I give no high and low. For nations releasing figures on both the total number of POWs captured and the number dying in Japanese captivity, the POW death rate averaged nearly 29 percent.
The table next lists estimates of the total Asian forced laborers who died from Japanese maltreatment. The most notorious case of indifference to the health and welfare of prisoners and forced laborers was the building of the Burma-Thailand railroad in 1942 to 1943. Estimates of those killed, including POWs, are given (lines 97 to 104) in the table. I already included these POW deaths under the POW total (line 93). As for Asian forced laborers working on the railroad, 30,000 to 100,000 died, probably 60,000 (line 105).
I also list forced labor deaths for specific countries, beginning with Indonesia (Dutch East Indies, at the time). How many Indonesian forced laborers were actually conscripted by the Japanese is unknown. Estimates run as high as 1,500,000 (line 110a); even more speculative is the death toll. This varies in the sources from 200,000 to 1,430,000 deaths, with perhaps the most likely figure being 300,000 (the figure "accepted" by the United Nations--line 114).
Information on Korean deaths under Japanese occupation is difficult to uncover (Korea was not invited to participate in the War Crimes Trial). We do know that 5,400,000 Koreans were conscripted for labor beginning in 1939 (line 119), but how many died can only be roughly estimated. Apparently Koreans were better treated than were laborers from other countries, but still their work hours, food and medical care were such that large numbers died (even Japanese coolies forced to work in other countries were so maltreated that many died). This is clear from the 60,000 Korean laborers that died in Japan out of the near 670,000 that were brought there in the years 1939 to 1945 (line 119a). To estimate what the total Korean death toll might be, I give the forced labor death rates for Koreans and Chinese in Japan and forced laborers from or in Indonesia (lines 119b-121). With these as the upper bounds, my reading of Korean history for this period suggests a possible range in the Korean death rate of 5 to 15 percent, with a mid-estimate of 7 percent. These should be conservative rates, given that near 9 percent died in Japan where work conditions can be assumed better than in Korea or Manchuria; and that the rates are much less than half those for China and Indonesia. Even at these low rates, however, the forced labor toll for Korea comes to 270,000 to 810,000 dead in seven years.
Data is equally sparse for Manchuria. From diverse sources it is clear that Japan conscripted over a 1,000,000 forced laborers from Manchuria, which is thereby made the low (line 126); but how many died is unknown. I use the same approach here as for Korea, assuming the death rate for Manchurian laborers to be closer to that for the Chinese forced to labor in Japan (line 127). This gives (line 128) a probably conservative range of 100,000 to 200,000 Manchurian dead over seven years.
For the Burma-Thailand railroad, and for Indonesia, Korea, and Manchuria, 600,000 to 1,610,000 Asian forced laborers died (line 131). Note that this is probably very conservative, even were some of the estimates too high for a few of the countries included. No figures, even a basis for rough estimates, are available in the sources for Malaysia, Indochina, and Burma (except for those dying while working on the Burma-Thailand railroad). Yet, based on Japanese behavior in other countries, many forced laborers from these countries also must have died elsewhere.
The table next presents estimates on Japanese massacres and atrocities in occupied countries and territories. I make two listings of these. The first (lines 134-217) is of those countries or territories for which no total or subtotal is available or can be calculated; the second (lines 228-289) is of countries and places for which a country total can be determined. Considering now the first list, in most cases the existence of a massacre was alleged, without any estimate of the number killed being given. Where such estimates were available, they add up to 8,089 killed (line 223), or an average slightly over 1,300 per incident.
A problem is how to handle the forty-three massacres for which there is a question mark (line 221). For the six massacres in this list for which there are estimates, the average is 1,348 killed. In China, where many more reports of the number massacred were available, the average killed for all the low estimates was 800.
Turning to the next list, there is enough information given about the countries or territories included here for me to a country-by-country estimate of those killed. The first territory tabulated is Indochina (lines 229-240). From information (line 244) that 5.5 percent of the European population died we can estimate for the French population (lines 242-243) at that time that at least 1,320 were killed.
Similarly, from the Indochinese (Vietnamese, Laotians, and Cambodians) population (lines 247 and 248) and a report that 2.5 percent died, we get a high estimate of 575,000 dead (line 250). This is a very large number, and appears to grossly exaggerated the total massacred (and is thus made a high). Many may have died from other causes, such as local famines, for which the Japanese were not wholly responsible. In the Philippines, where the Japanese were especially prone to massacre inhabitants due to the pervasive guerrilla war being waged against them, the democide rate high was almost 14 murdered per thousand by the Japanese (line 348); for China a high of near 30 per thousand of the population in occupied territory probably were similarly killed;
From information available in the sources, there appears no special reason to believe that the Japanese treated Indochinese with greater brutality than they did the Chinese or Filipinos; indeed, overall, they may have been much better treated than Filipinos. Accordingly, I assume the low based on the Chinese democide (21 percent that of the Philippines) is that for Indochina; I base the high on the native population dead (line 250); and I calculate the mid-value as the average between this and the Chinese and Filipino bases. This gives (line 250c) a range of 68,000 to 575,000 Indochinese killed, with a most probable estimate of 207,000, considerably under 2.5 percent of the population.
I treat Indonesia next (lines 253-284). Numerous massacres were reported in Indonesia, and those for which estimates of the number of people killed are available total 13,100 to 15,290 dead (line 285).
Given the population of Indonesia (line 289), this estimate can be checked by calculating an overall toll based on the Japanese democide in China and the Philippines (lines 292 and 293). Moreover, we have the one estimate that a total of 4,000,000 Indonesians died in the war from all causes (line 296). Finally, adding the forced labor and massacre deaths (line 297) gives a total that can be compared with these above figures. Clearly, juxtaposed to lines 292 to 296, the total seems well in line and I thus take the estimate of massacre and atrocity deaths (line 286b) as reasonable.
The next territory to consider is Malaya (lines 301-311). The Chinese living in Malaya particularly suffered from the Japanese occupation, at least 37,000 of them being executed (lines 301, 302, 304). In order to get some overall figures, estimates based on the China and Philippine democide rates are given (line 318 and 319). Also included with these bases is a high of 100,000 killed given by Malayan officials. Noting this high and that the few available figures already total 38,000 killed (line 312), I consolidated these figures into a range of 55,000 to 100,000 killed, with a mid-estimate of 83,000.
Manchuria, that follows (lines 326-329), is considered separately from China. It had already been taken over and administered as a Japanese colony (in effect) prior to the Sino-Japanese War. Very little information is available in the sources on massacres and atrocities in the territory during the war, although the Tokyo War Crimes Trials and Japanese behavior elsewhere suggest that many such took place.
As to Okinawa (line 333), we know about how many Okinawan civilians lost their lives during the American invasion of the Island, and some of these intentionally were killed by Japanese troops or ordered to commit suicide, but the democide is unknown and cannot even be guessed.
On the Philippines (lines 336 to 342), better estimates than for any other territory are available. After the Japanese defeat on the Islands, special American units tried to document the massacres committed by Japanese forces and secret police. Still, different and inconsistent figures are given (lines 336-340), taking into account the number of American civilians (line 336) and American and Filipino POWs (lines 73, and 78-82) captured and killed. Most likely this is due to the difficulty of estimating the toll of many recorded and unrecorded massacres and atrocities. In any case, a minimum of 90,000 Filipino civilians killed seems solid. No high is readily suggested, so I invoke the procedure of doubling the low and taking the mid-value as a third higher than the low (line 343). For the high and mid-value, these are prudent procedures.
No information is available in the sources on Philippine forced labor, yet judging from Japanese occupation behavior elsewhere, perhaps tens, if not hundreds of thousands of such laborers must have been conscripted, many probably dying. I assume these numbers to be absorbed into the democide's high and mid-values, while presuming that the low involves no forced labor deaths at all. All this gives a most probable democide rate of two to three Filipinos per thousand per year (line 349).
Next to consider is Saipan (line 352), where an unknown number of Japanese civilians were killed by Japanese troops or ordered too commit suicide. No basis for estimating these numbers is given in the sources.
Estimates of democide in Singapore's follows (lines 355-361). The best figure is of 150,000 Asians killed by the Japanese secret police (line 356) and this is made the low. I make the high twice the low, and the mid-value a third higher. If anything, this procedure may underestimate the real total. The low excludes at least 5,000 Chinese rounded up by the Japanese Army and killed in February, 1942 (line 359); and other Army massacres undoubtedly occurred, some of which are listed in the table (lines 355, 357, 358, and 360). And no forced labor deaths are included (although some may have been picked up by the Asian toll on the Burma-Thailand railroad--line 105).
Finally, there were 590 American civilian victims (line 365).
Adding together all these massacre and atrocity figures (line 369) gives a total of 413,000 to 841,000 killed.
Here and there in the sources are hints of local Japanese caused famines in one territory or another, but only for India and Indochina are estimates of famine deaths given. That for India is blamed on Japanese policies in Burma that upset the rice supply, but there is not enough information to assume that these policies were pursued with a reckless or knowing disregard of a famine that might be produced. For Indochina, when the food supply was disrupted by US air raids and a naval blockade, the Japanese knowingly diverted to their forces rice needed by the inhabitants for survival. Without more information, however, how much of this famine to blame on the Japanese is a guess. Accordingly, a low of 25 percent responsibility is estimated (line 378), which seems prudent enough.
From all the assumptions, consolidations and calculations made, the overall Japanese democide in World War II can now be estimated (lines 381-384), and Japanese democide in China included (line 386). This gives a total democide of 3,056,000 to 10,595,000 with a likely mid-total of 5,964,000 people killed.
How credible is this range and most probable democide? To assess this, the total population controlled by Japanese forces is first calculated (line 400), and after comparing this to one such figure (line 401) given in the sources, a range of population figures is consolidated (line 402). These population figures are then used to calculate the death toll using the Chinese and Filipino democide rates as the basis (lines 405 and 406). Since totals are now being compared, these bases are calculated for the full 1937 to 1945 period. The total democide figures are reproduced below the two resulting ranges (line 409) for comparison. As can be seen, the overall democide total for Japan is close to that one would get estimating it from Japan's democide in China or the Philippines. This implies that the total democide figures are not inconsistent from one territory or country to another, but that there was a pattern of Japanese democide throughout that is captured by these results. This pattern is there regardless of the many assumptions, estimates, and calculations involved, and even taking into account that in some cases a China and Philippine bases was used to estimate a country or territory's massacres and atrocities (most forced labor and all POW deaths were determined independently). And this relative consistency lends credibility to the democide totals.
With these totals I calculated (lines 412 and 413) the overall and annual democide rate (for the occupied population, at its greatest extent). As can be seen, nearly one out of every one-hundred people controlled by Japan was murdered, or almost three per thousand people per year.
* From the pre-publisher edited manuscript of Chapter 3 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. Pritchard and Zaide (1981).
2. See Rummel (1991, Table 6.A, line 83).
3. See Rummel (1991, Table 6.A, line 128).
4. Aside from the Tokyo War Crime Trial testimony, I had help in compiling these statistics from J. A. Willinge, Secretary-General of the Netherlands' Institute for the Tropics and B. R. Immerzeel, the scientific assistant to the Dutch Committee on the Resistance in the Dutch East Indies, who consulted for me the Dutch National Archives and other sources unavailable to me.
Go to top of document