APPENDIX ONE
Scholars have long debated the precise numbers of victims of the terror of the 1930s. Their efforts have produced a wide gap, often of several millions, between high and low estimates. Using census and other data, they have put forward conflicting computations of birth, mortality, and arrests in order to calculate levels of famine deaths due to agricultural collectivization (1932–33), victims of the Great Terror (1936–39), and total “unnatural” population loss in the Stalin period. Some have posited relatively high estimates while others working from the same sources have put forth lower totals.1 Each side has accused the other of sloppy or incompetent scholarship, and the conversation has often been marked by an unseemly and harsh tone.
Soviet secret police documents are now available that permit us to narrow the range of estimates. These materials are from the archival records of the Secretariat of GULAG, the Main Camp Administration of the NKVD/MVD (the USSR Ministry of the Interior), housed in the formerly “special” (closed) sections of the State Archive of the Russian Federation (GARF). The data are summarized in table 1, page 67.2
Archival data show that the total camp and exile population seems to have been slightly below 4 million before the war. Were we to extrapolate from the fragmentary prison data we do have, we might reasonably add a figure of 300,000–500,000 for each year as well as an additional contingent of about 200,000 exiles other than “kulaks,” to put the maximum total detained population at around 4.5 million in the period of the Great Purges.3
Mainstream published estimates of the total numbers of arrests in the late 1930s have ranged from Dmitri Volkogonov’s 3.5 million to Robert Conquest’s 5–8 million to Olga Shatunovskaia’s nearly 20 million.4 The new archival materials suggests that Volkogonov was closest to the mark. A 1953 NKVD statistical report shows that 1,575,259 people were arrested by the security police in the course of 1937–38, 87.1 percent of them on political grounds. Of those arrested by the secret police, 1,344,923, or 85.4 percent, were convicted. (The contrast is striking with the period 1930–36, when 61.2 percent were arrested for political reasons and 61.7 percent of those arrested were eventually convicted, and especially with the years 1920–29, when 58.7 percent of security police arrests were for political reasons but only 20.8 percent of those arrested were convicted.)5
To be sure, the 1,575,259 people in the 1953 report do not include all 1937–38 arrests. Court statistics put the number of prosecutions unrelated to “counterrevolutionary” charges at 1,566,185, but it is unlikely that all persons in this cohort count in the arrest figures.6 Especially if their sentences were noncustodial, such persons were often not formally arrested. 53.1 percent of all court decisions involved noncustodial sentences in 1937 and 58.7 percent in 1938, so the total of those who were executed or incarcerated gives 647,438 persons in categories other than “counterrevolution.”7 It therefore seems likely that the total number of arrests in 1937–38 is not too far from 2.5 million.
Although we do not have exact figures for arrests in 1937–38, we do know that the population of the GULAG camps increased by 175,487 in 1937 and by 320,828 in 1938 (it had declined in 1936). The population of all labor camps, labor colonies, and prisons on 1 January 1939, near the end of the Great Purges, was 2,022,976. This gives us a total increase in the camp and prison population in 1937–38 of 1,006,030. One must add to this figure the number of those who had been arrested but not sent to camps either because they were released sometime later, or because they were executed.
Popular estimates of executions in the Great Purges of 1937–38 vary from 500,000 to 7 million.8 We do not have exact figures for the numbers of executions in these years, but we can now narrow the range considerably. According to a Gorbachev-era press release of the KGB, 786,098 persons were sentenced to death “for counterrevolutionary and state crimes” by various courts and extrajudicial bodies between 1930 and 1953.9 According to the NKVD archival material currently available, 681,692 people were shot in 1937–38 (compared with 1,118 persons in 1936).10 These archival figures, coming from a statistical report “on the quantity of people convicted upon cases of NKVD bodies,” include victims who had not been arrested for political reasons, whereas the KGB press release concerns only persons persecuted for “counterrevolutionary offenses.”11 In any event, the data available at this point make it clear that the number shot in the two worst purge years was more likely a question of hundreds of thousands than of millions. The only period between 1930 and the outbreak of the war when the number of death sentences for non-political crimes outstripped those meted out to “counterrevolutionaries” was from August 1932 to the last quarter of 1933.12
Aside from executions in the terror of 1937–38, many others died in the regime’s custody during the 1930s. If we add the figure we have for executions up to 1940 to the number of persons who died in GULAG camps and the few figures we have found so far on mortality in prisons and labor colonies, then add to this the number of peasants known to have died in exile, we reach a figure of 1,473,424 deaths directly due to repression in the 1930s.13 If we put at hundreds of thousands the casualties of the most chaotic period of collectivization (deaths in exile, rather than from starvation in the 1932 famine), plus later victims of different categories for which we have no data, it is likely that “custodial mortality” figures of the 1930s would reach 2 million: a huge number of “excess deaths.” The figures we can document for deaths due to repression are inexact, but the available sources suggest that we are now in the right ballpark, at least for the prewar period.
Accurate overall estimates of numbers of victims are difficult to make because of the fragmentary and dispersed nature of recordkeeping. Generally speaking, we have runs of quantitative data of several types: on arrests, formal charges and accusations, sentences, and camp populations. But these “events” could have taken place under the jurisdiction of a bewildering variety of institutions, each with its own statistical compilations and reports. These agencies included the several organizations of the secret police (NKVD troikas, special collegia, or special conferences (osobye soveshchaniia), the procuracy, the regular police, and various types of courts and tribunals. No single agency kept a master list reflecting the totality of repression, and great care is necessary to untangle the disparate events and actors in the penal process.
Further research is needed to locate the origins of inconsistencies and possible errors, especially when differences are significant. One must note, however, that the accuracy of Soviet records on much less mobile populations does not seem to give much hope that we can ever clarify each problem. For example, the Central Committee gave quite different figures in two documents that were compiled about the same time for the total party membership and its composition as of 1 January 1937.14 Yet another number was given in published party statistics.15 The conditions of perpetual movement in the camp system created even greater difficulties than those posed by keeping track of supposedly disciplined party members who had just seen two major attempts to improve the bookkeeping practices of the party.16
At times tens of thousands of inmates were listed in the category of “under way” in hard regime camp records, though the likelihood that some of them would die before leaving jail or during the long and torturous transportation made their departure and especially their arrival uncertain.17The situation is even more complicated with labor colonies, where, at any given moment, a considerable proportion of prisoners were being sent or taken to other places of detention, where a large number of convicts served rather short terms, and where many people had been held pending investigation, trial, or appeal of their sentences.18
Moreover, we do not yet know whether camp commandants, in an effort to receive higher budgetary allocations, inflated their reports on camp populations to include people slated for transfer to other places, prisoners who were only expected to arrive, and even the dead. Conversely, though, they may have been equally well advised to report as low a figure as possible in order to secure easily attainable production targets.
Because of these uncertainties, there is still controversy about the accuracy of these data, and no reason to believe them to be final or exact.19One cannot stress enough that with our current documentation, we can posit little more than general ranges (although narrow ones). Still, these are the only data currently available from police archives. Moreover, there are good reasons for assuming that they are not wildly wrong because of the consistent way numbers from different sources compare with one another.20 These figures are analogous to those presented in secret reports to the Soviet Politburo in 1963 and 1988.21 Figures produced by researchers using other archival collections of different agencies show close similarities in scale. Documents of the People’s Commissariat of Finance discuss a custodial population whose size is not different from the one we have established.22 Similarly, the labor force envisioned by the economic plans of the GULAG, found in the files of the Council of People’s Commissars, does not imply figures in excess of our documentation.23 Last but not least, the “NKVD contingent” of the 1937 and 1939 censuses is also consistent with the data we have for detainees and exiles.24