Glasnost', the opening to scholars of hitherto secret archival material, has made possible a marked reduction of our area of ignorance. Combined with new demographic data, especially that relating to the suppressed census of 1937, it enables us to make with fair confidence an estimate of the likely number of abnormal deaths in the thirties and of the numbers in prisons and camps at various dates through 1950. Two articles of mine published in Soviet Studies (April and October 1990) presented some of the data. New information enabled me to modify some of the conclusions of the first article, and since then still more archival material has seen the light of day. So this would seem to justify a new and more comprehensive (and comprehensible) paper.

Before plunging into detail, several warnings are in order. Firstly, the archives themselves for the years of the terror, even those intended to be secret, by no means always provide a reliable source of information. A good example of this is an official letter about the 1937 census from Kurman, a census official, written shortly before his own arrest; in this letter he did not dare mention the word famine as having affected the population in 1933 (though, as we shall see, he used suggestive circumlocution) and, in referring to Kazakhstan, he plainly exaggerated the number of Kazakhs who fled from the USSR to conceal the number who died (or he himself was the victim of false reporting from Kazakhstan).