Hellenic Resources Network
Thursday, 12 August 2004
STATISTICS OF TURKEY’S DEMOCIDE
ESTIMATES, CALCULATIONS, AND SOURCES*
By R.J. Rummel
The infamy of executing this century’s first full scale ethnic cleansing
belongs to Turkey’s Young Turk government during World War I. In their
highest councils Turkish leaders decided to exterminate every Armenian in
the country, whether a front-line soldier or pregnant woman, famous
professor or high bishop, important businessman or ardent patriot. All
2,000,000 of them.
Democide had preceded the Young Turk’s rule and with their collapse at the
end of World War I, the successor Nationalist government carried out its own
democide against the Greeks and remaining or returning Armenians. From 1900
to 1923, various Turkish regimes killed from 3,500,000 to over 4,300,000
Armenians, Greeks, Nestorians, and other Christians.
This wholly genocidal killing is difficult to unravel. During this period
Turkey fought five wars, forcefully changed governments several times,
endured major revolutionary changes, and was occupied by foreign powers.
Suffering deportations, famine, exposure, war, genocide, and massacres,
millions of Turkish Moslems, Armenians, Greeks, and other Christians died.
Moreover, current Turkish governments utterly reject any claim that Turkey
committed genocide, and scholars specializing in the study of Turkey must
avoid the topic or follow the Turkish official line if they hope to do
research in the country. This line is that the government had to deport the
Armenians from the eastern war zone because of, or for fear of, their
rebellion. Many died in the process regardless of Turkish attempts to
protect and care for them; others died in communal strife or in a civil war
between Armenians and Moslems.[1] On the other side, Armenian scholars may
have exaggerated the size of the Armenian population in Turkey, the number
killed, and Turkish brutality and genocidal intentions.
Then there are the third-party reports, commentaries, and studies, published
during World War I. Since Turkey fought on the side of Germany, it was in
the interest of the French and British, who during the war years widely
disseminated anti-German propaganda, to put the worst face on events in
Turkey. Moreover, Armenians themselves may have falsified high level Turkish
documents and reports on the killing in order to win sympathy and support
for restoration, reparations, or the independence of Armenia.
Nevertheless, I do not doubt that this genocide occurred. Extant
communications from a variety of ambassadors and other officials, including
those of Italy, the then neutral United States, and Turkey’s closest ally
Germany, verify and detail a genocide in process. Moreover, contemporary
newsmen and correspondents documented aspects of the genocide. Then, two
trials were held. One by the post-war government that replaced the Young
Turks, which gathered available documentation and other evidence on the
genocide and found the leaders guilty.[2] The second trial was of the
Armenian who assassinated the former Young Turk leader Talaat in Munich in
1920.[3] Although the Germans were still friendly toward the Young Turks
they had supported during the war, the evidence on the genocide presented at
the trial convinced the court that the assassination was justified. Finally,
Turkish government telegrams and minutes of meetings held by government
leaders establish as well their intent to destroy all the Armenians in
Turkey. In my related Death By Government[4] I have quoted selections from
this vast collection of documents and need not repeat them here.[5] The
sheer weight of all this material in English alone, in some ways as diverse
and authoritative as that on the Holocaust, is such that the invalidity or
falsification of some of it can hardly effect the overall conclusion that a
genocide took place.
The problem, then, is somehow to cut through the exaggerations and
propaganda to make some reasonable estimates of the number of Armenians and
others killed. Tables 5.1A and 5.1B organizes this attempt, along with the
relevant estimates from the literature, their sources, and my calculations
and checks. Note that throughout the tables I use the specific term genocide
where appropriate, rather than the more general democide. Here, the people
were murdered simply because they were Christians, Armenians, Greeks, or
Moslems.
I divide the tables into four major periods. The first covers the last years
of Sultan Abdul Hamid’s rule, 1900 to April 1909 (lines 1 to 4 of Table
5.1A). Then there is the Young Turk rule before World War I (lines 5 to
72–the six-month period when the Young Turks were out of power is
irrelevant here and ignored) and that during the war (lines 74 to 274). The
final major division comprehends the post-WWI interregnum (lines 276 to 436)
until the internationally accepted establishment of a sovereign and
independent Turkey (Treaty of Lausanne). In the following two sections I
summarize the results for genocide (lines 438 to 488 of Table 5.1B) and
total dead 1900 to 1923 (lines 490 to 504), and then present estimates for
refugees (lines 508 to 539) and populations (lines 540 to 632). Finally, I
calculate the overall genocide rate (lines 634 to 641).
Possibly two massacres took place during the first period, but there is no
evidence in the sources that these were democidal (lines 2 to 3 of Table
5.1A).
Turning to the first years of the Young Turk period, first I list the three
wars that Turkey fought (lines 7 to 26–one was started while the Young
Turks were out of government). Although the sources record the military dead
for these wars, they usually ignore the civilian war-dead. I assumed a total
low of 20,000 civilian war-dead (line 30) for the three wars, but the
sources are not adequate to estimate a mid-value or high. This low added to
military war-dead (line 31) gives at least 84,000 overall dead in these
wars.
As to the 1909 massacres of Armenians in the Cilicia region, particularly
Adana, there are a variety of estimates shown in the table (lines 35 to 61).
Most notable is that these massacres occurred when the Young Turks had just
overthrown the government and even pro-Armenian sources differ as to their
complicity in the massacres. I therefore treat these as nondemocidal, and
consolidate them into a likely 30,000 killed (line 64).
Hints in the sources suggest that some genocide did occur elsewhere and
subsequently. Turk authorities apparently did kill Armenians and Greeks in
pogroms and expulsions from their villages, at least in 1913 (lines 67 to
68). Lacking more information, I can only give a conservative low estimate
of 5,000 killed in genocide for the whole period.
The table recapitulates the various totals for this period (lines 71 to 71b)
and sums them (line 72). Overall, some 109,000 to 152,000 people died, the
vast majority in wars.
Considering next the World War I period, and the resulting war-dead (lines
76 to 90), a problem is separating from the estimates those for civilian
war-dead, versus those including massacres and genocide. I could include
confidently only one estimate for war-dead (line 86). When this is added to
the probable 400,000 consolidated battle-dead (line 83), we find that some
650,000 Turkish soldiers and civilians died from the war (line 90).
Of greatest importance are the estimates of the Young Turk’s genocide during
the war. In the table I organize these into several categories. The first
gives and consolidates those of the number deported (lines 93 to 102), and
then also does this for the estimates of their toll (lines 104 to 121). I
calculate an alternative total (on line 122) from the estimated percentages
of those killed during deportation (notes on lines 105, 116, and 118) and
the consolidated number deported (line 102). From these two alternative
ranges (lines 121 and 122) I determine a total (line 123) in the usual way.
Next I list the estimates of Armenians that the Turks killed (lines 125 to
146). These I classified by soldier or civilian and by place killed and then
consolidate or sum them (lines 131, 138, and 147), and total them overall
(line 148).
Finally, the table presents the many estimates of the overall genocide’s
toll during 1915 to 1918 (lines 151 to 186). These I order from the lowest
to the highest figures. As can be seen, they vary from a low of 300,000
(lines 151 to 152) to a high of 2,000,000 (line 163), which anchor the
consolidated range (line 187). Consistent with the estimates 1,000,000 dead
(see lines 157, 160, 164 to 178) appears the most prudent mid-value.
Next I independently check this consolidation against the sum (line 188) of
those Armenians murdered during the deportations (line 123) and otherwise
(line 148). As can be seen, the alternative totals (lines 187 and 188) are
divergent, the mid-value alone being off by 808,000 dead. To compensate for
this, I give the final genocide range (line 189) the lowest low and highest
high of the two and average their mid-values. Thus, given all these
estimates, the Turks murdered most likely 300,000 to 2,686,000 Armenians,
probably 1,404,000 of them. A critical question is then whether this is
consistent with the Armenian population, itself a contentious estimate. This
I will later consider.
Not only did the Turks murder Armenians, but Greeks as well. Estimates of
this are far fewer (lines 201 to 203), but we do have assessments of those
deported (lines 193 to 197) from which to calculate the possible toll (line
198). The actual percentages from which I make this calculation reflect the
relevant historical bits and pieces in the sources.[6] Combining this
calculation and the sum of the estimates (line 204) suggest a likely
genocide of 84,000 Greeks.
Sometimes the sources would refer to Christians killed (lines 207 to 207b),
which most likely included Armenians or Greeks, but could also refer to the
relatively small number of Turkey’s Nestorians, Bulgarians, or Cossacks.
These are totaled separately (line 208).
During the war the British navy blockaded Turkey, including the Turkish
Levant. No food was allowed in by sea. The resulting famine in Lebanon and
Syria (with consequences shown on lines 208a to 208d) would not have become
as deadly as it did had not the Turks commandeered available food supplies
and refused to help the starving. As a result they bear the greater
responsibility for the famine, which I calculate as probably around 75
percent of the total dead (line 208i).
The Young Turks did not confine their democide to Turkey. When they invaded
Caucasia, their soldiers massacred Armenians and other Christians and also
encouraged Kurds and Azerbaijanis to do so. Overall, Turks possibly killed
(lines 212 to 220) 10,000 Christians, most of them probably Armenians–there
were very few Greeks in Caucasia. (It is difficult to keep this number in
perspective when other figures are in the tens and hundreds of thousands;
but imagine the contemporary enraged and horrified outcry were the highest
American, British, or French authorities to be responsible for the murder of
10,000 Moslem citizens–the responsible government would fall or be
impeached.) For this genocide the table also lists some specific estimates
(lines 224 to 227). These I consolidated (line 228) and then add (line 229)
an assumed 4/5ths of the Christian dead determined above. The table then
sums the two ranges (lines 228 and 229) to get the genocide (line 232).
As noted, the Turks also massacred Nestorian Christians, for which there are
also a few estimates (lines 235 to 238). From my assumption that 1/5th of
the Christian dead previously determined (line 218) were Nestorians, I
calculate a final genocide (line 241).
Only one estimate of Moslem Azerbaijanis killed is available (line 244).
I now can calculate the overall foreign genocide (line 249), which probably
ranges from 105,000 to 157,000 killed, most likely 131,000.
Turkey’s Armenians also massacred Moslems. Claims that this may have
amounted to at least 1,000,000, or even 1,500,000 Moslem dead (table 5.1A,
lines 106b and 106e) however, have no substantiation beyond former Young
Turks or their officials. Had the Armenians indeed massacred even half this
number, the Young Turks surely would have given it wide publicity,
photographs and all. They had no better way to counter sympathy for the
Armenians they were killing. In any case foreign newsmen and diplomats in
the country surely would have noted the massacres. Moreover, the Turkish
statistician Ahmed Emin, who was hardly sympathetic to the Armenians, gave
(table 5.1A, lines 105 and 106f) an upper limit of 40,000 Moslem Turks
killed by Armenians (including possibly by Armenian-Russian troops) in the
area occupied by Russian forces after the Russian Revolution in 1917, and at
least 128,000 for the 1914-1915 period.[7] Given the other estimates and the
overall populations involved, I estimate that from 128,000 to 600,000 Moslem
Turks and Kurds were killed. Since this was done by Armenian irregulars
serving with Russian forces, I split responsibility for these deaths in
Turkey between the Russians and Armenians, and show in Table 5.1A (line 255)
the Armenian half–probably 75,000 murdered.
Many Moslem Turks also died from famine and disease during the war (lines
258 to 262). Most estimates mix up the toll from these causes with the
number killed from combat. To compensate for this, I first consolidate the
estimates (line 263) and then subtract the war-dead previously determined
(line 264) to get an overall famine and disease range (line 265).
Finally, I can bring together these various totals (lines 268 to 271).
Domestically and during their foreign military actions and occupations, the
Young Turks probably murdered at least 743,000 and perhaps as many as
3,204,000 people, probably 1,883,000 Armenians, Greeks, Nestorians, and
other Christians (line 273). Altogether, likely 3,947,000 died or were
killed during the war (line 274). When I add this to the toll I will
determine below for the next period, we will be able to test the overall
total against the population deficit and unnatural death).
The next division in the table covers the interregnum period after WWI.
Turkish Nationalist forces fought three wars during this time (lines 279 to
303). Estimates for the Greco-Turkish war give two ways of determining
war-dead (lines 302 and 303), from which I select a final war-dead range in
the usual way.
There is one incredibly low estimate of the overall war and massacre dead
for this period (line 307) and a reasonable one for the Muslim male war-dead
from 1914 to this period’s end (line 308). From the latter I subtract the
WWI war-dead to get an estimate of the post-WWI war-dead (line 310). Since
it largely excludes female dead, this is a conservative result. Nonetheless,
as can be seen by comparing this to the war-dead sum for the three wars
(lines 311), the mid-value and high are significantly greater than the sum.
Departing from the usual approach because of the incredible low of zero (on
line 310–this implies that less than 500 were killed), I take the low of
line 311 for the low (line 312), the high of line 310 for the high, and
average the two mid-values.
Following this I list the estimates, consolidations, and sums for the
Nationalist genocide of Christians (lines 315 to 329), Armenians (lines 334
to 359), and Greeks (lines 366 to 375). Regarding the Christian genocide,
one estimate (line 322) of those killed in Izmir could refer to the former
city of Smyrna, or to the Izmir peninsula next to Smyrna. I cannot determine
which is meant (the estimate is only cited in Gross[8] and his source is in
Armenian), and I thus conservatively assumed that it largely duplicates
those already given for Smyrna. Virtually all the total domestic Christian
democide (line 329) took place in the Aydin Administrative District, of
which Smyrna was a part. Since almost all the Christians in this area were
either Greeks or Armenians, and in 1914 Greeks made up 94 percent of the
total of the two,[9] we then can assume that the Armenians were 6 percent
(line 330) and Greeks 94 percent (line 331) of the Christian toll. I later
employ the resulting ranges (lines 353 and 373) to determine the total
number of these two groups that the Turks killed.
For the Armenian toll (lines 334 to 359) I include the refugee deaths (lines
358 to 359). Armenia, which became temporarily independent during this
period, and adjacent areas contained hundreds of thousands who had fled the
Young Turk genocide. Within a few years they also had to flee before the
genocidal massacres of invading Nationalist forces and their
Kurdish-Azerbaijani tribal allies. These refugees died from famine, disease,
and exposure–deaths surely the responsibility of the Nationalists. The
sources give one estimate of these deaths (line 358), and based on this and
the estimates of the number of refugees I consolidate elsewhere in the table
(lines 509 to 522), I estimate the range of deaths shown (line 359). To
display the effect of these assumed refugee deaths on the Armenian genocide
total, I sum the deaths for non-refugees (line 362) and then list one
estimate of the overall number of returning deportees killed in Turkey (line
362a), which understandably is much lower than the non-refugee sum. Note,
however, that it is the same as the low for those killed in Turkish Armenia
(line 350). Adding the lowest of line 362a and 350 to the low for refugee
deaths (line 359) gives us the low for the Armenian genocide (line 363), and
summing all the estimates, including refugees, gives us the mid-value and
high. Most likely then, in total during this period the Turks killed from
325,000 to 545,000, most probably 440,000 of their Armenians–these along
with those murdered during WWI.
In the table I next list partial estimates (lines 367 to 374) for the
genocide of the Greek. There is one calculation of Turkey’s Anatolian (Asia
Minor) Greek population deficit during 1912 to 1922, taking into account
emigration and deportation from Turkey (line 378). Subtracting from this the
WWI Greek genocide I calculated from previous totals (line 379), I get the
range of post-WWI losses shown (line 380). This then provides an alternative
to the sum of the specific mortality estimates (line 381). From these
alternative ranges I calculated a final Greek genocide for this period in
the usual way (line 382). Most probably, the Nationalists Turks murdered
264,000 Greeks; 703,000 Greeks and Armenians together in the post-WWI years
(line 385).
Nationalist forces also committed similar genocide during their invasion of
Armenia, particularly in Kars and Alexandropol (lines 389 to 398). Many
Armenians also died during flight to escape the massacres and tribal Kurdish
and Azerbaijanis allies (lines 405 to 408). One source provides the overall
Armenian toll in Caucasia from 1914 to 1922 (line 412), which gives us a
total for this period (line 414) when we subtract those killed during WWI
(line 413). There is one estimate we can compare to this result (line 415),
which we find within its range. I also repeat the result (line 418) so that
we may compare it to an alternative total (line 419) that I summed from the
previous consolidations. The two ranges differ enough for me to calculate a
final genocide toll (line 420) as for previous such cases.
The Greek Army before and during the Greco-Turkish War massacred Moslem
Turks or permitted such to take place by Greek villagers. I show some
specific estimates of the democide in the table (lines 424 to 427). From
these and material in the sources, particularly Housepian [10] and
Toynbee[11], I believe a minimum number of killed is 15,000 (line 428).
Finally, I pull together the various totals (lines 431 to 434). In this
post-WWI period the Turks killed overall probably 878,000 Armenians and
Greeks, or at least 665,000 and even perhaps as many as 1,156,000 in total
(line 435). Including war-dead, 1,031,000 Turkish citizens or those under
Turkey’s rule or fleeing from it died during these years (line 436).
The table’s next section in Table 5.1B sums up the various sub-totals and
compares them to overall estimates in the sources and demographic
calculations. The first of these concerns the Armenian domestic genocide
(lines 441 to 449). I consolidate these (line 450) and compare the result to
one population based calculation of the Anatolian Armenian dead (line
451–relatively few lived in European Turkey) 1912 to 1922. Clearly this is
way below that of the various estimates. Moreover, it also is under the low
of the Armenian toll that I calculated in the previous sections (line 452),
even when I omit refugee deaths (line 453). This suggests caution in
accepting the totals.
To further check on this, I did my own demographic analysis and calculated
the likely Armenian unnatural deaths (line 454–see lines 601 to 606). Given
that this is calculated independently from the estimate-based totals, the
range is remarkably close to that for the relevant non-refugee total
(compare line 454 to line 453). Accordingly, I accept the totals previously
calculated and restate their sum (line 455).
To get the foreign genocide of Armenians in Caucasia, I sum the previous
totals (line 458) and compared the range to that of the Armenian-Russian
population deficit (line 459) I calculated separately (lines 608 to 611). As
can be seen, the summed range (line 458) is conservative and therefore
acceptable (line 460), even keeping in mind that Armenians were also killed
in WWI, in the Turkish invasion of Caucasia, in Armenia’s war against
Georgia, and in military conflict with Azerbaijan. Moreover, thousands
probably immigrated from the region.
Next I add together the Turkey and Russian Armenian population deficits and
compared them to the sum of domestic and foreign Armenian genocide (lines
463 to 466). The result is acceptable: the low is below that of the combined
deficit, the high is close, and the mid-value is also close and below that
of the deficit. This helps further establish confidence in the figures
determined here.
As to the genocide of the Greeks, I sum the previous totals I calculated
(line 470) and show beneath it a partial estimate of the Greek dead (line
471) and the Anatolian Greek population deficit (line 472). The deficit is
well within the range that I independently calculated and I therefore adopt
it as the final genocide (line 473).
After summing or displaying various totals (lines 475 to 485f), I show
Tashjian’s estimate of those killed or deported 1822-1922 (line 486). Now,
as noted in Death By Government, the Ottoman Empire committed numerous
genocidal massacres of Armenians in the previous century, particularly in
1894 to 1896 when Turks murdered perhaps 100,000 to 300,000 Armenians. Were
I to add to this 100,000 for other pre-1900 genocides, and then reduce
Tashjian’s estimate by the sum to compensate for these deaths, and by
another 10-15 percent to account for those surviving deportation (for the
sources of the percentages, see line 122 of Table 5.1A), the resulting
figure (line 486a) would still be within the range calculated here. Adding
all the sub-totals (line 488) gives us the grand total genocide in turkey or
committed by it: 1,428,000 to 4,380,000 murdered, likely 2,781,000
Armenians, Greeks, Nestorians, Moslem Turks, Azerbaijanis, and others.
Besides the tests of the genocide totals shown above (lines 451, 454, 459,
466, 471, 472, 486), we can also check the table’s total domestic dead. The
table first lists and consolidated three independent, overall dead estimates
or calculations for the years 1912 (or 1914) to 1922 (lines 492 to 495), and
then presents together the various totals (lines 498 to 501a) that I
previously determined and sums them (line 502) to get the total dead, and
next the overall domestic dead (line 503). Beneath this I show for
comparison the consolidation of the estimated domestic dead (line 504). The
comparison is as it should be: the low of line 503 is lower than line 504,
the high is higher, and the mid-value is slightly below by about 5 percent.
Because of this, there appears no need for me to reconsider the various
calculations going into this total.
I next show the estimates and consolidations for refugees from Turkey’s wars
and genocides (lines 510 to 537). There is nothing unusual in their
presentation and their consolidations figure in the calculation of
population deficits and unnatural deaths (e.g., line 606).
In order to calculate population deficits I give population estimates and
consolidations for Turkey as a whole (lines 542 to 551) in 1914 to 1915. To
determine a population deficit later, I also calculate the population for
1920 to 1921 (line 552) from the minority population estimates given next
for Armenians (lines 556 to 596), Greeks (lines 615 to 625), and Muslims
(lines 628 to 630). Moreover, I had to calculate an average population
controlled by the Nationalists (line 553) for later use in the genocide
ratios (lines 640 to 641). I could not find any information on what this
proportion was, even for a particular year, and therefore from narrative
histories of this period [12] I estimated it to vary from 40 to 75 percent,
with a mid-value of 50 percent, taking into account that French and Greek
forces occupied a portion of south-western Anatolia during this period.
The table lays out the calculation of the Armenian population deficit and
unnatural deaths (lines 600 to 611). From the consolidated estimates of the
Armenian’s population growth rate, I projected what the population should
have been in 1923 (line 604) and subtracted from it the actual population
(line 589). Subtracting from this the number of refugees that escaped the
genocide (line 522–this is conservative, since many refugees returned to
later be killed by the Nationalists) gives an estimate of those Armenians
who died unnatural deaths (line 606). I did the same for Armenian-Russians
(lines 609 to 610). I also sum the two ranges of unnatural deaths (lines 606
and 610) to get the number of unnatural deaths for Russia and Turkey’s
Armenians together (line 611). And I also give or calculate the population
deficits for the Greeks and Muslims (lines 626 and 632).
Finally, in the remainder of the table I calculate the democide rates for
the Young Turks (lines 636 to 637) and the local Nationalists (lines 640 to
641). Per year the Young Turks killed almost 1 out of every 100 of their
population (line 637). The Nationalists, however, were far more vicious. For
the population they controlled they murdered 1 out of every 38 per year
(line 641).
NOTES
From the pre-publisher edited manuscript of Chapter 5 in R.J. Rummel,
Statistics of Democide, 1997. For full reference to Statitisics of Democide,
the list of its contents, figures, and tables, and the text of its preface,
click note [5].
1. See McCarthy (1983), who in analyzing the change in the Armenian
population from before to after WWI manages to avoid any hint that Armenians
were killed by the government. McCarthy credits their population loss to war
conditions or a civil war they fought with Moslems. See also Shaw and Shaw
(1977), who in the three pages they devote to the Armenians allege that only
200,000 of them died, and these from war, famine, and disease in spite of
the attempts by Turkish authorities to protect them.
2. For relevant documentation and discussion, see Dadrian (1991a, 1991b,
1991c).
3. For a report on this trial, see Alexander (1991).
4. Rummel (1994, Chapter 10).
5. For primary sources and analysis the work of Dadrian (1986, 1991b,
and 1991c) is particularly useful.
6. As for example in McCarthy (1983), Miller (1966), Toynbee (1922), and
Ladas (1932).
7. Emin (1930, pp. 218-219, 222).
8. Gross (1972, 47n.6).
9. Calculated from the population statistics in Karpat (1985, p. 188).
10. Housepian (1966).
11. Toynbee (1922).
12. For example, Miller (1966).