November 30, 1916 
LAND & WATER 
The German Reserves 
A Reply by Mr. Hilaire Belloc to his numerous Correspondents 
I HAVE rocoivcd witii regard to the figures of the 
German reserves printed in this jonrnal three weeks 
ago, a great quantity of correspondence, and I 
have noted the criticism of the Press npon these 
same figures. What has thus been printed and written 
demands, I think, a few notes from me in expansion of 
that article and in explanation of certain points in it. 
I must also ask my numerous private correspondents to 
take these lines as a reply. 
I. — Not a few of the criticisms made upon these 
figures are to the effect that this sort of calculation has 
been made very frequently during the course of the war 
and (according to the writers' conception) has either 
proved inaccurate or irrelevant to the main issue. 
Such a judgment is the product only of a mood. 
It surely hardly needs repeating that to treat detailed 
calculation in this fashion is worthless. 
When you take an observation at sea you do not do so 
in order to buck yourself up and to feel that you are 
near port, you do so to find out exactly where you are. 
When you have established your latitude and longitude 
in a sailing ship you cannot therefore prophesy when you 
will be in port, because no man can prophesy the weather. 
But at least you have the only basis of calculation avail- 
able to you, and you are certain upon the main point, 
which is your true position. Only a man ignorant of 
navigation can regard the daily observation as irrelevant. 
Now, 'as to whether figures of this sort are accttrate 
or not — which is quite another point — I can answer that 
very briefly. 
A continuous analysis of figures, close, detailed and 
co-ordinated has appeared, I believe, in no other journal, 
but only in L.\nd & Water. The common phrases else- 
where, " a very large number," " a considerable pro- 
portion of class 1917," " we shall do well not to allow 
less than two million reserves," are not worth the paper 
they are written on. One must have exact calculation, 
a known margin of error, and proof. All else is foolish 
assertion — usually political in its object. 
The early calculations made here (as I have repeatedly 
said) were tentative and most insecure. In the first 
months of the war the margin of error in such calculation 
was very great, and the tendency to exaggerate unknown 
losses stronger than the tendency to minimise them. 
The reasons for this uncertainty of calculation in the 
early part of the war are obvious. The establishment 
of averages, under novel conditions especially, takes 
time. Tlie longer the time elapsed the closer the average 
arrived at. The establishment of an Intelligence System 
takes time. The discovery of what forms of intelligence 
are reliable takes time ; so does the co-ordination of 
reports, of documents, of examination of prisoners, etc. 
The- coefficient by which the number of known dead 
must be multiplied to arrive at the total casualty list 
was in the wars of movement and under the old con- 
ditions of fighting, very much larger than the coefficient 
to be used in this trench fighting. Therefore, in the 
first months of the war a given number of dead upon the 
enemy's side made one exaggerate his total probable 
casualty list. A further element of error was introduced 
when it was discovered that the enemy was falsifying his 
lists, for it took a long time to find out by exactly what 
percentage he concealed his real losses, and there was a 
tendency to exaggerate the amount which he thus con- 
cealed. We did not even known at first hdw the trick 
was done, and there was a tendency to believe that it 
was mere arbitrary falsification practised upon a large 
scale. Then we surmised — and grew more and more 
certain — that encouraging as large as possible a pro- 
portion of " doubtfuls " in each unit was the method 
established and the publication of " certains " only in 
the daily lists : leaving the rest for later correction or 
private notice. 
In the course of some months of observation the ele- 
ments of error were gradually eliminated. The mass of 
information through the Intelligence Departments of the 
various 'A Hies increased enormously; it became possible 
to establish averages, at first uith a considerable margin of 
error, later with a much smaller one. Our information 
about the internal conditions of the enemy:; his hospital 
population, his hospital discharges, etc., which had hardly 
existed at all in the first months of the war, became very 
full and complete, and the Allies' record of their own 
proportionate losses in killed and wounded, etc., and of 
their own hospital rates, confirmed the increasing know- 
ledge they had of the enemy. With the beginning of the 
open season of 1915 the system from which calculation is 
now made was fairly established, though not yet per- 
fected. And from that day on it has been more and more 
certain in its results. 
Each total was made, of course for the moment only. 
Time is an essential factor. Thus in the last article (of 
three weeks ago) we were dealing only with a definite future 
of nine months and excluding class 1919. But, with 
such a caution, the figures are accurate and authoritative. 
Proofs of Accuracy 
For instance, we were able to say fairly early in 1915 
that the normal recruitment would come to an end with 
the first year of the war ; that abnormal recruitment 
(that is, the summoning of the younger classes and the 
call of men hitherto rejected as inelficients) would begin 
in the second half of 1915. We were even able to give 
suggested dates. We said that the revision of inefificients 
would take place not later than November, and the calling 
of the 1917 class not later than the beginning of the 
next year, while the 1918 class would first be called in 
Ciermany, we said, in the summer, and probably about 
June, 1916. It is surely worth remarking that these 
suggestions were exactly verified. The revision of the 
rejected class took place during the course of October. 
Class 1917 began to be called even a little before the New 
Year, in the month of December, and class 1918 was called 
(the first batch of it in Saxony) last June. 
The figures we can print now, therefore, after 
so long an experience of the campaign, are really 
authoritative — say within a margin of error of 10 per 
cent. This margin of error is, of course, an average. 
In some categories, in the prisoners, for instance, 
there is no margin of error at all, the exact number 
is known. In others for example, the dead, the 
margin of error is smaller than in such a category as the 
" useful returns from hospitals," if only because the 
latter figure will differ with the definition of the term 
" useful." Thus, when we say that by the end of October 
1916, the military deaths of the German Empire— the. 
deaths of men in uniform who died as a consequence of the 
campaign — were about a million and a half, we know 
that the margin of error is, in this category, small. The 
margin of error for the men who will return to duty from 
hospital is much larger. I can conceive its fluctuating 
round a figure of 100,000 between its superior and inferior - 
limits. 
II. — Very many of my correspondents have noted the 
apparent discrepancy between the number of German 
divisions now organised and the total number of the 
German " acting army "—that is, the total number of 
men in uniform, but neither convalescent nor in depots. 
The number of divisions given was 203. Even if they 
were at full strength that would only allow for just over 
4,000,000 men, but they are hardly any of them at full 
strength and many of them are no more than three- 
quarters of that. Some of them even below that. All 
the infantry of the new divisions — save six — were, as 
we know, formed simply by taking men from older 
divisions, and the only new men in any numbers were 
the doctors, engineers- gunners and drivers formed for 
each new division — for there was no new cavalry. How 
then, it is asked, can 203 divisions account for a total 
army of five million men ? 
There are two causes at work to account for this appa- 
rent discrepancy. The first is the expansion of the 
auxiliary services. The proportion of men behind the 
line to men in line varies very much according to the 
kind of work the army is doing, the distance it is from its 
