May 1 6, 191 8 
Land & Water 
This, again, must only be taken for what it is worth : In 
a country full of good roads' a main road is not the absolute 
essential which it is in Eastern Europe, and under conditions 
of a long war railways come into existence of which the 
peace map knew nothing. But, still, the comparative weak- 
ness in main communications makes a difference and again, 
other things being equal, favours attack upon this quarter. 
There is no apparent disadvantage in an attack upon this 
sector compared with an attack upon any other. It is true 
that the communications immediately behind it form a belt 
of devastated territory, but it is a belt narrower than that to 
the south. The roads are in good condition, and there has 
been ample time for the enemy to lay dowii light railways 
and to accumulate all that he needs for such a blow. 
Meanwhile, pending that blow — if, indeed, the enemy 
intends to deliver it in full force, and to make of the third 
phase of the great battle a repetition of the first — the issue, 
as I have said, is still a question of numbers. Should he 
effect a breach and get a large result, reaching Doullens, for 
instance, a great expense is still worth his while. Should he 
effect a breach but be early checked, the expense would not 
be worth his while, and he would be weaker after the effort 
than before. Should he make a determined and prolonged 
attack and fail altogether to affect a breach, he is defeated. 
He cannot, if he attacks in the near future, and if he proposes 
an attack upon the largest scale, stand losses as great as 
those which he has already sustained. The sum of loss for 
which he has budgeted is now more than half exhausted, 
and though it is true that prolonging the business gives time 
for the rettirn of hospital ca.ses, yet it is manifest it also 
gives time for the strain at home to increase and for the 
steady arrival of the American recruitment to prepare for a 
full establishment some months hence. 
I have said previously in these columns that time would 
necessarily help to solve the chief unknown factor in our 
present military problem, which is the enemy's rate of 
expetise. One has to wait for evidence ; Knd now that 
evidence is beginning to arrive. 
We have not indeed as \'et any direct information on the 
enemy's losses, for there has been no appreciable re-advance 
over territory upon which those losses have occurred, and 
therefore no exact enumeration of dead in a particular 
action, no considerable nmnber of prisoners, and no discovery 
of documents. We have an increasing number of details 
by which we may judge the rate of loss in the heavily pun- 
ished units, especially where a small local recovery of ground 
is effected ; but we need very many more of these and the 
addition of average losses, and losses in units below the 
average, before we can get anything like accurate statistics. 
Indirectly, however, the efflux of time has given us an 
exceedingly useful piece of evidence in the number of imits 
the Gennans nave empIo\-ed, and that evidence not only 
points to higher losses than the minimum recently quoted 
(probably for some p)olitical purpose) in the Press, but to a 
particular method of warfare which necessarily involves the 
most rapid losses. 
Last week I said that the nvunber of divisions the Germans 
had put in (by the end of April) approximated to the equiva- 
lent of at least 182, and more probably 190. That was 
counting from 136 to 140 used, of which some 40 had appeared 
twice, and not less than 6 three* times. That was a' very 
conservative estimate of the target the esemy had presented 
up to the end of April in his great offensive ; but, even so, it 
made it likely that he had lost 400,000. 
Now, we have had in the last week information w^hich 
gives us far more precision in this matter, and shows a much 
higher bulk of units employed. 
The identification has been published of German divisions 
up to May ist — that is, for the first six weeks. This identi- 
fication presents us with a number of at least 140. You 
cannot have exact precision, you can only establish a mini- 
mum, and this because of numerous factors of error, though, 
luckily, these affect but a very small proportion of the total. 
For instance, you may identify the presence of a new division 
on a particular day without finding out its number, and 
there may be a discussion as to whether it is a division drawn 
from some other jiart of the Hne or not. When we say that 
140 is the number of original divisions used in the period 
between St. Quentin and May Day, we arc really taking a 
minimum. The total is something more, but only slightly- 
more, than 140. 
Of these 140 divisions (or a little more) it turns out that 
not 40, but at least 50 were put in twice over, and not 6, 
but somewhere about 20 — let us say 18 for a minimum — were 
put in three times ; further, we learn that one di\'ision was 
actually used four times. 
We are dealing, then, not with the equivalent of at least 190, 
but of at least 229 divisions as a target presented by the enemy 
during the first six weeks of the great offensive, and more 
probably over 230. A division is withdrawn after losing 
such and such a proportion of its effectives : say, 3,000 men. 
When a division reappears after fiUing its gaps it is equivalent 
to a new division for the purpose of calculating losses. I 
withdraw a division that has lost 3,000 men. I replace 
these, and send it in again made up. It loses another 3,000. 
I withdraw it and fill the gaps. The losses are 6,000, i.e., 
the same as those of two divisions used for the first time. 
Now, to say that the enemy losses of all kinds were as low 
during those first six weeks as even 450,000 is to say that 
the average loss of a division during the period it was put 
through the mill was less than two thousand of all arms. 
The equivalent of 230 divisions losing an average of only 
2,000 men each would give you 460,000. 
That is not credible. It is not in the past history of the 
enemy's method of action ; it is not consonant with the 
intensity of the great actions now engaged ; it is not, by any 
sane rule, economic of material. To be perpetually with- 
drawing divisions after a comparatively low standard of loss, 
to put a corresponding strain on your communications and 
on all your staff work, to advertise, as it were, to your own 
men your doubt of their standing a strain vastly inferior to 
what they have stood in the past — these and twenty other 
considerations surely make it certain that a divisional strength 
of 14,000 or 15,000 with an establishment of 7,000 to 8,000 
infantry is not thought to have done its work for the moment 
when its total losses of slightly wounded, sick, and all the 
rest of it included, come to less than 15 per cent, of its total, , 
and its infantry losses to perhaps 20 per cent. 
Moreover, we have positive evidence to guide us as weU as 
this consideration of common sense. 'We have first the 
extremelj' heavy losses discoverable whenever the data axe 
available — in a few samples only, it is true, but in samples 
fairly uniform. 
We have next the very rapid rate at which divisions are 
recruited and put in again and the large proportion which 
these bear to the whole. 
A year ago (on the defensive, it is true) the Germans were 
sending back one-eighth in the first six weeks of heavy fighting ; 
in this year they are sending back more than a third. 
Still more striking is the rate of using divisions three times 
in so very short a period ; and these the best quality — the 
3rd Division of the Guard, for instance. iQuite extraordinary 
is the use of one division no less than four times in so very 
short a period as six weeks. 
Proportion of Loss 
Now, it is true that this intensive repetition in use cuts 
both ways. On the one hand, it is argued that the mill 
working at such a rate is necessarily grinding down material 
very much faster than ever before. On the other hand, it is 
arguable that if divisions can be used again so quickly they 
are withdrawn after much less loss than formerly. 13ut of 
these two considerations the first is much the weightiest. 
There would be no point in withdrawing a division and 
sending it back almost immediately, then withdrawing it 
again and sending it back, again after sHght losses upon each 
occasion. It must be admitted that the proportion of loss 
suffered before a division is withdrawn is smaller than it was 
in the fighting on the Somme or at Passchendaele, but there 
is an obvious limiting minimum to the losses which make it 
worth your while to rest a division at all. Over and above all 
this, there is the obvious governing fact which conditions 
the wliole affair, from beginning to end, that the enemy, 
from his first tremendous attack of Ma#ch 21st up to his 
local defeat in Flanders upon April 29th, was pressing with 
the utmost energy, and had unclertaken a task which of its 
nature demanded and budgeted for very rapid loss in the 
hope that such a rate of expense would prove fruitful in the 
long nm. and that some 'decision would be really achieved. , 
It is difficult, with the evidence before us, to put the total 
losses of those six weeks at less than half a million. 
There .is just one other httle point worth noting in this 
connection. An official reply was given in the (lerman 
Parliament, after about a month of the fighting, and referring, 
therefore, j)robably to completed statistics of about the 
first three weeks, that 20,000 light cases had already re- 
turned—or, at any rate, had already been discharged from 
hospital. 
Now, these parliafnentary statements, designed to soothe 
civilians at home, are nearly always of great value to an 
opponent ; that is why they should never be made. The 
number of 20,000 cured put positively, without relation to 
the forces employed, sounds like a very large and satisfactory 
figure, showing a rapid recov.cry of men, and no doubt the 
statement did its work, which was purely political. But 
