There was nothing very mysterious about this; there was a certain known margin of error, and therefore a certain known maximum and minimum. For instance, it could be calculated from all sorts of sources which confirmed each other—prisoners' reports, published lists, rolls of honour, the rate of retirement of divisions, the analogy of known losses upon our side, and so on. But there came a time after which people were impatient and preferred to believe the whole thing to be fantastic guesswork, because they were in a hurry, and at the same time half despairing. There were hundreds and perhaps thousands of expert soldiers up and down Europe engaged in arriving at these numbers, their conclusions were centralised in the various staffs; and it was surprising to see by how little the various estimates differed. I remember once at Chantilly comparing a set of figures arrived at by one of the national staffs with another set of figures based upon different national methods, and remarking that there was not five per cent difference between them. Yet the ordinary member of the public had by that time become completely suspicious of all figures.
The German casualties on the Somme are a good example. We do not know them to a single unit, and shall never so know them. But we know them within a certain margin of error, and we knew them shortly after the cessation of the battle within a rather larger margin of error. But the ordinary member of the public at home was not in a mood for them, and would not follow them. I got shoals of insulting letters about them from honest people who were disappointed of a miracle. Yet these numbers were of supreme importance. For it was the usury of the Somme, coming after that of Verdun, which began the process of decline upon the enemy's side, politically and morally even more than militarily. The plain man could only see that the Somme had not succeeded in breaking the German line; his expectation was disappointed; he thought all this talk of numbers beside the mark, and his annoyance with it took the natural though very illogical form of belabouring the poor dumb figures that had done him no harm.
I remember very well how, after I had published the approximate figure of enemy losses of 1916 in "Land and Water" (supplied of course from Staff figures), the Harmsworth Press fully persuaded the mass of Englishmen that the German recruitment was inexhaustible. The figures showed a loss of about one-sixth of the total enemy recruiting power. The general reader could not be bothered with anything so precise. First he thought I had said the enemy were all gone, then he thought that the enemy never could be worn down. He disliked numbers, did the general reader, and every demagogue will do well to play upon that dislike of numbers. It is one of the main ways of deceiving the silly populace.
After all these more or less disreputable reasons for the dislike of numbers one comes to a series of much better reasons, and first among these undoubtedly is the truth expressed in the epigram, "Lies, Damned Lies, and Statistics." You can prove anything almost by Statistics, and therefore if one lives in a time when statistics are a plague (because there is such facility for gathering them under our modern system of arbitrary, all-powerful and highly-centralised government) one comes to suspect their use and even to detest it. Now that suspicion and that detestation are well founded.
The reason they are well founded is this.
Judgment, the most valuable of all the rational qualities, is essentially an integration of an almost infinite number of differentials. The mind is seized with a very large number, an almost infinite number (or, as mathematicians would say, an indefinitely large number) of impressions which it combines, and on the combined effect of which it bases certitude. I have no doubt of an oak tree when I see it, although I do not look at each leaf nor measure the exact outline even of one leaf to note within what margin of error it conforms to the type or form of an oak-leaf. I recognise a voice or a face by integrating a vast number of small differentials, and no amount of mathematical or mechanical argument drawn from impressions far less numerous and not combined by a human mind at all would convince me of error. That is why we say that a good portrait is always better than a photograph, and that is why we often say of a photograph that it is not like the original at all. A photograph is only the record of one selected, very limited set of impressions, highly restricted in number and in time, which cannot compare in value with the general conclusion of an observer using all the human faculties over a considerable time, and endowed with the organic power of unification.
In the same way a piece of statistics, however accurate, concerns only one of an infinite number of factors. You get nearer the truth as you combine one set of statistics with a second, a third, a fourth; but though you were to have a thousand sets they would not outweigh your general judgment based on observation.
For instance, a man knows perfectly well what he means by a wet day. Your statistician may come along with his figures of rainfall and may prove by them that a very fine day was a wet one. He says, "Half an inch of rain fell in twenty-four hours." But if the half-inch of rain all fell in one hour just before the dawn of a hot and cloudless summer's day he is wrong and you are right. To give even a rough impression through mere statistics of what common sense calls "a wet day" you would have to compare many sets of figures. There would be first of all the inches of rainfall, then the number of hours during which perceptible rain was falling, then the number of hours during which the sky was overcast, then the rate of evaporation, then figures showing what proportion of rainfall came in the waking and in the sleeping hours, then a table showing how far the rain was intermittent, for a day during which you have rainfall in the first half and none in the second is not at all the same thing as a day in which rain falls every half-hour. And even when you had all those statistics combined, no sane man would accept them against his own general impression.
There is a parallel here with the very legitimate suspicion people are beginning to feel toward what is called "scientific" argument in social and domestic matters. If a man has worked hard all the morning in the open air and is very hungry he looks forward to a beefsteak and a pint of beer at luncheon, and if some science fellow comes along with a tabloid and a little water out of a tank it is no good telling the man that such a meal will be "scientifically the equivalent" of what he was expecting. He knows very well that the word "equivalent" here is a lie; and it is hard lines that the noble title of Science should have been degraded as it has been in our generation by nonsense of this sort. They are still at it, but I do not think it will last long, for it is provoking anger. There are people who come and tell country folk that the rooms in which they sleep must have a certain "cubical capacity," so that a great healthy man as strong as an ox and sleeping as no townsman can sleep is "scientifically" proved to be in a very parlous way when he lives as his English fathers have lived before him for a thousand years. And "I with these mine eyes have seen" an inspector insisting that the window-space of a room should be not less than one-tenth or one-quarter or whatever it was of the floor-space, without noting that the window in one case looked on to a blank wall, and in another on to the infinite spaces of the sea.
The use of statistics in argument is essentially deduction from insufficient premises, and though the mass of men who are getting more and more exasperated with the results of such argument are not yet conscious of the flaw in their oppressors' reasoning, they only know instinctively that it is bad reasoning—and they are right.