But no empirical science, consequently also no theory of the Art of War, can always corroborate its truths by historical proof; it would also be, in some measure, difficult to support experience by single facts.  If any means is once found efficacious in War, it is repeated; one nation copies another, the thing becomes the fashion, and in this manner it comes into use, supported by experience, and takes its place in theory, which contents itself with appealing to experience in general in order to show its origin, but not as a verification of its truth.

But it is quite otherwise if experience is to be used in order to overthrow some means in use, to confirm what is doubtful, or introduce something new; then particular examples from history must be quoted as proofs.

Now, if we consider closely the use of historical proofs, four points of view readily present themselves for the purpose.

First, they may be used merely as an explanation of an idea.  In every abstract consideration it is very easy to be misunderstood, or not to be intelligible at all:  when an author is afraid of this, an exemplification from history serves to throw the light which is wanted on his idea, and to ensure his being intelligible to his reader.

Secondly, it may serve as an application of an idea, because by means of an example there is an opportunity of showing the action of those minor circumstances which cannot all be comprehended and explained in any general expression of an idea; for in that consists, indeed, the difference between theory and experience.  Both these cases belong to examples properly speaking, the two following belong to historical proofs.

Thirdly, a historical fact may be referred to particularly, in order to support what one has advanced.  This is in all cases sufficient, if we have only to prove the possibility of a fact or effect.

Lastly, in the fourth place, from the circumstantial detail of a historical event, and by collecting together several of them, we may deduce some theory, which therefore has its true proof in this testimony itself.

For the first of these purposes all that is generally required is a cursory notice of the case, as it is only used partially.  Historical correctness is a secondary consideration; a case invented might also serve the purpose as well, only historical ones are always to be preferred, because they bring the idea which they illustrate nearer to practical life.

The second use supposes a more circumstantial relation of events, but historical authenticity is again of secondary importance, and in respect to this point the same is to be said as in the first case.

For the third purpose the mere quotation of an undoubted fact is generally sufficient.  If it is asserted that fortified positions may fulfil their object under certain conditions, it is only necessary to mention the position of Bunzelwitz(*) in support of the assertion.