Book 6, Chapter 5, Footnote #01
Favourable cases for approximate generalizations
The most favourable cases for making such approximate generalisations are what may be termed collective instances, where we are fortunately enabled to see the whole class respecting which we arc inquiring in action at once, and, from the qualities displayed by the collective body, are able to judge what must be the qualities of the majority of the individuals composing it. Thus the character of a nation is shown In its acts as a nation; not so much in the acts of its government, for those are much Influenced by other causes; but in the current popular maxims, and other marks of the general direction of public opinion; in the character of the persons Or writings that are held in permanent esteem or admiration; in laws and institutions, so far as they are the work of the nation itself, or are acknowledged and supported by it; and so forth. But even here there is a large margin of doubt and uncertainty. These things are liable to be influenced by many circumstances: they are partly determined by the distinctive qualities of that nation or body of persons but partly also by external causes which would influence any other body of persons, in the same manner. In order, therefore, to make the experiment really complete: we ought to be able to try it without variation upon other nations: to try how Englishmen would act or feel if placed in the same circumstances in which we have supposed Frenchmen to be placed; to apply, in short, the Method of Difference as well as that of Agreement. Now these experiments we cannot try, nor even approximate to.