been proposed (in England) to give precision to the comparative method by applying "statistics"; this would mean the systematic comparison of all known societies and the enumeration of all the cases where two usages are found together. This is the principle of Bacon's tables of agreement; it is to be feared that it will be no more fertile in results. The defect of all such methods is that they apply to abstract and partly arbitrary notions, sometimes merely to verbal resemblances, and do not rest on a knowledge of the whole of the conditions under which the facts occur.
We can conceive a more concrete method which, instead of comparing fragments, should compare wholes, that is entire societies, either the same society at different stages of its evolution (England in the sixteenth, and again in the nineteenth century), or else the general evolution of several societies, contemporary with each other (England and France), or existing at different epochs (Rome and England). Such a method might be useful negatively, for the purpose of ascertaining that a given fact is not the necessary effect of another, since they are not always found together (for example, the emancipation of women and Christianity). But positive results are hardly to be expected of it, for the concomitance of two facts in several series does not show whether one is the cause of the other, or whether both are joint effects of a single cause.
The methodical investigation of the causes of a fact requires an analysis of the conditions under which the fact occurs, performed so as to isolate the necessary condition which is its cause; it presupposes, therefore, the complete knowledge of these conditions. But this is precisely what we