rather than actual facts the starting point, (b) In defining and grouping phenomena such as crime, the family, etc., use at first external marks only, not what may be deemed the essential characteristic. This last should be left to emerge as the result of our investigation, not assumed at the outset, nor should we fix on merely a part of the class of facts and test the rest by a standard so derived, (c) Study especially the consolidations of social facts in law, proverbs, modes, etc., e. g., the law of succession for determining the actual view of the family relations which has found a permanent form.
(2) Rules for distinguishing the pathological from the normal. It is a much mooted question whether sociology shall rest with a scientific ascertainment of facts, or whether it shall attempt to become of practical value by telling what should be. It is of course evident that social health is desirable; can we point out in what it consists? The ordinary method is to assume some criterion of health at the start (freedom from suffering, adaptation to environment), but these are either arbitrary or impracticable. The true objective method is to take for our criterion at the outset only an external mark, viz., define the normal, merely as the general. Hence normal = the mean, diseased = the exceptional. But it is evident that greater frequency must ordinarily be due to superiority, to health. (Durkheim here obviously either must leave out the notion of moral or æsthetic superiority, or make them equivalent to physical superiority, or, assume that survival of the fittest will secure the morally fittest—assumptions as serious as those to which he objects); hence (a) we may control our results by seeking the cause of the generality of a given phenomenon in its relation to the general conditions of life in the type in question, and (b) this becomes especially important in the case of a social species in a transitional stage.
(3) For making the classification into social types the objective principle to be adopted as our standard is that of simplicity, i. e., we ask whether a given group is made up of units which enclose other units more simple than itself.
