As a matter of fact, no real group, not even a centralised society, is a homogeneous whole. For a great part of human activity--language, art, science, religion, economic interests--the group is constantly fluctuating. What are we to understand by the group of those who speak Greek, the Christian group, the group of modern science? And even those groups to which some precision is given by an official organisation, States and Churches, are but superficial unities composed of heterogeneous elements. The English nation comprises Welsh, Scotch, and Irish; the Catholic Church is composed of adherents scattered over the whole world, and differing in everything but religion. There is no group whose members have the same habits in every respect. The same man is at the same time a member of several groups, and in each group he has companions who differ from those he has in the others. A French Canadian belongs to the British Empire, the Catholic Church, the group of French-speaking people. Thus the different groups overlap each other in a way that makes it impossible to divide humanity into sharply distinct societies existing side by side. In historical documents we find the contemporary names of groups, many of them resting on mere superficial resemblances. It must be made a rule not to adopt popular notions of this kind without criticising them. We must accurately determine the nature and extent of the group, asking: Of what men was it composed? What bond united them? What habits had they in common? In what species of activity did they differ? Not till after such criticism shall we be able to tell what are the habits in respect of which the group in question may be used as a basis of study. In order to study intellectual habits (language, religion, art, science) we shall not take a political unit, the nation, but the group consisting of those who shared the habit in question. In order to study economic facts we shall choose a group united by a common economic interest; we shall reserve the political group for the study of social and political facts, and we shall discard _race_[193] altogether. Even in those points in which a group is homogeneous it is not entirely so; it is divided into sub-groups, the members of which differ in secondary habits; a language is divided into dialects, a religion into sects, a nation into provinces. Conversely, one group resembles other groups in a way that justifies its being regarded as contiguous with them; in a general classification we may recognise "families" of languages, arts, and peoples. We have, then, to ask: How was a given group sub-divided? Of what larger group did it form a part?