ing these conditions we may modify a phenomenon or even eliminate it.
These previsions have a character which is clearly scientific. Examples appear in my Lois sociologiques, pp. 46-52, 112. The previsions which we have expressed in each of the successive editions of this work have in each case been verified. Yet they might not have been verified. Then it would have been necessary to investigate the new conditions which may have modified the phenomenon. However, the modifications can be only variations of intensity, that is to say, quantitative variations, so long as the conditions, although variable, are the same. This point of view, which is mainly qualitative, becomes predominant in the study of the synthesis of the functions of the social organs, groups of organs, and systems of particular societies, and in the study of civilization considered as a whole, independently of time and space. Thus, although in the examples cited in my Lois sociologigues the apparent advance of nominal wages might easily have corresponded to stationary wages, or even to a real fall of wages resulting from an increase of the price of commodities necessary for the subsistence of the coal-miners, in abstract sociology the question of illegitimate births or of wages can no longer arise, since these phenomena are purely historical forms of natality and of reward for labor; but there will always remain the corresponding abstract law that the relations of parents and children are closer as the standard of life of the former is elevated.
Knowledge of the causes or conditions favorable to the appearance or disappearance of certain social phenomena is the basis of the experimental method in sociology and in political science. The individual savant can in general only point out, observe, and describe the experiments, and show how certain phenomena may be modified; but the real modifications in the social world must, in the main, be accomplished by collective forces.
Social variations and accidents are always confined to limits which become narrow in proportion as our observation extends over a long period of time, over a great extent of space, and to a great number of cases. In this connection, the theory of proba-
