is still a difficulty in the way; for although the formulation of the relations may work either way in time, we cannot ignore the one-directionality of time in concrete experience. The relations, though symbolic of a dynamic process, are themselves static. They simply assert that on given assumptions such as uniformity, there is logical dependence of the present on the future, just as there is logical dependence of the present on the past; but they fail to comprehend concrete experience fully, in that they ignore the actual fact that time progresses in one direction only. Thus, if all that these relations imply were true, there is no reason why the crime should not sometimes follow the punishment which is its due, nor why the determination to build a house should not follow the appearance of that house on the scene. Perhaps the existence of purpose and consequent action leading to realization illustrates best of all the hopelessness of the attempt to replace the notion of causal efficiency by the notion of mere logical dependence. For there is certainly a sense in which it can be said, for example, that the house was built because Jack determined to build it, in which it cannot be said that Jack determined to build the house because the house was built. Moreover, we are indubitably aware that our actions determine their ends in a sense altogether different from that in which the ends determine the actions. This could not be so if the relation between them were purely logical.
The matter may therefore be summed up somewhat as follows: The true meaning which causality has for us is rooted in the realization of our own efficiency, as active individuals. The active individual is the 'cause.' The end which his (generally purposive) activity accomplishes is the 'effect.' The scientific method, however, takes the sequences which occur in experience as they stand and determines what may truly be said of them per se. In the first place, it finds that sequences continually recur sufficiently similar in nature to admit of a considerable degree of general characterization. Secondly, it follows that a general proposition may be affirmed with regard to each recurring sequence, whereby the occurrence of one event may be inferred from the occurrence of another event. Thirdly, there is no
