the sexuality of surrounding human beings. They may help to explain some curious points in the psychology of a crowd which have not yet received sufficient attention. In short, bi-sexuality cannot be properly observed in a single moment, but must be studied through successive periods of time. This time-element in psychological differences of sexuality may be regularly periodic or not. The swing towards one pole of sexuality may be greater than the following swing to the other side. Although theoretically possible, it seems to be extremely rare for the swing to the male side to be exactly equal to the swing towards the female side.
It may be admitted in principle, before proceeding to detailed investigation, that the conception of sexually intermediate forms makes possible a more accurate description of individual characters in so far as it aids in determining the proportion of male and female in each individual, and of measuring the oscillations to each side of which any individual is capable. A point of method must be decided at once, as upon it depends the course the investigation will pursue. Are we to begin by an empirical investigation of the almost innumerable intermediate conditions in mental sexuality, or are we to set out with the abstract sexual types, the ideal psychological man and woman, and then investigate deductively how far such ideal pictures correspond with concrete cases? The former method is that which the development of psychological knowledge has pursued; ideals have been derived from facts, sexual types constructed from observation of the manifold complexity of nature; it would be inductive and analytic. The latter mode, deductive and synthetic, is more in accordance with formal logic.
I have been unwilling to pursue the second method as fully as is possible, because every one can apply for himself to concrete facts the two well-defined extreme types; once it is understood that actual individuals are mixtures of the types, it is simple to apply theory to practice, and the actual pursuit of detailed cases would involve much repetition and bring little theoretical advantage. The second method,
