A further defect of memory has been made manifest to me by the study of perversions. Processes which in childhood were entirely devoid of any sexual tinge, but which later became associated with sex-feelings, very readily acquire false sexual associations also when they are revived in memory. Consider, for instance, the case of a homosexual man. He remembers that, as a small boy, he was very fond of sitting on his uncle's knees, and he believes that the pleasure he formerly experienced was tinged by sexual feeling. In reality this was by no means the case. His uncle took the boy on his knee in order to tell him a story. Possibly, also, the riding movements which the uncle imitated by jogging his knees up and down gave the child pleasure, which, however, was entirely devoid of any admixture of sexual feeling. But in the consciousness of the full-grown man, in whom homosexual feeling has later undergone full development, all this becomes distorted. The non-sexual motives are forgotten; he believes that even in early childhood he had homosexual inclinations, and that for _this_ reason it gave him pleasure to ride on his uncle's knees. Nor is observation in any way adapted to furnish us with a clear picture of the sexual life of the child. So little can be directly observed, that in the absence of reports much would remain entirely unknown. From the moment when the children gain a consciousness, however obscure, of the nature of sexual processes, they almost invariably endeavour to conceal their knowledge as much as possible, so that we shall discover its existence only by a rare chance. None the less, the results of direct observation are often important; sometimes because we are able to watch children when they are unaware of our attention, and sometimes because they do not as yet fully understand the nature of the processes under observation, and for this reason are less secretive. The third method, that of experiment, is available to us only in the form of castration. I need not dilate on the inadequacy of this application of the experimental method, even apart from the fact that it subserves our purposes almost exclusively in respect of the male sex--for in the case of young girls, castration (oophorectomy) is almost entirely unknown.