But not without a few words of critical reflection in conclusion. I might be asked whether I am myself convinced of the views here set forward, and if so how far. My answer would be that I am neither convinced myself, nor am I seeking to arouse conviction in others. More accurately: I do not know how far I believe in them. It seems to me that the affective feature ‘conviction’ need not come into consideration at all here. One may surely give oneself up to a line of thought, and follow it up as far as it leads, simply out of scientific curiosity, or—if you prefer—as advocatus diaboli, without, however, making a pact with the devil about it. I am perfectly aware that the third step in the theory of instinct which I am taking here cannot claim the same certainty as the two former ones, viz. the extending of the conception of sexuality and the establishing of narcissism. These innovations were direct translations of observation into theory, subject to no greater sources of error than is inevitable in anything of the kind. The assertion of the regressive character of instinct rests also, it is true, on observed material, namely on the facts of the repetition-compulsion. But perhaps I have over-estimated their significance. At all events there is no way of working out this idea except by combining facts with pure imagination many times in succession, and thereby departing far from observation. We know that the final result becomes the more untrustworthy the oftener one does this in the course of building up a theory, but the precise degree of uncertainty is not ascertainable. One may thereby have made a brilliant discovery or one may have gone ignominiously astray. In such work I trust little to so-called intuition: what I have seen of it seems to me to be the result of a certain impartiality of the intellect—only that people unfortunately are seldom impartial where they are concerned with the ultimate things, the great problems of science and of life. My belief is that there everyone is under the sway of preferences deeply rooted within, into the hands of which he unwittingly plays as he pursues his speculation. Where there are such good grounds for distrust, only a tepid feeling of indulgence is possible towards the results of one’s own mental labours. But I hasten to add that such self-criticism does not render obligatory any special tolerance of divergent opinions. One may inexorably reject theories that are contradicted by the very first steps in the analysis of observation and yet at the same time be aware that those one holds oneself have only a tentative validity. Were we to appraise our speculations upon the life and death-instincts it would disturb us but little that so many processes go on which are surprising and hard to picture, such as one instinct being expelled by others, or turning from the ego to an object, and so on. This comes only from our being obliged to operate with scientific terms, i. e. with the metaphorical