occasions of change. In accordance with the principle which is directing our attempt, we shall therefore ascribe these two functions to two different systems. We assume that an initial system of this apparatus receives the stimuli of perception but retains nothing of them- that is, it has no memory; and that behind this there lies a second system, which transforms the momentary excitation of the first into lasting traces. The following would then be the diagram of our psychic apparatus: (See illustration.)

We know that of the percepts which act upon the P-system, we retain permanently something else as well as the content itself. Our percepts prove also to be connected with one another in the memory, and this is especially so if they originally occurred simultaneously. We call this the fact of association. It is now clear that, if the P-system is entirely lacking in memory, it certainly cannot preserve traces for the associations; the individual P-elements would be intolerably hindered in their functioning if a residue of a former connection should make its influence felt against a new perception. Hence we must rather assume that the memory-system is the basis of association. The fact of association, then, consists in this- that in consequence of a lessening of resistance and a smoothing of the ways from one of the mem-elements, the excitation transmits itself to a second rather than to a third mem-element.

On further investigation we find it necessary to assume not one but many such mem-systems, in which the same excitation transmitted by the P-elements undergoes a diversified fixation. The first of these mem-systems will in any case contain the fixation of the association through simultaneity, while in those lying farther away the same material of excitation will be arranged according to other forms of combination; so that relationships of similarity, etc., might perhaps be represented by these later systems. It would, of course, be idle to attempt to express in words the psychic significance of such a system. Its characteristic would lie in the intimacy of its relations to elements of raw material of memory- that is (if we wish to hint at a more comprehensive theory) in the gradations of the conductive resistance on the way to these elements.

An observation of a general nature, which may possibly point to something of importance, may here be interpolated. The P-system, which possesses no capacity for preserving changes, and hence no memory, furnishes to consciousness the complexity and variety of the sensory qualities. Our memories, on the other hand, are unconscious in themselves; those that are most deeply impressed form no exception. They can be made conscious, but there is no doubt that they unfold all their activities in the unconscious state. What we term our character is based, indeed, on the memory- traces of our impressions, and it is precisely those impressions that have affected us most strongly, those of our early youth, which hardly ever become conscious. But when memories become conscious again they show no sensory quality, or a very negligible one in comparison with the perceptions. If, now, it can be confirmed that for consciousness memory and quality are mutually exclusive in the Psi-systems, we have gained a most promising insight into the determinations of the neuron excitations. *

* Since writing this, I have thought that consciousness occurs actually in the locality of the memory- trace.

What we have so far assumed concerning the composition of the psychic apparatus at the sensible end has been assumed regardless of dreams and of the psychological explanations which we have hitherto derived from them. Dreams, however, will serve as a source of evidence for our knowledge of another part of the apparatus. We have seen that it was impossible to explain dream- formation unless we ventured to assume two psychic instances, one of which subjected the activities of the other to criticism, the result of which was exclusion from consciousness.

We have concluded that the criticizing instance maintains closer relations with the consciousness than the instance criticized. It stands between the latter and the consciousness like a screen. Further, we have found that there is reason to identify the criticizing instance with that which directs our waking life and determines our voluntary conscious activities. If, in accordance with our assumptions, we now replace these instances by systems, the criticizing system will therefore be moved to the motor end. We now