What is in question here is not memory, in the stricter sense in which past experience is dated and placed, but free reproduction. To be able to revive an experience is not to remember when and where and how it occurred, but merely to have that peculiar state of mind available. Why some experiences are available and others not is unfortunately still a matter for conjecture merely. The difficulty upon most accounts, Semon’s for example, is to explain why _all_ our past experience is not being revived all the time. But some plausible conjectures are not difficult to make, and the absence of clear evidence or conclusive proof should not prevent our making them if they are recognised for conjectures. How far an experience is revivable would seem to depend in the first place upon the interests, the impulses, active in the experience. Unless similar interests recur its revival would seem to be difficult. The original experience is built upon a number of impulses; it came about only through these impulses. We may even say that it is those impulses. The first condition for its revival is the occurrence of impulses similar to _some_ of these. The patient in the asylum occupied in reliving the same piece of experience indefinitely does so (if he does) because he is limited very strictly in the range of his possible impulses, other impulses not being allowed to intervene. Hence the completeness with which he is said to reconstruct the past. Most revival is distorted because only some of the original impulses are repeated, new impulses being-involved and a compromise resulting. The impulses implicated in experiences may be many and varied or few and alike. An experience which has a very simple impulse structure will, we may suppose, tend to come back only when these impulses are again relatively dominant. Other things being equal it will have less chance of revival than an experience with a more complex structure. Recalling the illustration used in Chapter XIV, the broader the facet the more numerous are the positions from which the polyhedron will settle down on that facet. It is a first principle of psychology that the partial return only of a situation may reinstate the whole, and since most impulses have belonged in the past to many varied wholes there must evidently be much rivalry as to which wholes do actually recur. What seems to decide the dispute more than anything else is the character of the original connections between the parts. As has recently been emphasised by the exponents of _Gestalt-psychologie_, mere original contiguity or simultaneity is comparatively powerless to control revival. Compare the learning of a geometric theorem by heart with understanding it, or even a brief study of some building with mere daily familiarity.