Imagine an energy system of prodigious complexity and extreme delicacy of organisation which has an indefinitely large number of stable poises. Imagine it thrown from one poise to another with great facility, each poise being the resultant of all the energies of the system. Suppose now that the _partial_ return of a situation which has formerly caused it to assume a stable poise, throws it into an unstable condition, from which it most easily returns to equilibrium by reassuming the former poise. Such a system would exhibit the phenomena of memory; but it would keep no records though appearing to do so. The appearance would be due merely to the extreme accuracy and sensitiveness of the system and the delicacy of its balances. Its state on the later occasion would appear to be a _revival_ of its state on the former, but this would not be the case any more than a cumulus cloud this evening is a revival of those which decorated the heavens last year. This imaginary construction can be made more concrete by imagining a solid with a large number of facets upon any one of which it can rest. If we try to balance it upon one of its coigns or ridges it settles down upon the nearest facet. In the case of the neural system we are trying to suggest each stable poise has been determined by a definite set, or better, context of conditions. Membership of this context is what corresponds to nearness to a facet. The partial return of the context causes the system to behave as though conditions were present which are not, and this is what is essential in memory. That this suggestion in the form here presented is unsatisfactory and incomplete is evident. It is wildly conjectural no doubt, but so are the Archival and Pathway Theories. Yet it does avoid the chief deficiencies of those theories, it does suggest why only some conjunctions of experiences become ‘associated’, those namely which yield a stable poise. And it suggests why a thing should be recognised as the same though appearing in countless different aspects; every time it appears different conditions occur which, none the less, lead to one and the same stable poise, as the polyhedron we imagined may settle down on one and the same facet from all the surrounding ridges. One of the collateral advantages of such a view is that it removes some of the temptations to revert to animism from which psychologists, and especially literary psychologists, suffer. Dissatisfaction with current hypotheses as to the mechanism of reflex arcs is a main cause for the scientifically desperate belief in the soul. And apart from this, the special emotive factors which disturb judgment on this point are less obtrusive when this account is substituted for the usual story of the conditioned reflex, that sacrilegious contrivance of the mechanists.