When I call up an image of the year just closing, what really happens is a rapid movement of imagination over a series of prominent events, among which the succession of seasons probably occupies the foremost place, serving, as I have remarked, as a framework for my retrospective picture.  Each of the events which I thus run over is really a long succession of shorter experiences, which, however, I do not separately represent to myself.  My imaginative reproduction of such a period is thus essentially a greatly abbreviated and symbolic mode of representation.  It by no means corresponds to the visual imagination of a large magnitude, say that of the length of sea horizon visible at any one moment, which is complete in an instant, and quite independent of a successive imagination of its parts or details.  It is essentially a very fragmentary and defective numerical idea, in which, moreover, the real quantitative value of the units is altogether lost sight of.

Now, it seems to follow from this that there is something illusory in all our recallings of long periods of the past.  It is by no means strictly correct to say that memory ever reinstates the past.  It is more true to say that we see the past in retrospect as greatly foreshortened.  Yet even this is hardly an accurate account of what takes place, since, when we look at an object foreshortened in perspective, we see enough to enable us imaginatively to reconstruct the actual size of the object, whereas in the case of time-perspective no such reconstruction is even indirectly possible.

It is to be added that this constant error in time-reproduction is greater in the case of remote periods than of near ones of the same length.  Thus, the retrospective estimate of a duration far removed from the present, say the length of time passed at a particular school, is much more superficial and fragmentary than that of a recent corresponding period.  So that the time-vista of the past is seen to answer pretty closely to a visible perspective in which the amount of apparent error due to foreshortening increases with the distance.

In practice, however, this defect in the imagination of duration leads to no error.  Although, as a concrete image answering to some definite succession of experiences a year is a gross misrepresentation, as a general concept implying a collection of a certain number of similar successions of experience it is sufficiently exact.  That is to say, though we cannot imagine the absolute duration of any such cycle of experience, we can, by the simple device of conceiving certain durations as multiples of others, perfectly well compare different periods of times, and so appreciate their relative magnitudes.