We may conveniently proceed in our detailed examination of illusions of memory, by distinguishing between three facts which appear to be involved in every complete and accurate process of recollection. When I distinctly recall an event, I am immediately sure of three things: (1) that something did really happen to me; (2) that it happened in the way I now think; and (3) that it happened when it appears to have happened. I cannot be said to recall a past event unless I feel sure on each of these points. Thus, to be able to say that an event happened at a particular date, and yet unable to describe how it happened, means that I have a very incomplete recollection. The same is true when I can recall an event pretty distinctly, but fail to assign it its proper date. This being so, it follows that there are three possible openings, and only three, by which errors of memory may creep in. And, as a matter of fact, each of these openings will be found to let in one class of mnemonic illusion. Thus we have (1) false recollections, to which there correspond no real events of personal history; (2) others which misrepresent the manner of happening of the events; and (3) others which falsify the date of the events remembered. It is obvious, from a mere glance at this threefold classification, that illusions of memory closely correspond to visual illusions. Thus, class (1) may be likened to the optical illusions known as subjective sensations of light, or ocular spectra. Here we can prove that there is nothing actually seen in the field of vision, and that the semblance of a visible object arises from quite another source than that of ordinary external light-stimulation, and by what may be called an accident. Similarly, in the case of the first class of mnemonic illusions, we shall find that there is nothing actually recollected, but that the mnemonic spectra or phantoms of recollected objects can be accounted for in quite another way. Such illusions come nearest to hallucinations in the region of memory.