If the observer is asked what are the letters that he is sure of having seen, these may be, of course, the letters really written, but there may be also absent letters, either letters that we replaced by others or that have simply been omitted. Thus an observer will see quite distinctly in full light a letter which does not exist, if this letter, on account of the general sense, ought to enter into the phrase. The characters which have really affected the eye have been utilized only to serve as an indication to the unconscious memory of the observer. This memory, discovering the appropriate remembrance, _i.e._, finding the formula to which these characters give a start toward realization, projects the remembrance externally in an hallucinatory form. It is this remembrance, and not the words themselves, that the observer has seen. It is thus demonstrated that rapid reading is in great part a work of divination, but not of abstract divination. It is an externalization of memories which take advantage, to a certain extent, of the partial realization that they find here and there in order to completely realize themselves. Thus, in the waking state and in the knowledge that we get of the real objects which surround us, an operation is continually going on which is of quite the same nature as that of the dream. We perceive merely a sketch of the object. This sketch appeals to the complete memory, and this complete memory, which by itself was either unconscious or simply in the thought state, profits by the occasion to come out. It is this kind of hallucination, inserted and fitted into a real frame, that we perceive. It is a shorter process: it is very much quicker done than to see the thing itself. Besides, there are many interesting observations to be made upon the conduct and attitude of the memory images during this operation. It is not necessary to suppose that they are in our memory in a state of inert impressions. They are like the steam in a boiler, under more or less tension.