His manifestations may differ very considerably from the automatist's normal personality. Yet in one sense it is a process of selection rather than of addition; the spirit selects what parts of the brain-machinery he will use, but he cannot get out of that machinery more than it is constructed to perform. The spirit can indeed produce facts and names unknown to the automatist; but they must be, as a rule, such facts and names as the automatist could easily have repeated, had they been known to him:--not, for instance, mathematical formulæ or Chinese sentences, if the automatist is ignorant of mathematics or of Chinese. After a time the control gives way, and the automatist's spirit returns. The automatist, awaking, may or may not remember his experiences in the spiritual world during the trance. In some cases (Swedenborg) there is this memory of the spiritual world, but no possession of the organism by an external spirit. In others (Cahagnet's subject) there is utterance during the trance as to what is being discerned by the automatist, yet no memory thereof on waking. In others (Mrs. Piper) there is neither utterance as a rule, or at least no prolonged utterance, by the automatist's own spirit, nor subsequent memory; but there is writing or utterance during the trance by controlling spirits. Now this seems a strange doctrine to have reached after so much disputation. For it simply brings us back to the creeds of the Stone Age. We have come round again to the primitive practices of the shaman and the medicine-man;--to a doctrine of spiritual intercourse which was once œcumenical, but has now taken refuge in African swamps and Siberian tundras and the snow-clad wastes of the Red Indian and the Esquimaux. If, as is sometimes advised, we judge of the worth of ideas by tracing their _origins_, no conception could start from a lower level of humanity. It might be put out of court at once as unworthy of civilised men. Fortunately, however, our previous discussions have supplied us with a somewhat more searching criterion. Instead of asking in what age a doctrine originated--with the implied assumption that the more recent it is, the better--we can now ask how far it is in accord or in discord with a great mass of actual recent evidence which comes into contact, in one way or another, with nearly every belief as to an unseen world which has been held at least by western men. Submitted to this test, the theory of possession gives a remarkable result. It cannot be said to be inconsistent with any of our proved facts. We know absolutely nothing which negatives its possibility.