Believers in spirit photographs generally consider that they are playing their trump card when they point out that thousands of “extras” have been definitely recognised by sitters as portraits of their deceased friends or relatives. But this card, impressive as it looks, will not really take the trick. If it could be shown (i.) that a given “extra” was _unmistakably_ recognisable as a portrait of a deceased--or even of a living--person, and (ii.) that the medium concerned could not possibly have obtained a likeness of that person to work from, then we should be obliged to attach great weight to this factor, even if the conditions were not otherwise such as to exclude fraud. For such a result could not be fraudulently produced. But in spite of the perfectly honest assertions of many investigators, it seems very doubtful whether this state of affairs has ever been realised. There are two ways in which evidence based on recognition may be defective. First, the recognition may be perfectly well founded, but the “extra” may have been derived from an existing photograph of the deceased; second, and more frequently, the recognition is illusory and exists only in the sitter’s imagination. As regards the first of these points, it should be remembered that most people are photographed at one time or another, some of them frequently, and that it is not very difficult to obtain a photograph of a given person if one goes about it in the right way. A spirit photographer with an extensive _clientèle_ will find it well worth his while to take the necessary steps to secure photographs appropriate to at any rate his more regular sitters, from whom, in the course of a few _séances_, it will not be difficult to glean enough information to put him on the right track. It is, of course, particularly easy if they happen to be well-known people, photographs of whose relatives may have appeared from time to time in the press. But although this method may sometimes be employed where circumstances lend themselves thereto, or when there is some reason which makes a first-rate “test” especially desirable, I do not think that it is responsible for more than a small percentage of the recognitions which are claimed. By far the greater proportion appear to be due to the operation of subjective factors which lead the sitter to “recognise unmistakably” an extra which bears no more than a vague general resemblance to the person whom it is claimed to represent. Recognition can scarcely be assessed objectively; it is essentially a subjective affair, and as such liable to all the distorting factors which affect every mental process.