I have based my discussion of mental telegraphy almost wholly upon the occurrence of coincidences (using that term not as the equivalent wholly of chance occurrences, but including suggestive or interesting conjunctions of circumstances in general), for the reason that coincidences--both those of a commonplace character and those that seem to possess a striking personal significance--have prepared the popular mind for the acceptance of the telepathic hypothesis, and still constitute the most formidable array of evidence presented for that hypothesis. The other class of evidence to be considered is the experimental, which may be said to include as its most distinctive type the results of tests in which intentional attempts at mental telegraphy were made under definite conditions and usually with specially selected subjects; and as another type, the precise verification and registration of presentiments and peculiar and startling "psychic experiences" with reference to their coincidence with death, accidents, and other serious events in life. It may be admitted that the experimental data are equally worthy with the others of a logical analysis, and indeed that they present in some respects different and more favorable conditions for the application of such an analysis. In general, however (and I desire to confine this discussion to the general principles involved and not to the analysis of special cases), the considerations that determine the logical value, or the lack of it, of the one type of evidence are applicable without undue modification to the other. Nor do I consider that the experimental data have seriously modified the logical status of the problem as a whole; nor that they have, except in relatively few cases, been of themselves sufficient to make converts to a belief in telepathy. They have undoubtedly very much strengthened and disseminated that belief; but this implies that a favorable disposition to the belief was already present. It is because it seems to me that the presence of this favorable disposition, albeit in suppressed or half-acknowledged form, is in most cases due to some phase of the argument from coincidences, that I have made it central in my discussion. I must not fail to point out, however, that experiments in thought-transference have one important, and that a logical, advantage over observations of coincidences; this is the possibility which they present of quite accurately allowing for the effect of chance. In coincidences the estimate of chance as the source of the conjunction of events is frequently, if not always, a matter of complex judgment over which serious differences of opinion will occur. Some of the published quantitative estimates made by serious and able students of such problems, of the probabilities that certain coincidences have been due to chance have been pronounced altogether wide of the mark and even absurd by others. In experiments arranged with due precautions there can be no uncertainty