(4) _Thought-forms._--There remains one other source of error to be guarded against. An image--whether of an object, diagram, or name--which is _chosen_ by the agent may be correctly described by the percipient simply because their minds are set to move in the same direction. It must be remembered that, however unexpected and spontaneous they may appear, ideas do not come by chance, but have their origin mostly in the previous experience of the thinker. Persons living constantly in the same physical and intellectual environment are apt to present a close similarity in their ideas. It would not even be _prima facie_ evidence of thought-transference, for instance, if husband and wife, asked to think of a town or of an acquaintance, should select the same name. And investigation has shown that our thoughts move in grooves which are determined for us by causes more deep-seated and more general than the accident of particular circumstances. Thus it is found that individuals will show a preference for certain figures or certain numbers over others; and that the preference for some geometrical figures tends to be tolerably constant. The American Society for Psychical Research[11] made some interesting observations on this point in 1888. Blank cards were issued to a large number of persons, with the request that the recipients would draw on the card "ten diagrams." 501 cards were returned, and the diagrams inscribed on them were carefully tabulated. It was found that of the 501 persons no less than 209 drew circles, 174 squares, 160 equilateral triangles and crosses, while three only drew wheels, two candlesticks, and one each a corkscrew, a ball, and a knife. It was found that the simpler geometrical figures[12] occurred not only most frequently but as a general rule early in each series of ten. It follows, therefore, that in an experiment the success of the percipient in reproducing a circle, a square, or a triangle raises a much fainter presumption of thought-transference than if the object reproduced had been a corkscrew or a pine-apple. But so much was perhaps obvious even without a detailed investigation. From a similar analysis of the guesses made, it can be shown that some percipients have decided preferences amongst the simple numerals. And in the same way it seems probable that others have a preference for particular cards. An important illustration of the working of the "number-habit" has been brought forward by Professor E. C. Pickering of the Harvard College Observatory, U.S.A.[13] A revision of part of the Argelander Star-Chart had been undertaken by several observatories, of which the Harvard Observatory was one. For the purposes of the revision the assistant had the Argelander chart before him, whilst the observer, who was in ignorance of the magnitude assigned in the chart, made an independent estimate of the magnitude of each star. If no thought-transference or other disturbing cause