PHYSIOLOGICAL FACTOES OF THE ATTENTION-PEOCESS. 345 blackened and is set up in the dark room so that the observer, seated at a distance of 2 metres, sees only the four brass- rods (and a portion of the steel axis) rotating in a plane a little oblique to his sagittal plane, against a bright background formed by a sheet of parchment paper illuminated from behind. If no other light falls on the windmill and the observer keeps one eye closed, it is impossible for him to discover in which direction the rotation takes place, whether the wheel is, as it were, rolling towards him or away from him. If he is unaware of the nature of the experiment he is equally likely to perceive at first either kind of motion, and when after a period, which varies much in duration with different observers, he perceives a sudden change in the direction and plane of rotation he, in most cases, assumes it to be an ob- jective change, the rotation in one direction being just as clearly perceived and seemingly real as that in the reverse direction. The change of phase having once taken place and the rotation in both directions having been observed, the two phases then alternate at approximately regular intervals so long as the subject continues to observe the wheel with one eye only. If the wheel is looked at with both eyes open the subject sees the actual direction of rotation only, because binocular perception of the relative distances of the spokes of the wheel tends to prevent the illusory perception. On prolonged observation with both eyes I have observed occa- sional brief appearances of the illusory phase of notion. No one of the eight subjects who have kindly lent themselves to these experiments has observed the illusory phase during binocular vision, but that is probably owing to the fact that they have in no case been asked to prolong the observation beyond a period of two minutes. In order to study more accurately the alternation of phases graphic records were made by the help of the apparatus described above (vol. xii., p. 478). These records showed the following results : (1) the rate of alternation during passive observation varies considerably among the eight subjects, the periods of the phases varying from three seconds to about twenty seconds, in most subjects being about five seconds. 1 (2) The periods become briefer as observation with, one eye is prolonged. (3) All the subjects were able to exercise some degree of voluntary control of the phases, i.e., they could prolong the periods of one phase, but this prolongation was
- It seems probable that these differences indicate some important
differences of cerebral constitution hitherto overlooked.
