Visual Diffusivitij. 



29 



made, that the eye be steadily fixed at the centre of rotation 

 of the disks. If the eye follows the image, as its tendency 

 is, because of the faintness of the phenomenon, the after- 

 image is shifted forward, by an amount which can entirely 

 vitiate the measurement. The manner in which this shift 

 occurs is rather interesting, as it shows clearly that the 

 eye- ball does not rotate, but keeps its axis parallel as it 

 follows the bright spot on the disk. The after-image thus 

 does not merely move in closer, but appears to stand at an 

 angle to the primary image, and if the eye-movement is 

 rapid enough it may actually lie diagonally across this, 



as shown in fig. 8. 



Fig. 



Appearance of after-image (shown dashed) ; a, with proper fixation 

 at centre of rotation : b, with improper fixation. 



Accurate fixation at the centre of rotation is assured by 

 having this faintly illuminated. With accurate fixation, 

 however, the difficulties due to the low visibility of the 

 image are rather enhanced, for one's attention is distracted 

 from the part of the field, in which the after-image appears. 

 From one cause or another the illumination region in which 

 any measurements at all can be made on the after-image is 

 quite limited. Thus at a rather high illumination the after- 

 image disappears entirely, as though the rods ceased to 

 function. At an illumination somewhat below this the 

 after-image is fairly bright, and well separated from the 

 primary at moderate speeds. It is here at its best visibility. 

 From here down it becomes less bright, at the same time 

 moving in toward the primary, both changes reducing its 

 visibility. A faint image close to a bright one, whatever 



