28 Dr. H. E. Ives on 



course varied greatly with the illumination, due to the 

 Purkinje phenomenon. 



It is of considerable importance that the eye be properly 

 fixated at the centre of rotation o£ the disks. The best means 

 found for insuring this was to faintly illuminate the chalk- 

 marked centre of the axle by a beam of light from a distant 

 light and mirror. 



The observations were made by the chief observer in a 

 carefully darkened booth. Speeds of rotation were regulated 

 by an adjustable resistance in series with the disk motor. 

 Speeds were read and lamp positions changed by an 

 assistant *. 



In this work the artificial pupil heretofore consistently 

 used was omitted because it was found that, owing to the 

 chromatic aberration of the eye, slight unavoidable eye- 

 movements cause relative movements of the red and blue 

 images similar in appearance to those due to the diffusivity 

 differences, and hence likely to be confused with them in 

 measurement. 



(b) Measurements on red and blue light. 



No measurements were made on two lights of different 

 intensity (case 1 above), as the case of two colours practi- 

 cally includes it, and is more striking and easily measurable, 

 because of the approximate equality of brightness which may 

 be arranged for in the two colours under observation. 



A series of values of the time-interval between the red 

 and blue light used, for different illuminations, is shown in 

 fig. 7. In making this series the angle between Oa and Ob 

 was one degree at the higher intensities and two degrees at 

 the lower. The employment of larger angles and hence 

 higher speeds, which would at first thought appear prefer- 

 able, was actually found to be rather less conducive to 

 accuracy because the images begin to spread out and so 

 lose in definiteness of outline. 



The predicted behaviour of the two colours with respect 

 to each other is borne out by the measurements. 



(c) Measurements on after-images. 



Measurements on the positive after-image, or slowly 

 transmitted rod image as it is here considered to be, are 

 extremely difficult. Two kinds of difficulty may be men- 

 tioned — those of fixation and those of visibility. It is 

 absolutely necessary, if accurate measurements are to be 



* Acknowledgment is due Mr, E. R. Morton for assistance in the 

 ■experimental work. 



