Visual Diffusivity. 25 



intensities this phenomenon is less marked, as it is again 

 at very low ones. The prediction of theory is confirmed. 



(c) Bed and blue light. 



The slot Oa is covered by a good blue glass, which trans- 

 mits no red, and the slot Ob by a red signal glass. On 

 rotation the blue image lags behind by several degrees, 

 presenting the appearance already depicted by fig. 5 b, 

 substituting the difference of colour for the difference of 

 intensity which the figure indicated when referring to the 

 last section. 



This lagging of one colour behind the other is a very 

 striking phenomenon, well suited to demonstration to a small 

 audience. The effect is most extraordinary if the disk, 

 •instead of being rotated, is rapidly oscillated through five 

 or ten degrees, when the blue image may be made to vibrate 

 entirely out of phase with the red, appearing as though 

 attached to it by a cord. It is not an effect that has to be 

 glimpsed after painful preparation by a trained eye, but an 

 optical delusion of most convincing verisimilitude. On its 

 being exhibited without comment to the mechanic who made 

 the apparatus, he immediately assumed that he had been 

 called in to correct a mechanical slip between the disks ! 



At the very high intensity obtained by placing a carbon 

 arc lamp about 35 centimetres behind the flashed opal glass 

 of the disk, the lag of blue behind red is practically absent; 

 it was not found possible with the intensities at our disposal 

 to demonstrate a reversal of the red and blue positions, such 

 as theory indicates for soj£\e excessively high intensity, 

 beyond the crossing point m the red and blue diffusivity 

 lines of fig. 1. 



(d) Purple light. 



If the various colour impressions are transmitted by 

 entirely independent channels, it would follow that a narrow 

 line of a mixed colour such as purple ought to be resolved 

 into its constituents when passed across the field of vision. 

 Dr. Karrer informs me that he has observed a moving bright 

 light spread out into a spectrum in a way which would meet 

 with explanation in this way. I was not able, however, to 

 obtain a clear cut and unmistakable resolution of any purple 

 I tried with this apparatus. This agrees with the conclusion 

 drawn from the previous work with the flicker photometer 

 with unequal exposures, namely, that the difference in the 

 diffusivities peculiar to the two colours is much reduced 

 when they are simultaneously transmitted by the same 

 retinal area, although it is still present to some degree, as 



