1911.] 



Eye to Variations of Wave-length. 



119 



Then, the lens being removed and the slits in the two spectra adjusted so 

 that their width corresponded to 3 /jl,u, the movement of the first slit 

 necessary to produce an observable difference in hue, first in one direction 

 and then in the other, was determined. In this way the two sets of observa- 

 tions were made under exactly the same conditions as to illumination, size 

 of image on the retina, state of dark adaptation, etc. "When determining 

 the change in wave-length required to produce an observable difference in 

 hue, the slit was always moved till the difference in hue was quite distinct, 

 and no hesitation was felt as to which patch was, say, on the red side. In 

 general, the observer did not move the slit, so that he did not know in which 

 direction the change in colour would take place. 



Fourteen observers were tested in the above manner, and their readings, 

 which agree very well together, give the following mean values : — 



Difference in wave-length easily detected as a change in hue 1*4 fi/x. 



It will be observed that there is a very marked difference, and that when 

 the eye is not dealing with a continuous variation in hue, as is the case when 

 a portion of the spectrum is observed, a very much smaller difference in 

 wave-length is apparent as a difference in hue, and this even when the 

 conditions are as nearly as possible identical. 



This point was also investigated in a somewhat different manner. The 

 monochromatic patch having been projected on the screen, a rod was inter- 

 posed so as to cut out a portion corresponding to 1*9 fifi difference in wave- 

 length from the middle, and the two remaining portions were then brought 

 together by means of a Fresnell biprism. When this was done the difference 

 in hue was most marked, though on removing the rod the patch again looked 

 monochromatic. 



An attempt to repeat Dr. Edridge-Green's experiment, using a biprism, 

 failed ; since there was sufficient polarisation of the light produced by the 

 prisms of the spectroscope to cause such a difference in brightness of the two 

 images as to mask any difference in hue. 



To investigate what effect, if any, the presence of white light mixed with 

 the colours would produce on the perception of a difference of hue, arrange- 

 ments were made by which a known amount of white light fell on both halves 

 of the screen. Different quantities of white were added, and in the following 

 table 100 parts of white were of the same luminosity as the colour in each 

 case. Observations were made with yellow (D), red and green light. No 



Sodium Light. 



Width of monochromatic patch 



4"5 fi/n. 



