Polarization Flicker Photometer. 379 



the mean o£ the red and green light critical speed points, 

 and the lower line is the critical speed plot for red and 

 green alternated (the mean illumination being the same as 

 for each alone) at their equality setting. The relationship 

 between critical speed and log brightness shows a well- 

 marked deviation from the rectilinearity which would result 

 were both the above assumptions correct. This is not sur- 

 prising, for hue difference discrimination for steady observa- 

 tion varies quite rapidly with brightness, and in the flicker 

 photometer the working speed at low intensities will be 

 larger than that predicted from these assumptions owing to 

 the pronounced lag of one colour impression behind another * 

 and the consequent lack of exact dove-tailing. 



In spite of the failure of relation (14) to hold exactly, it 

 gives the order of magnitude of &)m so closely that the result 

 of using (15) with £ H as a constant will be correct to within 

 the probable errors of any experimental test. When used 

 for calculations similar to those made in the first paper with 

 ay assumed constant, the resultant predicted behaviour of the 

 flicker photometer is quite similar, the deviations from the 

 equal brightness position of setting being of course smaller, 

 but decreasing with increased intensity as before. It may, 

 in fact, be questioned whether the calculations made on the 

 constant speed assumption do not more nearly represent 

 the experimental conditions holding when the data with 

 which they were compared were obtained, since some 

 mechanical flicker undoubtedly prevented the use of as 

 low speeds as the colour differences used would warrant. 

 It is to be hoped that data of this sort collected in future 

 will be made with the polarization flicker photometer. 



It may be noted, in passing, that the use of a variable 

 instead of a constant speed in the calculation of the be- 

 haviour of the flicker photometer helps in the discussion of 

 the unequal exposure instrument treated in the second paper. 

 For by reducing the value of the constant F in equation (14) 

 of that paper the relative intensities necessary for a setting 

 of the instrument are brought nearer together, in agreement 

 with the experimental data. 



10. Summo.ry. 



The results of this investigation may be summarized as 

 follows : — 



1. A polarization flicker photometer has been constructed, 

 in which the transition from one field to the other is gradual 



* See "Visual Diffusivity," Ives, Phil. Mag. Jan. 1917. 



