316 Messrs. H. E. Ives and E. F. Kingsbury on the 



disks should be avoided in favour of opaque disks over a 

 white background, where such troubles are absent. 



Another serious effect of accidental dissymmetry was found 

 in our experiments upon the nicker photometer with unequal 

 exposures. In passing over from measurements with equal 

 exposures to ones with a ratio of 11 to 1, as described under 

 the previous section, it was found that the nicker setting 

 loitli no colour difference was changed by as much as 15 per 

 cent. The correct setting with no colour difference must, of 

 course, be the same with all openings, since it is the same as 

 the equality of brightness or non-flicker setting. 



Here, again, it was found that the fault lay in non-uni- 

 formities of the rotating-disk surface. Some of the experi- 

 mental results obtained in searching for the explanation are 

 shown in fig. 13. First of all consider the points marked 

 by circles. These were obtained with a carefully smoked 

 180-degree disk. The centre line is the equality of bright- 

 ness point as obtained by using one edge of the disk. It 

 will be seen that the flicker settings are displaced, as a whole, 

 slightly to the left from this centre line. After these readings 

 were made one of the 90-degree sectors of the rotating disk 

 was unsmoked, that is, the magnesia was rubbed off, leaving 

 only the white paint base, of probably 10 per cent, lower 

 reflecting-power. Critical speed readings were then made 

 as before, with the rather striking result of showing that 

 no minimum of flicker was present. In short, the arrange- 

 ment could no longer be used as a flicker photometer. The 

 critical speed for the disk unilluminated was as before 

 (necessarily) 32'5 cycles per second, from which it rose 

 uniformly, being 42 cycles per second at the equality of 

 brightness point of setting. 



The points marked by crosses show the more extensive 

 data obtained from the 330-degree white disk with which 

 the effects were first noticed. The lowest curve shows the 

 points found, using a very carefully smoked disk. The speed 

 at the minimum of flicker is quite high, and the median of 

 the figure is much displaced from the point which would 

 have been selected by equality of brightness setting at one 

 edge. (The equality of brightness setting is probably of 

 little significance owing to the small portion of the tota] 

 disk which figures in making that setting.) The next curve 

 above this, which of course starts from the same point at the 

 left of the diagram, was obtained from the same disk re- 

 smoked in an attempt to better the surface. Actually the 

 condition was made worse, as shown by the result. The 

 speed at the minimum is much greater, and the minimum 



