1897.] 



NEW YORK ACADEMY OF SCIENCES. 



211 



very slightly the more violent of the two. The discrepancy can 

 be readily explained if we assume that the measured differences 

 in luminosity are in error by a little less than one per cent. 

 This error, since it can be distributed among the determined 

 luminosities of the three disks, would give a probable error of 

 less than 0.5 per cent, for the determined luminosity of each 

 disk. This is within the accuracy claimed for the determined 

 luminosities of the disks. 



That the above discrepancy is due to slight errors in the de- 

 termined luminosities of the colors is shown by the following 

 experiment. A large red disk, Red 1 (See Fig. VI.), and a large 

 blue disk. Blue 3, were selected so that the difference in their 

 measured luminosities was 2.1 per cent. A small red disk. Red 

 2, was found which differed in luminosity from the disk. Red 1, 

 by 1.7 per cent. These three disks were combined as is shown 

 in Fig. VI., and rotated on the axis of the machine. The inside 

 flicker was found to be more violent than the outside flicker, al- 

 though on the inside there is no difference in color and a less dif- 

 ference in measured luminosity' than on the outside. The discrep- 

 ancy is, however, readily explained by the assumption of a slight 

 error in the determined luminosities of each of the three disks. 



The last three experiments seem to show, beyond a doubt, that 

 under daylight illumination, difference in color cannot, of itself, 

 produce an appreciable llicker and that for the production of 

 this plienomenon there must be a difference in luminosity. 



Fig. VI. 

 Summary. 



The results of the present research may be summed up briefly 

 as follows : 



In the application of the phenomenon of flickering to photo- 

 metric measurements it was found necessary to distinguish care- 

 fully between true flickering which occurs only on a background 

 •of uniform hue and the allied phenomenon of flashing which pre- 



