212 TRANSACTIONS OF THE [mAR. 1, 



sents no uniform back-ground. True flickering was found to be 

 no more tiring on the eyes than ordinary photometric work. 

 The phenomenon of flashing, on the other hand, was quite painful. 



It has been proved that under ordinary daylight illumination 

 true flickering cannot be produced by differences in color. For 

 the production of this phenomenon a diff'erence in luminosity 

 was found to be necessary, and the experiments also showed that 

 as the difljerence in luminosity was decreased the violence of the 

 flicker was also diminished. 



Furthermore, the experiments showed that when two flickers 

 are produced, the one by the rotation of a surface B, in combi- 

 nation with a surface A, more luminous than B, the other by the 

 rotation of the same surface B, in combination with a surface C, 

 less luminous than B by the same amount, the two flickers will 

 be of practically the same violence so long as the diflference in 

 luminosity between A and C is not greater than about 20 per 

 cent. 



Upon the last two conclusions are based two general methods 

 for the application of the phenomenon of flickering directly to 

 photometric work. The two methods are : the method of Least 

 Flicker and the method of Equal Flickers, respectively. The re- 

 sults obtained by the two methods have been shown to be har- 

 monious. The relative advantages of the two methods will de- 

 pend largely upon the nature of the work in hand. 



The experimental proof of Grassman's law which the new 

 flicker photometry enables one to make has been verified. The 

 application of this law to the determination of the luminosities 

 of colored surfaces by the aid of a few standardized disks was 

 suggested but not tried by Professor Rood. This method was 

 tested experimentally and found to give results which corre- 

 sponded with those obtained by either of the direct flicker meth- 

 ods. Consequently, investigators are enabled by the use of a few 

 standard disks to determine the luminosity of any colored surface. 



The experiments also indirectly showed that the flicker 

 method can, with great convenience and delicacy of results, be 

 applied to the comparison of the luminosities of surfaces of the 

 same color. In brief, the results of the experiments show that 

 the new flicker photometry will enable us to compare the lumi- 

 nosities of differently colored surfaces with the same ease and 

 exactness as was previously only attainable in the comparison of 

 surfaces of like hue. 



I wish here to express my thanks to Professor Rood and Pro- 

 fessor Hallock for assistance very kindly furnished during the 

 investigation. 



Phsyical Laboeatory of Columbia University, 

 April 2, 1897. 



