56 



STUDIES IN LUMINESCENCE. 



a straight line, and that there is nothing to indicate any systematic vari- 

 ation from linearity. In the case of one curve, namely, that for 0.483 /z 

 in Table 14, the points located in this way did not lie even approximately 

 on a straight line. This curve, therefore, does not have the form corre- 

 sponding to eq. (4). But in the case of the other curves, including one 

 taken at this same wave-length (Table 13), the agreement between the 

 observed and computed values of / is satisfactory, the differences being 

 only such as might well result from accidental errors. It seems probable, 

 therefore, that the observations at 0.483 ju in Table 14 were subject to some 

 unusual source of error. 



In Fig. 44 it will be noticed that one point of the line A ' (indicated by a 

 question mark) falls below the line, i. c, the observed time of decay was 

 considerably greater than that computed from an assumed linear relation 

 between 7 _i and /. Discrepancies of the same kind will be found for the 

 smallest values of / in the first curve of Table 13 and for the first and last 



00 



80 



60 



40 



20 



Fig. 45- 



Curves A and B show the decay of phos- 

 phorescence at 0.512 m for two dif- 

 ferent intensities of the exciting light. 

 In the curves A' and B' I '*- is plotted 

 instead of the intensity /. 



3 4 



Seconds 



curves of Table 15. No such discrepancies are present in any of the four 

 curves taken at 0.512 /j. i. e., near the maximum of the phosphorescence 

 spectrum. If these cases of disagreement between the observed and com- 

 puted values of t are not due to accidental errors, they show the beginning 

 of a systematic variation which will be dealt with at length in a later 

 paragraph. 



The theory discussed in Chapter XV leads us to expect that under ideally 

 simple conditions the decay of phosphorescence should be in accordance 

 with the equation 



where 



7= = a-\-bt 

 V/ 



= 



no^lka 





It is to be remembered that if / refers to the intensity at some particular 

 part of the phosphorescence spectrum, k must be regarded as a function 

 of the wave-length. 



The fact that the experiments indicate a linear relation between the time 

 and /~- has already been pointed out, and reference to the tables on pages 



