1408 
Finally all the 83 observations may be combined; then 0,242 
+ 0,004 is found. 
When a result is calculated for each of the 24 plates separately, 
and then the mean is taken, the effect is found to be 0,243 + 0,006. 
The number after the + sign indicates the mean error, and as 
was stated before all the results have been reduced to the velocity 
1000 cm./sec. 
Theoretical value of the effect for flint-glass. 
The firm ScHorr und GenossEN, Jena, gives the following indices 
of refraction for the flint-glass 0.103 used : 
wa’ = 1.6099 417677 A° 
he == 10193 6563 
up’ = 1.6202 5893 
up = 1.6324 4862 
" ugi= 6428 4341 
The effect can be calculated according to the formula derived 
before (Il, 9) 
l 
By the aid of the given data, values for u and = are derived 
for the effective wave-length 4750. The final result is: 
ee 4.120. 1000 
~ 4750 x 10-8 x 3.1010 
This value is almost in perfect agreement with the value 
yielded by our experiments. 
It is of interest to note that the dispersion term contributes to 
the value of the effect — 0,242 by an amount of 0,028. 
If the dispersion had not been taken into account, 0,214 would 
have been found for the effect, which is incompatible with the 
experiments. 
(1, 6834—1 + 0,084) = 0,242, 
APPENDIX. 
1. The publication of the above communication, which was already 
laid before the meeting of the Academy of April 23rd 1920, has 
been delayed through particular circumstances. 
This affords me an opportunity to add a few remarks to the paper. 
Our collaborator, Mr. W. pr Groot, phil. docts. orally informed 
me of another derivation of the formula for the optical effect (II, 9), 
