1246 
the absolute values of the differences in phase, observed in Fizrau’s 
experiment. 
From the table in my second communication on the experiment 
of Fizeav (These Proceedings Vol. 18, 405. 1915), I take these data: 
p = 2,14 K.G /em? v, == 465 em./sec. Umar == 5538.6 em. /sec. 
EE | i: 
in ACE. PE AL Dexp 
4500 0.786 0.825 | 0.826 + 0.007 
4580 0.771 0.808 0.808 + 0.005 
5461 0.637 0.660 | 0.656 + 0.005 
Under Ap, en A; the shifts of the interference fringes are given 
here without and with the dispersion term. Under A, are to be 
found the observed shifts together with the probable error in the 
final determination. 
Now the values under A; and Ap, have been calculated with 
the value y—0.840. For other values of g the results are given in 
the following table: 
4 = 4500 
| il 
AN Ar P | Aap 
0.771 | 0.810 | 0.856 
0.779 0.817 0.848 
0.786 0.825 0.840 0.826 + 0.007 
0.794 | 0.833 0.832 
0.802 0.841 0.824 
From this table it is evident that no plausible value of can be 
indicated for which the values under Ap, would agree with the 
results of the experiment. With the theory of Lorentz, however, for 
the neigbourhood of gy —0.843 the measured differences in phase 
are in extremely good agreement as to their absolute values. Already 
in my above cited communication I came to the conclusion of the 
necessity of LoreNtz’s dispersion term. This was based upon the ratio 
of the measured shifts independently as well of the value of p as of the 
length of the whole tube. We may say however that now the 
