1018 
the correct one. Then the density follows from equation (6). I have 
carried out the calculations, and collected the results in table Il. The 
density appears to be about 11 on an average, much higher, therefore, 
than according to Targonski. Probably a layer of a lighter substance 
forms at the surface, but the chief component remains mercury. 
In fig. 4 the points indicate the places that the observations occupy 
in the whole of the experiments, when I start from the supposition 
that the density is 11. In fig. 5 I have enlarged the first part of 
Fig. 5. 
the preceding figure in order to show where the points would lie 
when we had to do with pure mercury. This has been indicated 
by points. The crosses give the place for @=7, which is still 
too high according to TARGONSKI. 
7. To examine whether the formula of EINSTEIN-CUNNINGHAM holds 
also numerically, we can calculate the value of N from curve 2 of 
fig. 4 with the formula: 
=i eter eee 
A Fant NS 
i; find tor sab N= 698 102% 
Among the different calculations of N that of SOMMERFELD, from 
the theory of quanta of the spectral lines, may be considered as the 
