950 
smaller than the smallest percentage assumed thus far in our calcu- 
lations i.e. 0.00055 °/,. 
This gives as composition : 
0 20 40 60 80 100 120 140 KM. 
N 78.03 81:8 875 90.8 826 28:9 2.5 0.2 
O 20.99 ded 123 8.3 4.9 tst 0.1 — 
Ar 0.94 0.5 0.2 = = 
H 0.00055 —— — OTO ORG ASIO RIS 
He 0.00015 — 0.1 L.4 6.8 7.9 6.7 
and for the sound rays: 
Y- u H. D. 
42°33’ 100 184 
27°41’ 110 139 
21° 8’ 120 142 
195204 130 223 
The limit is found here at 136 KM. for an initial angle of 25°. 
It is this calculation which served as a basis for fig. 8. 
By a rough interpolation we find that the limit of 160 KM. would 
be obtained at a percentage of hydrogen, amounting to 0.0001 at 
the surface. It would be useless to make this estimate more accurate, 
as in all preceding calculations certain influences, e.g. the curvature 
of the earth’s surface, have been left out of consideration, and the 
results therefore have no absolute value. Moreover it appears that 
at present a larger percentage of Helium is assumed than we 
derived from Hann and others. When Prof. E. ConeN was so kind 
as to draw my attention to the research of Craupr, time failed to 
make another calculation with a percentage of Helium about three 
times larger. The order of magnitude of our result, however, will 
not be altered by this circumstance, and the percentage of hydrogen 
calculated will remain very near that determined by Craupr. 
§ 8. We have treated the calculations about the sound orbits in 
air at rest somewhat at length, in order to make clear that the 
data about the constitution of the atmosphere are by no means so 
accurately known, that one would be in a position to reject VON DEM 
BorNeE's explanation only because the limit of the silent region does 
not oceur at 114 KM., as von pem BorNe's calculation would have 
it. It appears to us, that given the uncertainty existing in this respect, 
more weight ought to be attached to the shape of the silent region 
and the distribution of sound intensity at its limit than to the absolute 
dimensions, and we have pointed out already that in this respect 
