868 
(B) and (C) is purely a matter of taste. There is no physical crite- 
rion as yet available to decide between them. It is true that the 
systems (B) and (C) do not satisfy Macn’s postulate that inertia 
must be traceable to a material source. But this postulate is a purely 
metaphysical one, and has no physical foundation whatever. It 
appears to me to be the last remnant of the desire for a purely 
mechanical interpretation of nature, which logically and historically 
is based on the belief in forces at a distance, and the impossibility 
of which has been so clearly demonstrated by EinsreiN in his Leiden 
address. 
The three systems differ however in their physical consequences 
at large distances, and an experimental discrimination between them 
may be possible in the future. The decision between (2) on the one, 
and (A) and (C) on the other hand may be brought about by the 
study of systematic radial motions of spiral nebulae *). The distinction 
between (A) and (C) is more difficult, since they both have 
is = 1, and differ only in the gi; with 7 and j different from 4, 
the values of which at great distances it is not so easy to ascertain. 
The decision between these two systems must, I fear, for a long 
time be left to personal predilection. 
infinity; two straight lines have only one (and not two) point of intersection, 
which may be situated at infinity; if we go to infinity along one branch of a 
hyperbola, we return along the other branch on the other (and not on the same) 
side of the asymptote. All these are properties of the elliptical as contrasted with 
the spherical space. The spherical is only a quite unnecessary reduplication 
of the elliptical one. 
1) See pe Sirrer, |. c. pp. 27—28. At that time (1917) the radial velocities of 
only three spirals were known, of which one was negative; the mean being 
+ 600 km/sec. Now the radial velocities of 25 spirals are known (see Mount 
Wilson Publications, Nr. 161, p. 19) of which only three are negative, the mean 
being + 560 km/sec (or + 677 km/sec if the four brightest are omitted). The 
system (B) requires a (spurious) positive radial velocity for distant objects. 
