362 
only the type. Consequently the reduced ®'s are to be 
used for the solution of the constants. 
Unfortunately the number a is not known; neither is 
the factor __ by which the empirical ©’'s are to 
be multiplied. But as soon as we assume a definite value 
for a, the solution can be carried through. Different 
values of a will of course yield different curves; the higher 
the value, the steeper the curve. We may now fix the 
condition that ihe empirical and theoretical curve shall 
coincide at a third point, whereby a becomes a sort of 
parameter to be determined. 
I first tried the value a — 10, thus ee A 4 
This gave: 
Values of x used dé -4012 Ka = MUR 
Empirical O's @, = 0.274 ©, — 0.779 
Reduced @'s (1.1 *X @) 0.302 0,857 
Values of R from Table 16 KR, = — 0.367 KR, =+0/755 
Hence: h — ie re — 9.63 
log. 7 
0.367 
M = 963 log. 13 = 1.15205. 
With these constants Ï| computed the theoretical value 
of © for x — 15, a point equidistant between x, and x. 
R = h(log.x — M) = + 0.232 
Reduced © = 0,628; © — 0.571. 
Empirical ® — 0.566. Difference Emp. — Theor. — — 0.005. 
The original logarithmic curve had given at this point 
a difference + 0.014 between Empirical and Theoretical ©. 
[ expected that the value for a obtained by interpolation, 
namely a — about 7, would yield the desired harmony. 
[ got: 
h = 9.18; M= 15575; 
As seen in Table 3 and 4, the series of computed © 5, 
