360 
PHYSICS: DAVIS AND TERRILL 
Proc. N. a. S. 
we have to pass through these hnes a curve whose ordinates are propor- 
tional to the coefficients of b. Now the values found do not lie well on any 
curve of this type, the second order being entirely too low. The curves 
that fitted most nearly gave a value of 5 = 3 X 10 ~^ while the error limits 
would permit of values ranging from zero to several times this. It will 
be seen that this corresponds to a shift of the first order of 5", so that for 
this wave-length, the effect of refraction is very slight. 
This value 5 = 3 X 10~^ necessarily contains large possible errors. It 
is, however, of the same order of magnitude as that obtained by the method 
of total reflection. 
An independent determination of b was obtained by measuring the angle 
of total reflection as suggested by the experiments of A. H. Compton^ on 
total reflection from glass and lead. The ionization chamber was shielded 
from the direct beam and the spectrum searched at angles close to grazing 
incidence. Total reflection appeared as a line, approximately 4' wide at 
half maximum, sHghtly unsymmetrical and of an intensity about 25 
times the surrounding values which coincided with the leak. To test if this 
line consisted principally of the monochromatic radiation corresponding to 
the Ka line, a zirconium screen was introduced. This diminished the 
intensity without materially changing the shape of the line or shifting its 
position. 
Measurements were made on each side of the zero position, and the mean 
of the readings gave 6' 30 the value being good to 30'^ Now for total 
reflection, putting 6' = 0 in (1) we have 
. . sin"^ d 
V = cos B or b = — - — 
whence 
5 = 1.7±.5 X 10-6 
Comparing this with the theoretical value from the Lorentz equation 
" - n e"^ 1 
g = . - 
2 TT W V 
we take the density of calcite as 2.71 giving 81.5 X 10^*^ electrons per cubic 
centimeter, and place v equal to the frequency of the molybdenum i^a 
line, whence 6 = 1.85 X 10-«. 
Determinations of the total reflection made with a second crystal gave 
results which checked with the preceding. Incidently it developed that 
the face of this second crystal had been ground at a slight angle to the re- 
flecting planes, the crystal zero determined by the angle of total reflection 
differing from that obtained from the line spectra by an angle of 44'. 
This suggests an extension of the method of determining the refraction 
