Vol.. 8, 1922 
PHYSICS: DAVIS AND TERRILL 
359 
It should be remarked that if the crystal is not mounted so that the mean 
reflecting plane is directly over the center of rotation, and the face of the 
crystal be slightly curved, a shift of the lines will be obtained, that might 
be taken for refraction. To eliminate this possibility, the crystal was 
mounted on a slide with a micrometer screw attached, so that it might be 
moved in a direction perpendicular to its face. The crystal being mounted 
as nearly as possible over the center of rotation by mechanical methods, 
the adjustment was changed by small fractions of a millimeter, reading 
each time on both sides the crystal and chamber angles of the ai line. If 
curvature is present, the crystal angle will change as the crystal is moved. 
Twice the crystal angle is plotted against the readings of the micrometer 
head and on the same sheet, the chamber angle is also plotted against 
these readings. The intersection of the two lines gives the proper microm- 
eter setting, since the crystal must be over the center, when the chamber 
angle is exactly twice the crystal angle. With the crystal used, it was 
found that the curvature was very slight, a shift of less than one second of 
arc being found within the range of adjustment. This was checked for 
each order. 
We proceed then to reduce the angles measured to the corresponding 
first order angles. Now ^„ = ^„ + A where A includes the slight in- 
crease due to refraction and also the error of observation. Since A is 
small 
sin On = sin -\- A cos $„ 
arc sm 
y w y n 
n 
cos 
cos I 
6' = arc sin 
fsin_dn\ __ ^ cos On 
\ n ) ncosB 
and the coefficient of A can be computed for each order, it being 
.49 for the second and .31 for the third. Multiplying the errror limits by 
these coefficients and introducing the refraction from (3) which is also mul- 
tiplied by them, we have the following values for 
e' = 6° 42' 43'' ± 10" - 1".76 - b • W 
= 6° 42' 33" =«= 5" - 0".63 • b ' W 
= 6° 42' 38".^ ± 3" - 0".27 • 5 • 10^ 
and we have to determine 5 from these equations. 
Plotting the values of the angles against the ordet; using lines covering 
the limits of error instead of points to locate the extremities of the ordinates> 
