52 The Spectroscope : [January, 
We have further the following arithmetical progression :— 
h, —by =139 
6) — 8, — 123 
— 16 | 
fe—d, =171I | 
hz —b, =139 ; Common difference 16, 
fie- dy, = I71I 
Cze—Lx —123 
ee ee 
thus corresponding to the laws. By calculating backwards 
from these adjusted quantities we obtain the corrected in- 
dices of refraction, which will be found to differ but slightly 
from those derived from observation, as shown in the 
following table :— 
Observed. Calculated. Difference + Difference — 
B. rO22570 1°623571 0°000001 
C. = 1°625477 2: 1°025477 
IDy 1°630585 1°630582 0°000003 
E. 1°637356 1°637350 0°000006 
FE. 1°643466 1°643463 0°000003 
G. 1°655406 1°655398 0°000008 
isle 1°666072 1°606061 O*OOOOII 
Also the value of « becomes 1°570518, and of a 0°0265908. 
These small differences are far within the limits of probable 
errors of observation. 
Reverting to the normals, as wave-length is in the dif- 
fracted spectrum inversely equivalent to refrangibility, it 
follows that the positions assigned to the different lines in 
M. Angstrom’s Atlas being fixed according to their wave- 
length, these must be their true positions in the normal 
spectrum, and that in so far as—in the spectrum produced 
by any set of prisms composing a spectroscope—the relative 
positions differ from those assigned to them in M. Angstrom’s 
Atlas, such differences must be due to the action of the 
prisms. 
To show to how great an extent the lines are displaced in 
M. Kirchhoff’s spectrum, it is needful to compare the rela- 
tive positions there assigned to the lines with those in 
M. Angstrom’s Atlas. Fortunately this is not difficult ; for 
the interval between the more and less refrangible lines 
marked D, which in M. Angstrém’s Atlas occupies six divi- 
sions, in M. Kirchhoff’s occupies four; and as the relative 
positions of those two lines cannot differ appreciably in the 
