1872.] tts Imperfections and their Remedy. 55 
In Mr. Grubb’s spectroscope recourse has been had to a 
combination of flint and crown-glass in the construction of 
the prisms; there being a central prism of dense flint-glass 
and large angle, having cemented on each of its two re- 
fracting sides, in the reverse position, a crown-glass prism 
of one-fourth of the angle. Without knowing the exact 
refractive indices of the two kinds of glass employed for 
each of the fixed lines, it is impossible todo more than show 
approximately the effect of this combination on the inter- 
vals ; but such an approximation may be made by assuming 
the indices. Suppose, then, that the flint-glass prism has 
the same indices as Fratinhofer’s No. 30, and that the 
crown-glass has the same indices as Fraiinhofer’s No. 9, 
which, when corrected by the laws, are as under :— 
B. C. D. 1B, F, G: lel 
1°525832, 1°526849, 1°529587, 1°533000, 1°536052, 1°541659, 1°546566. 
The differences between these and the observed indices are 
still more trifling than in the case of the flint-glass. Sup- 
pose the refracting angle of the flint-glass prism to be go”, 
and that of each of the crown-glass prisms to be a fourth 
of this, or 22° 30, the ultimate differences of deviation 
arising from such a compound prism would be in seconds as 
under :— 
B—C. C—D. D—E. E—F. F—G. G—H. 
754, 1862, 2548, |2360,' 4728, 4372. 
The difference of deviation between the two lines marked D 
is 32”; consequently, if we take one-fifth of the above devi- 
ations, we shall have the intervals, according to M. Ang- 
strom’s scale, sufficiently near for the present purpose. 
They will then, in comparison with the normal intervals, 
stand as under :— 
Normals. Comp. Prisms. Diff. + Diff. — 
B—C 305 I51 154 
C—D. 667 B72, 295 
D—E. 627 510 117 
E—F. 408 407 I 
PG. 553 946 393 
G—H. 374 874 500 
It is thus evident that, while the character of the irration- 
ality is much altered by this combination, it is far from 
being removed. The almost perfect coincidence of the 
intervals E—F, in the above spectra, is remarkable. 
It might nevertheless be possible, by the expenditure of a 
great amount of time and skill, to find a combination of 
