54 The Spectroscope : (January, 
0'000065. This corresponds to a difference of deviation 
of about 62” for the four prisms. As these 62” of deviation 
correspond to six degrees of Angstrom’s scale, if we take 
a tenth of the deviations we shall have the intervals for this 
flint glass according to M. Angstrom’s scale, sufficiently near 
for the present purpose. The intervals between the lines 
resulting from these four flint glass prisms, as compared 
with those from M. Kirchhoff’s four, and with the normals, 
will accordingly stand thus— 
Normals. Flint Glass. Kirchhoff's. Dien Berto 
B—C. 305 183 153 —122 +30 
C—D. 667 490 463 Savi ary 
D—E. 627 652 781 + 25 —129 
E—F. 408 503 834 +185 — 241 
F—G. 553 1164 1162 4 611 + 2 
G—H. 374 1046 1542 +672 — 4096 
It will be perceived from this table that the intervals 
of the spectrum, produced by the flint glass, No. 30, of 
Fratinhofer, approach more nearly in character to those of 
the normals than do those of M. Kirchhoff’s, and that the 
increased dispersion of this last is in a great measure due 
to an excessive enlargement of the interval between G and 
H. The peculiar features of M. Kirchhoff’s spectrum are 
probably traceable to his having given to his prisms curved 
faces, which, while increasing the dispersion, has augmented 
the irrationality in a still higher degree. 
As all media tend to alter the intervals between the lines 
to a greater or smaller extent, it is evident that, if prisms 
are to be retained in the construction of the spectroscope, 
some device must be adopted in order to give those intervals 
their true value before the instrument can be regarded as 
satisfactory. The means that appear available for this 
purpose are the combining of different media, the giving of 
one or more of the faces of one or more of the prisms some 
peculiar curvature, or the introduction of both of those 
means of correction to render it more perfect. If, as in 
M. Kirchhoff’s instrument, curvature of the faces of the 
prisms has so greatly increased the irrationality of the in- 
tervals, it seems probable that a similar device applied in a 
different way might be rendered available to restore them 
to their true values. The aim has hitherto been to obtain 
a very large spectrum by means of a great amount of dis- 
persion, irrespective of the displacement of the lines; 
whereas the correct principle of construction should be to 
obtain an accurate spectrum, corresponding exactly with 
that obtained by diffraction, trusting to magnification by the 
telescope for increase of size. 
