of Edinburgh, Session 1870 - 71 . 
411 
made several trials with hollow prisms and prismatic vessels, using 
various substances, such as oils of cassia and turpentine, toluol, 
alcohol, saturated solutions of salts, &c., with the view of imitating, 
with nearly transparent substances, the singular results obtained 
by Talbot, Christiansen, and Kundt. The observations are cer- 
tainly very easy in one sense, though very laborious in fact ; but I 
have already produced a spectrum doubled on itself, and have no 
doubt that with patience I shall be able to produce one with two 
and even more inversions; though, of course, the more numerous 
are the inversions the smaller is the scale of the whole phenomenon. 
The easiest method seems to be to put into a hollow prism a mix- 
ture of two substances of very different refractive powers, and to 
immerse it in a prism or trough containing a substance of inter- 
mediate refractive power. When a trough is employed, an external 
glass prism may vjith advantage be used along with the combina- 
tion. The sought phenomenon is, of course, obtained best near the 
point of adjustment for achromatism, and is in fact very closely 
connected with the investigations of Dr Blair in his attempts to 
improve the achromatic telescope by using fluid lenses. 
One of my hastily set-up combinations (of two liquids only) gave 
me a direct-vision spectroscope complete, more powerful than one 
of Browning’s excellent instruments with five glass prisms, and I 
have little doubt that in this way very good results may be obtained. 
But, if it be needful to examine only a small region of the spectrum 
at a time, practically unlimited dispersion may be obtained by using 
so very simple a combination as two approximately isosceles flint 
prisms of small angle with their edges together and their adjacent 
faces inclined at an angle approaching to 180°, so as to form a hollow 
prism to be filled with oil of cassia. In fact, the dispersion is in 
this case easily seen to be nearly proportional to the tangent of 
half the angle of the oil prism. If two kinds of glass, of very 
different dispersive powers, but of nearly equal mean refractive 
powers, could be obtained, a permanent combination might be 
easily formed on this plan, giving as much dispersion as a very 
long train of ordinary prisms, and losing scarcely any light. A 
slight inclination of the ends to one another will enable us to use 
ordinary flint and crown for the purpose, except in so far as total 
reflection may interfere. Such a combination, adjusted for the red 
