USE OF SPECTROSCOPE IN ASTRONOMICAL OBSERVATION. 143 
stars is called stellar spectroscopy ; and, in fine, the determina- 
tion of the nature of the spectra of celestial objects is the aim 
of all the various contrivances made use of in the application of 
spectral analysis to astronomy. 
The first point to be considered is the foj'mation of a pure 
spectrum. From fig. 1 we see that a light-pencil, which would 
form a spot, of definite magnitude, if not interfered with, 
would form a series of images along the streak rv. The figure 
shows the formation of one red image, one orange image, and 
so on. But in reality there would be an infinite number of 
images along rv; and although the beauty of the spectrum 
would not be at all affected by this circumstance, it would be 
impossible to detect the absence of rays of definite refrangibi- 
lities, simply because the images formed by rays of neighbouring 
refrangibilities would conceal the want. The spectrum, how- 
ever beautiful, would be in fact impure in the sense in which 
the spectroscopist uses the term. 
Therefore, when the object to be observed has a sensible mag- 
nitude, it is necessary to let its light pass through a narrow 
slit parallel to the refracting edge of the prism. The same is 
true whether the object itself is observed, or the image of the 
object obtained by means of a lens, or by any combination of 
lenses. But, of course, any optical contrivance by which the 
apparent dimensions of an object can be reduced, is serviceable 
to the end we are considering. 
Next, as to the mode by which the spectrum is viewed. 
An eye placed so as to receive any of the light forming the 
spectrum is sensible of the nature of the particular part of the 
spectrum to which that light belongs. The eye may be aided 
by a telescope, or by an eye-piece, or may be used without 
such aid ; but for delicate researches a telescope is always used 
for the examination of the spectrum. 
Beginning with the simplest form of observation — suppose 
an observer wished to view the spectrum of a star with the 
prism p. Here no slit or darkened room would be necessary. 
But the observer would meet with a difficulty. Supposing the 
star to be towards A (fig. 1), and the prism held as shown, the 
observer would have to look from i' towards b, which is not the 
direction in which the star lies. 
To obviate this inconvenience in the case of a simple obser- 
vation of this sort, direct-vision prisms have been devised. The 
property on which they are founded is that called the “irra- 
tionality of dispersion ” ; — 
In fig. 1 the angle between the emergent ray from b to i' 
and the line iA is called the deviation of the ray. In the 
position of the prism shown in the figure the deviation for this 
ray has its minimum value ; and if the ray is supposed to be 
VOL. VIII. — NO. XXXI. L 
