ACCORDING TO THE UNDULATORY THEORY. 491 
der G H as the limit of perceptibility for the eye, we obtain a spectrum 
exactly like that above described. 
The phznomena which appear in the spectrum of coloured flames 
stand undoubtedly in connection with the present object, and may be 
explained in the same manner as the phenomena of absorption. But 
besides the presupposition of retardations, we must here still make an- 
other, namely, that certain flames can only produce light of a certain 
length of wave, or at least that the produced light is inclosed in certain 
limits, which lie closer to one another than the red and violet. Various 
phenomena in the spectra of coloured flames arise only from this 
cause. This, for example, is the case with the bright orange-coloured 
stripe which is formed in the spectrum of a common candle light. If 
we consider the light of a candle, we find it to consist of several divi- 
sions, differing from one another. The inner and lighting flame con- 
tains, as is well known, heated particles, which undergo a real combus- 
tion then only when they arrive at the outermost border, or where they 
come into contact with the air. The outer flame is therefore of quite 
a different consistence from the inner; it has also quite a different 
appearance ; it lights feebly, and possesses a faint orange colour. The 
broadest part of the flame has, on the contrary, a blue colour, and re- 
sembles in every respect that produced by a slow combustion of coals. 
As this flame originates at the point where even the wick comes into 
contact with the air, I consider it quite probable that it arises froma 
slower combustion of it. 
If we place a convex lens between the flame and the opening through 
which we allow the light to fall on the prism, so as to produce a mag- 
nified image of flame on the prism, we are then able to bring to the 
Opening any part of the image of the flame by sliding the lens, and 
in that manner to examine any particular part of it. Suppose we then 
move the image so that only its outermost border lies on the aperture, 
and consequently only the light of the most external flame can pass 
through it, we find that the spectrum contains nothing else than a part 
of the orange-coloured stripe. If we change the form or breadth of 
the aperture, we find that the orange-coloured stripe undergoes just 
the same change, so that it always remains a complete copy of the aper- 
ture. If we slide the image of the flame so that the inner lighting part 
arrives at the aperture, we obtain a complete spectrum ; and the nearer 
the middle part of the flame comes to the aperture the greater is the 
brightness the spectrum acquires, while the orange-coloured stripe de- 
creases more and more. Hence I conclude that the inner flame pro- 
duces light of all possible lengths of wave; the outer flame, on the con- 
trary, only light of a single length of wave, that is, of a completely ho- 
mogeneous light. If we view a flame of light through a prism, without 
letting the light pass through a minute aperture, we naturally obtain an 
