MR SWAN ON THE CONSTITUTION OF FLAME. 23 
or inner surface, each conoid always produced its own peculiar spectrum, or rays 
of the same refrangibility. In describing my experiments, however, I purposely 
made no reference to Professor Drarer’s peculiar views, for I had not then had 
leisure to examine them with sufficient care. The object of the present paper is 
to state the result of a careful repetition of the principal experiments by which 
Professor DRAPER conceives he has demonstrated the existence, in flame, of con- 
centric layers producing light of varying degrees of refrangibility. 
Professor DraPrr’s method of analysing the light of a flame, which is exceed- 
ingly ingenious, is thus explained by him. “The instrumental arrangement I have 
employed is as follows:—The rays of the flame, of which the examination is to 
be made, pass through a horizontal slit one-thirtieth of an inch wide*and one inch 
long in a metallic screen, and are received at a distance of six or eight feet on a 
flint-glass prism, the axis of which is parallel to the slit. After passing the prism, 
they enter a small telescope, which has a divided micrometer, and also parallel 
wires in its eye-piece. Through this telescope the resulting spectrum is viewed.’’* 
The reason why the slit is placed horizontally, and the light refracted in a vertical 
plane, he states to be as follows :—‘In this arrangement the slit should be hori- 
zontal, and not vertical. So far from its being immaterial which of the two posi- 
tions is selected, very great advantages arise from the former. If the slit be ver- 
tical, the prism, it is true, will separate the constituent colours from one another ; 
but it fails to show their relative position. If it be horizontal, the relative positions 
of the different colours can be demonstrated; and it can be proved that a hori- 
zontal section of aflame is in reality, as has been already remarked, a coloured ring, 
the red being the innermost colour, and the violet outside; for if this is the order 
in which the colours occur, the red ring must necessarily have a less diameter 
than the green, and the green than the violet ; and when the prism, set in a hori- 
zontal position, separates those colours from each other, the sides of the resulting 
spectrum ought not to be parallel, but inclined to one another, the breadth being 
least in the red, and increasing as we pass to the violet end.” “This being under- 
stood, I may illustrate the facts now to be brought forward by an example of the 
prismatic analysis of a horizontal element of the flame of a spirit-lamp, it being 
understood that the prism is at its angle of minimum deviation, and the spectrum 
seen through the telescope. All the prismatic colours, in their proper order, are 
visible, the sides of the spectrum not being parallel, the inclination being quite 
rapid towards the red extremity, the rays of which come from the interior of the 
flame where the diameter is less. Mere inspection is sufficient to show the rapid 
approach of the red sides to each other; and I satisfied myself that even in the 
more refrangible regions there is the same want of parallelism, by rotating the 
telescope on its vertical axis, so that the vertical wires in-its eye-piece might coin- 
* Tbid., pp. 101-2, 
