354 On the Refraction of Light. 
figure of the pyramid was varied, and the base on the paper describ- 
ed, according to the position of the prism, a square, a rhomb, a 
rhomboid, a trapezium, or a parallelogram. All the figures were — 
quadrilateral, and terminated by. right lines. ‘These phenomena 
would lead us to consider the spectrum in an entirely new light; for 
if we. admit the Newtonian hypothesis, that it consists of circular 
images of the sun, the appearance of ‘these. luminous pyramids must 
remain inexplicable. The idea that the spectrum consists of mono- 
omatic circles, the diameter of one of which forms its breadth, 
and whose circumferences bound it on all sides, is a beautiful con- 
ception, but it appears inconsistent with the luminous pyramids, 
and with. the following fact, which I accidentally discovered. 
Whilst experimenting on the effects of transmitting the analyzed 
rays emerging from a prism, in a dark room, through a bi-convex 
lens, I chanced to interpose the lens between the sun and the prism, 
in such a manner that the focus fell on the prism near its refrac- 
ting angle. On the opposite .wall I was surprised to see a spec- 
_ trum of a very peculiar and novel appearance. ‘The proportion be- 
tween its length and breadth exactly corresponded to. that of ‘the 
Newtonian Spectrum, it being about five times longer ‘than it was 
broad ; but unlike the Newtonian image, its length lay in a direction 
parallel with the length of the prism. In this spectrum the colors 
~ are arranged longitudinally, in the order of the common spectrum, 
and consequently all the dimensions of the common one are’ reversed 
in this. , Newton, from the circumstance of his not having succeeded 
in attempts to increase the breadth of the spectrum, concluded that 
its breadth was the diameter of one monochromatic circle ; but if we 
grant that the breadth of this spectrum is formed by the diameter of 
a colored circle, we must admit that that circle is painted. with the 
seven primitive colors, which, from a consideration of both spec- 
tra, would be absurd... The lines bounding the extremities of the 
common spectrum, form the lateral boundaries of this new one5 and 
its. extremities are terminated by lines, which bound the long sides 
- 
of the Newtonian spectrum. The lines which bound the extremities 
of the common spectrum are so produced in this image as to terml- 
nate its long sides with straight lines ; and as it is certain that cet 
line on being produced: cannot forma straight line, so it is certain that 
all the boundaries of the solar spectrum are right lines. - In order to 
examine this spectrum minutely, the prism should be fixed on a stand, 
the lens mounted on a swivel, and the white surface placed at a sult- 
