Neiu Analysis of Solar Light. 405 



visible which are totally imperceptible in ordinary daylight. 

 Prisms of rock-salt were not in my possession, and I can there- 

 fore form no opinion as to their completeness. 



The second circumstance to be taken into account is the re- 

 peated reflexion of light in the prism. In the majority of prisms 

 used for experiments on dispersion the two refracting surfaces 

 alone are polished, the other three being ground dull. If such 

 a prism be placed upon a dark ground so that the dull surface 

 shall be illuminated, then within the prism a series of reflected 

 images of this surface is observed. The two polished sides act 

 like an angular mirror, which exhibits a series of circular images 

 of any object placed between its reflectors. In the case before 

 us the third surface occupies this position, and we look through 

 one of the reflectors into the interior. The reflected images of 

 the third surface appear in exactly the same direction as the 

 spectra which are observed on looking through the prism ; and 

 as a portion of the incident light usually falls upon the third 

 surface, illuminating it and its images, a weak white luminosity 

 is thus created which spreads itself over the spectrum. The 

 quantity of the reflected light is certainly very small, and in 

 general will not be at all observed beside the regularly refracted 

 light. To cut it off", it is necessary to blacken all the surfaces 

 well except the two refracting ones. 



When the coloured media arc introduced between the prism 

 and the eye, it must be remembered that if the polish of their 

 surfaces and the purity of their mass be not perfect, light will 

 also be dispersed by them. As coloured media, Brewster used 

 for the most part glass plates or coloured fluids, the latter 

 of course enclosed between glass plates. Regarding the purity 

 of the glasses I have just spoken ; but in the case of fluids also, 

 for example of distilled water, we know that through layers of 

 a certain depth the light which passes is cloudy, that is, a portion 

 of it is dispersed. Besides this, the reflexions which take place 

 between two surfaces of the coloured medium, and between them 

 and the cornea of the observer's eye, are also to be taken into 

 account. When the coloured plate has parallel surfaces, the rays 

 which have undergone repeated reflexion between them create 

 secondary images of the spectrum which almost completely coin- 

 cide with the original one, and cannot do much injury. If the 

 surfaces are not parallel it would be more suspicious, for here 

 the colours of the secondary images might fall upon other colours 

 of the primary. To this it must be added, that the incident 

 light is partly rctflectcd by the cornea, and this reflexion again 

 reflected by the glass plate ; the imag(! thus formed being too 

 near the eye must apjjear as a bright luminositj' in the held of 

 view. On account of these circumstances, I prefer placing the 



