50 Lord Bayleigh's Investigations in Optics, 



beam ; and then illumination suffers. If full illumination be 

 required, we must employ a thickness three or four times as 

 great as that defined by (5), § 3. 



Apart from the question of the area of the pupil occupied 

 by the beam, illumination suffers from the effects of absorption 

 and reflection. The first depends simply upon the thickness 

 traversed, and is therefore an invariable quantity when the 

 material and resolving-power are given. Some years since it 

 was laid down by Pickering* that in spectroscopes composed 

 of prisms of the same material and in the position of minimum 

 deviation (which disperse equally and admit the same amount 

 of light) the loss by absorption will be the same. In accord- 

 ance with what has just been proved, we are now able to 

 dispense with the restriction to minimum deviation. 



In powerful spectroscopes the transparency of the material 

 of which the prisms are made is a point of great importance. 

 Some specimens even of well-made flint and crown glass exa- 

 mined by Christie f stopped as much as half the light in a 

 thickness of 4 inches. Such a degree of absorption renders 

 the glass unsuited for instruments of more than moderate 

 power. From measurements by Robinson and Grubb J, how- 

 ever, it would seem that absorption need not stand in the 

 way of much more powerful instruments than any yet at- 

 tempted. One specimen of Chance's glass was of such trans- 

 parency that 111 inches would be necessary to reduce the 

 transmitted light in the ratio 2 # 7 to 1. 



The loss of light by reflection depends upon the number 

 of surfaces and upon the angles at which the rays are inci- 

 dent. It might be thought that a great multiplication of 

 surfaces was necessarily very unfavourable to brightness ; but, 

 as has been pointed out by Pickering in the paper referred to, 

 this difficulty may be overcome by using prisms of such angle 

 that the reflected light is perfectly polarized. Under these 

 circumstances half the light at least escapes reflection ; and 

 the necessary angles (64° for ordinary flint glass) are not 

 otherwise objectionable. The least loss of light is incurred 

 when the whole thickness is thrown into one prism of mode- 

 rate angle ; but the gain in brightness would rarely compen- 

 sate for the other disadvantages of such a construction. 



* " On the Comparative Efficiency of Different Forms of the Spectro- 

 scope," Phil. Mag. vol. xxxvi. p. 41, 1868. 



•j- " On the Magnifying-power of the Half-prism as a Means of obtain- 

 ing great Dispersion, and on the general Theory of the Half-prism Spec- 

 troscope," Proceedings of the Royal Society, March 1, 1877. 



% " Description of the Great Melbourne Telescope," Phil. Trans. 1869, 

 p. 160. 



