242 Transactions of the 



ghost object, long known as the " diffraction spectrum." I -will, 

 however, try another combination to act just opposite, and so make 

 a double-barrelled afiair of it, though I do not expect to gain much 

 except in increase of light, and this is scarcely needed. Finally, it 

 might be urged that we do not secure near the same average 

 amount of obliquity of the rays as with the parabolic condenser. 

 This is correct in appearance but not in effect, as there is a reason 

 why the most oblique rays on this system of illumination are 

 scarcely available. It is those nearest to 41° that are the most 

 serviceable. 



In my paper of 1856 I gave a diagram, showing that the use 

 of water hmited the angle to 160". In order to secure the extra 

 20^, I reconmiended the use of oil of cassia or cloves for patching on 

 the truncated lens ; but the Eev. S. L. Brakey, who has made 

 investigations on this principle of illumination, informed me that 

 he found that it caused not the slightest difference whether water 

 or a more refractive oil was used. Instead of waxing wroth at 

 what seemed to demolish my pretty bit of theory, I accepted it as 

 a fact to be reasoned upon, particularly as it allowed some latitude 

 for improvements, and soon found that for very oblique rays the 

 objects to appear luminous must be in absolute, or, indeed, in 

 strong adhesive contact with the glass, otherwise these rays would 

 not enter them ; and as we now deal so considerably in the principle 

 of total internal reflexion in microscopes and their apphances, I will 

 finally submit a few remarks on the theory for further consideration. 



When a ray of light falls on an internal surface of crown glass 

 at an angle of near 41°, it can no longer be refracted outwards ; its 

 passage coming parallel with the plane surface, the whole is reflected 

 back again at an angle equal to that of the incident ray. For 

 monochromatic light, the limiting line between total reflexion, and 

 transmission, or refraction, is extremely definite, but for ordinary 

 light not so, as the coloured rays are refracted and pass in the 

 order of their refrangibility. At about 40° a colour bow is seen 

 across the prism, caused by the reflexion of a portion of the suc- 

 cessive rays of compound light, but even then the boundary is very 

 narrow, and comprised within an incident angle of one degree five 

 minutes, consequently in all total reflecting arrangements the 

 incident angles of the rays must exceed 40°. It is a fact well 

 known to all those conversant with optical apparatus, that total 

 reflecting surfaces must be kept clean, like those for direct trans- 

 mission; any particles of dust, &c., on the former will abstract a 

 portion of light from internal reflexion. It is exactly on this 

 principle that the method of illumination described in this paper is 

 based, but it appears to involve some conditions which do not 

 hitherto seem to have been thoroughly investigated. I must call 

 attention to an error which has been handed down in most optical 



