Theories of Microscopical Vision. By A. E. Conrady. 631 



of the wave-fronts, the diffraction spectra refuse to move ; they 

 uniformly expand and contract agreeably to the effective cones of 

 light, but they rigorously preserve their distance apart ; they 

 absolutely decline to crowd together ! 



I have indeed failed to cause the diffraction spectra to dis- 

 appear or to essentially change in appearance, when, by the use 

 of an apochromatic objective as condenser (working of course at 

 correct tube-length and through a proper cover-glass), and by 

 choice of source of light, the theoretical conditions for getting rid 

 of the diffraction spectra seemed completely realised ; but nothing 

 would do away with them except cutting down the flame-image 

 to such a small size that it covered one slit only. 



It would seem that Dr. Johnstone Stoney was right when he 

 claimed that, under the conditions prevailing with " critical illu- 

 mination," a flame-image behaved as if it were composed of 

 innumerable plane waves. 



The only explanation that I can suggest for the phenomenon 

 which Mr. Gordon describes is that he must have observed the 

 diffraction spectra without making sure that the instrument was 

 at least approximately focussed upon the object. Only in this 

 case would it seem possible to get the diffraction spectra to 

 appear to crowd together, owing to their being observed in a 

 wrong plane. 



In cases where a coarse structure is illuminated by very wide 

 cones, we can observe that the different diffraction spectra expand 

 to such an extent as to overlap, and that indeed we may have 

 direct light and portions of a number of diffraction spectra of 

 several successive orders super-imposed in the centre of the objec- 

 tive. But it is easy to see that the portions thus super-imposed 

 are never capable of meeting in the plane of the image of the 

 structure in such a way as to produce an image. For, in the case 

 assumed by Prof. Abbe, i.e. that the illuminating cone is obtained 

 from a relatively distant extended source of light, these supei'- 

 imposed portions are derived from different luminous points, and 

 therefore incoherent or incapable of interference ; whilst in case 

 of an illuminating cone really consisting of spherical wave-fronts, 

 having their focus near but not in the plane of the object, the 

 super-imposed portions of different diffracted waves have such 

 inclinations to each other that they cannot possibly meet in the 

 plane of the true image. This latter case, which is the one really 

 under discussion, I will briefly deal with. 



Let A B, fig. 101, represent a grating with slits 1, 2 7, and 



let a luminous point C be brought close to that grating. Those por- 

 tions of the spherical waves sent out by C winch fall upon slits 

 are there diffracted according to the universal Huyghenian prin- 

 ciple, i.e. the first effect is as if the slits themselves had become 

 self-luminous. But the secondary waves thus set up in the indi- 



2x2 



