Theory of Diffraction- gratings. 85 



from a tap, afterwards flooding it with pure alcohol and setting- 

 it up to drain and dry spontaneously. Sometimes I have found 

 nitric acid useful ; but I always try to avoid the rubbing contact 

 of a solid. These precautions have been so successful that, after 

 several hundred copies have been taken, the originals have 

 scarcely, if at all, deteriorated. 



For the support of the photographic film it is no doubt most 

 satisfactory to use optically worked "parallel" glass. Ordinary 

 glass would fail, for two reasons. In the first place it would ge- 

 nerally be impossible to secure a sufficiently close contact in the 

 printing. But even if this difficulty could be surmounted, the 

 spectrum given by the copy would not bear the magnifying- 

 power which it is generally desirable to apply. It is indeed 

 evident that the glass support of the grating requires the same 

 precision of workmanship as the object-glass of the telescope 

 used in conjunction with it. 



Although ordinary glass taken at random is inadmissible, I 

 have done a great number of excellent gratings on selected 

 pieces of patent plate. In order to choose the best, I lay the 

 plates on a table in such a position that the bars of a window or 

 skylight are seen reflected in them. Each bar appears in general 

 double, one image corresponding to each surface. By sliding 

 the plate about, while the head is kept still, irregularities are 

 easily detected by the shifting or curvature of the images. From 

 a package of two dozen 5x4 plates as issued by photographic 

 dealers, three or four, often lying together, may usually be se- 

 lected as flat enough for the purpose, or at any rate decidedly 

 superior to the remainder. It is worth notice that the object 

 aimed at is flatness of the two faces, exact parallelism being of 

 much less consequence; for it is evident that the interposition of 

 a truly worked prism of very acute angle would produce no evil 

 result. A glass is therefore not to be rejected merely because 

 the two images of the bar seen reflected in it are decidedly sepa- 

 rated. The question is rather whether this separation remains 

 constant as the plate is moved about without rotation. I have 

 never seen a piece of patent plate that could not be at once dis- 

 tinguished from worked glass in the way described ; so that the 

 test is abundantly sufficient for the purpose. The more delicate 

 methods by which worked glass is examined would be less prac- 

 tically useful. 



Whatever kind of glass be used, if the photographic process 

 be at all complicated, there is considerable economy of labour in 

 preparing comparatively large pieces, to be afterwards cut with 

 the diamond to the required size. A 5 x 4, or even a 4| x 3^ 

 plate will do very well for four gratings. In the case of worked 

 glass economy is an object; but when patent plate is used I 



