896 



SCIENCE 



[N. S. Vol. XXXIV. No. 887 



been more or less successfully replaced, 

 the former by mirrors, and the latter by 

 diffraction gratings. 



These are made by ruling very fine lines 

 very close together on a glass or a metal 

 surface. The effect on the incident light 

 is to alter its direction by an amount which 

 varies with the wave-length — that is, with 

 the color; and a spectrum is produced 

 which may be observed to best advantage 

 by precisely the same form of spectrometer, 

 with a substitution of a grating for the 

 prism. 



The dispersion of a diffraction grating 

 depends upon the closeness of the rulings; 

 but the resolving power is measured by 

 the total number of lines. It is important, 

 therefore, to make this number as large as 

 possible. 



The first gratings made by Fraunhofer, 

 1821, contained but a few thousand lines 

 and had a correspondingly low resolving 

 power — quite sufScient, however, to sepa- 

 rate the sodium doublet. A considerable 

 improvement was effected by Nobert, whose 

 gratings were used as test objects for 

 microscopes, but these were still very im- 

 perfect as spectroscropic instruments, and 

 it was not till Rutherfurd, of New York 

 (1879), constructed a ruling engine with a 

 fairly accurate screw, that gratings were 

 furnished which compared favorably with 

 the best prisms in existence. 



With 30,000 lines (covering over 40 mm.) 

 the theoretical resolving power would be 

 30,000; practically about 15,000— suffi- 

 cient to separate doublets whose compon- 

 ents were only one fifteenth as far apart as 

 those of the sodium doublet. 



An immense improvement was effected 

 by Rowland (1881) Avhose gratings have 

 been practically the only ones in service for 

 the last thirty years. Some of them have 

 a ruled surface of 150 mm. X 60 mm., with 

 about 100,000 lines and can separate doub- 



lets whose distance is only one one hun- 

 dredth of that of the sodium doublet, in 

 the spectrum of the first order. In the 

 fourth order, it should resolve lines whose 

 distance is only one fourth as great. 



Practically, however, it is doubtful if 

 the actual resolving power is more than 

 100,000; the difference between the theo- 

 retical and the actual performance being 

 due to the defect in uniformity in the spac- 

 ing of the grating furrows. - 



The splendid results obtained by Row- 

 land enabled him to produce the magnifi- 

 cent atlas and tables of wave-lengths of the 

 solar spectrum which are incomparably 

 superior in accuracy and wealth of detail 

 to any previous work; so that until the 

 last decade this work has been the univer- 

 sally accepted standard. With these 

 powerful aids it was possible not only to 

 map the positions of the spectral lines 

 with marvellous accuracy, but many lines 

 before supposed simple were shown to be 

 doublets or groups; and a systematic rec- 

 ord is given of the characteristics of the 

 individual lines, for example, whether they 

 are intense or faint, nebulous or sharp, nar- 

 row or broad, symmetrical or unsymmetri- 

 eal, reversed, etc. — characteristics which 

 we recognize to-day as of the highest im- 

 portance, as giving indications of the 

 structure and motions of the atoms whose 

 vibrations produce these radiations. 



One of the most difficult and delicate 

 problems of modern astronomy is the 

 measurement of the displacement of spec- 

 tral lines in consequence of the apparent 

 change of wave-length due to "radial ve- 

 locity" or motion in line of sight. This is 



' This applies to all the Eowland gratings which 

 have come under my notice, with the exception of 

 one which I had the opportunity of testing at the 

 Physical Laboratory, University of Gottingen. 

 The resolving power of this grating was about 

 200,000. 



