OF ARTS AND SCIENCES. 277 



The dispersion, therefore, as shown above, would be about 2400. 

 Cohima Am. gives the lines seen by Dr. Amory, with a diiFractioa 

 ffratiiis: of Mr. Rutherford's, having 510 lines to a millimetre. As he 

 employed the sixth spectrum, the dispersion was 30 GO. For the most 

 recent and complete measurement I am indebted to Professor Young, 

 who has measured the E line with a very perft^ct grating by IVIr. Ruth- 

 erford, having 340 lines to the millimetre ruled on silvered glass. 

 As he used the eighth spectrum, the dispersion was 2720. These 

 results have been taken as a basis, and the resultant wave lengths 

 are given in the second column. My own observations are given in 

 the column marked P., and were made in 18G9 and 1870, with a 

 spectroscope in which the light traversed each prism twice, giving 

 a dispersion equivalent to 7, 9, or 11 flint glass prisms. This would 

 correspond to dispersions of 1700, 2200, and 2640; but the best 

 results were obtained with the two lower powers. All the observers 

 used telescopes about a foot and a half in length. 



We have thus four entirely independent maps, as neither observer 

 had at the time a copy of the work of any of the others. The simi- 

 larity of the results, with instruments differing so greatly in form and 

 power, seems to show that we have nearly reached the limit beyond 

 which an increase of dispersion is unadvisable ; and as if with our 

 largest instrument nearly all the lines really present in the spectrum 

 were visible. It is much to be desired, however, that these lines may 

 be compared with any other insti-umcnts of greater power, if such are 

 ever constructed. It is only essential that the measurements should be 

 made before comparison with the above results, since with the lines, as 

 with fiiiut stars, it is much easier to detect them when we know exactly 

 where to look. Various tests may be selected from these lines for an 

 insti'ument of any power. Tims to double the E hues, or to show 24 

 and 26 as two lines, is a good test for a one prism spectroscope of 

 large size. The five pairs, (27, 28), (29, 30), (31, 32), (35, 36), and 

 (37, 38), also form an excellent test for any but a very large instru- 

 ment. The great number of double lines in this group, and more 

 particularly in the B line, and in the electric spectrum of sodium, 

 seems to prove most conclusively, as in the case of double stars, 

 some real relation between the two components. 



The last two columns give the relative intensity and width of the 

 lines as estimated by Professor Young. Nos. 2, 16, and 27 are hazy, 

 and No. 35 is a mere shade. The numbers under the observers' names 

 give the approximate dispersion employed by each. 



