( 358 ) 



force these numbers for line 5791 become -4- 366 + 6.7 resp. 399 + 6.7, 

 the values preceded by + indicating the mean error. According to 

 our observations to t lie abscissa 415 correspond (he ordinates 368 

 and 403, hence a very good agreement. 



11. From the extremely small amount of the asymmetry viz. 

 0.036 A. U. one might infer after comparison with the width of' the 

 spectral lines in our figures that the asymmetry is only a small part 

 of the real width of the line Such a conclusion would however be 

 too rash. 



It is true that from our figures and from their originals follows 

 an apparent width of the outer components of about 0.190 A. U. 

 The negative of Fig. 1 was not taken however with extremely narrow 

 slit, but with a width of slit of 0.08 m. in. Olher photographs taken 

 with a width of slit of 0.02 m.m. gave a somewhat smaller apparent 

 width of the spectral line as the first result. 



To be sure however of the real width of the line, which is of 



some importance here, I made an independent determination by means 



of an echelon spectroscope of high resolving power, the mercury tube 



being under the same circumstances, as in the experiments under 



review. The width of the spectral line appeared to be the 7u th P art 



of the distance of two successive orders of the echelon. In the 



vicinity of the yellow mercury lines this distance is 0.(594 A. U. 



0.694 

 hence the width of these lines is about — = 0.063 A. U. 



11 



We may compare this result with a value we may deduce from 

 results obtained by Michei.son. Michelson's analysis ') by means of the 

 interferometer shows that in a field of 10000 Gauss the whole sepa- 

 ration of the yellow mercury lines is 0.36 A. IT. From his diagram 

 on pag. 354 1. c. we infer that the width of the spectral line was 

 under the circumstances of the case one fourth part of the separation 

 or 0.09 A. U. 



Hence taking a mean value for the .vidth of 0.07 A. U. we 

 conclude that the asymmetry amounts to about one half of the 

 width of the line or at any rate that width and asymmetry are of' 

 the same order of magnitude. 



l ) Michelson. Phil. mag. Vol. 45, p. 348. 1898. 



