( 864 ) 



capillary has a small diameter. Now the gap-width of the electrp- 

 magnet may be considerably diminislied; iit the same time the 

 electrical resistance of the vacuum tul)o is moderate. 



9. Measurements were made in the following manner. The slit 

 of the auxiliary spectroscope performing the preliminary analysis of 

 the light, was widened in order to obtain in the echelon spectroscope 

 light of the two yellow mercury lines simultaneously. The steps of 

 the echelon were placed parallel to the slit of the auxiliary spectro- 

 scope. The image of the vacuum tube projected on the slit was now 

 chosen in such a manner that only light from the uniform part of 

 the field was analyzed. By means of a suitable small screen placed 

 before the photographic plate its middle part could be exposed first 

 to light under magnetic influence; then the unmodified lines were 

 taken in the upper and lower parts of the |)late. The plates taken 

 confirmed the result ol)lained in k 6 as to llie shift of line 5791 

 towards the red. 



As to line 5770 the amoujit and even the existence of the shift 

 is not quite certain at present ^). All these measurements were not 

 further pursued howe\'er, because after the publication of the first 

 part of this paper ^) and during my measurements there appeared a 

 communication by Gmklin in the 1 April number of the physikalische 

 Zeitschrift. 



Independently of my paper and in another way our present 

 subject was taken up by Gmklin. The enormous resolving power of 

 the echelon spectroscope used by Gmelin apparently permits of greater 

 accuracy in the measurements than would have been possible for me. 



It may be remarked finally that, so far as the present results can 

 settle the question, the observation of the asymmetry in a direction 

 parallel to the lines of force, ") which first induced me to this 

 investigation, but which was given with some reserve, must have 

 been correct. 



1) The liisl part ot this cummLinicalion coiitaiiis an error, wliich I only noticed 

 after the printing off. fn the last division of § 4 the ehange of wavelength has 

 been calculated from the measured displacement in the same manner as must be 

 done, when the distance of two adjacent orders with the field off, is compared 

 with the distance separating the components towards red and towards violet with 

 the field on. Of course this proceeding is faulty in the case of § 4. Hence the 

 last sentence of § 4 and the last column of the table in § 5 have lost significance. 

 The further consideration of the peculiar change of the distance of orders noticed 

 in § 5 must be reserved foi- a future paper. 



~) Zeeman, These Proceedings 29 Febr. 1U08. 



«) Zeeman, § 7 in New Observations etc. These Proceedings :29 Febr. 1908. 



