( 568 ) 



The svsteni of rings was formed in liie fooal plane of a small 

 achromatic lens of 18 m.m. aperture and of J 2 cm. focus. This 

 focal plane coincided exactly with the plane of the slit of a one- 

 prism spectroscope. The width of the slit was so far reduced that 

 the rings of the two yellow mercury lines could be observed sepa- 

 rately. Reproductions of negatives (somewhat enlarged) are given 

 on the Plate, the first with field off; the second showing the first 

 coincidence (superposition of rings K and ^„) ; the third gives the 

 second 'coincidence, the rings ;.,. and ;.„ being now in coincidence 

 with ;„. The plate refers to coincidences for 5770 ; negatives showing 

 the coincidences for 5791 however scarcely present any difference 

 with those now given. 



By measurements on half a dozen of negatives concerning tlie 

 first coincidence, the result was obtained that a separation equal to 

 0.166 Angstrom units for line 5770, corresponds to a separation of 

 line 5791 towards the red of 0.160 A. U., towards the violet of 

 0.177 A. U. 



Naw a separation of 0.166 A. Ü. corresponds, according to the 

 data given in § 6 of my paper cited in § 1 above sub '), to a strength 

 of field of 9130 Gauss. 



Considering as the object of the investigation the determination 

 of the numerical ^'alue of the asymmetry we infer from the given 

 data that it is equal to 0.017 A. U. A discussion of the systematic 

 errors of observations to be feared, shows that the values 0.015 A. U. 

 and 0.019 A.U. are yet possible, that however the values 0.011 A. 

 U. and 0.023 A. U. are \QYy improbable. 



Some measurements made by means of the niethod of diameters 

 tend to show^ that the accuracy of results obtained by that method 

 is somewhat superior to that now found. 



The accuracy obtained is however in excellent accordance which 

 what might be expected from data given by Fabry and Perot ') 

 if applied to our case. 



By our experiments with the method of silvered plates two 

 points are clearly shown viz. first that the positive results concern- 

 ing asymmetrical resolution in strong fields obtained on a former 

 occasion by Rowland's method have a real significance, secondly 

 that also in lower fields the asymmetry remains and has an amount 

 such as to be expected, if strength of field and asymmetry are 

 nearly proportional. 



•) Fabry et Perot. Ann. de Chim. et de Phys. Janvier 1S02. 



