80 C. Barus — Coronas with Mercury Light. 



vals neither disc nor ring are quite green. Hence there is a 

 periodic term impressed on the diffractions, which may be iden- 

 tified as an interference similar to the case of the lamellar 

 grating referred to in another paper.* 



When monochromatic light is used it is necessary, therefore, 

 to observe both the edge of the disc and the inner edge of the 

 first ring ; for neither appear vividly at the same time. The 

 chord s on a radius of 30 cm for the minimum in terms of 

 the corresponding chords s' and s'\ of the first and second 

 edges specified, may then be written. 



s = -55 + 1*06 s', 

 s = — -50 + '93 s" 



Probably the ratio s/s' and s/s" should be constant and the 

 absolute term in these equations is an error of observation ; but 

 as it is small, little depends upon it, millimeters only being 

 significant. 



If we summarize all the observations for ds/dt, the agree- 

 ment as a whole is in keeping with the nature of the observa- 

 tions and reasonably satisfactory. Thus, in 



Series I, II, d'sjdz = *90, } s = '44 '-[- *85 s' (goniometer in front), 



III, 1-15,1 



IV, 1-13, I 



Y, 110, ! s= -54 + 1*06 s' (goniometer behind), 



VI, 1*13, [ mean ds r /dz = 1-12. 



VII, ' 1-13, | 



VIII, 107, J 



The feature of these data is the occurrence of linear loci for 

 s and nearly throughout the extent of the curves. It is as 

 difficult to even conjecture a reason for this, as it is easy to find 

 reasons against it. The presumptive equation for s is 



s = -004 (7r/6m)* n*, oc 3y^ 



and for n' z if II denotes the product, 



nz = n Q y z Ii. 

 Hence, 



s == -004 (7m /6m)* (y z II)^ 

 if we disregard the subsidence correction for large coronas 



ds I dz oc y z/3 



in which there is no suggestion of a sustained constancy of the 

 coefficient ds/dz such as the experiments show. 



To come to some conclusion as to the cause of the discrepancy 

 between the optic value of the nucleation n' and the presum- 

 able value n (geometric progression), we may compare the 

 * This Journal, xxv, p. 224, 1908. 



