254 HALE— SOLAR MAGNETIC PHENOMENA. [April 24, 



I need not dwell here on the other evidences of the Zeeman effect, 

 but the proof is very complete. The resolution of the spot lines is 

 not sufficiently perfect to permit the numerous components shown 

 in some cases by laboratory observations to be detected, but triplets 

 and quadruplets can be distinguished, and the resemblance of the 

 observed effects to those of a magnetic field is very close for all 

 lines. One of the most important tests is afforded by the steady 

 decrease in the average separation of the components toward the 

 violet, corresponding with the fact that in a magnetic field their 

 separation is proportional to the square of the wave-length. Here 

 we have a marked disagreement with the Stark effect, where the 

 separation of the components increases toward the violet. 



In the case of the sun's general magnetic field, my conclusions 

 are also based exclusively upon displacements due to circular or 

 elliptical polarization. This field, w^hich is about eighty times as 

 intense as that of the earth, but of only about one hundredth of the 

 intensity of the maximum sun-spot field, is quite insufficient to sepa- 

 rate the solar lines. In fact, the widening which it produces is much 

 too small to be detected, and it is only through the possibility of 

 cutting off one or the other component, and thus of producing a 

 slight shift, that it can be measured. 



In the Stark effect the absence of circular or elliptical polariza- 

 tion compels us to seek for evidence presented by changes in the 

 width of lines. The hydrogen lines H/3 and Hy, when observed for 

 the transverse effect, have been shown by Stark to have five com- 

 ponents, the three inner polarized at right angles, the two outer 

 parallel to the field. In the longitudinal effect the two outer com- 

 ponents are absent, while the three inner components are present but 

 unpolarized. In the general electric field of the sun the lines of 

 force may be regarded as radial. Hence all lines having Stark effects 

 similar to those of the hydrogen lines should be wider near the limb 

 than at the center of the sun, and their plane polarized outer edges 

 should be capable of extinction by a Nicol. Lack of symmetry in the 

 distribution of the components of a line, such as Stark has observed 

 in some cases, would cause a shift of the lines near the limit. 

 Tests made some time ago, in connection with the study of the 



