628 Professor Georr/e E. Hale [May 14, 



times that of the spot, the quantities obtained by dividing the separa- 

 tions in the second column by 5*1 are given in the third column. 

 These separations are directly comparable with the separations of the 

 corresponding lines in the spot, which are given in the fourth column. 

 The fifth column shows that the differences between the solar and 

 laboratory results are very small. As the strength of the field in the 

 laboratory was about 15,000 gausses, the strength of the field in this 

 spot would be about 15,000 -^ 5-1 = 21)00 gausses. The strongest 

 field hitherto measured on our photographs of sj:)ot spectra is about 

 4500 gausses, corresponding to a considerably greater separation of 

 the Hues (Fig. 8). 



When a similar comparison was made for various lines of titanium 

 and chromium, a much less perfect agreement between the spot and 

 laboratory results was found. It had already been observed that 

 such lines as D of sodium and 1) of magnesium, which undoubtedly 

 represent a much higher level than the great majority of lines in the 

 spot spectrum, are but very slightly widened. As these lines are 

 strongly affected by a magnetic field in the laboratory, it appeared 

 evident that the strength of the field in spots must fall off rapidly in 

 passing outward through the spot vapours. Under these circum- 

 stances lines of other elements, which represent levels higher than the 

 average, should show small separations in the magnetic field of the 

 spot. It seems probable that in this way the lack of perfect agree- 

 ment between the laboratory and solar results, observed in the case of 

 titanium and chromium, can be accounted for. 



A further important test was afforded by the well-known phe- 

 nomenon exemplified in Preston's law. According to this law, the 

 distance between the components of the lines split up by a magnetic 

 field varies directly as the square of the wave-length. This we found 

 to be true even in the case of a metal like iron, the lines of which 

 cannot be grouped into series, if the average separations of a sufficient 

 number of hues were considered. We should therefore expect that 

 the widening of lines in spots would rapidly decrease toward the 

 violet, and that the separation of spot doublets should diminish in a 

 similar way. A study of the spot spectrum shows that this actually 

 occurs. 



It soon appeared that the normal spot spectrum always contains 

 triplets as well as doublets (Fig. 8.) These are less easily recognised, 

 because the presence of the central line crowds the components so 

 closely together that they are not readily separated with the resolving 

 power available. As these triplets are photographed even when the 

 spot is very near the middle of the sun, it is evident that the spot 

 always sends out light which makes a considerable angle with the lines 

 of force. In a normal triplet the central line is of twice the intensity 

 of the side components, when observed at right angles to the lines of 

 force, and disappears altogether when observed parallel to the lines of 

 force. Thus, by determining the relative intensities of the central 



