334 ANNUAL EEPORT SMITHSONIAN INSTITUTION, 1908. 



of measurement. The other doublets, \ 5903.56 and \ 5938.04, show in the spot 

 spectrum but little more than one-half the separation that would be expected 

 on the assumption that the strength of the field is the sam^ for all of these lines. 



On consideration it will be seen, however, that the separation of the doublets 

 must depend, in some degree, on the distribution of the absorbing vapor in the 

 solar atmosphere, and on the coefficient of absorption of the particular line 

 employed. A striking instance of this kind, affecting lines of the same series, is 

 illustrated in the case of hydrogen, described in a previous paper.** Although 

 the H5 line extends to the upper part of the chromosphere and prominences, the 

 mean level represented by its absorption is much lower than that given by Ha. 

 The consequence is that Ha enables us to photograph the solar vortices, the 

 characteristic stream lines of which do not apiiear at the lower H5 level. Simi- 

 larly, if the intensity of a given titanium line falls off rapidly, the level repre- 

 sented by this line may be comparatively low. If, on the other hand, its 

 intensity curve is of such a form as to indicate that the absorption at higher 

 elevations plays an important part, the mean level represented by the line may 

 be considerably higher than in the previous case. To settle this question we 

 must know: (1) The range of elevation in the spot of the vapors of iron, 

 titanium, and other elements; (2) the intensities of the lines of these elements 

 at different levels; (3) the rate at which the strength of the field decreases 

 upward. 



In the absence of information regarding the first two points, we may inquire 

 as to the probable relative behavior of titanium, iron, and other elements if the 

 distribution of the vapors at different levels were the same as in the chromo- 

 sphere. From a discussion of a large number of photographs of the flash spec- 

 trum, made by different observers at several eclipses, Jewell has compiled a 

 table showing the heights above the sun's limb attained by various lines in the 

 blue and violet.^ The heights for titanium range from 100 miles (160 kilo- 

 meters) for \ 4466.0 to 3,500 miles (5,640 kilometers) for X 4466.7, while certain 

 strong enhanced lines in the ultra-violet reach elevations of 6.000 or s,(X)0 miles 

 (9,600 or 12,880 kilometers). For iron the minimum height is 200 miles (320 

 kilometers) for \ 4482.4 and the maximum 1,000 miles (1,610 kilometers) for 

 \ 4584.0. Chromium ranges from 100 miles for X 4280.2 to 1,200 miles (1.930 

 kilometers) for X 4275.0 ; manganese from " 100 miles or more " for X 4451.8 to 

 "800 miles (1,290 kilometers) or more" for X 4030.9; vanadium from 100 miles 

 for X 4390.1 to 200 miles for X4379.4. It thus appears that the range in level 

 represented by the titanium lines is much greater than for the lines of iron, 

 chromium, manganese, and vanadium. If the vapors were similarly distributed 

 in spots, the maximum strength of field indicated by tlie titanium lines should 

 therefore correspond with the maximum value for iron, but some titanium lines, 

 produced by absorption at higher mean levels, should give lower lield strengths. 

 Chromium should agree more nearly with iron. Vanadium, if the less refrangi- 

 ble lines reach no greater elevations, should give closely accordant (maximum) 

 values for the field strength. It will perhaps be possible, with the aid of the 

 30-foot spectrograi)h, to determine the relative levels in the chromosphere 

 attained by most of the lines in question, but it is a much more difficult matter 

 to do this for sun spots. I hope, however, that our new spectroheliograph of 

 30 feet focal length may throw some light on this subject. 



It is evident that these considerations will have no bearing on the present 

 problem, unless the field strength decreases very rapidly upward in spots. 



" Solar "Vortices, p. 3. 



^ " Total Solar Eclipses of May 28, 1!M)(), and May 17, 1901," Publications of the 

 United States Naval Observatory, second series, Vol. IV, Appendix I. 



