604 
MR, J. NORMAN LOCKYBR ON THE 
Turner in 1886.*' His observations were made under less favourable conditions than 
those in Egypt, and in the absence of statements as to the relative lengths of the lines 
observed, it is impossible to utilise them. 
This is one of the most important points in solar physics, but there is not 
yet a consensus of opinion upon it. Professor Young and others apparently still hold 
to the view first announced by Dr. Frankland and myself in the infancy of 
the observations, that the Fraunhofer absorption takes place in a thin stratum, lying- 
close to the photosphere. 
It is, therefore, of the highest importance to determine what light is thrown on the 
subject by the photographs of 1893. 
A study of the distribution of the lines recorded in Tables I., II., and III., 
throughout the different parts of the chromosphere and prominences as revealed at 
different phases, proves that the spectrum gradually becomes more complex as the 
photosjDhere is approached. 
A partial photographic record of the reversal of the lines at the beginning of 
totality was secured in Brazil (Photograph No. 2), and of the reversal near the end 
in Africa (Photograph No, 22). Owing to over exposure, the Brazilian photograph 
(Plate 12) is very dense, and the lines are only distinctly seen in the regions F to b, 
and that more refrangible than K. The African photograph, taken just after totality 
(No. 22, Plate 13), shows the ends of the long bright arcs visible at the end of totality, 
and we seem to be justified in taking these arcs as representing the reversals seen in 
former eclipses. The number of these bright arcs, as well as the Brazilian photo¬ 
graph No. 2, indicates a spectrum of considerable complexity near the base of the 
sun’s atmosphere. 
The complexity of the spectrum of the sun’s atmosphere in the neighbourhood 
of the photosphere is indicated in another way. In the African photographs, taken 
after totality, except at the southern cusp, the spectrum, as already pointed out, is due 
to the following four sources :—(1) The spectrum of the vapours of the cusps, giving- 
bright arcs ; (2) the spectrum of the visible crescent of photosphere, consisting of a 
bright continuous spectrum crossed by dark arcs ; (3) the spectrum of the vapours 
surrounding the uneclipsed part of the sun, giving bright arcs ; (4) the spectrum of 
the corona, consisting of a relatively feeble continuous spectrum and a number of bright 
rings. 
The results vary with the relative intensities of the different sources, the chief 
variable being the breadth of the photosphere, depending upon the interval from 
totalitv. 
When the crescent is very thin, as in the African Photograph, No. 22, the bright 
chromospheric arcs are relatively strong and are seen crossing the continuous 
spectrum of the photosphere, while the dark arcs due to the photosphere are absent. 
Bespigiii observed a similar absence of the Fraunhofer spectrum just after totality in 
* 1 Phil, Trans.,’ 1889, vol. 180, A, p. 391. 
