September 9, 1897] 



NATURE 



447 



minence No. 3 as it appears in the African Photographs 

 Nos. 1 1, 9, and 7 respectively, the first giving the spectrum 

 of the upper part only, while the last shows the spectrum 

 nearer the base. Accepting the time of commencement 

 of totality in Africa as 2h. 23m. 48s. by the deck watch, it 

 has been calculated that Spectrum i represents a part of 

 the prominence 22"-26" (9950 to 11,600 miles) above the 

 photosphere ; Spectrum 2, 6"7-8"*5 (3000 to 3800 miles) ; 

 and Spectrum 3, 3"7 (1660 miles) above the photosphere. 

 Strip ^is the spectrum of the base of the chromosphere 

 as represented by the cusp in the African Photograph 

 No. 22. 



These enlarged spectra have been obtained by covering 

 copies of the original negatives with tinfoil, leaving only 

 narrow strips showing the prominence spectra, and giving 

 them the necessary width by moving the photograph in 

 a direction at right angles to the length of the spectrum.^ 



The want of e.xact coincidence of lines common to 

 different horizons in the copies of the photographs 

 which I have given is due to the difficulty of obtaining 

 enlargements on exactly the same scale. The difference 

 in thickness of the same line in different photographs of 

 a prominence is due to the varying widths of the cor- 

 responding images of the prominence formed by the 

 prismatic camera at different stages of the eclipse. 



In order that the changes of intensity of the various 

 lines may be separated from the effects due to varying 

 exposures, the individual observations are arranged in 

 groups according to the time of exposure of the photo- 

 graphs. 



In contrasting the spectrum of the prominences with 

 the spectrum of the cusp, it should be borne in mind that 

 the cusp in the African photograph specially examined 

 (No. 22), does not represent the base of the chromosphere 

 immediately beneath either of the metallic prominences. 

 Still the cusp was not far from a prominence (No. 19), and 

 it is fair to consider the base of the chromosphere homo- 

 geneous. If so, the prominences cannot be fed from the 

 base of the chromosphere, since they contain different 

 vapours. 



TJie Spectrum of the Chrpinosphere at Different Heights. 



But we are not limited in these investigations to the 

 study of the prominences ; we can obtain similar in- 

 formation from the chromosphere itself. 



The spectrum of the chromosphere itself at different 

 heights can also be partially investigated in the eclipse 

 photographs. A considerable arc of chromosphere was 

 photographed in one of the African negatives (No. 21). The 

 photograph was taken about ten seconds before the end of 

 totality, so that the lower reaches of the solar atmosphere 

 within 1660 miles of the photosphere were hidden. The 

 bright arcs accordingly represent the spectrum of the 

 chromosphere above that height. None of the photo- 

 graphs give us any information as to the spectrum lower 

 down until we come to the part very near to the base 

 which is shown at the cusp in another photograph (22). 

 Most of the lines become relatively brighter as the base 

 of the chromosphere is approached, but some become 

 dimmer. 



The Evidence as to the Existence of Layers in the 1893 

 Photographs. 



The most direct evidence which the eclipse photo- 

 graphs give as to the separation of the solar atmospheric 

 vapours into layers is that afforded by the increased 

 relative brightness or;^f6n:^e- of the lines in passing to 

 higher levels. ' ''' *•"'",'" 



We have seen that a careful and impartial tabulation 

 of intensities has shown that both in the prominences 

 and in the chromosphere some vapours do seem to be 

 Ijrighter as they increase their distance from the photo- 

 sphere. 



> Phil Trans., 1893, vol. clxxxiv. A. p. 684. 



NO. 1454, VOL. 56] 



As we have to deal with the projection of a sphere and 

 not with a section of the sun's atmosphere, the spectrum 

 arcs would brighten in passing outwards from the photo- 

 sphere in consequence of the increased thickness of 

 vapour presented to us, even if the radiation per unit 

 volume remained constant. The spectroscopic differ- 

 ences studied and carefully recorded show, however, 

 numerous inversions even in the behaviour of the same 

 Hne in different prominences, so that the increased bright- 

 ness observed cannot always be due to this cause alone. 



Some of the lines are brightest at the base of the 

 chromosphere, while others are brighter at greater 

 elevations. As already explained, some of the lines which 

 are brightest above the photosphere are probably produced 

 by vapours existing in layers concentric with, but above 

 and detached from the photosphere. Those lines which 

 become dimmer in passing outwards must owe their 

 origin to vapours resting on the photosphere. 



It will be obvious to everybody that the more the idea 

 of the absorption which gives rise to the Fraunhofer lines 

 taking place in one thin layer is disproved, the more 

 certain it must be that it represents the integrated effect 

 of several layers. Hence this special examination of the 

 1893 photographs, to which I am now directing attention, 

 to see if there are any indications as to the localisation 

 of the absorbing vapours which are not represented in 

 the base of the chromosphere. 



It will be noted that the evidence is distinctly in favour 

 of such localisation above the chromosphere. 



But the matter is so important that it must not be 

 allowed to rest here, while photographs with higlier 

 dispersion are possible. Hence, then, the 1898 results 

 must be carefully studied from this point of view. 



The Chemical Constitution of the Suris Atmosphere. 



The results obtained with regard to the chemistry of 

 the sun will, of course, depend upon the results of the 

 accurate measurement of the arcs, both long and short, 

 obtained jn the prismatic cameras ; and of the lines — true 

 images of the slit— in slit spectroscopes ; these measure- 

 ments have for their object the determination of the 

 wave-lengths of the radiations, so that they can be com- 

 pared with the wave-lengths of terrestrial substances 

 observed in the laboratory. 



The dispersion employed in 1893 was only moderate 

 as compared with that now used in laboratory work, 

 though it was far greater than any employed in eclipse 

 observations before. The doubled dispersion proposed 

 for 1898 will necessitate additional precaution against 

 error, and in return it may land us in new discoveries. 

 To show that this remark is justified, I will first refer to 

 the method employed in the determination of wave- 

 lengths in the case of the photographs of the 1893 and 

 1896 eclipses. 



The wave-lengths are expressed on Rowland's scale. 

 In the region less refrangible than K, they have been 

 determined from the African photographs, by comparison 

 with the spectrum of Arcturus and other stars photo- 

 graphed with the same instrument, the wave-lengths of 

 the lines in which were determined by reference to Row- 

 land's photographic map. The spectrum of Arcturus is 

 almost identical with that of the sun, so that the com- 

 parison lines were sufficiently numerous for the purpose. 

 Stars like Bellatrix were employed as an additional 

 check in the case of bright lines not coincident with 

 prominent Fraunhofer lines. 



Micrometric measurements of the lines were also made 

 and reduced to wave-lenjJths in the usual way, by means 

 of a curve ; these furnished a check on the general 

 accuracy. In the case of the Brazilian negatives wave- 

 lengths were determined by means of micrometric mea- 

 sures and a curve, and checked by direct comparisons with 

 a solar spectrum, photographed with the same spectro- 

 scope while it was temporarily provided with a slit and 



