448 



NATURE 



[September 9, 1897 



collimator. For the reduction of the ultra-violet, in both 

 series of photographs the wave-lengths of the hydrogen 

 lines have been assumed as far as H</, from those given 

 by Hale,^ with the exception of Hv, which falls suffi- 

 ciently near the calculated wave-length to be accepted as 

 a hydrogen line. 



With these as datum lines, wave-length curves were 

 constructed, and the wave-lengths of the other lines found 

 by interpolation. 



The wave-lengths of the radiations more refrangible 

 than Hf were determined from extrapolation curves, 

 so that the degree of accuracy is necessarily less than in 

 the case of the remaining lines. 



The scale of intensities adopted is such that lo reprer 

 sents the brightest lines and i the faintest. This facili- 

 tates comparisons with Young's well-known list of chromo- 

 spheric lines, in which loo represents the maximum 

 frequency and brightness. The intensities have been 

 estimated by taking the strongest line in each negative 

 as 10, irrespective of length of exposure. 



This much being premised, let us next consider the 

 thing that is actually measured. If we study the actual 

 photographs, or such reproductions as have been given 

 in Figs. 17 and i8, it will be clear in a moment that the 

 arcs are of different widths because some of the vapours 

 and gases extend further above the photosphere, and i 

 therefore above the dark moon which covers it during 

 an eclipse, than others. Obviously, then, we must not 

 take the centre of the arc. It is also obvious that we 

 must not take that edge further from the dark moon. 

 If we did either of these things, the positions of the lines 

 thus recorded would depend not only on the wave-length 

 of the radiations of the vapours and gases which pro- 

 duced them, but also upon the thickness of the vapours. 



If, however, we take the edge of the arc at the moon's 

 edge, in every case we shall have a series of numbers 

 involving wave-length only, except under two conditions, 

 and this is a very important exception. 



The first condition which may vitiate the determin- 

 ation of wave-length in this way is that some of the 

 vapours or gases producing certain lines may be in 

 movement sufficiently rapid along the line of sight to 

 change the wave-lengths of the lines according to a well- 

 known law. Suppose, for instance, we have a stream of 

 iron vapour moving at the rate of fifty miles a second 

 towards the eye through a mass of hydrogen at rest ; 

 the lines of the iron spectrum will be shifted towards 

 the violet part of the spectrum, while those of hydrogen 

 will be in their normal position. The higher the dis- 

 persion employed, the more carefully must such matters 

 as this be studied. This cause, in fact, will in the case 

 of very violent motion change even the forms of the 

 prominences. 



The forms of monochromatic images of the prominences 

 being produced in part by the movement in the line of 

 sight of the vapours which give rise to them, regions 

 in which th^ vapours are approaching the earth will be 

 displaced to the more refrangible side of their true posi- 

 tions with respect to the sun's limb, and in the case of 

 receding vapours there would be displacements towards 

 the less refrangible end. Such distortions can be deter- 

 mined, if they exist, by comparing the monochromatic 

 images with those photographed at the same time with 

 the coronagraph. For this purpose, in dealing with the 

 eclipse of 1893 a photograph of the eclipsed sun was 

 enlarged to exactly the same size as the K ring shown in 

 Fig. 9, and the comparison could be made very exactly 

 by fitting a negative of one to a positive of the other. 

 No differences of form, however, could be detected, so 

 that the velocities in the line of sight must have been 

 comparatively small. Movements across the line of sight 

 will not affect the forms of the monochromatic images of 

 the prominences. 



1 " Astronomy and Astrophysics," 



NO. 1454, VOL. 56] 



3. 50, 602, 61 



This, is a true physical origin of the change of wave- 

 length which may be detected in eclipse photographs ; 

 but there is a second, as I have hinted. This, although 

 only an apparent change, has to be reckoned with, since, 

 on the one hand, it may be very misleading, while, on 

 the other, if properly dealt with, it may furnish us with 

 new knowledge. 



I have already pointed out that in the determination 

 of the wave-length of the arcs in the prismatic camera 

 photographs the edge of the line nearest to the dark 

 moon must be measured, rather than the other edge or 

 the middle of the arc. But this assumes that all the 

 arcs really rest on the dark moon. It is possible, hotv- 

 ever, that some of them do not extend down to it — that 

 they represent real upper layers, and in this case the 

 wave-length obtained by a reference to the dark moon 

 will not be the true one, and, by some means or another, 

 it will have to be corrected. This, though a difficult 

 problem, does not seem an impossible one. 



Eclipse Work in relation to the Dissociation Hypothesis. 

 In the course of the spectroscopic solar investigations 

 which have been going on since 1868, I have pointed out 

 over and over again that the phenomena observed could 

 be more easily explained on the hypothesis that the 

 chemical elements with which we are familiar here were 

 broken up by the great heat of the sun into simpler 

 forms, than in the ordinary one that the " elements " as 

 we deal with them in laboratories are incapable of 

 simplification, that is that they are indestructible. 



The recent work on the enhanced lines of several of 

 the metallic elements, really enables us to predict what 

 wc shall obtain in the Indian eclipse if the dissociation 

 hypothesis be true. 



With regard especially to the bearing of the 1893 work 



on this view, I may state that it is entirely in its favour. 



The preliminary discussion of individual substances has 



further abundantly shown that although some of the 



lines belonging to any particular metal may appear as 



dark lines in the solar spectrum on account of absorption 



by the chromosphere, other lines of the same substance 



are only represented among the dark lines because of 



absorption taking place elsewhere. This again is an 



; indication of the stratification of the sun's absorbing 



j atmosphere, which, if it exists, must furnish a very 



strong argument in favour of the dissociation of metallic 



I vapours at solar temperatures. In fact, the eclipse 



I phenomena have been found to be as bizarre, in relation 



i to the non-dissociation hypothesis, as those which I have 



I already discussed in relation to observations of sun- 



] spots, chromosphere, and prominences, made on the 



uneclipsed sun. 



The long-continued Italian observations of the quiet 

 solar atmosphere and the Kensington observations of 

 sun-spots have already been especially mentioned. Not 

 only is there no correspondence in intensity, but the 

 variation in the sun-spot spectrum from maximum to 

 minimum is enormous, while the Fraunhofer lines remain 

 constant. 



The view I expressed in 1879,^ ^.nd to which I adhere, 

 is therefore strengthened by the eclipse work. I then 

 wrote : "The discrepancy which I pointed out, six years 

 I ago, between the solar and terrestrial spectra of calcium 

 is not an exceptional, but truly a typical, case. Varia- 

 I tions of the same kind stare us in the face when the 

 minute anatomy of the spectrum of almost every one of 

 the so-called elements is studied. If, therefore, the 

 arguments for the existence of our terrestrial elements in 

 extra-terrestrial bodies, including the sun, is to depend 

 upon the perfect matching of the wave-lengths and in- 

 tensities of the metallic and Fraunhofer lines, then we 

 are driven to the conclusion that the elements with 

 which we are acquainted here do not exist in the sun." 



1 Roy. Soc. Proc, 1879, vol. xxviii. p. 13. 



