366 



NATURE 



\Atigit.st 1 8, 



The time has uow come, I think, to go into this question in 

 more detail. 



Let us consider the maps. Of the 96 iron lines in that 

 first region which we considered only 4 are seen in the flames ; 

 92 of those therefore must not be looked for in the flame region, 

 for the reason that twelve years of patient work have not divulged 

 their existence. Again of these same 96 lines only about 32 are 

 seen in the spots at all extensively affected. It is useless there- 

 fore to look for the remaining 60 lines or thereabouts in the same 

 spot region, for the reason that they have been looked for for a 

 long time without being seen equally widened. 



Of course it must be remembered that these changes are due 

 to change of intensity, and that otlier lines may be there of an 

 intensity so low that they have escaped the keen eyes of those 

 anxious to chronicle them. Still it will be acknowledged, I 

 think, that the method of treatment I have adopted is the best 

 open to us, and is a fair one on the whole. 



The facts being so, it looks really as if the origin of the 

 mass of the absorption to which the Fraunhofer spectrum of iron 

 is due is to be sought in a region of the solar atmosphere much 

 nearer to the place assigned to it by Kirchhoff originally than to 

 that lower region where we considered we were driven to place 

 it when the new method was first established. When the new 

 method had been working for some considerable time observers 

 recorded hydrogen with magnesium underlying it, and with 

 sodium underlying that. And since they were metals of Xciw 

 atomic weight and vapour density we were jastified in considering 

 them as occupying the highest levels — the very extreme limit of 

 the solar atmosptiere. 



It was therefore fair to argue that if the substance-, of the 

 lowest atomic weights were really close to the photosphere, those 

 of highest atomic weights were really in the photosphere itself, 

 and therefore, being in the photosphere, the absorption by means 

 of which we were able to determine their existence really took 

 place in or near the photosphere. 



This later work, I think, seems to show that that view re- 

 quires reconsideration ; and it may well be that subsequent 

 work will show that those Fraunhofer lines, which we do not 

 trace in flames and which we do not trace in the spots, are 

 probably absorbed in a cooler, higher region of the atmosphere, 

 much more nearly occupying the place assigned to the general 

 atmosphere by Kirchhoff than that which has been given to it by 

 later observers. If we accept this the work becomes a little 

 plainer, and the reason that we get such an excessively simple 

 spectrum in the lower reaches of the sun is because the more 

 complex vapours exist at a considerable elevation above them, 

 and as the interior of the sun must be hotter than any of its 

 envelopes, no cold substances — nothing approaching the solid state 

 which we have learnt for many years gives us the most complete 

 spectrum of the substance — nothing approaching a solid can enter 

 those charmed regions. 



Therefore we are also again driven to the view that these 

 cooler vapours — vapours much nearer the solid state, much more 

 condensed, much more complex than those which can exist alone 

 in .the hottest layer — probably originate the great mass of ab- 



sorption ; that is, many lines not traced either in spots or flames 

 are produced in the higher regions. 



If this be so, the Fraunhofer spectrum is really not the 

 spectrum of any particular part of the sun ; but because it con- 

 tains lines thickened in the spots, lines brightened in the flames, 

 and other lines about which we know nothing, it must really be 

 the summation of the absorption of the different strata which 

 compose the solar atmosphere ; so that chemically the solar atmo- 

 sphere, with regard to the iron spectrum, gets more and more 

 complex every mile we go upwards. Of course, too, if this is good 

 for iron it is good for every other substance which we believe to 

 exist in, or to have some connection with, the solar atmosphere. 



Further Test supplied by this Vietu 

 If this be so we really can go on with our tests; we can bring 

 the laboratory into the field, and we must leam in our labora- 

 tory experiments to make abstraction of those hues which are 

 due to the more complex masses reduced by the transcendental 

 temperature which we employ, if there is any truth in the view 

 that I am bringing before you. In a laboratory experiment, for 

 instance, when we want to observe the vapour of iron we have 

 to employ two poles of solid iron. We have no means, such as 

 are afforded us by the sun, of shielding the precise part we want 

 to observe by a considerable number of envelopes of gradually- 

 increasing temperature, so that even if we can get the highest 

 temperature in the laboratory this result of the highest tempera- 

 ture will be cloaked, masked, and hidden by all those results, 

 by all those simplifications which have been brought about to 

 produce that precise effect of the highest temperature. So that 

 the only thing we can do is to watch the intensities of the lines 

 when we considerably change our temperature. I am speaking 

 now of iron. I will show by and by that for some other substances 

 there is a method which enables us to get over this excessive 

 difficulty, for no doubt a very great difficulty it is ; but in the 

 case of iron, that really is the only thing that remains to us. 

 Fig. 38 will give an idea of the way in which we may be misled 

 if we do not examine our light source « ith the greatest care. It 

 is engraved from a photograph of the spectrum taken between 

 two poles of a Siemens machine, moistened with a salt of 

 calcium, an image of the vertical poles having been thrown on 

 the vertical slit. 



It is seen how wonderfully we get the simplifications brought 

 about by the electric current, depicting themselves in two per- 

 fectly distinct ways. The lower part gives the spectrum of the 

 positive pole, and the upper part of the spectrum of the nega- 

 tive pole. In the first place it will be seen that there is no axis 

 of symmetry for these lines ; some of them elongate considerably 

 in one direction ; others of them elongate considerably in the 

 other ; some of them are exceedingly short, and only appear 

 close to that region of the negative pole where the lines broaden ; 

 others again are brighter in the region much nearer the middle 

 of the field. Others of the lines start from a region far rem jved 

 from the arc ; others again seem to start almost in the arc itself. 

 Now this not only reminds one of what one sees in a solar storm, 

 but it shows us most distinctly that even in the electric arc, when 



Fig. 38.— Photograph of the spectrum of the poles, showing 



we have had time to study it sufficiently, these very simplifica- 

 tions which we have been so long in search of may be recognised 

 eventually and permanently recorded. 



Tests supplied by the Variations bdtoeen Solar and Terrestrial 

 Spectra 

 Attention has been called to Kirchhoff's statement that the 

 existence of the terrestrial elements in the sun is established by 

 the fact of the coincidence of wave-length and intensity between 

 the lines visible in our laboratories and the lines recorded as 

 existing in the solar spectrum. 



We have now arrived at a point wlien we can discuss this with 

 advantage. 



I propose to show first that the statement is not true ; and. 

 secondly, how the tests supplied by the variations from terrestrial 

 spectra can be explained on, and bring most valuable confirmation 

 to, my view. We are now able to say that at least two causes 

 are at work, and they will require to be discussed separately. 



But first as to the facts. We have already seen i*hat enor- 

 ninus differences there are in the spectrum of calcium under 

 different conditions. In the diagram of the calcium spectrum 

 (Fig. 28) we saw that H and K, the most important lines in 



