Feb. 1 8, 1886J: 



NA TURE 



371 



necessary to heat through the same range an atom of 

 barium, provided both substances are taken in the same 

 state — both sohds for instance. 



If then the specific heat of an elementary atom is 

 represented by 6, that of a compound atom of chloride of 

 barium will be represented by 18. This is in truth the 

 law which Kopp has found to hold with respect to the 

 atomic heat of compound bodies, and the theoretical con- 

 clusion to be derived from it, and that of Dulong and 

 Petit, is, not that the elements are essentially different 

 from the compound bodies, but that, if compound, the 

 forces which bind together their constituents are vastly 

 more powerful than those which bind together the so- 

 called constituents of bodies known to be compound. 



Again, if we compare together the atomic weights of 

 the so-called elements with those of compound bodies, we 

 shall find that as a whole the former are smaller than 

 the latter — that is to say, the family of ebments have on 

 the whole smaller atoms as well as smaller atomic heats 

 than the families of compounds. Now, if the elements are 

 in reality compounds we might expect in like manner 

 that those which have the smallest atoms should have 

 the smallest atomic heats. 



We have great reason for supposing that this is the 

 case, for, although we have not obtained the specific 

 heat of either oxygen or hydrogen in the solid state, 

 Kopp has found that his law with regard to compounds 

 will only hold good under the hypothesis that the atomic 

 heats of hydrogen and oxygen are decidedly less than 

 those of the great bulk of the elementary bodies, that 

 of hydrogen being likewise smaller than that of oxy- 

 gen. Furthermore carbon and boron are two elements 

 which have small atoms. Now if we make the ob- 

 servation at ordinary temperatures it will be found that 

 the atomic heats of these two elements are decidedly less 

 than those of the great bulk of the elements. In fine, 

 elements of small atomic weight and presumed simplicity 

 of structure appear to bear to those of great atomic 

 weight a relation similar to that which the elements as a 

 class bear to the compounds as a class, as far as atomic 

 heat is concerned. 



On the whole the result of this discussion appears to 

 be in favour of the so-called elements being in reality 

 compound structures the components of which possess 

 attractions for each other vastly greater than those 

 exhibited in ordinary chemical combinations. 



In connexion with this branch of my subject I may 

 allude to the peculiar family relation between certain 

 elements which all chemists are now agreed in recog- 

 nising. 



This means that the various members of a group of the 

 elements consisting, let us say, of A, B, C, and D, bear to 

 one another some peculiar relation different from that 

 which they bear to the other elements. Now this is pre- 

 cisely what happens in the case of groups of substances 

 which we know to be compound, and the impression is 

 thus conveyed that the elements themselves consist of 

 varied groupings of some still simpler substance. Indeed 

 it seems quite possible that there may be only one kind of 

 primordial atom, and the fact that the force of gravitation 

 bears a constant relation to mass quite independently of 

 chemical constitution seems to speak strongly in favour 

 of some such hypothesis. 



Let us now try to picture to ourselves what would have 

 happened had spectrum analysis been known as an in- 

 strument of research at the time when we were yet 

 unable to isolate the metal sodium. Under such circum- 

 stances chloride of sodium and caustic soda would both 

 be considered as separate elements and the spectra of 

 both these bodies exhibiting the same yellow line would 

 lead to the conclusion that these substances contained 

 some common principle which was momentarily dis- 

 sociated from its surroundings in the spectral flame. 

 This leads us to ask whether there are any such coincident 



lines in the spectra of the various so-called elements 

 besides those which may be caused by common im- 

 purities. Lockyer, who has greatly studied this subject, 

 tells us that short-line coincidences exist between many 

 metals, the impurities of which have been eliminated, or 

 in which the freedom from mutual impurities has been 

 demonstrated by the absence of the longest lines. Some 

 of his results are exhibited in Fig. 25, in which the lines 

 marked - are due to impurities, while those marked + 

 are common or basic lines. It would thus seem that these 

 short-line coincidences cannot be due to impurities, and 

 the question at once arises whether they do not indicate 

 the presence of some common principle in the spectra 

 before us momentarily dissociated from its surroundings 

 by the high temperature. 



It is important here to explain what we really mean 

 when we speak of a coincidence between two spectral 

 lines. We mean simply that there is no perceptible dif- 

 ference in their position when examined with an instru- 

 ment of a certain power. IVIr. Lockyer therefore did not 

 with his instrument perceive any such diflerence in the 

 spectral position of certain short lines given by various 

 elements. Messrs. Liveing and Dewar, however, applied 

 to some of these lines an instrument of greater power, 

 and succeeded in showing that in many cases there was 

 a slight difference between their spectral positions. 



This result raises a new question. We have now to 

 ask ourselves what, under these circumstances, is the 

 value that we can attach to this very near but not quite 

 absolute coincidence between certain short lines of various 

 elements ? 



Now, in comparing together certain absorptive spectra 

 of compound bodies which have some principle in com- 

 mon, w-e learn from the researches of Russell and others 

 that we can sometimes trace a band presumably due to 

 the common principle, the spectral position of which is, 

 however, slightly different in the various compounds. 

 The want of perfect freedom may, it is imagined, alter 

 slightly the time of vibration of the molecular groupings, 

 and thus displace the spectral position of the absorption 

 bands. I think we are justified in imagining that some- 

 thing of this kind may take place in the elements, in 

 which, the forces being so intense, our highest attainable 

 temperature may be insufficient to produce complete dis- 

 sociation. In this case the want of complete concordance 

 in the short lines common to various elements, or basic 

 lines, as these have been named by Lockyer, may denote 

 nothing else than this absence of power. 



It is imagined that these approximate coincidences are 

 too frequent and too near to be due to chance, but this is 

 a subject that will ultimately require mathematical in- 

 vestigation. In fine, we may conclude this short account 

 of the terrestrial evidence regarding the nature of the 

 elements by saying that — (i) there is no proof that they 

 form a class essentially different from compound bodies, 

 but much to the contrary ; and (2) that, if compound, the 

 forces which bind their constituents to one another must 

 be very great. 



Before discussing the spectra of the sun and stars it 

 may be well to pause for a moment and ascertain what 

 we mean when we say that sulphur, for instance, is an 

 element. It is quite clear that solid sulphur, liciuid sul- 

 phur, and gaseous sulphur are different things ; also, we 

 may have two kinds of solid sulphur, while if we take the 

 spectrum of gaseous sulphur there is little doubt that the 

 molecular groupings suffer vast changes as the tempera- 

 ture rises. 



Now, in all its variotis states we still call the substance 

 sulphur, because if we firing it down to the temperature 

 of our laboratories it will combine with other bodies as 

 sulphur and as nothing else. Thus the word sulphur 

 does not in reality mean a definite arrangement of matter. 

 Similar remarks apply to other elements, several of which 

 are in the form of gas and give us their spectra in 



