140 



KNOWLEDGE. 



[June 1, 1900. 



the unit of atomic weights. This was the first appear- 

 ance of the modified or '' elastic " Prout, as it has been 

 sarcastically named. A little later, Dumas proposed 

 one-quarter of the atomic weight of hydrogen. Someone 

 suggested ether as the hypothetical protyle, and gave 

 it an atomic weight of .0001, a figure which is, of couj-se, 

 quite beyond the reach of analytical chemistry. 



A theory that is being continually changed like this 

 needs no disproof; but the magnificent work of Stas 

 was destined to give the death-blow to both the original 

 and the elastic Prout, as far as it could be given by exact 

 analysis. Yet Stas himself began his work as a firm 

 believer in Prout. He wrote: — "When I began my 

 researches, I had an almost absolute confidence in the 

 exactness of Prout's Law." His " absolute confidence ' 

 soon vanished, and he afterwards described the hypo- 

 thesis as a pure illusion. 



In a paper on the atomic weight of aluminiitm, pub- 

 lished in 1880, Mallet criticised Stas' work on the ground 

 that, though accidental errors had been practically 

 abolished, there might still remain undetected constant, 

 errors. This is highly improbable, considering the 

 variety of methods used by Stas; and, even if true, it 

 would prove little or nothing. For while the detection 

 of a constant error may bring one element nearer the 

 required value, it is just as likely to take an'^ther 

 element farther away from it. Oxygen is a case in 

 point. Since the date of Mallet's paper it has fallen out 

 of the list of elements whose atomic weights are approxi- 

 mate whole multiples of that of hydrogen. The work of 

 Keiser, Cooke and Eichards, Rayleigh. Scott, Morley and 

 others, leaves little doubt that the atomic weight of 

 oxygen is less than 15.9, and is probably very near io 

 15.88. 



There is nevertheless a real and striking approxima- 

 tion of many atomic weights to whole multiples of *^hat 

 of hydrogen. In the paper already refeiTed to. Mallet 

 gave a list of eighteen elements whose atomic weights 

 have been most acctirately determined. Ten of these 

 atomic weights were within 0.1 of whole numbers. (As 

 already stated oxygen has fallen out of this list.) The 

 chances against this occiuTing accidentally are more than 

 1,000 to 1. F. W. Clarke extended this argument to 

 sixty-six elements, of which forty have atomic weights 

 falling within the 0.1 limit of variation. He says that 

 forty agreements include nearly all the t•rustwor^.hy 

 determinations. 



The case seems to stand thus: — Prout's hypoth?sis 

 in its original form, and in the modifications proposed 

 by Maiignac and Dumas, is untrue. But there is, never- 

 theless, a certain approximation to it, which is scarcely 

 likely to be quite accidental. 



All this, however, does not really affect the wider 

 question, whether the elements are primordially distiuct 

 bodies, or whether they are derived by aggregation from 

 a simpler form or forms of matter. For a long tim? 

 there was only negative evidence on both sides; but for 

 a good many years now positive evidence has been 

 pccumulating in favour of the evolution of the elemen*^s. 



To begin with, the only evidence in favour of the 

 elementary nature of the elements is ptirely negative. 

 The definition of an element is based, not on any 

 attribute of the thing defined, but on the limitdtions 

 of human power. It is merely a confession of im- 

 potence. 



The progress of science is, in general, a process of 

 simplification. Larger and larger groups of facts aie 

 brought under more and more general laws. Thus 



chemists have, during the last two centuries, reduced the 

 millionfold chemical complexity of heaven and earth 

 to, say, a seventyfold complexity. Is there to be at 

 this point a solution of continuity, or is the simplifi- 

 cation to go on to its logical conclusion ? 



Physics, it may be noted, takes small account of 

 chemical differences. AU forms of matter alike obey the 

 laws of phvsics in the same way. Graham, who was 

 both phvsicist and chemist, was strongly impressed by 

 this physical unity underlying chemical diversity, fie 

 wrote in 1863: — "It is conceivable that the various 

 kinds of matter, now recognised as different elementary 

 substances, may possess one and the same ultimate or 

 atomic molecule existing in different conditions of move- 

 ment. The essential unity of matter is an hypothesis 

 in harmony with the equal action of gravity on all 

 bodies." 



Again, the relations between the atomic weights of 

 the elements render their complete independence of one 

 another hardly supposable. The gi-eatest generalisatiou 

 made in chemistry since the atomic theoi^y is " the pro- 

 perties of the elements are functions of their atomic 

 weights " ; and the prediction of the properties of 

 gallium, scandium and germanium was a greater 

 triumph to chemistry than was the prediction of 

 Neptune to astronomy. 



The work of Sir William Grookes on the rare earths 

 is of the highest significance. He has shown that a 

 substance like yttrium, which from every chemical point 

 of view behaves as an element, can by repeated fractional 

 precipitations be split up into several groups having 

 different spectra, and presumably different atomic 

 weights. The process of fractionation implies differenras 

 smong the " elementary " atoms, and a possibility of 

 selection. Crookes writes thus about didymium, after 

 splitting it up into neodymium and praseodymium : — 

 " Didymium is certainly a compound. It has a definite 

 atomic weight and well-defined salts, and has been 

 closely scrutinised by some of the ablest chemists in the 

 world. But it emerged as a seeming element from every 

 trial." 



The distribution of the elements in the earth also 

 deserves consideration. How can this be explained 

 apart from the theory of evolution? Why, for instance, 

 should nickel and cobalt be always found together ? 

 Why should the platinum group of metals and the rare 

 earths be so localised and so rare ? Why should meteoric 

 iron always contain nickel and cobalt, and very often 

 manganese and chromium as well — all elements of 

 similar properties, and nearly equal atomic weights? 

 These facts are just what one would expect if these 

 elements had been formed under nearly identical con- 

 ditions from simpler forms of matter. 



The remainder of the evidence is more positive in 

 character, and is chiefly due to the spectroscopic re- 

 searches of Sir Norman Lockyer. So long ago as 1876, 

 Lockyer showed that the spectrum of calcium varies at 

 different temperatures, and that the changes brought 

 about by rising temperature are exactly parallel to the 

 changes in the spectrum of a compound as it is gradually 

 dissociated by heat. In short, this element behaves, 

 spectroscopically speaking, as a compound. 



The spectrum of iron is well known to be enormously 

 complex; and it ought to be noticed, in passing, that 

 this complexity is in itself some evidence against the ele- 

 mentary nature of iron ; it is difficult to imagine an 

 indivisible atom vibrating in so many hundreds of dif- 

 ferent ways. But the iron lines in the solar spectinm 



