August 28, 1896.] 



SCIENCE. 



251 



cause it is incapable of proof or disproof by 

 experiment. The reason is obvious. If we 

 suppose that all atomic weights are divisi- 

 ble by one tenth of the atomic weight of 

 hydrogen, then, in case the theory is er- 

 roneous, the average deviation of the actual 

 atomic weights from those required by the 

 theory is only the fortieth of the unit. The 

 man who supports a theory which has no 

 physical basis would assert that all such 

 ascertained deviations were due to errors 

 of experiment. Others would reply that 

 you cannot prove that a man is a good 

 marksman by crowding the targets so near 

 each other that not even his random shots 

 can miss them all. But his backers might 

 make so uncritical a claim. 



No, Prout's hypothesis, if subdivided far 

 enough, may be true for all which can be 

 proved with the balance ; but in such new 

 form it is of no use and of no interest, for 

 it cannot be proved so as to become a safe 

 basis for further inference. In its present 

 form there is no root of truth in it. 



So far I have argued that Prout's hypoth- 

 esis is not true as heretofore enunciated, 

 and that, if some further modification of it 

 is true, we cannot know it. This conclu- 

 sion has been sustained by the evidence of 

 the chemist's balance. A conclusion sup- 

 ported by a single kind of evidence may 

 command the confidence of one who has 

 been long familiar with the evidence and 

 who has become capable of weighing it. 

 But for others the concurrence of evidence 

 of different kinds rightly adds greatly to its 

 cogency. In this case there is such con- 

 current evidence. There is other proof that 

 the atoms of some well studied elements 

 are not additive structures. Let me briefly 

 describe the nature of this evidence. 



When certain elements are volatilized in 

 a colorless gas flame, or in the electric arc, 

 their molecules are made to vibrate, so as 

 to produce light. By the study of this light 

 we can in time learn much of the nature of 



the vibrating system. The observed facts 

 are gradually reducing to order, and one 

 result is very striking. In the case of three 

 closely similar elements before mentioned, 

 lithium, sodium and potassium, the com- 

 plexity of vibration is precisely similar in 

 all, and the numerical relations among the 

 component vibrations are precisely similar 

 in all. Therefore we are compelled to as- 

 sume that the complexity of structure is 

 the same in all, and that the relations of 

 the component parts, and of the forces act- 

 ing between them, are the same in all. To 

 illustrate the nature of the argument: the 

 complexity of vibration and the numerical 

 relations among the component vibrations 

 in the case of a large church bell are pre- 

 cisely similar to those in the case of a bell 

 only one third as large. Then, even with- 

 out the direct evidence of other senses, we 

 must presume that the two bells are similar 

 structures, having similar parts, similarly 

 related. We cannot believe that the larger 

 bell is made of a small bell loaded with 

 weights, nor of three small bells bound 

 closely together. The larger and the smaller 

 are of the same order. The larger is not 

 made of more parts than the smaller ; it is 

 made of more metal. So with the atoms of 

 these three elements ; the larger are not 

 made up by the addition of parts which 

 preserve their identity and remain undi- 

 vided. But all we know of chemical com- 

 bination relates to structures which are 

 made by the addition of parts which pre- 

 serve their identity and remain undivided. 

 Then Prout's hypothesis assumes an anal- 

 ogy which does not exist ; and deductions 

 from an imaginary analogy will themselves 

 differ from the truth, much as fairy tales 

 difier from history. 



There are still other sources of evidence 

 drawn from the specific heats of the 

 elements ; the evidence is of the same kind 

 and leads to the same conclusion, but I 

 simply allude to it. 



