August 28, 1896.] 



SCIENCE. 



247 



manipulative skill. Our experiments de- 

 termine most atomic weights, not with ref- 

 erence to hydrogen, but with reference to 

 oxygen. Experiment, for instance, does 

 not determine directly that the atomic 

 weight of lithium is seven times that of hy- 

 drogen, but that it is seven sixteenths that 

 of oxygen. If the atomic weight of oxygen 

 is uncertain, the atomic weights of the 

 other seven elements, with reference to hy- 

 drogen, are all uncertain in the same pro- 

 portion, although with reference to oxygen 

 they are now determined with very small 

 uncertainty. Accordingly the third condi- 

 tion imposed on us in attempting to learn 

 the truth about Front's hypothesis is that 

 the atomic weight of oxygen must be well 

 determined. 



It may be remarked that it would be a 

 grekt gain, as all chemists will see, if sev- 

 eral other atomic weights could be de- 

 termined by direct comparison with hydro- 

 gen, provided the precision attainable was 

 of the degree which I have called admi- 

 rable, or even excellent. ISTow, methods 

 have been devised by which the atomic 

 weights of lithium, sodium and potassium, 

 as well as of several other metals, could be 

 referred directly to hydrogen, by experi- 

 ments which present no great difficulty 

 and which are capable of the required pre- 

 cision. Further, a method has been de- 

 vised by which the atomic weight of chlo- 

 rine can be determined with direct ref- 

 erence to hydrogen, by experiments capable 

 of the required degree of precision, but in- 

 volving considerable difficulty in manipula- 

 tion. But, until some such methods shall 

 have been employed by some one, we must 

 be content with the inferences which can be 

 drawn from data of the kind now available, 

 which depend on our knowledge of the 

 atomic weight of oxygen as the corner stone 

 of the system. 



Our knowledge of the atomic weight of 

 oxygen ten years ago depended largely on 



the experiments of Dumas. His results 

 differed from the whole number 16.00 by 

 one four-hundredth part; he himself judged 

 that the uncertainty remaining might be 

 one two-hundreth part. If we accept this 

 estimate of uncertainty, we may say that 

 he proved that the atomic weight of oxygen 

 is included between the limits 15.88 and 

 16.04. No one could assert that the true 

 numlber is, or that it is not, the whole num- 

 ber 16.00. A proportionate uncertainty, 

 therefore, existed in the other seven atomic 

 weights just mentioned. Accordingly, ten 

 years ago. we could not well discuss the 

 question whether these atomic weights were 

 divisible, without remainder, by the atomic 

 weight of hydrogen. 



The atomic weight of oxygen is, accord- 

 ingly, doubly important for our purpose. 

 The atomic weight is a small one, well 

 adapted to aid in the solution; and, further, 

 many other atomic weights, also well 

 adapted to aid in the solution, depend on a 

 prior knowledge of this constant. It is for 

 this twofold reason that the work done 

 since our last meeting at Buffalo is impor- 

 tant and interesting. The members of this 

 Association have not failed to take upon 

 themselves a fair proportion of the consid- 

 erable labor involved. 



Since that time not less than ten or 

 eleven independent determinations of the 

 atomic weight of oxygen have been success- 

 fully concluded. 



Cooke and Richards were the first to 

 complete and publish their result ; they 

 used a new and ingenious process. Keiser 

 was next ; he employed a method for 

 weighing hydrogen which he had independ- 

 ently invented (though it had been previ- 

 ously invented elsewhere) which is the 

 best yet used. In both these series of ex- 

 periments the hydrogen was combined 

 with oxygen by manipulation something 

 like that of Dumas ; but the improvement 

 which permitted the direct weighing of the 



