152 PROCEEDINGS OP THE AMERICAN ACADEMY 



not fail to produce on the mind the impression that there is some influ- 

 ence which tends to bring about this result. 



It may be that the discrepancies are due to unknown constant errors, 

 which every experimentalist knows are greatly to be feared. Or it 

 may be that there is in nature a tendency to whole multiples ; and this 

 last view, if not compatible with our present conception of the atomic 

 theory, may hereafter appear as one of the phases of a broader 

 philosophy. 



The force of evidence which such a distribution of values as the 

 above table presents was brought home to the writer in his investiga- 

 tion of the atomic weight of antimony.* After eliminating various 

 causes of error, he was enabled to determine with great accuracy the 

 atomic weights of antimony, silver, and bromine, in one and the same 

 series of experiments ; and it appeared that this ratio was 



120.00:108.00:80.00, 



with a probable error of less than one in the last decimal place. Here 

 then is a ratio of whole numbers within the one hundredth of a single 

 unit. Since the ratio of the atomic weights of silver and oxygen have 

 been determined with great accuracy, we can extend the above propor- 

 tion to a fourth term, the atomic weight of oxygen, which appears also 

 as a whole number, perhaps with a somewhat larger probable error. 

 Still, we have not reached the unit of the system, and when we attempt 

 to extend the ratio to the atomic weight of hydrogen, we find that the 

 most probable value from all experiments hitherto made gives the ratio 

 not of 16 to 1, but of 16 to 1.0025. 



If now we wished to refer to the hydrogen unit the atomic weights 

 of antimony, silver, bromine, and oxygen whose ratios of whole num- 

 bers had been determined as above, it was only necessary to divide all 

 the terms of the above proportion by 1.0G31, when we obtain the series 

 of values given below the others, and all semblance to the hypothesis 

 of Prout disappears, although of course the second series of numbers 

 bear the same ratios to each other as the first : — 



The numbers in the lower of the two proportions appear as uncom- 

 mensurable as Stas maintained that they were, and the same is true 



* Additional Experiments on the Atomic Weight of Antimony, Am. Acad. 

 Proa, vol. xvii. p. 13, by Josiah Parsons Cooke. 



