August 28, 1896.] 



SCIENCE. 



243 



result of the discussion between Bertliollet 

 and Proust, a discussion well worth recall- 

 ing for the dignified courtesy and simple 

 love for truth shown by both the disputants. 

 A second of these laws of Dalton is the law 

 of equivalent proportions : if two elements, 

 which combine with each other, combine 

 also with a third, then the ratio in which 

 they combine with each other (or a simple 

 multiple of it) is also the ratio of the 

 quantities of those which combine with the 

 same quantity of the third. That this was 

 true, at least in some cases, was known be- 

 fore Dalton. The third law is the law of 

 multiple proportions : if two bodies com- 

 bine in more than one ratio, those ratios are 

 simple multiples of each other. This truth 

 was discovered by Dalton. 



These three laws are statements of facts. 

 Careful and multiplied experiments have 

 convinced us that, if these statements are 

 not rigorously exact, their deviation from 

 accuracy is less than the accidental errors 

 of the best experiments used to test them. 



Perhaps it is worth while to delay for a 

 moment, in order to state to what degree of 

 precision such experiments have been 

 brought. The degree of precision with 

 which any supposed law can be verified de- 

 pends on the skill of the investigator, on the 

 instrumental equipment available, and on 

 the conditions of the problem. Often the 

 conditions of the problem impose very 

 stringent limitations on the precision of 

 our experiments. For instance, the truth 

 known as Ohm's law has been verified, in 

 the case of metallic conductors, to one part 

 in a million millions; but in the case of 

 liquid conductors, the conditions are such 

 that the precision attainable so far has been 

 only a millionth as much. Huyghens' law, 

 relating to double refraction, has been veri- 

 fied to one part in half a million, and there 

 seems to be no possibility of attaining any 

 considerable increase in the precision of tlie 

 observations. These are examples of the 



very highest degree of precision which has 

 been secured in the verification of supposed 

 la^ys of nature. 



The precision which can be attained in 

 chemical analysis, even of the most elaborate 

 kind, is much less than in the cases just 

 mentioned. The determination of atomic 

 weights is the chemical process in which 

 the highest degree of precision is demanded. 

 If we denote the precision of such deter- 

 mination by the words ' good,' ' excellent,' 

 ' admirable,' ' consummate,' then we may 

 fairly say that in a good series of determina- 

 tions the average difference from the mean 

 of all will be less than one thousandth part 

 of the ratio sought; in an excellent series, less 

 than one three-thousandth part; in an ad- 

 mirable series, less than one ten-thousandth 

 part; and in a consummate series, less than 

 one fifty-thousandth part. 



Now the work of Stas was all admirable 

 in precision, and much of it was consum- 

 mate, and he made experiments expressly 

 intended to verify the law of definite pro- 

 portions. The average error in this series 

 of experiments was not more than one part 

 in thirty thousand; and his result was, that, 

 if the composition of the compounds exam- 

 ined is not rigorously constant, the varia- 

 tions are too small to be detected. The law 

 of equivalent proportions was verified with 

 the same degree of precision ; the accuracy 

 of the law of multiple proportions has been 

 thought to be deducible from the truth of 

 the two other laws. 



To some such degree of precision, then, 

 Dalton's laws are the expression of facts. 

 With these facts for a guide, and with no 

 theory founded on the facts and explaining 

 the facts, all chemical computations could 

 be made, and chemical formulae could be 

 established. And, if a theory should be de- 

 vised, and accepted, and finally overthrown, 

 these facts would remain, unchanged for 

 our perpetual guidance. Some of Dalton's 

 contemporaries accepted the facts as a suf- 



