444 



SCIENCE, 



[N. S. Vol. XI. No. 273. 



ratus of both these classes stands at the 

 command of the modern investigator in 

 natural science ; literally numberless are 

 the conditions under which matter is ever 

 anew compelled to yield up its secrets. 



Let us now look more closely at the work 

 of the chemist, that we may learn how far 

 the results attained by the methods at his 

 disposal may lay claim to exactitude. 



The task which we have proposed may 

 be treated in different ways. We could 

 from a general standpoint, discuss the 

 criteria by which the accuracy of an ex- 

 perimental method is to be judged. A 

 method will thus be valued as exact accord- 

 ing to the concordance of the results attained 

 by it on many repetitions or according to 

 the concordance of these results with those 

 attained by other methods. The fulfilment 

 of this last condition is especially of value, 

 and we shall therefore later recur to an 

 especially interesting case of this kind. 

 The value of a method moreover will be 

 further increased if it is fitted to broaden 

 the field of experimentation ; just as a piece 

 of apparatus is of increased value, when it 

 is adaptable for different experiments. 



I prefer, however, to pass by these more 

 general considerations and to discuss the 

 exactitude of experimental methods in their 

 bearing upon a few problems, whose funda- 

 mental importance is recognized. Of these 

 I select three. 



First we will consider the proof of the law 

 of indestructibility of matter, as an example of 

 the degree of exactness for which experi- 

 mental data can be obtained, where the de- 

 termination of atomic masses is concerned. 



A second problem will be the delicacy of 

 chemical reactions, under which it is to be 

 considered, how far the accuracy of our ap- 

 paratus allows us to detect the existence of 

 definite chemical substances. And if both 

 of these cases shall lead us to increased con- 

 fidence in our methods, and we come to 

 feel that the exactness of our experimental 



methods is very great, our third study on 

 the other hand will serve to shatter this 

 confidence. This study is that of the abso- 

 lutely pure substance, and brings us at once 

 to the point where the experimenter is 

 wholly convinced of the justice of the con- 

 clusion that the hands of man can never 

 bring into existence the unconditionally 

 perfect. 



Before we enter upon a consideration of 

 these three examples, however, it will be 

 well to notice the case which has already 

 been spoken of, which reveals, as hardly 

 any other could, the fact that those views, 

 which have been attained by very different 

 ways, and which we form from facts and 

 phenomena upon the basis of sentient percep- 

 tion, may coincide to a remai'kable degree ; 

 this is particularly the case if we take into 

 account the possibility that our observa- 

 tions may not correspond to the realitj' 

 quite as closely as the investigator com- 

 monly claims. In determining the den- 

 sity of the vapor of mercury, if one pro- 

 ceeds on the supposition that all gaseous 

 bodies under the same conditions of tem- 

 perature and pressure contain the same 

 number of molecules in the same space, he 

 comes to the conclusion that the vapor of 

 mercury contains in the molecule only one- 

 half as many atoms as that of hydrogen. 

 If, as appears from the volume relation in 

 the analj^sis and sj'nthesis of hydrochloric 

 acid, this hydrogen molecule contains two 

 atoms, it then follows that in the vapor of 

 mercury the smallest particles which exist 

 in a free state are single atoms. N"ow it is 

 possible to reach this result by a wholly 

 different method. It is a' familiar fact that 

 the velocity of sound may be used to deter- 

 mine the relation between specific heats of 

 gases at constant volume and constant pres- 

 sure. For this determination the knowl- 

 edge of volume and weight is necessary, 

 and besides this a measure of length, that 

 is the length of the sound waves. The ex- 



