146 EMPIRICAL TOOLS AND EMPIRICISM 



that X and P melt at the same temperature, the chemist prepares a 

 mixture of X and P and determines its melting point. If this is un- 

 changed he concludes that X and P are identical. To determine the 

 melting points a thermometer is ordinarily used, but this is super- 

 fluous in principle. The chemist need only prepare three similar tiibes 

 containing respectively X, P, and a mixture of X and P. If he finds 

 that they all melt at very nearly the same time in a slowly-heated and 

 well-stirred "melting point bath," he then feels well justified in iden- 

 tifying X as a specimen of P. With or without use of the thermometer, 

 the denotation of the concept of P is firmly established by the one 

 reference specimen of P and the technique of mixed melting point. 



Techniques as standards. Even in the absence of a reference 

 "thing," a technique or group of techniques can constitute an ade- 

 quate external standard— as when Lavoisier established the denota- 

 tion of "element" as follows : 



The last term at which analysis arrives, all the substances which we 

 have as yet been unable to decompose by any means, are elements 

 as far as we are concerned. 



What is to pass as an element is here fixed in terms of the techniques 

 for analysis available at the time. What is taken to be an element may 

 thus become a function of time, but at any one time we well know 

 what does pass as an element. Invented in antiquity, the concept 

 "element" developed its full usefulness only after Lavoisier thus pro- 

 vided a firm link between the abstract concept and the world of ex- 

 perience. Often deceptively simple in retrospect, the forging of such 

 a link can, as here, represent a major step forward. A similar if less 

 spectacular advance was wrought by Proust, who first suggested that 

 techniques of analysis can serve also to establish the denotation of 

 the concept "compound": a material is a compound if, wherever and 

 however it is obtained, analysis finds it to contain its component ele- 

 ments in an invariant proportion. 



Like all other denotations, those established in terms of techniques 

 fall short of absolute clarity. Ordinarily the modern chemist still 

 chooses to treat as an "element" a material that analysis, broadly 

 conceived, finds resolvable into distinct (isotopic) components; and 

 as a "compound," a material slightly variable in composition for this 

 and other reasons. These usages present no real problems, but 

 borderline cases severely strain the denotative precision of some 



