142 EMPIRICAL TOOLS AND EMPIRICISM 



denotations, we may think to do so with the aid of more elaborate 

 "self-sufficient" instruments. Imagine a device in which insertion of 

 a sample at one end is followed by automatic production at the other 

 of a pointer reading giving "the result." The operating procedure is 

 then reduced to plugging in the instrument and inserting the sample; 

 the need for corrections is eliminated because the instrument itself 

 provides stabilization of the conditions of measurement, or contains 

 compensatory mechanisms for making the corrections automatically; 

 and the result given as a pointer reading seems to be the non plus 

 ultra of denotative clarity. Have we then at last arrived at a fully 

 categorical denotation? 



Alas, we have not. We now require a protocol for detecting and re- 

 pairing malfunction of the "self-sufficient" instrument. Far from elim- 

 inating all auxiliary protocols, we have simply substituted a new one 

 for the old— and the new one is ordinarily much the more difficult. 

 This protocol cannot safely be left to the "good judgment" of the in- 

 vestigator. Through an elaborate instruction manual, the manufac- 

 turer of a complex instrument supplies an operating protocol as ex- 

 plicit as he can make it and, beyond this, he must make available also 

 the services of the specialists required to rectify malfunctions of the 

 instrument. In the long term, then, such an instrument may make 

 demands on human judgment not less but greater than those imposed 

 in use of simpler devices. 



Through instruments we may first gain access to realms of ex- 

 perience otherwise outside our ken. The microscope "amplifies," the 

 NMR spectrometer "com^erts," hypothetical signals to which our 

 senses are dead into signals to which they are very much dlive. In 

 addition, the automatic pointer-reading instrument may offer great 

 advantages in speed, in sensitivity, and in simplicity of routine op- 

 eration—which can often be entrusted to unskilled technicians. These 

 are immense gains, justifying every effort to achieve them. But, as 

 Bernard well recognized long ago, the introduction of elaborate in- 

 struments where they are not needed is positively disadvantageous. 



. . . we need to learn that the more complicated the instrument, the 

 more sources of error does it create. Experimenters do not grow great 

 by the number and complexity of instruments. Quite the contrary. 

 The great experimenters, Berzelius and Spallanzani [and Bernard, 

 Pasteur, and Rutherford], made great discoveries by means of simple 

 instruments. 



