Weaver • The Imperfections of Science 307 



has no business to ask, "What is the numbers ground out by our black box 

 real nature of physical phenomena"; will depend not only on the numbers 

 or to ask, "Are there really precise de- we insert, but also on our dial settings, 

 terministic laws behind the statistical And then by changing these dial set- 

 data which I obscR'e"; or to ask, "How tings one can answer such a question 

 can light be both a wave motion and as "How will the result change if I 

 a beam of particles"; or to ask, "What vary one or more of the circumstances 

 sense does it make for a particle to of the experiment?" Under useful cir- 

 have electric charge but zero mass?" cumstances such as these, the theory 

 These are, to the one who accepts has brought the phenomenon under 

 the formal procedure, senseless assort- control. You know just what to do in 

 ments of words. For the formal pro- order to modify the result in the de- 

 cedure makes no pretense whatsoever sired way. 



of "explaining." The formal procedure, Finally, the black box, if it is a 



in fact, says "It is impossible to explain really good one, must be able to grind 



phenomena, and it is in fact senseless out numbers which will prove to corre- 



to try. All you can do— and this is a late properly with numbers which you 



triumph of great dimensions— is to deal will obtain in an experiment or observa- 



successfully with phenomena. tion not yet made. That is to say, the 



The equations, or more generally theory should be able to predict. 



the theory, are a sort of "black box." We can now state in more compact 



You can feed one set of numbers into summary what the modern scientist 



this black box, turn the crank, and out calls a good theory. It is a theory which 



comes a second set of numbers. If this is general and elegant, which puts us 



second set correlates properly with in control of the phenomena in ques- 



numbers which can be determined, fol- tion, and which can predict. But notice 



lowing given rules, from nature, then that I have not said one single word 



you have a successful theory. about explaining. The advocates of ab- 



This idea of not explaining, but of stract theories have to agree that the 



dealing successfully with phenomena scientist understands a phenomenon 



deserves a few further words. What, to when he can control and predict it, 



the scientists, constitutes a really satis- and that as a product of his creative 



factory sort of success for a theory? imagination he appreciates and admires 



The answer lies largely in the words the theory the more, the more general 



generality, elegance, control, and pre- and elegant it is. 



diction. If one single theory— one black It is essential to my general argu- 



box— is capable of grinding out results ment to point out at once that many 



that relate to a wide range and a large scientists enthusiastically disagree with 



apparent diversity of experience, then this position. 



the theory has the obvious practical ad- The general and popular reputation 



vantage of generality. If in addition the of science rests largely on its success at 



theory is stated in compact form, then control, and to a lesser degree upon 



it possesses the illusive but lovely trait its ability to predict. Unfortunately, 



which the scientist calls "elegance." only scientists themselves, and a few 



Suppose that the black box of our others who make a real effort, achieve 



theor}' has certain dials on one of its the knowledge that makes generality 



faces, and that we set these dials, be- important, and elegance lovely, 



fore inserting input data, to values When we restrict attention to mod- 



which are characteristic of the particu- erate-scale phenomena, involving, say, 



lar experiment in question. Then the objects above electron-microscopic size, 



