The Domain of Physical Science 201 



physical experiment involves readings of a scale or an equivalent 

 estimate made in cruder fashion. (Of course, the proof of this 

 statement requires a survey of the operations of physics far more 

 extended than we can undertake here ; and it requires a strict 

 definition of the type of datum included under the general heading 

 pointer-reading ; the student of relativity will be familiar with the 

 argument that every observable result reduces to a determination of 

 the intersections of world-lines.) It is natural, therefore, that the 

 branch of knowledge created on this basis should be wholly covered 

 by exact science. But in most subjects exact science goes a little 

 way and then stops, not because of the limitations of our ignorance, 

 but because we are dealing with something which includes both 

 metrical and non-metrical aspects. The theory of music starts as 

 though we were about to build up a science every whit as exact as 

 physics ; the integral relations of the notes of the scales, the distinc- 

 tion of concord and discord, the time relations of crotchets and 

 quavers, all belong to a metrical scheme. But it is unimaginable 

 that any system of measurements could be correlated to tune in such 

 a way that knowledge of these measurements would be accepted 

 as equivalent to knowledge of tune. That harmony which is 

 metrical and melody which is non-metrical both play a part 

 in determining the pleasing effect of music, suggests that there 

 is no strong opposition between measurable and non-measurable 

 agencies ; measurableness is a specialisation which is relevant when 

 we are studying certain aspects and irrelevant in others. 



These considerations also indicate the limits to the method of 

 microscopical analysis so universally employed in physics, except 

 possibly in some of the most recent developments. It is not a 

 microscopic analysis of the entities of the external world, but of the 

 pointer-readings accepted as equivalent to knowledge of these 

 entities. We can see that the equivalence of this substituted 

 knowledge becomes more and more remote and formal the farther 

 the dissection is carried. That is no drawback to its use in physics ; 

 the physicist is concerned only in working out the exact scheme of 

 interconnection of the pointer-readings and is not professionally 

 interested in the entities which these have replaced. When he has 

 arrived at a theory such that all the pointer-readings work out 

 correctly, he has reached the extreme limit of his task. 



Perhaps this breakdown of microscopic analysis may be made 

 clearer by an illustration. The operator at a telephone exchange 



