4/13 



STABILITY 



engine. He need know nothing of the nature of the ultimate 

 physical linkages, but he would observe the fact. Then, still 

 keeping 4 velocity of flow of steam ' constant, he would try various 

 distances between the weights, and would observe the effect of 

 such changes on the speed of the engine; he would find them to 

 be without effect. He would thus have established that there is 

 an arrow from left to right but not from right to left in 



Speed of 

 engine 



Distance 

 between 

 weights 



This procedure could then be applied to the two variables 

 ' distance between weights ' and ' velocity of flow of steam ', 

 while the other variable ' speed of engine ' was kept constant. 

 And finally the relations between the third pair could be established. 



The method is clearly general. To find the immediate effects 

 in a system with variables A, B, C, D . . . take one pair, A and 

 B say; hold all other variables C, D . . . constant; note B's 

 behaviour when A starts at A x \ and also its behaviour when A 

 starts at A 2 . If these behaviours of B are the same, then there is 

 no immediate effect from A to B. But if the J5's behaviours are 

 unequal, and regularly depend on what value A starts from, 

 then there is an immediate effect, which we may symbolise by 

 A-+B. 



By interchanging A and B in the process we can then test for 

 B — > A. And by using other pairs in turn we can determine all 

 the immediate effects. The process consists purely of primary 

 operations, and therefore uses no borrowed knowledge (the process 

 is further considered in S. 12/3). We shall frequently use this 

 diagram of immediate effects. 



4/13. It should be noticed that this arrow, though it sometimes 

 corresponds to an actual material channel (a rod, a wire, a nerve 

 fibre, etc.), has fundamentally nothing to do with material con- 

 nexions but is a representation of a relation between the behaviours 

 at A and B. Strictly speaking, it refers to A and B only, and not 

 to anything between them. 



That it is the functional, behaviourial, relation between A and 

 B that is decisive (in deciding whether we may hypothesise a 

 channel of communication between them) was shown clearly on 



51 



