QUANTITY OF VARIETY 7/25 



correspondence. This point of view would be more appropriate 

 if we were studying some very complex transducer, making fresh 

 experiments on it each day. If it contained great numbers of rather 

 inaccessible parts, there might be difficulty in bringing it each morn- 

 ing back to some standardised state ready for the next experiment. 

 The theorem says that if its input is taken, in the early morning, 

 through some standardised routine, then the longer the routine, the 

 more certain is it that the machine will be brought, ready for the 

 experimenter, to some standard state. The experimenter may not 

 be able to name the state, but he can be confident that it tends to 

 be reproducible. 



It should be noticed that mere equality of the set's parameter at 

 each step of the sequence is not sufficient; if the effect is to be more 

 than merely nominal (i.e. null) the parameters must undergo actual, 

 non-zero, change. 



The theorem is in no way dependent for its truth on the size of 

 the system. Very large systems are exactly as subject to it as small, 

 and may often be expected to show the effect more smoothly and 

 regularly (by the statistical effect of largeness). It may therefore 

 be usefully applicable to the brain and to the social and economic 

 system. 



Examples that may correspond to this process are very common. 

 Perhaps something of this sort occurs when it is found that a number 

 of boys of marked individuality, having all been through the same 

 school, develop ways that are more characteristic of the school they 

 attended than of their original individualities. The extent to which 

 this tendency to uniformity in behaviour is due to this property of 

 transducers must be left for further research. 



Some name is necessary by which this phenomenon can be referred 

 to. I shall call it the law of Experience. It can be described more 

 vividly by the statement that information put in by change at a 

 parameter tends to destroy and replace information about the 

 system's initial state. 



139 



