8/10 DESIGN FOR A BRAIN 



of the first suitable new value, the homeostat would act to stabilise 

 instead of to overthrow ; it would return the plane to the hori- 

 zontal ; and it would then be ordinarily self-correcting for dis- 

 turbances. 



There is therefore some justification for the name ' ultrastable ' ; 

 for if the main variables are assembled so as to make their field 

 unstable, the ultrastable system will change this field till it is 

 stable. The degree of stability shown is therefore of an order 

 higher than that of the system with a single field. 



Another difference can be seen by considering the number of 

 factors which need adjustment or specification in order to achieve 

 stability. Less adjustment is needed if the system is ultrastable. 

 Thus an automatic pilot must be joined to the ailerons with care, 

 but an ultrastable pilot could safely be joined to the ailerons at 

 random. Again, a linear system of n variables, to be made stable, 

 needs the simultaneous adjustment of at least n parameters 

 (S. 20/11, Ex. 3). If n is, say, a thousand, then at least a thou- 

 sand parameters must be correctly adjusted if stability is to be 

 achieved. But an ultrastable system with a thousand main 

 variables needs, to achieve stability, the specification of about 

 six factors ; for this is approximately the number of independent 

 items in the specification of the system (S. 9/9). A large system, 

 then, can be made stable with much less detailed specification 

 if it is made ultrastable. 



8/10. In S. 6/2 it was shown that every dynamic system is 

 acted on by an indefinitely large number of parameters, many of 

 which are taken for granted, for they are always given well- 

 understood 4 obvious ' values. Thus, in mechanical systems it 

 is taken for granted, unless specially mentioned, that the bodies 

 carry a zero electrostatic charge ; in physiological experiments, 

 that the tissues, unless specially mentioned, contain no unusual 

 drug ; in biological experiments, that the animal, unless specially 

 mentioned, is in good health. All these parameters, however, 

 are effective in that, had their values been different, the variables 

 would not have followed the same line of behaviour. Clearly 

 the field of an absolute system depends not only on those para- 

 meters which have been fixed individually and specifically, but 

 on all the great number which have been fixed incidentally. 

 Now the ultrastable system proceeds to a terminal field which 



100 



