17/9 ANCILLARY REGULATIONS 



homologous and equivalent; but F comes into the whole in a 

 different way. To see how, suppose that it is most desirable (for 

 some major essential variable S) that success, on some lesser 

 essential variable E, be achieved in fewer than a hundred trials 

 (i.e. F is to be less than 100). E, making trials, will cause change 

 after change to occur on its corresponding step-mechanisms; at 

 the same time F (increasing exhaustion perhaps) is steadily 

 mounting to its limit. What is to happen if F passes its limit 

 of 100 ? If & is such that the organism dies, nothing remains 

 to be said ; but if & is not totally essential, the organism is in the 

 condition of having made many trials in some way that has failed 

 to bring success quickly (the situation discussed in Chapter 11). 

 What is to be done ? By the method of ultrastability, F's passing 

 beyond its limit must induce changes, but clearly these changes 

 should not be simply in the same step-mechanisms that E has 

 been working on, or the action by F is no different from a hundred- 

 and-first trial by E. For F to have an appropriately effective 

 action, its passage beyond the limit must induce changes in those 

 conditions that have continued unchanged throughout E's hundred 

 trials. jE's trials must not consist of further samples from the 

 same set, but must change to samples from a new set. Thus if 

 the organism is a cat in a box, and if it has made 100 trials of 

 manipulating the levers and objects without success, now is the 

 time for it to make trials from a new statistical population — to 

 change perhaps to various forms of mewing and calling. 



Thus the improvement of the E's speed of adaptation by the 

 selection of an appropriate value for the step-mechanisms under 

 F's control is not the same as making a selection on the step- 

 mechanisms that E itself should make. Providing an examinee 

 with pen, paper, and a quiet room may be called ' helping the 

 examinee ', but it is clearly quite distinct from the ' help ' that 

 would show him how to answer the individual question. F 

 4 helps ' the E's only in the first sense, not the second. 



Thus the conclusion of S. 17/7 — that if an organism is to adapt 

 with reasonable speed, certain parameters will have to be brought 

 within certain limits — does not involve a circular appeal, for the 

 two selections are working at different levels, i.e. on different sets. 



17/9. It is not for a moment suggested that all naturally occurring 

 organisms have essential variables that divide neatly into distinct 



225 



