THE EVOLUTION OF SCIENTIFIC THEORIES 293 



mean that the road map is not, of its kind, an unexceptionable map of 

 the region. Providing that it is not thought of as having irrelevant 

 pretensions, there is nothing wrong with it: indeed, for some applica- 

 tions one will be able to discover the things one wants to know, 

 e.g., distances by car, more easily from the road map than from the 

 physical one. 



The topographic map is useful in a wider range of applications; and 

 the road map is derivable from it, as it is not from the road map. 

 The road map is thus subordinate to the topographic map. But no 

 greater degree of illusoriness then attaches to the road map, which 

 eflFectively discharges precisely those functions it was designed for. 



Let me suggest a diflFerent analogue. Flying at a considerable alti- 

 tude, I observe some "objects" in a field below. Guided by earlier 

 experiences, my companion the truck-driver declares that the objects 

 are rectangular blocks, and my companion the farmer asserts that 

 they are aggregates of strands of hay. Loading his truck, the truck- 

 driver does indeed treat such objects as rectangular blocks; feeding 

 his cattle, the farmer treats them as hay. Perhaps the concept of 

 rectangular block is subordinate to the concept of hay: knowing 

 something of the properties of hay, I can infer the possibility of form- 

 ing and handling it in bales; knowing only the presence of rectangu- 

 lar blocks, I cannot infer that I am in the presence of cattle feed. 

 But it does not then follow that there is anything deceptive about 

 the concept of rectangular blocks. 



I find only one non-fundamental distraction that makes the situa- 

 tion of Eddington's tables apparently diflFerent from the proposed 

 analogues. The analogues suggest paired views of which one is 

 directly subordinate to the other. Eddington's tables are the two 

 extremes of a whole series of theoretic views and, as Conant observes, 

 the omission of all reference to the intermediate members is what 

 makes the extremes seem so irreconcilably far apart. Given the view 

 of electrons and nucleons, I cannot pass at once to the everyday 

 table; but I can understand how, in certain contexts, atoms of various 

 kinds may be formed. Given such atoms, I can similarly understand 

 the possibility of forming molecules of cellulose, lignin, etc. And so 

 on from molecules to fibrils to fibers to wood to, at the very last, the 

 familiar table of everyday experience. 



All these views are, directly or indirectly, subordinate to the view 

 of electrons and nucleons— which, in principle, encompasses all the 



