138 PAUL WEISS 



less manifestly, from the beginnings of ontogeny. But experimental 

 embryology has thoroughly discredited this concept by proving that 

 mature cells which differ profoundly in their physico-chemical consti- 

 tution may be derived from cells that are demonstrably equivalent and, 

 to all practical purposes, identical. When and by what means have their 

 developmental courses then become divergent? This is the crucial ques- 

 tion of cytodifferentiation. Fortunately, though analytical techniques 

 fail us, we can answer it by a test that might be called "behavioral." 

 This is where we must draw on logics rather than gadgets. 



Suppose we are to compare two closed material systems, whether 

 they be molecules or cells or organisms, about whose character and 

 composition we know nothing. How are we to tell whether they are 

 the same or different? The only way of telling is by watching their 

 behavior. If both behave identically under identical conditions, we de- 

 cide that they are alike. However, if both, again under identical con- 

 ditions, behave differently, then we must conclude that they have been 

 intrinsically different from the very beginning of the test. 



Against the cogency of this postulate even the most persistent fail- 

 ures to detect more tangible signs of difference would prevail nothing. 

 Two seeds giving rise to different plants in the same soil, or two eggs 

 giving rise to different animals in the same pond, would have to be 

 acknowledged as intrinsically different even if they were otherwise 

 indistinguishable. Similarly, if two cells under identical conditions 

 behave and respond differently in any respect (e.g. movement, growth 

 rate, chemical activity, sensitivity to radiation or drugs, etc.), then they 

 are different, which implies that if they have descended from a com- 

 mon source of identical cells, they must have become different, i.e. have 

 gone through a process of differentiation, no matter whether they show 

 it by morphological signs or not. The behavioral test is as compelling 

 as any morphological test, and far more pertinent; for morphological 

 criteria are but residues and convenient indicators of prior activities. 



Early differences of this type, which are not immediately recogniza- 

 ble and can only be deduced from later formative activities, are com- 

 monly described as differences of "potency." If this were to imply 

 that they are virtual rather than real (in the sense of Driesch), then 

 it would mean reinstating the old "vital spirits" under another name. 

 Actually there can be no doubt that we are dealing with differences of 

 equipment and conditions which are real, yet too subtle to be detectable 

 by present techniques. 



In conclusion, we propose to define "differentiation" as the unidirec- 



