268 Nature of Segregation [ch. 



serious extension of the powers of specific substances as 

 usually imagined. 



But disregarding for the present these more obscure 

 cases in which meristic phenomena play the chief part, we 

 >may draw from Mendelian observations the conclusion that 

 in at least a large group of cases the heredity of characters 

 consists in the transmission of the power to produce some- 

 thing with properties resembling those of ferments. It is 

 scarcely necessary to emphasise the fact that the ferment 

 itself must not be declared to be the factor or thing trans- 

 mitted, but rather, the power to produce that ferment, or 

 ferment-like body. 



Nature of Segregation, 



Next we have to recognize that this antecedent power 

 must be of such a nature that in the cell-divisions of gameto- 

 genesis it can be treated as a unit, being included in one 

 daughter-cell and excluded from, the other at some definite 

 cell-division. As we have no knowledge as to the actual 

 nature of the. factof^and only a conjecture as to whether it 

 .Ts^ material substance, or a phenomenon of arrangement — 

 we are not in a position to hazard so much as a guess re- 

 specting the physical process of segregation. We may be 

 fairly confident that segregation is ^6>/^ a process-of~chefm€aL_ 

 separation. I ts features point rather to mechanical analogies. 

 If some mental picture be demanded, I would for the purpose 

 of illustration suggest a comparison with the separation of 

 a fairly heavy precipitate from a filtrate by decanting. The 

 precipitate is to represent the factor, and the filtrate the 

 recessive product of division, lacking the factor. The 

 analogy is probably quite erroneous, and small purpose 

 would be served by developing it further. 



Whatever that process may be, it must be one which is 

 applicable to an extraordinary variety of factors. No con- 

 clusion to which genetic research has led is so surprising 

 as the fact that the same system of transmission should be 

 foUow^ed by characters which, by whatever test they are 

 judged, must be supposed to be most diverse in physio- 

 logical causation. Even if it were fairly easy to conceive 

 of all the dominant characters of animals and plants as due 



