It should clearly be understood what is implied in the construction of such a hypothesis. Certain processes are observed to take place when a somatic cell divides: these processes we have regarded as having for their object the exact division of all the parts of the cell into two halves. This process of somatic cell division is modified when a germ cell divides prior to maturation (the process fitting it to become fertilised). Then the cell nucleus divides into four daughter-nuclei. One of these remains in the cell substance which is to become the ovum, and the other three, each of them invested in a minimal quantity of cytoplasm, are eliminated as the “polar bodies.” Also the number of chromosomes in the mother-cell becomes halved, so that the mature ovum, or spermatozoon, possesses only one-half of the number of chromosomes which are present in the ordinary somatic cell. Now let the reader puzzle out for himself what may be meant by this behaviour of the germ cells, and he will certainly see that several interpretations are possible. But suppose that the chromatin consists of an incredibly large number of bodies differing in chemical structure from each other, and occupying definite positions with regard to each other; and suppose that there is a mechanism of unimaginable complexity in the cell capable of rejecting some of these chemically individualised parts, and of “assembling” or arranging the others in much the same way as an engineer assembles the parts of a dynamo when he completes the machine. _Then_ we may regard the hypothetical discrete bodies which form the hypothetical nuclear architecture as the material carriers of Mendelian characters. It is strange that the correspondence of such a logically constructed mechanism with the effects which it would produce if it existed should be regarded as a proof that it does exist, yet biological speculation has actually made use of such an argument. “It seems exceedingly unlikely that a mechanism so exactly adapted to bring it” (the separation from each other of the Mendelian material “factors” of inheritance) “about should be found in every developing germ cell if it had no connection with the segregation of characters that is observed in experimental breeding.” Put quite plainly this argument is as follows: there is a certain segregation to be seen in experimental breeding, and certain processes may be observed to occur in the developing germ cell. Add to these processes many others logically conceivable, and add to the observed material structure of the cell another structure also logically conceivable. Then the assumed mechanism and structure is “exactly adapted” to produce the effects which are to be explained. Therefore the mechanism and structure do actually exist!