148 SUMMARY AND CONCLUSION. 
tion of some existing species or the setting apart of new groups that 
grow into new species unless reabsorbed by crossing or exterminated 
by competition. 
The whole process of bionomic evolution, whether progressive or 
retrogressive, whether increasingly ramified and divergent, or increas- 
ingly convergent through amalgamation, is a process by which the 
limitations of segregate breeding are either set up and established or 
cast down and obliterated. But, as we have already seen, on the side 
of amalgamation an impassable barrier is in time reached in the 
physiological and psychological incompatibilities of long-established 
types, while on the side of advancing segregation the possibilities are 
constantly increasing. The general result is that new isolations 
and incompatibilities are constantly arising, forming new races and 
species, which in time become so divergent that it is impossible for 
them to coalesce under any conditions. 
2. Unbalanced Propagation. 
If we wish to find a principle which, if continued from generation to 
generation, will steadily tend towards the transformation of type, it 
is unbalanced propagation continuously of the same sign. That is, 
if the result desired is increase of the character under consideration, 
the selection in successive generations must be of those individuals 
that possess the character in more than the average degree; and such 
selection may be said to be continuously of the plus form. If the in- 
dividuals selected in each generation depart from the type, but are so 
selected that those above the average are exactly sufficient to balance 
those from below the average, the average character of the mixing 
mass will be the same as the average of the original stock; and again, 
if the selection is plus in one generation and equally minus in the next 
generation, the result will be uncertain, even though long continued, 
for the effects of selection in one generation will be balanced by the 
effects of selection in the next generation, and we shall have one form 
of balanced selection. Witha definition of balanced propagation that 
includes balancing of both the kinds just mentioned, we may say with 
confidence that unbalanced propagation, if continuous, will produce 
transformation, and that balanced propagation of the type, if con- 
tinuous, will produce stability of type, and that balanced propagation 
of forms, some of which are considerably above the type and others 
of which are considerably below the type, will produce fluctuating 
variation. 
We may next ask, how is unbalanced propagation brought about? 
The answer is that, in both natural and artificial breeding, it may be 
