5 26 



APPENDIX 



distance : and it is then clear that his limit will be reduced from 24 inches 

 to 1 8 inches. So long as he is short of the 18 inch mark he will be 

 set back behind the 6 inch mark, and will thus have more than a 

 foot to do to reach 18 inches. He will thus never get there, though 

 he will continually approach it. The effect of ancestral inheritance 

 is to reduce the limits of deviation from type which we arrived at 

 in 10. 



29. Stability of type. We have thus arrived at a very definite notion 

 of the stability of a type when no cause such as natural selection interferes 

 with it. Mating alone would suffice to prescribe certain limits of 

 deviation from the ancestral type which could not be exceeded. 



30. These limits are narrow a simple analogy suggests as average 

 limit twice the average deviation of a single generation : and though a 

 more suitable but more complex calculation might give a rather different 

 result, the limits are not likely to be many times greater than the 

 deviations from one generation to the next. 



31. Similarly the tendency of descendants to resemble their ancestors 

 causes a reversion to the type in each generation, which narrows the 

 limits still further. 



32. We shall presently point out an important difference between 



the action of these two agencies, the 

 sexual and the ancestral ; but at present 

 we are concerned with their con- 

 currence. Together they restrict the 

 deviation within narrow limits and 

 render the type stable. 



3 3 . Paring down of irregularities. 

 It is perhaps worthy of note that the 

 two agencies combine to destroy, not 

 only deviations of the whole path OV 

 -jj from OP, but also irregular deviations 

 such as OSQR, which may cross the 

 type path. The portion, OSQ, which 



lies above OP, will tend to be replaced in the next generation by OsQ, still 



above but less above : at Q there is no 



tendency to change, and in the portion, 



QTR, the influence is to raise the path 



towards OP. Thus the path of the child 



tends to straighten out irregularities 



in those of the parents. Indeed it is . 



easily seen that irregularities of this 



kind have, owing to the sexual effect, 



far less chance of surviving than have 



deviations of the path as a whole, 



either above or below, for the paths of 



the parents may easily both deviate 



in one direction as a whole, but their 



irregularities would be unlikely to coincide, and would tend to annul one 



FIG. 12. 



FIG. 13. 



