From J. D. Hooker   [after 26 March 1862?]¹

the old type as away from it— this is conceivable of course, but I maintain it is not so in nature— it is opposed to my idea of centrifugal variation.—² Now variation is not centrifugal because of any repulsive or repellent power, but simply because in fact it is a million chances to one that identity of form once lost will be returned to. Each character we estimate is a compound of lesser characters each too minute to estimate in the gross, but missed if with-drawn. A petal if further varying after it has once varied has a thousand characters of form color consistence nervation &c &c &c to chuse amongst (so to speak) besides those its grandparents bore. After all, experience is our best guide, & we do not find in the human race any reversion so strong as would lead us to confound a man with his ancestor, a Yankee with an Englishman— Whoever saw a Grandfathers portrait that would really pass for his Grandsons—& if we do not find reversion amongst individuals so close in kin & in time, how can we expect them in organisms that have reached the specific term of divergent development—

True you may have a 10^th removed lineal descendent (who supposing the name of Wedgwood to be lost to genealogists—except by a stray portrait) may so resemble the Wedgwood portrait that an Ethnologist would call him an anomalous Darwin representing the lost Wedgwood type.—but the chances are a hundred to one in favor of both Wedgwood & Darwin physiognomy being wiped out long before the 10^th. generation—³ How often do we consider 2 people strikingly alike till we see them together & then consider them wholly unlike; this is because one minute character common to both was alone carried away, in the mind, but which, though no doubt present, is not perceived when again sought for amongst the thousand other characters of the 2 faces.