THE CAUSE OF SPONTANEOUS VARIATIONS 97 



reversion in any species as regards any particular character 

 depends mainly on two factors, (a) reversion towards the 

 ancestry caused by defective recapitulation, and (6) reversion 

 towards the specific mean caused by regression to the 

 parental mean. The former is an instance of normal re- 

 gressive variation ; the latter is a necessary consequence of 

 bi-parental reproduction ; the extent of the former depends 

 usually on the speed of the antecedent evolution ; the extent 

 of the latter on the amount of the difference between the 

 parents. 



162. Darwin showed that, in many instances, in-breeding 

 is injurious and cross-breeding beneficial, the former causing 

 a lack of health and vigour, the latter an increase of it. 1 

 The elaborate devices evolved by many plants to secure 

 cross-fertilization lend strong support to his contention. If 

 the views set forth in the present work are correct, it follows 

 that the benefits of cross-breeding are due to the rapid 

 elimination of useless and injurious characters, which often 

 accumulate as a result of too close in-breeding. These 

 characters may not be tangible or visible, but may be mere 

 tendencies and predispositions, which act as checks to healthy 

 development. The worst effects of in-breeding are observed 

 under artificial selection, not only because the in-breeding is 

 generally closer than under natural conditions, but also 

 because man is unable to detect and eliminate many in- 

 jurious variations which nature would infallibly eliminate. 

 Occasionally the increase of vigour which follows crossing 

 results in an improvement in some particular character of 

 such a kind that it is apt to be mistaken for a progressive 

 variation. For instance, Darwin mentions instances of the 

 fruit of melons and apples as improved by crossing varieties. 2 

 But the massive nature of the evidence pointing to the 

 regressive action of bi-parental reproduction renders it more 

 than probable that the improvement is due rather to the 

 improved health of the plants, combined possibly with ex- 

 clusive inheritance, or with progressive variations in characters 

 in which the crossed species agree. Certain domesticated 



1 Animals and Plants, chap, xviii. 



2 Ibid., vol. ii, p. 110. Many years ago I grew several varieties 

 of melons in a small garden in New Zealand. There was of course 

 extensive cross-fertilization through the agency of insects. The fruit 

 from the hybrid plants was always quite worthless, having apparently 

 lost every quality which renders a melon agreeable. Every year it was 

 necessary to procure fresh seed from gardens in which one variety only 

 was grown. When seed was required it was the practice in the country 

 to grow the varieties separately. 



H 



