BROIVN-SEQUARUS EXPERIMENTS 237 



different from experiments on decapitated earthworms, curtailed 

 lizards, crabs with lost limbs, and the like, for there the investi- 

 gator is in touch with injuries which frequently occur in natural 

 conditions. 



The case is certainly a difficult one, but from what we have 

 said it must be evident that it cannot be cited without quaUfica- 

 tion in support of the thesis that somatic modifications are 

 transmissible. It is illegitimate to conclude, as Debierre does 

 (1897, p. 4) : " II est done incontestable que des caracteres 

 acquis artificiellement pendant I'age adulte de 1' animal ou 

 acquis naturellement pendant la vie embryonnaire peuvent etre 

 transmis par I'heredite." 



Our general conclusion is that the results of Brown-Sequard's 

 experiments do not strengthen the affirmative position ; and 

 that their probable interpretation is that the artificially induced 

 epilepsy liberated a toxin which affected the germ-cells in some 

 cases, the germ-ceUs and the foetus in other cases. 



§ 12. Negative Evidence in favour of the Affirmative Answer 



In support of the affirmative answer Herbert Spencer ad- 

 duced what he called negative evidence — namely, those ' ' cases in 

 which traits otherwise inexplicable are explained if the structural 

 effects of use and disuse are transmitted." 



(i) First he referred to the co-adaptation of co-operative parts. 

 With the enormous antlers of a stag there is associated a large num- 

 ber of co-adaptations of different parts of the body, and similarly 

 with the giraffe's long neck and the kangaroo's power of leaping. 

 Spencer argued that the co-adaptation of numerous parts cannot 

 have been effected by natural selection ; but that it might be effected 

 by the hereditary accumulation of the results of use. 



It must be admitted that co-adaptations are difficult to account 

 for in terms of the ordinary selection formula, but it is also dif&cult 

 to accept the use-inheritance interpretation. We do not really 

 know to what extent deep-seated co-adjustment can be effected by 



