Arber — Atavism and Law Irreversibility. 31 



they appear to correspond to the outer perianth members 

 of a normal flower, the three inner being absent. In 

 other words, the abnormal form of perianth may be inter- 

 preted as due to the chorisis or dedoublement of the 

 normal outer whorl. According to Errera, the six sta- 

 mens are to be interpreted as incuding the three members 

 of the outer whorl, and, in addition, the three members of 

 that inner whorl which in the Iridaceae is normally sup- 

 pressed. But it seems to me more reasonable to suppose 

 that these three extra stamens are not the avatars of the 

 defunct inner whorl, but have originated through the 

 doubling of the members of the existing outer whorl ; this 

 suggestion has the advantage of postulating the same 

 type of variation for both perianth and stamens. The 

 important part which such secondary dedoublement may 

 have played in the phytogeny of Angiosperms, has 

 received full recognition from certain botanists who have 

 tried to elucidate the history of the flower. Wernham 11 

 for instance holds that when indefinite stamen numbers 

 occur within the Archichlamydese, but outside the more 

 primitive families, this condition may perhaps be inter- 

 preted as due to secondary branching; the Opuntiales 

 illustrate the extremest expression of this tendency to 

 chorisis. In this connection it is possibly suggestive 

 that — as Professor Punnett 12 has pointed out — in more 

 than one of the rare cases in which the evolution of 

 domestic races appears to have come about by the addi- 

 tion rather than the loss of factors, the interpolated fac- 

 tor is of such a nature as to cause reduplication. There 

 may be some relation between such reduplicating factors 

 and the chorisis to which we have just alluded. 



The view that such forms as a Linaria with a five- 

 chambered ovary, 13 a five-stamened 8temodia 14: or a four- 

 toed guinea-pig, can be classed as ' reversions, ' is also 

 open to criticism on more general grounds. Such abnor- 

 malities are only claimed as atavistic if they happen to 

 correspond to those forms which on morphological or 

 palaeontological evidence we suppose to be ancestral. 

 But other similar cases, such as the various well known 

 examples of polydactylism in man, are passed over in 



"Wernham, H. F., Floral Evolution; with particular reference to the 

 sympetalous Dicotyledons, New Phyt., vol. 10, 1911 and vol. 11, 1912; see 

 especially p. Ill, vol. 10. 



12 Punnett, R. C, Mendelism, 4th Ed., p. 80, 1912. 



13 Crepin, F., Recueil de f aits teratologiques, Bull. Soc. Roy. Bot. de 

 Belgique, 4, pp. 276-8, 1 pi., 1865. 



14 Errera, L., Pentstemon gentianoides et Pentstemon Hartwegi Bull. 

 Soc. Roy. Bot. de Belgique, 17, pp. 182-248, 1878. 



