30 Arber — Atavism and Laiv Irreversibility. 



primaeval flower was of the eu-anthostrobilus type with 

 a perianth. It is not necessary to labour this point, 

 since Errera himself, at a later date, became disposed to 

 regard the Ranales and Alismacese, rather than the Ape- 

 talae, as primitive forms. And even if Errera 's original 

 view be accepted, this case, though it might be then inter- 

 preted as an exception to the 'Law of Irreversibility, ' 

 cannot be claimed as affecting the validity of the 'Law of 

 Loss,' with which we are here more especially concerned. 



The other examples which Errera cites are the occa- 

 sional development of a fifth stamen in the normally 

 four-stamened Scrophulariaceae, and also the case of 

 Heinricher 's 9 curious "Iris pallida Lam, var. abavia." 

 In this Iris, by the selection of spontaneous variations, 

 Heinricher obtained a form in which all six perianth 

 members were alike and bearded, while there were six 

 stamens instead of the normal three. This condition of 

 the stamens was interpreted as an atavistic return to the 

 type of androecium characteristic of the liliaceous stock 

 from which the Iridaceae are almost certainly derived, 

 and in which there are two whorls of stamens, each with 

 three members. These botanical cases may be compared 

 on the animal side with Castle's 10 race of four-toed guinea 

 pigs, which in this respect approached the ancestral 

 form— presumably five-toed — more nearly than does the 

 ordinary modern cavy, with its three-toed hind foot. 

 The occasional appearance of three-toed colts has also 

 been interpreted as an example of the reversionary 

 recovery of lost organs. 



It immediately becomes obvious, even on a casual 

 scrutiny of these cases, that they possess one striking 

 common characteristic — a charcteristic which seems to 

 me to annul their significance as evidences of reversion. 

 They all relate to meristic variations in which certain 

 organs, of which at least one already exists, suffer an 

 increase in number. 



The case of Iris pallida, var. abavia, is particularly 

 interesting from this standpoint and seems susceptible of 

 a different explanation from that given by Errera. The 

 six perianth members are all alike and all bearded, i. e., 



9 Heinricher, E., Iris pallida Lam., abavia, das Ergebnis einer auf Grund 

 atavischer Merkmale vorgenommenen Ziichtung und ihre Geschichte, Biol. 

 Centralbl., 16, pp. 13-24, 2 text-figs., 1896. 



10 Castle, W. E., The Origin of a Polydactylous Eace of Guinea-Pigs, 

 Carnegie Institution of Washington, Publication No. 47, pp 17-29, 1906. 

 It should be noted that Castle does not describe this polydactylism as 

 reversionary, but significance in this connection has been attributed to it 

 by others, e. g., Walter, H. E., Genetics, New York, 1913. 



