XXIII.] THE CAUSES OF DEVELOPMENT. 303 



nibranchiate. But the axolotl as hitlierto known, tliough 

 it is a mature form by the usiial test of being able to pro- 

 pagate,^ proves to be a larva in the sense of having yet 

 to undergo metamorphosis. The specimens observed in 

 captivity in France, from which alone our knowledge of 

 their metamorphosis is derived, sometimes undergo meta- 

 morphosis and sometimes not, without any assignable reason. 

 Like otlier BaJ:rachians, they lose their branchige when 

 they undergo metamorphosis, in addition to other changes ; 

 and if the branchiae are removed (under which operation 

 the animal does not appear to suffer in health), the chance 

 of the metamorphosis taking place is increased, though it 

 is still very uncertain.^ 



These facts are most interesting, partly as showing that 

 an animal which has not undergone its final metamor- 

 phosis may yet propagate ; and partly as showing how 

 variable is the fact of fianl metamorphosis, as between in- 

 dividuals of the same species. This last is an instance of 

 the general law, that those characters which are variable as 

 between allied species are also variable as between indi- 

 viduals of the same species : for no one doubts the real and 

 near affinity between the Perennibranchiates, which never 

 lose their branchiae and are in fact permanent tadpoles, and 

 the Caducibranchiates, such as the newt and the frog, 

 which lose their branchiae when they cease to be tadpoles.^ 

 The irregularity of the metamorphoses of the axolotl sug- 

 gests also, that the first individual of a Perennibranchiate 

 species that lost its branchife, and gave origin to a race 

 of Caducibranchiates, may have done so accidentally, in 

 consequence probably of the branchiae drying up and 

 withering for want of water, 



1 I.e., sexually. No vertebrate ever propagates in any other way. 



2 Memoir by Auguste Dumeril, translated in the Annals of Natural 

 History, December 1867. 



3 I do not mean to imply that it would be a natural classification to 

 divide the Batrachia into Perennibranchiates and Caducibranchiates. 

 The right division of the class is into (1) the newt or salamander tribe, 

 having four limbs and a tail ; this includes the Perennibranchiates : 

 (2) the frog tribe, having four limbs but no tail : (3) the Csecilia genus, 

 which are like serpents in form, having no limbs. 



