[page] 46

I will first give a case among the Reptilia from some of Mr. Darwin's unpublished MSS., which have been kindly lent me by Mr. Francis Darwin.

"M. Milne Edwards (Annales des Sci. Nat, 1 ser., tom. xvi. p. 50) has given a curious table of measurements of fourteen specimens of Lacerta muralis ; and, taking the length of the head as a standard, he finds the neck, trunk, tail, front and hind legs, colour, and femoral pores, all varying wonderfully ; and so it is more or less "ndth other species. So apparently trifling a character as the scales on the head affording almost the only constant characters."

¹ The quotations are from Natural Selection chapter IV, folios 25 to 33. Wallace noted in the preface "I wish to express my obligation to Mr. Francis Darwin for lending me some of his father's unused notes".

[page] 68

The next example is taken from Mr. Darwin's unpublished MSS.

[page] 69

"In some species of Shrews (Sorex) and in some field-mice (Arvicola), the Rev. L. Jenyns (Ann. Nat. Hist., vol. vii. pp. 267, 272) found the proportional length of the intestinal canal to vary considerably. He found the same variability in the number of the caudal vertebrae. In three specimens of an Arvicola he found the gall-bladder having a very different degree of development, and there is reason to believe it is sometimes absent. Professor Owen has shown that this is the case with the gall-bladder of the giraffe."

[page] 79

The following examples of variation in important parts of plants were collected by Mr. Darwin and have been copied from his unpublished MSS.:—

"De Candolle (Mem. Soc. Phys. de Genève, tom. ii. part ii. p. 217) states that Papaver bracteatum and P. orientale present indifferently two sepals and four petals, or three sepals and six petals, which is sufficiently rare with other species of the genus."

"In the Primulacae and in the great class to which this family belongs the unilocular ovarium is free, but M. Dubury (Mem. Soc. Phys. de Genève, tom. ii. p. 406) has often found individuals in Cyclamen hederaefolium, in which the base of the ovary was connected for a third part of its length with the inferior part of the calyx."

"M. Aug. St. Hilaire (Sur la Gynobase, Mem. des Mus. d'Hist. Nat., tom. x. p. 134), speaking of some bushes of the Gomphia oleaefolia, which he at first thought formed a quite distinct species, says: 'Voilà donc dans un même individu des loges et un style qui se rattachent tantôt a un axe vertical, et tantôt a un gynobase; donc celui-ci n'est qu'un axe veritable; mais cet axe est deprimé au lieu d'être vertical." He adds (p. 151), 'Does not all this indicate that nature has tried, in a manner, in the family of Rutaceae to produce from a single multilocular ovary, one-styled and symmetrical, several unilocular ovaries, each with its own style.' And he subsequently shows that,

[page] 80

in Xanthoxylum monogynum, 'it often happens that on the same plant, on the same panicle, we find flowers with one or with two ovaries;' and that this is an important character is shown by the Rutaceae (to which Xanthoxylum belongs), being placed in a group of natural orders characterised by having a solitary ovary."

"De Candolle has divided the Cruciferae into five sub-orders in accordance with the position of the radicle and cotyledons, yet Mons. T. Gay (Ann. des Scien. Nat., ser. i. tom. vii. p. 389) found in sixteen seeds of Petrocallis Pyrenaica the form of the embryo so uncertain that he could not tell whether it ought to be placed in the sub-orders 'Pleurorhizée' or 'Notor-hizée'; so again (p. 400) in Cochlearia saxatilis M. Gay examined twenty-nine embryos, and of these sixteen were vigorously 'pleurorhizées,' nine had characters intermediate between pleuro-and notor-hizées, and four were pure notor-hizées."

"M. Raspail asserts (Ann. des Scien. Nat., ser. i. tom. v. p. 440) that a grass (Nostus Borbonicus) is so eminently variable in its floral organisation, that the varieties might serve to make a family with sufficiently numerous genera and tribes—a remark which shows that important organs must be here variable."