CA8TANEACEM. 249 



lection of characters. It is scarcely separable from the Ulmaceœ to 

 which, as we have seen, Adanson had united it. Only normally 

 among the Gastaneacece, there are not the polygamous flowers of the 

 Elms, nor the stipules characteristic of the Artocarpeœ, nor the 

 peculiar disposition of the staminal filaments of the Morece, nor the 

 opaline or milky latex of the two latter groups. Moreover, at adult- 

 age, the greater part of the Castaneacece preserve in the ovary more 

 than one ovuliferous cell, which is the case with no one of the 

 Ulmaceœ. On the other hand, by the Betulece, the family before us 

 borders on the amentaceous groups of the EitpJiorbiacece, such as the 

 Scepeœ and Antidesmeœ ; and by the Gonjleœ, to the series of Saxifra- 

 gacece which comprise the Platanece and Hamamelidece. In fact, as 

 Ave have elsewhere said, 1 it is not simply a resemblance of foliage and 

 of habit that is found between the Alders and certain Fothergilla or 

 Parrofin, or between Corylopsis and Gorylus ; for these latter, with 

 their inferior ovary and descending ovules, perfectly definite in number, 

 in cells at first incomplete, seem to be only amentaceous and apc- 

 talous representatives of Corylopsis and neighbouring Hamamelidece. 

 Hence an analogy between the Quercineœ and Gorylece and the 

 Comacece, which themselves have so many points of agreement with 

 the Hamamelidece. Take away the involucre and all those accessory 

 organs of tardy growth, which form the cupules and spinous sacs of 

 the Gorylece and Quercineœ, and the flower with inferior ovary of 

 the Oaks, Chestnuts, &c, is altogether, in construction, that of the 

 apetalous Combretacece, notably of Termiiialia, which often also 

 have apetalous, diclinous flowers in spikes, or amentiform capitules 

 (Anogeissus, Bamatuella, Conocarpus), and the placentas of which, 

 parietal at first, like those of Quercus or Castanea, but remaining 

 so to the end, bear in like manner ovules definite in number, de- 

 scending, with micropyle exterior and superior. Finally, by the 

 Myriceir, this family approaches the Juglandece, the unilocular ovary 

 of which likewise encloses a single orthotropous and erect ovule ; 2 

 but the independence of the gynseciuin in Myrica suffices to dis- 

 tinguish it immediately from them. 3 



collateral to the superior Oupuliferce and to the considers Myrica as intermediate between Amen- 



Aquilarineoe, pointing out also, in the same («e<e and Urticecè. 



work, their affinity with the Setuleœ. 3 Leitneria. seems to unite the Amentaceœ to 



i See Admisonia, x. 137. the Willows. Balanops has a fruit and habit 



- M. Clakkb [Ann. Nat. Hist. (1858), 100) resembling the Sapotaceœ ; it represents perhaps 



an apetalous and amentaceous form of it. 



