26 THE CEKATOPHYLLINAL DIVISION. 



extends from the base of the style to the ovary, passing over the dorsal side of the funiculus of the ovule into its cavity. In the position of the carpel, as indicated by the position of the raphe, it 

 does not materially differ from its allies, as it is always lateral, as it is in Laurus nobilis. 



5. The embryo of Hernandia may be regarded as imperfectly plaited, being crumpled all over the surface, and in addition to this longitudinally 6-furrowed on its lower half, but it is not at 

 all convolute. I have already noticed that the embryo of Sparattanthelium is 6-furrowed in consequence of being longitudinally plaited, so that in this character it forms a connecting link between 

 Hernandia and Gyrocarpus. 



The genera may be naturally and conveniently distinguished by their fruit and embryos, although Sparattanthelium further differs in having the stamens alternate with the sepals, and 

 Hernandia in its anthers. 



„ . , , C Fruit winged laterally .... Illigera. 



Cotyledons convolute . < _ . . ° . 1 ' ~ 



C r ruit winged only at the apex . . . (jryrocarpus. 



Cotyledons plicato-convolute, making the embryo 6-furrowed . . * Sparattanthelium. 



Cotyledons crumpled, each of them 3-furrowed at the base . . Hernandia. 



Laurace^e 



The flower of Laurus is described as tetramerous with the stamens alternate the sepals, but a careful examination has convinced me that it is trimerous like the generality of the genera, and 

 it serves to show that the 3 inner sepals of this family are sepaloid petals. The sepals are 4, but 1 is smaller and internal to the others, and is, I doubt not, 1 of 3, the other two of which have 

 become stamens, and this circumstance has occasioned this genus to be described as having the stamens alternate the sepals. This appears to me to be clearly proved by the occurrence of flowers 

 with 6 sepals, 3 inclosing 3 (the petals) much more than in most of the genera, and by the stamens in these cases being opposite the sepals ; — one good instance I met with had 6 stamens opposite 

 6 sepals, 3 (the petals) entirely within 3, and 1 of the internal still retained its antheriferous character so far as to bear half an anther on one side. It is common to find flowers with five sepals, and 

 it was sufficiently evident in all the instances examined that 1 of the stamens was missing having become the additional sepal, i.e., a sepaloid petal. 



Litscea is dimerous, the 2 sepals in aestivation inclosing the 2 petals, from which however they cannot well be distinguished after expansion. The stamens are 6, the outer 4, which are 

 usually glandless, being opposite and adherent to the sepals and petals ; and the remaining 2 which have glands at their base, are opposite the sepals, being within the first 2 and like them adherent 

 to the middle of the sepal near its base. The female flower has 6 barren stamens and their position is more obvious. 



Calycanthace^e. 



This family is so near both to Rosacea and Magnoliacece, that they have been regarded as osculant between them, but still they may not improbably prove to be decidedly nearer to one of 

 them than to the other. It appears preferable to place them here, as having opposite leaves and a calyx composed of more than the usual number of sepals, in which they agree with Monimiacece ; 

 and anthers turned outwards, in which they agree with some genera of Anonacea, but decidedly differ both from Rosacea; and their allies; and should the remarkable character of the wood, 

 exhibiting discoid markings on its tissue prove to be identical with that which occurs in Illiciece, the approach to the Anonal Alliance will become the more evident. As a near ally of Magnoliacem 

 they also, m common with that family, approach Nymphaaeea and even Nelumbiacea, agreeing with the former in the gradual transition from petals to stamens, which occurs in Calycanthus 

 occidentalism where the inner row of petals are tipped with precisely the same kind of gland that forms the crest of the anthers, and in the margins of the smaller ones becoming more or less 

 antheriferous; and with the latter in the gradual transition of the sepals to petals, the glandular crest of the stamens being also the same, that of Nelumbium being rather more elongated. These 

 affinities being admitted, Calycanthacece would be far removed from Rosacea, and their convolute cotyledons, opposite leaves, and other characters, would find a parallel in Combretacece near which they 

 are placed in the Laurinal Subdivision. 



