THE CERATOPHYLLINAL DIVISION. 



25 



Hernandia belongs to the apetalous series of families, and Sparattanthelium has the stamens alternate with the sepals (which are not biseriate when only four,) being of the same number, and is therefore, 

 when the sepals are only four, in the strictest sense apetalous ; this however may serve to indicate that when two whorls are present as in the former genus the inner may be petals. 



In the descriptions of Hernandia the ovary is either obscurely referred to, or said to be superior, or to contract a partial adhesion with the calyx at the base of the style ; but on examining 

 specimens from different localities, both of H. sonora and ovigera, (the flowers being in various stages of growth,) the ovary proved to be uniformly inferior and completely so (fig. 2) ; from which 

 it might be supposed that its fruit only had been examined, in which apparently an almost complete separation of the calyx or superficial integument of the pericarp takes place. There is, however, 

 a tendency to this separation in the fruit of Gyrocarpus, and in that of Sparattanthelium it takes place (except perhaps at the apex) quite as much as in Hernandia. Gcertner describes the fruit of 

 Hernandia as clothed with the inflated calyx ; I believe, however, that the neck of the calyx still continues adherent to the apex of the ovary, as in Sparattanthelium, unless accidentally broken away ; 

 but this description may possibly refer to the involucre which enlarges and almost covers the fruit. 



Some time afterwards I examined some imperfectly hermaphrodite flowers, and in two of them I found a cavity on one side only of the ovary, scarcely extending to half its circumference, 

 the neck of the calyx being completely adherent with the base of the style and upper part of the ovary. This cavity appeared to me to be only an irregular separation of the calyx from the ovary, 

 which might be owing to the imperfect development of the ovary in consequence of the flowers becoming hermaphrodite ; but even supposing this to be the constant structure in one or more 

 species, it would be only another instance of "flos superus" with "ovarium liberum," which occurs in a much more marked degree in the Dipsacece.* 



The anthers of Hernandia may perhaps be, in some degree, analogous to those of Illigera and Gyrocarpus, the dehiscence consisting in the valves being cut out of the integument of the 

 anther lobes, leaving a thick connective, the difference being that they are turned laterally instead of directly upwards (figs. 4 & 5). This mode of dehiscence however as far as the valves being 

 turned laterally, occurs in families of very remote affinity, as Sabiacece ( S. parmfiora ) but the connective is wanting. 



In all other characters, except the involucre, which may be compared to that of Myristicacece and Chlcenacece, Hernandia differs but little from Gyrocarpus, having nearly the same habit 

 as one of the species of Gyrocarpus has undivided leaves. 



1 . The calyx is biseriate, consisting usually of three external and three internal sepals, or of four within four, the highest number occurring in the imperfectly hermaphrodite flowers, which 

 sometimes have five external and four internal.f When in Sparattanthelium the calyx consists of eight or nine sepals,^ it so closely resembles that of Hernandia, that they could be with 

 difficulty distinguished, except by their size and by that of the latter being biseriate. 



2. The stamens are three in flowers with six sepals, being opposite the outer three (thus agreeing with Illigera in their position), and the anthers open inwardly. In such flowers the barren 

 stamens are six, in pairs, opposite the three inner sepals, so that each stamen appears as if it had one on each side of it, and in the hermaphrodite flowers this proved to be the case, each of them 

 having a pair of glands attached to the filament close to its base, as in some genera of Lauracece. These barren stamens or glands are either solid, or consist of sacs on short stalks opening 

 outwardly, or are small scale-like leaves, which can be unfolded so as to resemble the small petaloid barren stamens internal to the fertile occurring in Anonacece. They may in all probability be 

 regarded as analogous to the glands of Illigera, and in being turned outwards while the anthers open inwards they resemble the barren stamens of Sassafras, and so nearly as to show I believe a very 

 close and undoubted affinity between Illigeracece and Lauracece. 



3. The tissues of the ovary of H. ovigera contain very numerous glandular dots like those of Gyrocarpus, and so much so, as with the other characters, to suggest a very near affinity. The 

 expanded stigma is deeply divided on one side, the fissure extending to the centre, and being continued down the style on the same side distinctly shows the position of the carpel (fig. 2). 



4. The ovule has a strongly marked dorsal raphe,§ as in Illigera, and a comparatively large foramen, through which the secundine sometimes projects (fig. 3), and, as in Illigera, a canal 



* I have observed this separation of the calyx very distinctly in Lpacridece, — the cavity passed all round the ovary. 



f Since the above was written, I have examined imperfectly hermaphrodite flowers of two species, which fully agreed with those of Illigera, the sepals being ten, biseriate, and the stamens five, opposite the external sepals. On this account 

 and from the close resemblance of the flower to Lauracece, I regard the inner series as petals. The ovaries were very small and mostly empty. It seems also very probable that Sparattanthelium has petals when the perigonium is 7-9 parted. 



1 It may appear questionable whether these flowers belonged to a distinct species or were only those of S. Tupiniquinorum become monstrous ; -the ordinary number of sepals in this species is four or five, when increased to eight or nine 

 the stamens are, 1 believe, not equally increased, seven being the highest number I have seen. I believe, however, there are two very distinct species included under this name, the sepals being much longer in the 6-9 sepaled specimens; their 

 anthers like those of Gyrocarpus and Lauracece have the valves turned directly upwards. 



§ In every flower I have examined I found the ovule attached to that side of the ovary (although very near the apex) which corresponded with the fissured side of the style and stigma, so that the attachment of the ovule, together with 

 the stigmatic canal always entering the ovary towards the opposite side; proved beyond doubt that the raphe is dorsal. 



