96 M. Alphonse De Candolle's Review of 



nating with the lobes of the corolla, so that, as in Frimulacece, there remain 

 only a number of stamens equal to the lobes of the corolla, and opposite to 

 them. In this respect Sapotece are but a regular state of Mi/rsinecc and Pri- 

 mulacece. Without this character of a double or simple verticil of stamens, no 

 positive distinction would remain between these orders, as the direction of the 

 embryo, erect or transverse, has been shown by Mr. Brown to be of no great 

 consequence. 



From Primitlacece, the only distinctive character seems to be in the fruit 

 not being dehiscent; the habit of the two orders is besides very different, 

 Primulacece being herbaceous, and Myrsinece more or less ligneous, some- 

 times even forming large trees. There may be also some difference in the 

 shape of the grains of pollen. In Primula grandiflora they appear under 

 the microscope to be rectangular ; in Primula sinensis and Primula Auri- 

 cula they are oblong, but with some irregularity, and with a disposition to 

 show sometimes angular extremities and a quadrilateral form. In Ardisia 

 humilis, anceps, crenulata and cuhaiia, I saw nothing angular in them ; but they 

 are ovoid and very obtuse. In both orders they have no asperities by which 

 cohesion takes place. Dr. Martins represents the grains of pollen as really 

 round in Cyhianthus and Conomorpha {Cyb. laxijlorus, Mart.) ; but I suppose 

 they have been observed in water, which makes elliptic grains become round. 



Some difficulty arises from tlie genus Mccsa {Bceohotrys) having a great 

 number of seeds, as in many Primulacece, and an inferior ovarium, as in the 

 well-known and anomalous genus Samolus. But this last differs more from 

 the true Primulacece than Maesa does from Myrsinece, because it has five small 

 filaments alternating with the lobes of the corolla, so as to show the natural 

 state of Primulacece and the constant abortion that prevails in them. When 

 Dr. Bartling constituted a distinct family of Samolus and Mccsa among his 

 extensive class of 3Iyrsinece, where Primulacece are also included, he omitted 

 the fact of these five rudiments of stamens existing in Samolus and not in Mcesa. 

 After all, were Samolus an extensive genus, and not limited to a few species 

 only, it would have been considered worthy of forming a distinct order, inter- 

 mediate, as BIyrsinece, between Sapotece and Primulacece. At present, the best 

 classification, I suppose, is to include among Myrsinece the tribe of Mcesece, 

 and in Primulacece that of Samolece. 



