214- THE ANDR<ECIUM OF MENTZELIA. 
of the centrifugal and centripetal modes of evolution in compound sta- 
mens be admitted, it will be seen that the difference between the Loa- - 
sece and Mentzeliece is not so great as might at first sight be supposed. 
At all events, it seems to me that systematists Aould hesitate, in ab- 
sence of other diagnostic characters of any value, to break up the old 
Order Loasacece, 
Having stated my reasons for holding the androBcium of the Ment- 
zelie^ to consist of five confluent compound stamens, I shall now con- 
siuer some of the arrangements in the Rosacece^ in which family Payer 
has pointed out an analogy in staminal evolution witb that of the Mmt- 
zeliem. I am induced to do so more particularly, because the expla- 
^ ven of the androeciura of Mentzelia and Bartorda seems 
to throw considerable light upon the morphology of that of the Romcem, 
I have constructed diagrams, in accordance with Payer's figures, of the 
principal forms of staminal aiTangement met with in this family, to 
which I shall refer in the following remarks. I have myself carefully 
examined the development of the androecium of Rubus Idaus and Rosa 
spinosissima ; and to a certain extent observed that of Comarum and 
Geum. 
In Ruhus^ soon after the appearance of the petals^ the hemispherical 
termination of the receptacle comes to be surrounded by a flattish 
cushion. The external outline of the cushion is that of a pentagon 
with gently concave sides ; and the petals form the angles. An ex- 
amination of the earliest development of this cushion, gives one the 
impression of its being produced by an extension and fusion of the 
bases of the petals. On the surface of the cushion, the stamens are 
developed in the following manner : — First, ten stamens appear, one 
on either side of each petal, and closely approximated to it. From 
these points the evolution of the other numerous stamens extends cen- 
tripetally. This development of stamens is accurately described by 
Payer as " uue eruption de petits mamelons, qui commen9ant pres les 
petales, s'en eloignerait pevi a peu,"* In Rubm tomentoms, the species 
which Payer has figured, the series of stamens approaching each other 
from any tw^o adjacent petals do not coalesce untU the stamens of the 
fourth degree, as I have represented in the diagram (Pig. 8). In i?. 
Id(EU8 I find that this coalescence usually takes place in the stamens 
of the third degree, I am satisfied that, in this plant, the carpels 
* ' Organogenie,' p. 504. 
