THE ORIGIN OF ANGIOSPERMS. 49 
all existing Angiosperms are descended from forms with a conspicuous 
perianth, and that, in those plants where it is not present, its absence is 
due to abortion. We are thus inclined to postulate for the group a primitive 
perianth, which was completely differentiated, from the sporophylls on the 
one hand, and from the foliage leaves on the other, before the existing 
Angiosperms came into being. Hence we can hardly seek for its origin 
among their present-day representatives. At the same time we agree 
that the modern perianth may have, in certain instances, acquired additional 
members, either from above by the sterilisation of mierosporophylls, or from 
below by the modification of foliar organs. 
On this supposition, let us endeavour to arrive at some idea, from a study of 
living forms, of what may have been the characters of this primitive perianth, 
possessed by the immediate ancestors of the existing Angiosperms. 
Naturally we turn first to the Ranales, for in this group of families, as we 
have already seen, a number of primitive features appear to be retained in the 
andrecium and gyneceum. Тһе special points presented by the Ranalian 
perianth as а whole, which strike us as primitive, are the inconstant 
number and spiral arrangement of its members, as well as the absence 
of a marked separation into calyx and corolla. Тһе perianth of the 
Magnoliaceze is of special interest from this point of view, as showing 
transitions towards a definite number of tepals, arranged in whorls, and 
a differentation into a calyx and corolla. In Zlliciwm, there is a gradual 
passage from sepaloid to petaloid tepals. In Drimys, the distinction 
between the protective and attractive parts of the perianth is more marked, 
though they are still spirally arranged. In Magnolia, and its close allies, 
the perianth tends to assume a cyclic arrangement, and the tepals become 
reduced to a definite number. M. grandiflora, L., and М. stellata, Maxim., 
for example, have as many as thirteen perianth members, all much alike. 
In M. Yulan, Desf., they are reduced to nine, arranged in three fairly similar 
whorls. M. obovata, Thunb., and M. glauca, Linn., have likewise the same 
number, but the three outer ones are quite small. The tendency then in 
this family may be said to be towards a definite perianth, comprised in three 
whorls. Such a floral envelope, in which as a rule the outer whorl is 
sepaloid and the two inner petaloid, is a constant feature of the closely 
allied, but more highly evolved family, the Anonaces. 
The Ranunculaceæ are perhaps the next most interesting Ranalian family 
from this point of view, but the perianth here, on the whole, appears to be 
hardly so primitive, and is complicated by the occurrence of the so-called 
* honey-leaves," which we regard, in agreement with Prantl*, as recent 
modifications of some of the outer stamens. We should be inclined to view 
* Prantl (1888). 
LINN. JOURN.—BOTANY, VOL. XXXVIII. Е 
