78 
H. F. Wernhant. 
The step from this “ flower ” to the flower of many members 
of the cohort Ranales’ appears at first sight to he but small. But 
the very definite angiospermy of the latter and the equally definite 
gymnospermy of the former are separated by a gulf which research 
has failed hitherto to bridge, and at which speculation must halt 
for the present. The peltate, closely approximated inter-ovular 
scales in the bennettitean “ flower ” may, it is true, represent 
carpels which are on the way to the angiospermous condition ; but 
in any case these “ carpels ” are highly specialized, and very 
different from the typical female sporophylls of Angiosperms. 
We can venture to say no more in this connection than that 
angiospermy, like other biological characters, may have been 
acquired in more ways than one in descent; and that the 
Angiosperms represent the type so far successful that only the 
merest traces of other types— e.g., Bennettiteae—remain to us in 
the fossil record. 
Our purpose, however, is to follow the history of the flower 
subsequently to the establishment of angiospermy, and not to 
enquire into the origin of the latter phenomenon. Cycadeoidea 
affords a useful analogy, if nothing more, from the point of view of 
the principle of economy in the number of reproductive parts; for 
it is, relatively to the Angiosperms, a primitive form with an 
undoubted filicinean ancestry, and has, at the same time, indefinite 
numbers in its reproductive organs. 2 
In the Magnoliaceae 2 the arrangement of the flower is essentially 
the same, typically, as in the case of the Cycadeoidea described 
above; the perianth comprises an indefinite number of leaves, and 
the stamens and carpels are borne, spirally arranged, upon an 
elongated axis. 
The progression along the line of reduction in the number of 
parts may be well illustrated by the typical floral formulae for the 
largest 3 natural orders of Ranales exhibited in the following table:— 
1 The names of the cohorts and natural orders used throughout 
this paper are those of Engler’s classification, unless other¬ 
wise specified ; and Willis’s Flowering Plants and Ferns will be 
found convenient for purposes of reference to the details of 
this system. 
2 See Newell Arber and Parkin. Journ. Linn. Soc. xxxviii. 29, 
1907. 
3 Excluding Lauracese, which Engler groups with Ranales, and 
in which the component parts of the flower are definite in 
number. 
