of Bennettites gibsonianus , Carr. 449 
which contain the ovules is due to the cohesion of the apices 
of the interstitial organs, and in this, as in all similar cases, 
we see a contrivance for protection. Nor will it seem too 
fanciful to appeal, as to analogous conditions, to the closing 
of the carpels in forming the ovary, or to the position of the 
inferior ovary. If we suppose the enveloping coherent inter- 
stitial organs or leaves to spring from the seed-stalk itself, we 
should actually have an ovary, in rudimentary form indeed, 
but still of essentially similar character to the ovary in Angio- 
sperms ; we need only add a stigma to make it complete. But 
if the cohering members do not belong to the seed-stalk, but 
originate wholly from the axis of the spadix, this would 
certainly imply so primitive a stage in the formation of the 
flower that we could not point to any fitting analogue amongst 
existing plants. Still we might form an idea of it from the 
capitulum in Compositae, if we were to imagine the scales of 
the involucre and those on the receptacle all fully developed 
and coherent above, whilst the flowers were reduced to the 
simplest condition, to naked ovules. To realise this idea to 
some extent, we might call to mind the female capitula in the 
genus Xanthium. The effect, which is so often produced in 
our existing plants by intercalary growth of the tissue of the 
axes (Monimiaceae, Dorstenia ), would result in the present case 
from the cohesion of the last leaves of the main axis which 
bears the flowering shoot. The wall of the ovary thus formed 
would therefore belong to an older shoot-generation than the 
ovule ; and if we are prepared to adopt the standpoint of some 
morphologists and maintain that the ovule is formed from the 
leaf, the only change of importance required would be a change 
in the terminology, for the entire shoot would then become 
a flower with numerous leaves, some of which are sterile and 
serve to form the involucre, while the others bear the ovules. 
If in the course of the above account the Bennettiteae have 
constantly been compared with Cycadeae, it is because I am 
convinced that Cycadeae, notwithstanding all differences, are 
the nearest known allies of Bennettiteae, and on this point I 
shall be at one with nearly all palaeontologists and botanists. 
