54 PnOCEEDlNfiS OF Tiir, 



able to evulve in a diivciioii ca|)al)le ot usiii;; insect -aji;ency ior 

 the purpose oF cross-pollination. This could he and may have 

 heen brought about by the aggregation of both kinds of sporo- 

 phylls into one and the same cone. 



It is not dithcult to advance a reason why the anthostrobilus — 

 the su|)po-;ed insect-pollinated strobiius — has invariably the male 

 s|)orophylls situated on the axis below the female. JJy this 

 arrangement there would be less likelihood of self-pollination. 

 In the reverse sequence pollen would be apt to fall on the ovules 

 below. Further, it may be conjectured that a pollen-seeking 

 insect would alight on the apex of the cone and, in the event of 

 the microsporopliylls being on the upper part of tlie strobdus, 

 the niegaspdrophylls below would not be traversed and no cross- 

 pollination would ensue. 



As regards the evolution of the anthostrobilus from the 

 Pteridospern)ous arrangement of sporophylls, probably on a 

 given axis a series of one kind of s|)oropliyll (prelerably male) 

 was followed by a series of the other kind. A primitive antho- 

 strobilus woulil result from such an axis ceasing further growth 

 apically. Each batch of sporophylls may be imagined to have 

 been protected by a series of bracts. The lower series became 

 the perianth of the anthostrobilus. The upper series aborted as 

 the two sets of sporophylls drew closer together on the axis, 

 allowing the lower series to take on the protective function of 

 both. The length of bare axis which separated the male and 

 female parts of the cone in some Bennettitales may signify that 

 the two kinds of sporophylls were originally some distance apart. 

 It is tempting in this connection to see some ancestral significance 

 in the gyncphore of Michdla, a genus separated from MiujaoUa 

 on account of possessing this feature. 



At the time of the publication of our paper perhaps the 

 weakest point in our theory lay in the lack of any similarity 

 between the vegetative features of Bemuttitcs and the Dicotyle- 

 donous tree. The former was so Cycadean in leaf and stem as to 

 bear no resemblance to the latter, and had besides apparently 

 axillary fructitications. AVe postulated a solitary Hower as a 

 jirimitive Angiospermous character, and the writer has shown 

 since (38) that this was probably borne terminally to a leafy 

 shoot. Now a striking feature brought to the front in recent 

 years respecting Bennettitalean genera other than Bennettites 

 (Cycadeoidea) itself resides in the fact that tlie strobili were 

 borne terminally. Such cones occurred in Willidmsoiiia, Wii'Iand- 

 iella, and WiUlanisonidla. Further, the vegetative features of the 

 above three genera had other points in common with the Dicoty- 

 ledonous tree. The following may be mentioned: — (1) Marked 

 internodes : (2) slender stems ; (3) free branching; and (4) small 

 foliage leaves. The evolution of the Dicotyledonous tree-habit 

 from that of the Pteridosjierm is thus rendered less impr()babl<,\ 

 The Bennettitales went jxirfc of the way only, the llemiangio- 

 spertns on parallel lines the full way. 



