In the seeds, however, was an _embryo_. In this they differ from all known seeds of an earlier date, which, as has been already noted (see p. 77), are always devoid of one. This embryo is one of the most important features of the plant. It had two cotyledons which filled the seed space (see fig. 73), and left almost no trace of the endosperm. Reference to p. 112 will show that this is an advance on the Cycad seed, which has a small embryo embedded in a large mass of endosperm, and that it practically coincides with the Dicotyledonous type. The seed with its embryo suggested comparison with the Angiosperms long before the complete structure of the fructification was known. The fern-like nature of the pollen-bearing structures is another very important point. Were any one of these leaflike “stamens” found isolated its fern-like nature would not have been questioned a year or two ago, and their presence in the “flower” of _Bennettites_ is a strong argument in favour of the Fern-Pteridosperm affinities of the group. Had the parts of this remarkable fructification developed on separate trees, or on separate branches or distinct cones of the same one, they would have been much less suggestive than they are at present, and the fructifications might well have been included among those of the Gymnosperms, differing little more (apart from the embryo) from the other Gymnosperm genera than they do from each other. In fact, the extremely fern-like nature of the male organs is almost more suggestive of a Pteridosperm affinity, for even the simplest Cycads have well-marked scaly cones as their male organs. The female cone, again, considered as an isolated structure, can be interpreted as being not vitally different from _Cordaites_, where the seeds are borne on special short stalks amidst scales. The embryo would, in any case, point to a position among advanced types; but it is so common for one organ of a plant to evolve along lines of its own independently, or in advance of the other organs, that the embryo structure alone could not have been held to counterbalance the Cycadean stems and leaves, the Pteridosperm-like male organs, and the Gymnospermic seeds. But all these parts occur on the same axis, arranged in the manner typical of Angiosperms. The seed-bearing structures at the apex, the “stamens” below them, and a series of expanded scales below these again, which it takes little imagination to picture as incipient petals and sepals; and behold—the thing is a flower! And being a “flower”, is in closest connection with the ancestors of the modern flowering plants, which must consequently have evolved from some Cycadean-like ancestor which also gave rise to the Bennettitales. Thus can the flowering plants be linked on to the series that runs through the Cycads directly to the primitive ferns!