CARPOLITHES. 103 



mutua angulosa extus laevia vel longitudinaliter striata costataque, 

 basi semper rotundiore insertionis cicatrice notata apice autem plus 

 minus ve attenuata." l 



As regards the difference between fossil specimens of cycadean 

 and coniferous seeds, it would seem that we cannot trust to any 

 convenient method of distinguishing, in all cases, between the 

 two groups of plants. The large oval, or almost spherical seeds 

 of certain cycads may generally be distinguished from the typical 

 forms of coniferous seeds, but in the latter group we have such 

 forms as Ginkgo, Cephalotaxus, and others, in which the size 

 approaches more closely to that of the cycadean ovule, than to 

 the smaller seeds of such conifers as Larix, Pinus, and many 

 others. In many Coniferce the presence of a membranous wing 

 affords a ready means of identification, at least as regards their 

 separation from Cycadacece, but the seeds of many conifers are 

 without any winged appendage, and even in the case of winged 

 seeds, the thin membrane might readily become detached before 

 the seed had been permanently enclosed in a mass of sediment. 

 Another obvious source of difficulty, worth referring to in this 

 connection, is the very great difference in size exhibited by the 

 seeds of the same plant at different stages of growth. The ripe 

 seeds of such a genus as Ci/cas, preserved with the wrinkled 

 reddish - brown outer coat intact, present a very different 

 appearance from those in which this coat has become detached, 

 thus exposing the perfectly smooth inner coat ; and a still greater 

 contrast is afforded by the more spherical kernel (nucellus), with 

 its surface traversed by branching grooves marking the position 

 of vascular bundles. 2 On the whole, it would seem advisable to 

 follow the example of Schenk in his Flora, der Grenzschichten* 

 and make use of the old term Carpolithes for gymnospermous 

 Mesozoic seeds. In certain cases the character of the seeds, or 

 their frequent juxtaposition with cycadean fronds, may enable 

 us to speak of them as cycadean with reasonable certainty; 

 when such is the case it will be well to give expression to our 

 more accurate knowledge, either by adopting Saporta's genus, 



1 Saporta (A. 2), Pal. FranQ. vol. ii. p. 235. 



2 This is well shown in the seeds of Cycas circinalis. 



3 PI. xxxiii. figs. 5-9. 



