38o 



NATURE 



[August 20, 1903 



of the spore-sacs, which has much in common with that 

 of the pollen-sacs and ovules of Phanerogams, and this, 

 combined with the occurrence of well-marked hetero- 

 spory in some genera, appeared to point to a relationship. 

 But the former character (the development of the spore-sac 

 from a group of cells instead of from a single one) is now 

 known to be common to certain ferns, and to just those 

 ferns (the Marattiaceai, &c.) which prove to be the most 

 ancient, so that this argument has lost its weight. It has 

 lately been found, indeed, that some of the Carboniferous 

 Lycopods produced seed-like organs, presenting the most 

 striking analogies with true seeds, but the plants which 

 bore them were in all other respects Lycopods pure and 

 simple, and the case appears to have been one of homo- 

 plastic modification. There is no indication, as yet, of any 

 forms really transitional between the Lycopods and the 

 Spermophyta. 



The one line which, so far, has yielded truly intermediate 

 types is that of the ferns. 



Among recent plants, the Cycads, as we have seen, offer 

 some points of agreement with ferns, sufficient to have led 

 certain distinguished botanists, for example Sachs and 

 Warming, strongly to maintain- their fern-ancestry. The 

 chief points of agreement are : — 



(i) The fern-like foliage in some Cycads, and in many 

 the mode of folding of the leaflets in the bud. 



(2) The arrangement of the pollen-sacs in groups on the 

 underside . of the cone-scales, like that of the spore-sacs 

 of ferns on the underside of the leaves. 



(3) The carpels or fertile leaves of Cycas, which, though 

 bearing true seeds, are more like fertile fern-fronds than 

 any other reproductive leaves. 



By themselves, these characters, though suggestive, would 

 be inconclusive ; the anatomy is not directly comparable 

 with that of any living ferns. 



What, then, do we know of the history of this family in 

 past times? The Cycads are now a small and isolated 

 group ; in the Mesozoic period, from the Trias to the Lower 

 Cretaceous, they were one of the dominant types of vegeta- 

 tion, and spread all over the world. Of the fossil 

 species recorded from the. Oolite of the Yorkshire coast 

 and from the Wealden of the south of England, one-third 

 are referred to Cycads, and they were equally abundant in 

 the Mesozoic floras of North America, India, and other 

 countries. If they existed in the same proportion now as 

 then, they would have about 35,000 species instead of 70 ! 

 The Cycads of the Mesozoic, however, were not, as they are 

 now, a single family, but a great class (the Cycadophyta of 

 Nathorst) embracing very diverse types, often with organs 

 of reproduction widely different from those of their 

 surviving relatives, and showing a certain parallelism with 

 angiospermous fructifications. But with all this there was 

 on the whole a remarkable uniformity in habit, just as we 

 find a general similarity in outward characters among 

 so many dicotyledonous trees of the present day, though 

 belonging to the most diverse families. 



In the Mesozoic rocks we also find a certain number of 

 plants (known only from their foliage) as to which it 

 remains doubtful whether they belonged to Cycads or ferns, 

 or to some intermediate group. 



Besides the Cycadophyta, seed-plants were represented in 

 Mesozoic days by a great number of Coniferae, more or 

 less allied to those still living, and by various forms akin 

 to the maidenhair-tree, perhaps the more ancient type 

 surviving in the recent flora. 



When we go further back, to the Palaeozoic rocks, it is 

 only in their uppermost strata that we find forms clearly 

 referable to Cycads or Conifers. 



The best known seed-bearing plants of the older rocks 

 are those of the family Cordaiteae, which stretches back to 

 the Devonian. They were tall, branched trees, bearing 

 great simple leaves, sometimes a yard long. The anatomy 

 of stem and root resembled that of an Araucarian Conifer, 

 but the leaves had just the structure of the leaflets of a 

 Cycad. Male and female flowers were borne in little spikes 

 or catkins, and may best be compared with those of the 

 maidenhair-tree. The seeds, of which the structure is 

 known, closely resemble those of that plant, or of recent 

 Cycads. 



The Cordaiteae, however, ancient as they are, were 

 already pronounced gymnospermous seed-plants — by them- 

 selves they give no direct clue to the origin of Spermo- 

 NO. 1764, VOL. 68] 



phvta. We must look elsewhere for the key to our main 

 problem. 



The vast number and variety of fern-like remains through- 

 out the Palaeozoic strata, wherever land-plants are known, 

 is familiar to all. Almost every form of recent fern-frond 

 can be matched from the impressions in the Carboniferous 

 and Devonian rocks. A considerable number of these fossil 

 fern-fronds are known to have really belonged to ferns, for 

 typical fern-fructifications are found upon them. An 

 experienced collector of Coal-measure plants, Mr. Heming- 

 way, once told me that he reckoned on finding about 20 

 per cent, of the specimens of any true fossil fern in the fertile 

 state. When, therefore, a common fossil fern-frond (so- 

 called) is never found fertile, a strong suspicion is awakened 

 that the plant must have had some kind of fructification 

 other than that of an ordinary fern. This is the case with 

 a surprisingly large proportion of the Palaeozoic plants 

 commonly described as ferns, and holds good of certain 

 entire "genera"; the important genera Alethopteris, 

 Neuropteris, Mariopteris, Callipteris, Tasniopteris, and 

 others, have never yet been found, in any of their species, 

 with fertile fronds, if we except one or two specimens so 

 questionable and obscure that no conclusion can be drawn 

 from them. It is probably under the mark to say that one- 

 third of the so-called ferns of Palaeozoic age afford no 

 evidence from fructification that they were really ferns, as 

 we now define thein. 



The absence of recognisable fertile fronds may, it is true, 

 be partly accounted for by dimorphism. Many ferns, both 

 recent and fossil, bear their reproductive organs on modified 

 portions of the frond, or even on special fronds, very 

 different from the vegetative foliage. P'ossil remains are 

 usually fragmentary, and when the sterile and fertile fronds 

 are found isolated, there may be nothing to show that the 

 one belonged to the other. But, allowing for this, there are 

 very many " fern-fronds " which offer no evidence, even 

 from association, of any fern-like fructification, while the 

 fructifications actually associated with them are often any- 

 thing but fern-like. There are, in fact, a number of un- 

 assigned seeds from the Coal-measures, some of which are 

 commonly associated with certain of the quasi-ferns of 

 which we are speaking. 



On the whole, however, we have, up to this point, had 

 before us merely negative evidence, indicating that many 

 of the leaves, so familiar to palaeobotanists, classed on 

 account of their form and veining as fern-fronds may 

 really have belonged to some group different from the true 

 ferns. Negative evidence is notoriously weak ; at most it 

 only justifies us in taking up a position of philosophic 

 doubt, though in this case it was enough to induce the 

 distinguished Austrian palaeobotanist Stur to suspect that 

 the genera Alethopteris, Neuropteris, and others were not 

 ferns, but Cycads. 



During the last thirty years, however, positive evidence 

 has been accumulating proving that certain of the fern- 

 like Palaeozoic plants were at any rate something distinct 

 from true ferns, as we now know them. This evidence is 

 derived from a study of the anatomical structure, which in 

 Cycads and ferns, as they now exist, is sufficiently different 

 to prevent any possible confusion between the two groups. 

 A single section from the leaf-stalk of the fern-like Cycad 

 Stangeria would be enough to show that it is a true Cycad 

 and no fern, and conversely, a single section from the frond 

 of Lomaria, with which Stangeria was once confused, would 

 show it to be a true fern and not a Cycad. 



A common Coal-measure plant, named Lyginodendron 

 Oldhamium, was one of the first of the Paheozoic quasi- 

 ferns to be examined anatomically. We owe this work, 

 like so many other great advances in fossil botany, to the 

 late Prof. Williamson, who thus led the way to the solution 

 of the problem before us. 



Externally, the plant is wholly fern-like ; its characteristic 

 highly compound foliage is that of a Sphenopteris 

 (S. Honinghausi) with a Davallia-like habit. The large 

 fronds were borne, at intervals, on a somewhat slender 

 stem, which rooted freely. The slender proportions and 

 the presence of spines everywhere, on leaf and stem, suggest 

 that the plant may have been a scrambling climber like 

 Davallia aculeata, for example, among recent ferns. 



The structure of all the vegetative parts of the plant, 

 stem, leaf, and root, is known as perfectly, perhaps, 

 as in any plant now living. The leaves turn out. to be 



