seeds constructed on the Gymnospermous plan, but characterised by an architectural complexity far beyond that represented in the seeds of any modern Conifer or Cycad.

In such genera of Gymnosperms as Cordaites, Pitys, and others, we have examples of forest trees possessing wood almost identical with that of existing species of Araucaria, but distinguished by certain peculiarities which point to a relationship with members of the Cycadofilices, and suggest that Conifers as well as Cycads may have sprung from a filicinean stock.

These waifs and strays from the vegetation of an era incredibly remote, when strange amphibians were lords of the animal world, afford, as Newberry expresses it, fascinating glimpses of the head of the column of terrestrial vegetation that has marched across the earth's stage during the different geological ages.'

Two facts stand out prominently as the result of a general survey of what are practically the oldest records of plant-life. One is the abundance of types which cannot be accommodated in our existing classification founded solely on living plants.

The Devonian and Lower Carboniferous plants lead us away from the present along converging lines of evolution to a remote stage in the history of life; they bring us face to face with proofs of common origins, which enable us to recognise community of descent in existing groups between which a direct alliance is either dimly suggested or absolutely unsuspected if we confine our investigations to modern forms. We recognise, moreover, in such a plant as Archæocalamites an ancestor from which we may derive in a direct line the existing members of the Equisetales. In other types, by far the greater number, we see striking examples of Nature's many failures, which, after reaching an extraordinary complexity of organisation, gave place to other products of evolution and left no direct descendants.

Another fact that seems to stand out clearly is the almost worldwide distribution of several characteristic Lower Carboniferous plants. The accompanying table (page 830), based on the artificial divisions marked out on the map, to which reference has already been made, shows how widely some of the plants had migrated from an unknown centre far back in a still more remote age. We are, as yet, unable to follow these Devonian plants to an earlier stage in their evolution. We are left in amazement at their specialised structure and extended geographical distribution, without the means of perusing the opening chapters of their history.

Upper Carboniferous (Coal-Measures) and Permian Floras.

From the Lower Carboniferous formation we pass on to the wealth of material afforded by the Upper Carboniferous and Permian rocks. From the point of view of both botanists and geologists, the fossil plants obtained from the beds associated with the coal are of greater interest and importance than those of any other geological period. By a fortunate accident our investigations are not restricted to the examination of carbonaceous impressions and sandstone casts left by the stems and leaves of the Coal-period plants. By means of thin sections cut from the calcareous nodules of the coal-seams of Yorkshire and Lancashire, and from the silicified pebbles of France and Saxony, it is possible to make anatomical investigations of the coal-forest trees with as much accuracy as that with which we can examine sections of recent plants. The differences between the vegetation that witnessed the close of the Carboniferous era and that which flourished during the opening stages of the succeeding Permian epoch are comparatively slight. It has been demonstrated by Grand'Eury, Kidston, Zeiller, Potonié, and others, that it is possible both to separate the floras of the Coal-measures from those of Lower Permian age, and to use the plant species as trustworthy guides to the smaller subdivisions of the Coal-measures; but apart from these minor differences, the general facies of the vegetation remained fairly constant during the Upper Carboniferous and Lower Permian periods.

The vast forests of the Coal age occupied an extensive area of land on the site of the present United States of North America, stretching across Europe into

Eastern Asia; under the shade of their trees lived the stupid, salamander-like Labyrinthodonts, which pottered with much belly and little leg, like Falstaff in his old age.' The plants of these Paleozoic forests seem to be revivified, as we subject their petrified fragments to microscopical examination. Robert Louis Stevenson has referred to a venerable oak, which has been growing since the Reformation and is yet a living thing liable to sickness and death, as a speaking lesson in history. How much more impressive is the conception of age suggested by the contemplation of a group of Paleozoic tree-stumps exposed in a Carboniferous quarry and rooted where they grew! An examination of their minute anatomy carries us beyond the mere knowledge of the internal architecture of their stems, leaves, and seeds; it brings us into contact with the actual working of their complex machinery. As we look at the stomata on the lamina of a leaf of one of those strange trees, and recognise a type of structure in the mesophyll-tissues which has been rendered familiar by its occurrence in modern leaves, it requires but little imagination to see the green blade spreading its surface to the light to obtain a supply of solar energy with which to extract carbon from the air. We can almost hear the murmur of plant-life and the sighing of the branches in the wind as the sap courses through the wood, and the leaves build up material from the products of earth and air; products that are to be sealed up by subsequent geological changes, till after the lapse of countless ages the store of energy accumulated in coal is dissipated through the agency of man.

The minute structure of the wood of the Calamites, Lycopods, and other trees, agrees so closely with that of existing types that we are forced to conclude that these Paleozoic plants had already solved the problem of raising a column of water more than 100 feet in height. The arrangement of the strengthening or mechanical tissues in the long flat leaves of Cordaites is an exact counterpart of that which we find in modern leaves of similar form. The method of disposition of supporting strands in such manner as to secure the maximum effect with the least expenditure of material, was as much an axiom in plant architecture in the days of the coal-forests as it is now one of the recognised rules in the engineer's craft.

We need not pause to discuss the various opinions that have been expressed as to the conditions under which the forests grew; we may adopt Neumayr's view, and recognise a modern parallel in the moors of the sub-arctic zone, or find a close resemblance in the dismal swamp of North America. There is also the view expressed many years ago by Binney and warmly advocated by Darwin, that some at least of the Coal-period trees grew in salt-marshes, an opinion which receives support from several structural features suggestive of xerophytic characters recognised in the tissues of Palæozoic plants.

Time does not admit of more than the most cursory glance at the leading types of the Permo-Carboniferous floras. The general character of the preceding vege tation is retained with numerous additions. Archæocalamites is replaced by s host of representatives of the genus Calamites, an Equisetaceous type with stout woody stems and several forms of cones of greater complexity than those of modern Horsetails. Side by side with the Calamites there appear to have existed plants which, from their still closer agreement with Equisetum, have been described by Zeiller, Kidston, and others as species of Equisetites. The genus Sphenophyllum, a solitary type of an extinct family, was represented by several forms which, like the Galium of our hedgerows, may have supported their slender branches against the stems of stronger plants. Lycopods, with trunks as thick and tall as forest trees, were among the most vigorous members of the later Palæozoic forests. Although recent research has shown that several of the supposed ferns must be assigned to the Cycad-fern alliance, there can be no doubt that true ferns had reached an advanced state of evolution during the Permo-Carboniferous epoch. The abundance of petrified stems of the genus Psaronius, of which the nearest living representatives are probably to be found among the tropical Marattiacere, demonstrates the existence of true ferns. Others had more slender stems which clambered over the trunks of stouter trees, while some grew in the shade of Lepidodendron and Cordaites. The most striking fact as regards the Permo-Carboniferous ferns is the abundance of fertile fronds bearing sporangia which exhibit a more or less

close agreement with those of the few surviving genera of Marattiaceæ. The more familiar type of sporangium met with in our existing fern-vegetation is also represented, and we have recently become familiar with several genera bearing sporangia exhibiting a close resemblance to those of modern Gleicheniaceae, Schizæaceae, and Osmundaceæ. The sporangial characteristics of the different families of living ferns are many of them to be found among Palæozoic types, but there is a frequent commingling of structural features showing that the ferns had not as yet become differentiated into so many or such distinct families as have since been evolved.

Prominent among the Gymnosperms of the Palæozoic forests must have been the genus Cordaites: tall handsome trees, with long strap-shaped leaves, recalling on a large scale those of the kauri pine of New Zealand. This genus, which has been made the type of a distinct group of Gymnosperms, combined the anatomy of an Araucaria with reproductive organs more nearly allied to the flowers of Cycads, and exhibiting points of resemblance with those of the Maidenhair-tree. It is not until the later stages of the Permo-Carboniferous epoch that more definite coniferous types made their appearance. The genus Walchia, in habit almost identical with Araucaria excelsa, the Norfolk Island pine, with Ulmannia and Voltzia, are characteristic members of the vegetation belonging to the later phase of the PermoCarboniferous era. The Maidenhair-tree of the far East, one of the most venerable survivors in our modern vegetation, is foreshadowed in certain features exhibited by Cordaites and, as regards the form of its leaves, by Psygmophyllum, Whittleseya, and other genera. Psygmophyllum is known to have existed in Spitzbergen in the preceding Culm epoch, and Wittleseya occurs in Canadian strata correlated with our Millstone Grit. Leaves have been found in Permian rocks of Russia, Siberia, Western and Central Europe, referred to the genus Baiera, a typical Mesozoic type closely allied to Ginkgo. In the upper Coalmeasures and lower Permian rocks a few pinnate fronds have been discovered, such as Sphenozamites, from the Permian of France, Pterophyllum from France and Russia, and Plagiozamites from the Permian of Alsace, which bear a striking likeness to modern Cycadean leaves. Throughout the Permo-Carboniferous era the Cycadofilices formed a dominant group; Lyginodendron, Medullosa, Poroxylon, and many other genera flourished in abundance as vigorous members of an ancient class which belongs exclusively to the past.

One distinctive characteristic of the vegetation of later Permo-Carboniferous days is the occurrence of the Cycad-like fronds already referred to; also the appearance of Voltzia and other conifers with species of Equisetites, pioneer genera of a succeeding era that constitute connecting links between the Paleozoic and Mesozoic floras.

What we may call the typical vegetation of the Coal-measures, which continued, with comparatively minor changes, into the succeeding era, flourished over a wide area in the northern hemisphere, suggesting, as White points out, an almost incredible uniformity of climate. The same type of vegetation extended as far south as the Zambesi in Africa, and to the vast coal-fields of China; it possibly existed also in high northern latitudes, but, since Heer's record of Cordaites in Novaya Zemlya in 1878, no further traces of arctic Permo-Carboniferous plants have been found. Calamites, Lepidodendron (with its near relative Sigillaria), Ferus, Cycadofilices, Cordaites, and other Gymnosperms, constitute the most familiar types. We have already noticed the existence in the southern hemisphere of Lower Carboniferous and Devonian genera identical with plants found in rocks of corresponding age within the Arctic circle. This agreement between the northern and southern floras was, however, not maintained in the later stages of the Paleozoic epoch. Australian plant-bearing strata homotaxial with Permo-Carboniferous rocks of Europe, have so far afforded no examples of Sigillaria, Lepidodendron, or of several other characteristic northern forms; in place of these genera we find an enormous abundance of a fern known as Glossopteris, a type which must have monopolised wide areas, suggesting a comparison with the green carpet of bracken that stretches as a continuous sheet over an English moor. With Glossopteris was associated a fern bearing similar leaves, 1903. 3 H

known as Gangamopteris, and with these grew Schizoneura and Phyllotheca, members of the Equisetales. In addition to these genera there are others which bear a close resemblance to northern hemisphere types, such as Noeggerathiopsis, a member of the Cordaitales, and several species of Sphenopteris. Similarly, in many parts of India, Glossopteris has been found in extraordinary abundance in the same company with which it occurs in Australia. In South Africa an identical flora is met with which extends to the Argentine and to other regions of South America. A few members of this southern flora have been recorded from Borneo, and the genus Glossopteris is said to occur in New Zealand, but the latter statement has been called in question and requires confirmation. It is clear that from South America, through South Africa and India to Australia, there existed a vegetation of uniform character which flourished over a vast southern continent at approximately the same period as that which, in the northern hemisphere and in China, witnessed the growth of the forests whose trees formed the source of our coal-supply.

Since attention was drawn by Dr. Blanford and other writers to the facts of plant-distribution revealed by a study of the later Palæozoic floras, it has beer generally admitted that during the Permo-Carboniferous era there existed two fairly well-marked botanical provinces. The more familiar and far richer flora occupied a province stretching from the western states of North America acros Europe into China and reaching as far as the Zambesi; the other province was occupied by a less varied assemblage of plants, characterised by the abundance of Glossopteris, Gangamopteris, Neuropteridium, Noeggerathiopsis, Schizoneura, and other genera, stretching from South America through India to Australia.

Two questions at once suggest themselves: firstly, were these two botanical provinces defined by well-marked boundaries, or did they dovetail into one another at certain points? Secondly, is there any probable explanation of this difference between northern and southern floras, a feature not shown either by the preceding Devonian and Lower Carboniferous or by the succeeding Lower Mesozoic floras!

In Brazil, Professor Zeiller has recorded the occurrence of a flora including Lepidophloios, a well-known European member of the Lycopods, associated with such characteristic southern types as Gangamopteris and Noeggera thiopsis. Similarly from the Transvaal a European species of Sigillaria, with a Lepidodendroid plant, and another northern genus, Psygmophyllum, have been found in beds containing Glossopteris, Gangamopteris, Noeggera thiopsis, Neuropteridium, and other members of the so-called Glossopteris flora. In India, the Glossopteris flors exhibits an entire absence of Lepidodendron, Calamites, Sigillaria, and other common northern genera, while Sphenophyllum is represented by a single species. The Australian Permo-Carboniferous flora is also characterised by the absence of the great majority of the northern types. Until a few years ago the genus Glossopteris had not been discovered in Europe, but in 1897 Professor Amalitzky recorded the occurrence of this genus in association with Gangamopteris in Permian strata in Northern Russia.

We see, then, that in Brazil and South Africa the Glossopteris flora and the northern flora overlapped, but the former was the dominant partner. On the other hand, in rocks belonging to a somewhat higher horizon in Russia, we meet with a northern extension of the Glossopteris flora. The accompanying map (p. 835) serves better than a detailed description to illustrate the geographical distribution of these two types of vegetation in the Permo-Carboniferous era.

There is little doubt that the differences between the flora of the southern continent, that existed towards the close of the Carboniferous and during the succeeding Permian period, and that which flourished farther north have in some respects been exaggerated; geographical separation has played too conspicuous s part in influencing botanical nomenclature. Granting the existence of identical genera or representative types, there remains a striking difference between the two provinces into which the Permo-Carboniferous vegetation was divided. As regards an explanation of this fact, we can only hazard a guess; as Dr. Blanford and others have pointed out, there is a probable solution to hand. Briefly stated, the Upper Paleozoic plant-bearing strata of India, South America, Australia, and

MAP II.-Permo-Carboniferous Floras.

parts of South America, and Australia remain as comparatively insignificant remnants, was exposed to climatal conditions favourable to the accumulation of snow and to the formation of glaciers. One possible explanation, therefore, of the existence of a distinct vegetation in the southern area is that the climate was such as to render impossible the existence of those coal-forest plants that exhibited so vigorous

3

South Africa are in close association with boulder-beds of considerable extent. In some places, as for example in India and Australia, the boulder-beds rest on rocks bearing unmistakable signs of the grinding action of ice. There can be no reasonable doubt that the huge continental area of which India, South Africa,