4. The vegetation of coal accumulated for the most part where it grew; it was not generally drifted and deposited by waves and currents. Commonly the fire clay beneath the seam is penetrated with roots, and the shale above is packed with leaves of ferns and other plants as beautifully pressed as in a herbarium. There often is associated with the seam a fossil forest, with the stumps, which are still standing where they grew, their spreading roots, and the soil beneath, all changed to stone. In the Nova Scotia field, out of seventy-six distinct coal seams, twenty are underlain by old forest grounds. The presence of fire clay beneath a seam points in the same direction. Such underclays withstand intense heat and are used in making fire brick, because their alkalies have been removed by the long-continued growth of vegetation. Fuel coal is also too pure to have been accumulated by driftage. In that case we should expect to find it mixed with mud, while in fact it often contains no more ash than the vegetal matter would furnish from which it has been compressed. [Illustration: Fig. 303. Fossil Tree Stumps of a Carboniferous Forest, Scotland] These conditions are fairly met in the great swamps of river plains and deltas and of coastal plains, such as the great Dismal Swamp, where thousands of generations of forests with their undergrowths contribute their stems and leaves to form thick beds of peat. A coal seam is a fossil peat bed. =Geographical conditions during the Pennsylvanian.= The Carboniferous peat swamps were of vast extent. A map of the Coal Measures (Fig. 260) shows that the coal marshes stretched, with various interruptions of higher ground and straits of open water, from eastern Pennsylvania into Alabama, Texas, and Kansas. Some individual coal beds may still be traced over a thousand square miles, despite the erosion which they have suffered. It taxes the imagination to conceive that the varied region included within these limits was for hundreds of thousands of years a marshy plain covered with tropical jungles such as that pictured in Figure 304. On the basis that peat loses four fifths of its bulk in changing to coal, we may reckon the thickness of these ancient peat beds. Coal seams six and ten feet thick, which are not uncommon, represent peat beds thirty and fifty feet in thickness, while mammoth coal seams fifty feet thick have been compressed from peat beds two hundred and fifty feet deep. At the same time, the thousands of feet of marine and fresh-water sediments, with their repeated alternations of limestones, sandstones, and shales, in which the seams of coal occur, prove a slow subsidence, with many changes in its rate, with halts when the land was at a stillstand, and with occasional movements upward.