CHAPTER VII.

DRAINAGE SYSTEMS AND THE GLACIAL PERIOD.

We will begin the consideration of this part of our subject, also, with the presentation of the salient facts in North America, since that field is simpler than any field in the Old World.

The natural drainage basins of North America east of the Rocky Mountains are readily described. The Mississippi River and its branches drain nearly all the region lying between the Appalachian chain and the Rocky Mountains and south of the Dominion of Canada and of the Great Lakes. All the southern tributaries to the Great Lakes are insignificant, the river partings on the south being reached in a very short distance. The drainage of the rather limited basin of the Great Lakes is northeastward through the St. Lawrence River, leaving nearly all of the Dominion of Canada east of the Rocky Mountains to pour its surplus waters northward into Hudson Bay and the Arctic Ocean. With the exception of the St. Lawrence River, these are essentially permanent systems of drainage. To understand the extent to which the ice of the Glacial period modified these systems, we must first get before our minds a picture of the country before the accumulation of ice began.

_Preglacial Erosion._

Reference has already been made to the elevated condition of the northern and central parts of North America at the beginning of the Glacial period. The direct proof of this preglacial elevation is largely derived from the fiords and great lake basins of the continent. The word "fiord" is descriptive of the deep and narrow inlets of the sea specially characteristic of the coasts of Norway, Denmark. Iceland, and British Columbia. Usually also fiords are connected with valleys extending still farther inland, and occupied by streams.

Fiords are probably due in great part to river erosion when the shores stood at considerably higher level than now. Slowly, during the course of ages, the streams wore out for themselves immense gorges, and were assisted, perhaps, to some extent by the glaciers which naturally came into existence during the higher continental elevation. The present condition of fiords, occupied as they usually are by great depths of sea-water, would be accounted for by recent subsidence of the land. In short, fiords seem essentially to be submerged river gorges, partially silted up near their mouths, or perhaps partially closed by terminal moraines.

It is not alone in northwestern Europe and British Columbia that fiords are found, but they characterize as well the eastern coast of America north of Maine, while even farther south, both on the Atlantic and on the Pacific coast, some extensive examples exist, whose course has been revealed only to the sounding-line of the Government survey.

The most remarkable of the submerged fiords in the middle Atlantic region of the United States is the continuation of the trough of Hudson River beyond New York Bay. As long ago as 1844 the work of the United States Coast Survey showed that there was a submarine continuation of this valley, extending through the comparatively shallow waters eighty miles or more seaward from Sandy Hook.

The more accurate surveys conducted from 1880 to 1884 have brought to our knowledge the facts about this submarine valley almost as clearly as those relating to the inland portion of it above New York city. According to Mr. A. Lindenkohl,[CA] this submarine valley began to be noticeable in the soundings ten miles southeast of Sandy Hook. The depth of the water where the channel begins is nineteen fathoms (114 feet). Ten miles out the channel has sunk ninety feet below the general depth of the water on the bank, and continues at this depth for twenty miles farther. This narrow channel continues with more or less variation for a distance of seventy-five miles, where it suddenly enlarges to a width of three miles and to a depth of 200 fathoms, or 1,200 feet, and extends for a distance of twenty-five miles, reaching near that point a depth of 474 fathoms, or 2,844 feet. According to Mr. Lindenkohl, this ravine maintains for half its length "a vertical depth of more than 2,000 feet, measuring from the top of its banks, and the banks have a nearly uniform slope of about 14°." The mouth of the ravine opens out into the deep basin of the central Atlantic.

[Footnote CA: Bulletin of the Geological Society of America, vol. i, p. 564; American Journal of Science, June, 1891.]

With little question there is brought to light in these remarkable investigations a channel eroded by the extension of the Hudson River, into the bordering shelf of the Atlantic basin at a time when the elevation of the continent was much greater than now. This is shown to have occurred in late Tertiary or post-Tertiary times by the fact that the strata through which it is worn are the continuation of the Tertiary deposits of New Jersey. The subsidence to its present level has probably been gradual, and, according to Professor Cook, is still continuing at the rate of two feet a century.

Similar submarine channels are found extending out from the present shore-line to the margin of the narrow shelf bordering the deep water of the central Atlantic running from the mouth of the St. Lawrence River, through St. Lawrence Bay, and through Delaware and Chesapeake Bays.[CB] All these submerged fiords on the Atlantic coast were probably formed during a continental elevation which commenced late in the Tertiary period, and reached the amount of from 2,000 to 3,000 feet in the northern part of the continent.

[Footnote CB: See Lindenkohl in American Journal of Science, for June, 1891.]

To this period must probably be referred also the formation of the gorge, or more properly fiord, of the Saguenay, which joins the St. Lawrence below Quebec. The great depth of this fiord is certainly surprising, since, according to Sir William Dawson, its bottom, for fifty miles above the St. Lawrence, is 840 feet below the sea-level, while the bordering cliffs are in some places 1,500 feet above the water. The average width is something over a mile.

It seems impossible to account for such a deep gorge extending so far below the sea-level, except upon the supposition of a long-continued continental elevation, which should allow the stream to form a cañon to an extent somewhat comparable with that of the cañons of the Colorado and other rivers in the far West. Then, upon the subsidence of the continent to the present level, it would remain partially or wholly submerged, as we find it at the present time. During the Glacial period it was so filled with ice as to prevent silting up. The rivers entering the Pacific Ocean, both in the United States and in British Columbia, are also lost in submerged channels extending out to the deeper waters of the Pacific basin in a manner closely similar to the Atlantic streams which have been mentioned.

During this continental elevation which preceded, accompanied, and perhaps brought on the Glacial period, erosion must have proceeded with great intensity along all the lines of drainage, and throughout the whole region which is now covered, and to a considerable extent smoothed over, by glacial deposits, and the whole country must have presented a very different appearance from what it does now.

A pretty definite idea of its preglacial condition can probably be formed by studying the appearance of the regions outside of and adjoining that which was covered by the continental glacier. The contrast between the glaciated and the unglaciated region is striking in several respects aside from the presence and absence of transported rocks and other _débris_, but in nothing is it greater than in the extent of river erosion which is apparent upon the surface. For example, upon the western flanks of the Alleghanies the regions south of the glacial limit is everywhere deeply channeled by streams. Indeed, so long have they evidently been permitted to work in their present channels that, wherever there have been waterfalls, they have receded to the very head-waters, and no cataracts exist in them at the present time. Nor are there in the unglaciated region any lakes of importance, such as characterize the glaciated region. If there have been lakes, the lapse of time has been sufficient for their outlets to lower their beds sufficiently to drain the basins dry.

On entering the glaciated area all this is changed. The ice-movement has everywhere done much to wear down the hills and fill the valleys, and, where there was _débris_ enough at command, it has obliterated the narrow gorges originally occupied by the preglacial streams. Thus it has completely changed the minor lines of superficial drainage, and in many instances has produced most extensive and radical changes in the whole drainage system of the region. In the glaciated area, channels buried beneath glaciated _débris_ are of frequent occurrence, while many of the streams which occupy their preglacial channels are flowing at a very much higher level than formerly, the lower part of the channel having been silted up by the superabundant _débris_ accessible since the glacial movement began.

_Buried Outlets and Channels._

It is easy to see how the great number of shallow lakes which frequent the glaciated region were formed by the irregular deposition of glacial _débris_, but it is somewhat more difficult to trace out the connection between the Glacial period and the Great Lakes of North America, several of which are of such depth that their bottoms are some hundreds of feet below the sea-level, Lake Erie furnishing the only exception. This lake is so shallow that it is easy to see how its basin may have been principally formed by river erosion, while it is evident that such must have been the mode of its formation, since it is surrounded by sedimentary strata lying nearly in a horizontal position.

That Lake Erie is really nothing but a "glacial mill-pond" is proved also by much direct evidence, especially that derived from the depth of the buried channels of the streams flowing into it from the south. Of these, the Cuyahoga River, which enters the lake at Cleveland, has been most fully investigated. In searching for oil, some years ago, borings were made at many places for twenty-five miles above the mouth of the river. As a result, it appeared that for the whole distance the rocky bottom of the gorge was about two hundred feet below the present bottom of the river, while the river itself is two or three hundred feet below the general level of the country, occupying a trough about half a mile in width, with steep, rocky sides. These facts indicate that at one time the river must have found opportunity to discharge its contents at a level two hundred feet below that of the present lake, while an examination of the material filling up the bottom of the gorge to its present level shows it to be glacial _débris_, thus proving that the silting up was accomplished during the Glacial period.

As the water of Lake Erie is for the most part less than one hundred feet in depth, and is nowhere much more than two hundred feet deep, it is clear that the preglacial outlet which drained it down to the level of the rocky bottom of the Cuyahoga River must have destroyed the lake altogether. Hence Ave may be certain that, before the Glacial period, the area now covered by the lake was simply a broad, shallow valley through which there coursed a single river of great magnitude, with tributary branches occupying deep gorges. Professor J. W. Spencer has shown with great probability that the old line of drainage from Lake Erie passed through the lower part of the valley of Grand River, in Canada, and entered Lake Ontario at its western extremity, and that during the great Ice age this became so completely obstructed with glacial _débris_ as to form an impenetrable dam, and to cause the pent-up water to flow through the Niagara Valley, which chanced to furnish the lowest opening.

In speaking of the present area of Lake Erie, however, as being then occupied by a river valley, we do not mean to imply that it was not afterwards greatly modified by glacial erosion; for undoubtedly this was the case, whatever views we may have as to the relative efficiency of ice and water in scooping out lake basins.

In the case of Lake Erie, we need suppose no change of level to account for the erosion of its basin, but only that, since the strata in which it is situated were deposited, time enough had elapsed for a great river to cut a gorge extending from the western end of Lake Ontario through to the present bed of Lake Erie, and that here a great enlargement of the valley was occasioned by the existence of deep beds of soft shale which could easily be worn away by a ramifying system of tributary streams. Rivers acting at present relative levels would be amply sufficient to produce the results which are here manifest.

But in the case of Lakes Ontario, Huron, Michigan, and Superior, whose depths descend considerably below the sea-level, we must suppose that they were, in the main, eroded when the continent was so much elevated that their bottoms were brought above tide-level. The depth of Lake Ontario implies the existence of an outlet more than four hundred feet lower than at present, which, of course, could exist only when the general elevation was more than four hundred feet greater than now.

The existence of an outlet at that depth seems to be proved also by the fact that at Syracuse, where numerous wells have been sunk to obtain brine for the manufacture of salt, deposits of sand, gravel, and rolled stones, four hundred and fifty feet thick, are penetrated without reaching rock. Since this lies in the basin of Lake Ontario, it follows that if the basin itself has been produced by river erosion, the land must have been of sufficient height to permit an outlet through a valley, or cañon, of the required depth, and this outlet must now be buried beneath the abundant glacial _débris_ that covers the region.

Professor Newberry, who has studied the vicinity carefully, is of the opinion that there is ample opportunity for such a line of drainage to have extended through the Mohawk Valley to the Hudson River. But, at Little Falls, a spur of the Adirondack Mountains projects into the valley, and the Archæan rocks over which the river runs are so prominent and continuous that some have thought it impossible for the requisite channel to have ever existed there. Extensive deposits of glacial _débris_, however, are found in the vicinity, especially in places some distance to the north, and in Professor Newberry's opinion the existence of a buried channel around the obstruction upon the north side is by no means improbable.

The preglacial drainage of Lake Huron has not been determined with any great degree of probability. Professor Spencer formerly supposed that it passed from the southern end of the lake through London, in the western part of Ontario, and reached the Erie basin near Port Stanley, and so augmented the volume of the ancient river which eroded the buried cañon from Lake Erie to Lake Ontario. But he now supposes, though the evidence is by no means demonstrative, that the waters of Lake Huron passed into Lake Ontario by means of a channel extending from Georgian Bay to the vicinity of Toronto.

With a fair degree of probability, the basin of Lake Superior is supposed by Professor Newberry to have been joined to that of Lake Michigan by some passage, now buried, considerably to the west of the Strait of Mackinac, and thence to have had an outlet southward from the vicinity of Chicago directly into the Mississippi River. Of this there is considerable evidence furnished by deeply buried channels which have been penetrated by borings in various places in Kankakee, Livingston, and McLean Counties, Illinois; but the whole area extending from Lake Michigan to the Mississippi is so deeply covered with glacial _débris_ that the surface of the country gives no satisfactory indication of the exact lines of preglacial drainage.

Some of the most remarkable instances of ancient river channels buried by the glacial deposits have been brought to light in southwestern Ohio, where there has been great activity in boring for gas and oil. At St. Paris, Champaign County, for example, in a locality where the surface of the rock near by was known to be not far below the general level, a boring was begun and continued to a depth of more than five hundred feet without reaching rock, or passing out of glacial _débris_.

Many years ago Professor Newberry collected sufficient facts to show that pretty generally the ancient bed of the Ohio River was as much as 150 feet below that over which it now flows. During a continental elevation the erosion had proceeded to that extent, and then the channel had been silted up during the Glacial period with the abundant material carried down by the streams from the glaciated area. One of the evidences of the preglacial depth of the channel of the Ohio was brought to light at Cincinnati, where "gravel and sand have been found to extend to a depth of over one hundred feet below low-water mark, and the bottom of the trough has not been reached." In the valley of Mill Creek, also, "in the suburbs of Cincinnati, gravel and sand were penetrated to the depth of 120 feet below the stream before reaching rock." But from the general appearance of the channel, Professor J. F. James was led to surmise that a rock bottom extended all the way across the present channel of the Ohio, between Price Hill and Ludlow, Ky., a short distance below Cincinnati, which would preclude the possibility of a preglacial outlet at the depth disclosed in that direction. Mr. Charles J. Bates (who was inspector of the masonry for the Cincinnati Southern Railroad while building the bridge across the Ohio at this point) informs me that Mr. James's surmise is certainly correct, and that his "in all probability" may be displaced by "certainly," since the bedded rocks supposed by Professor James to extend across the river a few feet below its present bottom were found by the engineers to be in actual existence.

In looking for an outlet for the waters of the upper Ohio which should permit them to flow off at the low level reached in the channel at Cincinnati, Professor James was led to inspect the valley extending up Mill Creek to the north towards Hamilton, where it joins the Great Miami. The importance of Mill Creek Valley is readily seen in the fact that the canal and the railroads have been able to avoid heavy grades by following it from Cincinnati to Hamilton. As a glance at a map will show, it is also practically but a continuation of the northerly course pursued by the Ohio for twenty miles before reaching Cincinnati. This, therefore, was a natural place in which to look beneath the extensive glacial _débris_ for the buried channel of the ancient Ohio, and here in all probability it has been found. The borings which have been made in Milk Creek Valley north of Cincinnati, show that the bedded rock lies certainly thirty-four feet below the low-water mark of the Ohio just below Cincinnati, while at Hamilton, twenty-five miles north of Cincinnati, where the valley of the Great Miami is reached, the bedded rock of the valley lies as much as ninety feet below present low-water mark in the Ohio.

Other indications of the greater depth of the preglacial gorge of the Ohio are abundant. "At the junction of the Anderson with the Ohio, in Indiana, a well was sunk ninety-four feet below the level of the Ohio before rock was found." At Louisville, Ky., the occurrence of falls in the Ohio seemed at first to discredit the theory in question, but Professor Newberry was able to show that the falls at Louisville are produced by the water's being now compelled to flow over a rocky point projecting from the north side into the old valley, while to the south there is ample opportunity for an old channel to have passed around this point underneath the city on the south side. The lowlands upon which the city stands are made lands, where glacial _débris_ has filled up the old channel of the Ohio.

Above Cincinnati the tributaries of the Ohio exhibit the same phenomena. At New Philadelphia, Tuscarawas County, the borings for salt-wells show that the Tuscarawas is running 175 feet above its ancient bed. The Beaver, at the junction of the Mahoning and Shenango, is flowing 150 feet above the bottom of its old trough, as is demonstrated by a large number of oil-wells bored in the vicinity. Oil Creek is shown by the same proofs to run from 75 to 100 feet above its old channel, and that channel had sometimes vertical and even overhanging walls.[CC]

[Footnote CC: Geological Survey of Ohio, vol. ii, pp. 13, 14.]

The course of preglacial drainage in the upper basin of the Alleghany River is worthy of more particular mention. Mr. Carll, of the Pennsylvania Geological Survey, has adduced plausible reasons for believing that previous to the Glacial period the drainage of the valley of the upper Alleghany north of the neighbourhood of Tidioute, in Warren County, instead of passing southward as now, was collected into one great stream flowing northward through the region of Cassadaga Lake to enter the Lake Erie basin at Dunkirk, N. Y. The evidence is that between Tidioute and Warren the present Alleghany is shallow, and flows over a rocky basin; but from Warren northward along the valley of the Conewango, the bottom of the old trough lies at a considerably lower level, and slopes to the north. Borings show that in thirteen miles the slope of the preglacial floor of Conewango Creek to the north is 136 feet. The actual height above tide of the old valley floor at Fentonville, where the Conewango crosses the New York line, is only 964 feet; while that of the ancient rocky floor of the Alleghany at Great Bend, a few miles south of Warren, was 1,170 feet. Again, going nearer the head-waters of the Alleghany, in the neighbourhood of Salamanca, it is found that the ancient floor of the Alleghany is, at Carrollton, 70 feet lower than the ancient bed of the present stream at Great Bend, about sixty miles to the south; while at Cole's Spring, in the neighbourhood of Steamburg, Cattaraugus County, N. Y., there has been an accumulation of 315 feet of drift in a preglacial valley whose rocky floor is 155 feet below the ancient rocky floor at Great Bend. Unless there has been a great change in levels, there must, therefore, have been some other outlet than the present for the waters collecting in the drainage basin to the north of Great Bend.[CD]

[Footnote CD: For a criticism of Mr. Carll's views, see an article on Pleistocene Fluvial Planes of Western Pennsylvania, by Mr. Frank Leverett, in American Journal of Science, vol. xlii, pp. 200-212.]

While there are numerous superficial indications of buried channels running towards Lake Erie in this region, direct exploration has not been made to confirm these theoretical conclusions. In the opinion of Mr. Carll, Chautauqua Lake did not flow directly to the north, but, passing through a channel nearly coincident with that now occupied by it, joined the northerly flowing stream a few miles northeast from Jamestown.[CE] It is probable, however, that Chautauqua did not then exist as a lake, since the length of preglacial time would have permitted its outlet to wear a continuous channel of great depth corresponding to that known to have existed in the Conewango and upper Alleghany.

[Footnote CE: Second Geological Survey of Pennsylvania, vol. iii.]

The foregoing are but a few of the innumerable instances where the local lines of drainage have been disturbed, and even permanently changed, by the glacial deposits. Almost every lake in the glaciated region is a witness to this disturbance of the established lines of drainage by glacial action, while in numerous places where lakes do not now exist they have been so recently drained that their shore-lines are readily discernible.

An interesting instance of the recent disappearance of one of these glacial lakes is that of Runaway Pond, in northern Vermont. In the early part of the century the Lamoille River had its source in a small lake in Craftsbury, Orleans County. The sources of the Missisquoi River were upon the same level just to the north, and the owner of a mill privilege upon this latter stream, desiring to increase his power by obtaining access to the water of the lake, began digging a ditch to turn it into the Missisquoi, but no sooner had he loosened the thin rim of compact material which formed the bottom and the sides of the inclosure, than the water began to rush out through the underlying and adjacent quicksands. This almost instantly enlarged the channel, and drained the whole body of water oft 3 in an incredibly short time. As a consequence, the torrent went rushing down through the narrow valley, sweeping everything before it; and nothing but the unsettled condition of the country prevented a disaster like that which occurred in 1889 at Johnstown, Pa. Doubtless there are many other lakes held in position by equally slender natural embankments. Artificial reservoirs are by no means the only sources of such danger.

The buried channel of the old Mississippi River in the vicinity of Minneapolis is another instructive example of the instability of many of the present lines of drainage. The gorge of the Mississippi River extending from Fort Snelling to the Falls of St. Anthony at Minneapolis is of post-glacial origin. One evidence of this is its narrowness when contrasted with the breadth of the valley below Fort Snelling. Below this point the main trough of the Mississippi has a width of from two to eight miles, and the faces of the bluffs on either side show the marks of extreme age. The tributary streams also have had time to wear gorges proportionate to that of the main stream, and the agencies which oxidise and discolor the rocks have had time to produce their full effects. But from Fort Snelling up to Minneapolis, a distance of about seven miles, the gorge is scarcely a quarter of a mile in width, and the faces of the high, steep bluffs on either side are remarkably fresh looking by comparison with those below; while the tributary gorges, of which that of the Minnehaha River is a fair specimen, are very limited in their extent.

Upon looking for the cause of this condition of things we observe that the broad trough of the Mississippi River, which had characterised it all the way below Fort Snelling, continues westward, without interruption, up the valley of the present Minnesota River, and, what seems at first most singular, it does not cease at the sources of the Minnesota, but, through Lake Traverse and Big Stone Lake, is continuous with the trough of the Red River of the North.

Deferring, however, for a little the explanation of this, we will go back to finish the history of the preglacial channel around the Falls of St. Anthony. As early as the year 1876 Professor N. H. Winchell had collected sufficient evidence from wells, one of which had been sunk to a depth of one hundred and seventy-five feet, to show that the preglacial course of the stream corresponding to the present Mississippi River ran to the west of Minneapolis and of the Falls of Minnehaha, and joined the main valley some distance above Fort Snelling, as shown in the accompanying map.

This condition of things was at one time very painfully brought to the notice of the citizens of Minneapolis. A large part of the wealth of the city at that time consisted of the commercial value of the water-power furnished by the Falls of St. Anthony. To facilitate the discharge of the waste water from their wheels, some mill-owners dug a tunnel through the soft sandstone underlying the limestone strata over which the river falls; but it very soon became apparent that the erosion was proceeding with such rapidity that in a few years the recession of the falls would be carried back to the preglacial channel, when the river would soon scour out the channel and destroy their present source of wealth. The citizens rallied to protect their property, and spent altogether as much as half a million dollars in filling up the holes that had been thoughtlessly made; but so serious was the task that they were finally compelled to appeal for aid to the United States Government. Permanent protection was provided by running a tunnel, some ways back from the falls, completely across the channel, through the soft sandstone underlying the limestone, and filling this up with cement hard enough and compact enough to prevent the further percolation of the water from above.

_Ice-Dams._

The foregoing changes in lines of drainage due to the Glacial period were produced by deposits of earthy material in preglacial channels. Another class of temporary but equally interesting changes were produced by the ice itself acting directly as a barrier.

Many such lakes on a small scale are still in existence in various parts of the world. The Merjelen See in Switzerland is a well-known instance. This is a small body of water held back by the great Aletsch Glacier, in a little valley leading to that of the Fiesch Glacier, behind the Eggischorn. At irregular intervals the ice-barrier gives way, and allows the water to rush out in a torrent and flood the valley below. Afterwards the ice closes up again, and the water reaccumulates in preparation for another flood.

Other instances in the Alps are found in the Mattmark See, which fills the portion of the Saas Valley between Monte Rosa and the Rhône. This body of water is held in place by the Allalin Glacier, which here crosses the main valley. The Lac du Combal is held back by the Glacier de Miage at the southern base of Mont Blanc. "A more famous case is that of the Gietroz Glacier in the valley of Bagnes, south of Martigny. In 1818 this lake had grown to be a mile long, and was 700 feet wide and 200 feet deep. An attempt was made to drain it by cutting through the ice, and about half the water was slowly drawn off in this way; but then the barrier broke, and the rest of the lake was emptied in half an hour, causing a dreadful flood in the valley below. In the Tyrol, the Vernagt Glacier has many times caused disastrous floods by its inability to hold up the lake formed behind it. In the northwestern Himalaya, the upper branches of the Indus are sometimes held back in this way. A noted flood occurred in 1835; it advanced twenty-five miles in an hour, and was felt three hundred miles down-stream, destroying all the villages on the lower plain, and strewing the fields with stones, sand, and mud."[CF]

[Footnote CF: Professor William M. Davis in. Proceedings of the Boston Society of Natural History, vol. xxi, pp. 350, 351.]

In Greenland such temporary obstructions are frequent, forming lakes of considerable size. Instances occur, in connection with the Jakobshavn and the Frederickshaab Glaciers, and in the North Isortok and Alangordlia Fiords.

Frequently, also, bodies of water of considerable size are found in depressions of the ice itself, even at high levels. I have myself seen them covering more than an acre, and as much as a thousand feet above the sea-level, upon the surface of the Muir Glacier, Alaska. They are reported by Mr. I. C. Russell[CG] of larger size and at still higher elevations upon the glaciers radiating from Mount St. Elias; while the explorers of Greenland mention them of impressive size upon the surface of its continental ice-sheet.

[Footnote CG: See National Geographic Magazine, vol. iii, pp. 116-120.]

With these facts in mind we can the more readily enter into the description which will now be given of some temporary lakes of vast size which were formed by direct ice-obstructions during portions of the period.

One of the most interesting of these is illustrated upon the accompanying map, which will need little description.

While tracing the boundary-line of the glaciated area in the Mississippi Valley during the summer of 1882, I discovered the existence of unmistakable glacial deposits in Boone County, Kentucky, across the Ohio River, from Cincinnati.[CH]; These deposits were upon the height of land 550 feet above the Ohio River, or nearly 1,000 feet above the sea, which is about the height of the water-shed between the Licking and Kentucky Rivers. As the Ohio River occupies a trough of erosion some hundreds of feet in depth, and extending all the way from this point to the mountains of western Pennsylvania, it would follow that the ice which conveyed boulders across the Ohio River at Cincinnati, and deposited them upon the highlands between the Licking and Kentucky Rivers, would so obstruct the channel of the Ohio as to pond the water back, and hold it up to the level of the lowest pass into the Ohio River farther down. Direct evidences of obstruction by glacial ice appear also for a distance of fifty or sixty miles, extending both ways, from Cincinnati.

[Footnote CH: The existence of portions of this evidence had previously been pointed out by Mr. Robert B. Warder and Dr. George Sutton (see Geological Reports of Indiana, 1872 and 1878).]

The consequences connected with this state of things are of the most interesting character.

The bottom of the Ohio River at Cincinnati is 432 feet above the sea-level. A dam of 550 feet would raise the water in its rear to a height of 982 feet above tide. This would produce a long, narrow lake, of the width of the eroded trough of the Ohio, submerging the site of Pittsburg to a depth of 281 feet, and creating slack water up the Monongahela nearly to Grafton, West Virginia, and up the Alleghany as far as Oil City. All the tributaries of the Ohio would likewise be filled to this level. The length of this slack-water lake in the main valley, to its termination up either the Alleghany or the Monongahela, was not far from one thousand miles. The conditions were also peculiar in this, that all the northern tributaries rose within the southern margin of the ice-front, which lay at varying distances to the north. Down these there must have poured during the summer months immense torrents of water to strand boulder-laden icebergs on the summits of such high hills as were lower than the level of the dam.

Naturally enough, this hypothesis of a glacial dam at Cincinnati aroused considerable discussion, and led to some differences of opinion. Professors I. C. White and J. P. Lesley, whose field work has made them perfectly familiar with the upper Ohio and its tributaries, at once supported the theory, with a great number of facts concerning certain high-level terraces along the Alleghany and Monongahela Rivers; while additional facts of the same character have been brought to light by myself and others. In general, it may be said that in numerous places terraces occur at a height so closely corresponding to that of the supposed dam at Cincinnati, that they certainly strongly suggest direct dependence upon it. The upward limit of these terraces in the Monongahela River is 1,065 feet, and they are found in various places in situations which indicate that they were formed in still water of such long standing as would require an obstruction below of considerable permanence.

One of the most decisive cases adduced by Professor White occurs near Morgantown, in West Virginia, of which he gives the following description:

"Owing to the considerable elevation--275 feet--of the fifth terrace above the present river-bed in the vicinity of Morgantown, its deposits are frequently found far inland from the Monongahela, on tributary streams. A very extensive deposit of this kind occurs on a tributary one mile and a half northeast of Morgantown; and the region, which includes three or four square miles, is significantly known as the 'Flats.' The elevation of the 'Flats' is 275 feet above the river, or 1,065 feet above tide. The deposits on this area consist almost entirely of clays and fine, sandy material, there being very few boulders intermingled. The depth of the deposit is unknown, since a well sunk on the land of Mr. Baker passed through alternate beds of clay, fine sand, and muddy trash, to a depth of sixty-five feet without reaching bed-rock. In some portions of the clays which make up this deposit, the leaves of our common forest-trees are found most beautifully preserved.

"At Clarksburg, where the river unites with Elk Creek, there is a wide stretch of terrace deposits, and the upper limit is there about 1,050 feet above tide, or only 130 feet above low-water (920 feet); while at Weston, forty miles above (by the river), these deposits cease at seventy feet above low water, which is there 985 feet above tide. It will thus be observed that the upper limit of the deposits retains a practical horizontality from Morgantown to Weston, a distance of one hundred miles, since the upper limit has the same elevation above tide (1,045 to 1,065 feet) at every locality.

"These deposits consist of rounded boulders of sandstone, with a large amount of clay, quicksand, and other detrital matter. The country rock in this region consists of the soft shales and limestones of the upper coal-measures, and hence there are many 'low gaps' from the head of one little stream to that of another, especially along the immediate region of the river; and in every case the summits of these divides, where they do not exceed an elevation of 1,050 feet above tide, are covered with transported or terrace material; but where the summits go more than a few feet above that level we find no transported material upon them, but simply the decomposed country rock."

Other noteworthy terraces naturally attributable to the Cincinnati ice-dam are to be found in the valley of the Kanawha, in West Virginia, and one of special significance on the pass between the valleys of the Ohio and Monongahela, west of Clarksburg, West Virginia. According to Professor White, there is at this latter place "a broad, level summit, having an elevation of 1,100 feet, in a gap about 300 feet below the enclosing hills. This gap, or valley, is covered by a deposit of fine clay. The cut through it is about thirty feet, and one can observe the succession of clays of all kinds and of different colours, from yellow on the surface down to the finest white potter's clay at the level of the railway, where the cut reaches bed-rock, thus proving that the region has been submerged."[CI]

[Footnote CI: Bulletin of the Geological Society of America, vol. i, p. 478.]

Another crucial case I have myself described at Bellevue, in the angle of the Ohio and Alleghany Rivers, about five miles below Pittsburg, where the gravel terrace is nearly 300 feet above the river, making it about 1,000 feet above the sea. A significant circumstance connected with this terrace is that not only does its height correspond with that of the supposed obstruction at Cincinnati, but it contains many pebbles of Canadian origin, which could not have got into the valley of the Alleghany before the Glacial period, and could only have reached their present position by being brought down the Alleghany River upon floating ice, or by the ordinary movement of gravel along the margin of a river. Thus this terrace, while corresponding closely with the elevation of those on the Monongahela River, is directly connected with the Glacial period, and furnishes a twofold argument for our theory.

A still stronger case occurs at Beech Flats, at the head of Ohio Brush Creek, in the northwest corner of Pike County, Ohio, where, at an elevation of about 950 feet above the sea, there is an extensive flat-topped terrace just in front of the terminal moraine. This terrace consists of fine loam, such as is derived from the glacial streams, but which must have been deposited in still water. The occurrence of still water at that elevation just in front of the continental ice-sheet is best accounted for by the supposed dam at Cincinnati. Indeed, it is extremely difficult to account for it in any other way.

There are, however, two other methods of attempting to account for the class of facts above cited in support of the ice-dam theory, of which the most plausible is, that in connection with the Glacial period there was a subsidence of the whole region to an extent of 1,100 feet.

The principal objection heretofore alleged against this supposition is that there are not corresponding signs of still-water action at the same level on the other side of the Alleghany Mountains. This will certainly be fatal to the subsidence theory, if it proves true. But it is possible that sufficient search for such marks has not yet been made on the eastern side of the mountains.

The other theory to account for the facts is, that the terraces adduced in proof of the Cincinnati ice-dam were left by the streams in the slow process of lowering their beds from their former high levels. This is the view advocated by President T. C. Chamberlin. But the freshness of the leaves and fragments of wood, such as were noted by Professor White at Morgantown, and the great extent of fine silt occasionally resting upon the summits of the water-sheds, as described above, near Clarksburg, bear strongly against it. Furthermore, to account for the terrace described at Bellevue, which contains Canadian pebbles, President Chamberlin is compelled to connect the deposit with his hypothetical first Glacial epoch, and to assume that all the erosion of the Alleghany and Monongahela Rivers, and indeed of the whole trough of the Ohio River, took place in the interval between the "first" and the "second" Glacial periods (for he would connect the glacial deposits upon the south side of the river at Cincinnati with the first Glacial epoch)--all of which, it would seem, is an unnecessary demand upon the forces of Nature, when the facts are so easily accounted for by the simple supposition of the dam at Cincinnati.[CJ]

[Footnote CJ: See matter discussed more at length in the lee Age, pp. 326-350, 480-500; Bulletin of the United States Geological Survey, No. 58, pp. 76-100; Popular Science Monthly, vol. xlv, pp. 184-199. _Per contra_, Mr. Frank Leverett, in American Geologist, vol. x, pp. 18-24.]

We have already described[CK] the various temporary lakes and lines of drainage caused by the direct obstruction of the northward outlets to the basin of the Great Lakes. In connection with the map, it will be unnecessary to do anything more here than add a list of such temporary southern outlets from the Erie-Ontario basin.[CL] The first is at Fort Wayne, Indiana, through a valley connecting the Maumee River basin with that of the Wabash. The channel here is well defined, and the high-level gravel terraces down the Wabash River are a marked characteristic of the valley. The elevation of this col above the sea is 740 feet. Similar temporary lines of drainage existed from the St. Mary's River to the Great Miami, at an elevation of 942 feet; from the Sandusky River to the Scioto, through the Tymochtee Gap, at an elevation of 912 feet; from Black River to the Killbuck (a tributary of the Muskingum) through the Harrisville Gap, at 911 feet; from the Cuyahoga into the Tuscarawas Valley, through the Akron Gap, at 971 feet; from Grand River into the Mahoning, through the Orwell Gap, 938 feet; from Cattaraugus Creek, N. Y., into the Alleghany Valley through the Dayton Gap, about 1,300 feet; between Conneaut Creek and Shenango River, at Summit Station, 1,141 feet; from the Genesee River, N. Y., into the head-waters of the Canisteo, a branch of the Susquehanna, at Portageville, 1,314 feet; from Seneca Lake to Chemung River, at Horseheads, 879 feet; from Cayuga Lake to the valley of Cayuga Creek, at Spencer, N. Y., 1,000 feet; from Utica, N. Y., into the Chenango Valley at Hamilton, about 900 feet.

[Footnote CK: See pp. 92 seq., 199 _seq._]

[Footnote CL: See also accompanying map.]

Perhaps it would have been best to give this list in the reverse order, which would be more nearly chronological, since it is clear that the highest outlets are the oldest. We should then have to mention, after the Fort Wayne outlet, two others at lower levels which are pretty certainly marked by distinct beach ridges upon the south side of Lake Erie. The first was opened when the ice had melted back from the south peninsula of Michigan to the water-shed across from the Shiawassee and Grand Rivers, uncovering a pass which is now 729 feet above the sea. This continued to be the outlet of Lake Erie-Ontario until the ice had further retreated beyond the Strait of Mackinac, when the water would fall to the level of the old outlet from Lake Michigan into the Illinois River, which is a little less than 600 feet, where it would remain until the final opening of the Mohawk River in New York attracted the water in that direction, and lowered the level to that of the pass from Lake Ontario to the Mohawk at Rome.[CM]

[Footnote CM: Mr. Warren Upham, in the Bulletin of the Geological Society of America, vol. ii, p. 259.]

A study of these lines of temporary drainage during the Glacial period sheds much light upon the long lines of gravel ridges running parallel with the shores of Lake Erie and Lake Ontario. South of Lake Erie a series of four ridges of different elevations can be traced. In Lorain County, Ohio, the highest of these is 220 feet above the lake; the next 160 feet; the next 118 feet; and the lower one 100 feet, which would make them respectively 795, 755, 715, and 700 feet above tide.

These gravel ridges are evidently old beach lines, and indicate the different levels up to which the water was held by ice-obstructions across the various outlets of the drainage valley. The material in the ridges is water-worn and well assorted, and in coarseness ranges from fine sand up to pebbles several inches in diameter. The predominant material in them is of local origin. Where the rocks over which they run are sandstone, the material is chiefly sand, and where the outcropping rock is shale, the ridges consist chiefly of the harder nodules of that formation which have successfully resisted the attrition of the waves. Ordinarily these ridges are steepest upon the side facing the lake. According to Mr. Upham, who has driven over them with me, the Lake Erie ridges correspond, both in general appearance and in all other important respects, to those which he has so carefully surveyed around the shores of the ancient Lake Agassiz in Minnesota and Manitoba, an account of which will be given a little farther on in this chapter.

We are not permitted, however, to assume that there have been no changes of level since the deposition of these beaches surrounding the ancient glacial Lake Erie-Ontario. On the contrary, there appears to have been a considerable elevation towards the east and northeast in post-glacial times. The highest ridge south of Lake Erie, which at Fort Wayne is about 780 feet high, is now about 795 feet in Lorain County. The second of the ridges above-mentioned, which is about 740 feet above tide at Cleveland, Ohio, rises to 870 feet where the last traces of it have been discovered at Hamburg, N. Y. The third ridge, which is 673 feet at Cleveland, has risen to the height of 860 feet at Crittenden, about one hundred miles to the east of Buffalo, N. Y.

A similar eastern increase of elevation is discoverable in the main ridge surrounding Lake Ontario. What Professor Spencer calls the Iroquois beach, which is 363 feet above tide at Hamilton, Ontario, has risen to a height of 484 feet near Syracuse, N. Y.; while farther to the northeast, in the vicinity of Watertown, it is upwards of 800 feet above tide.

There is also a similar northward increase of elevation in the beaches surrounding the higher lands of Ontario eastward of Lake Huron and Georgian Bay.

All this indicates that at the close of the Glacial period there was a subsidence of several hundred feet in the area of greatest ice-accumulation lying to the east and north of the Great Lake region. The formation of these ridges occurred during that period of subsidence. The re-elevation which followed the disappearance of the ice of course carried with it these ridges, and brought them to their present position.[CN]

[Footnote CN: See Spencer, in Bulletin of the Geological Society of America, vol. ii, pp. 465-476.]

In returning to consider more particularly the remarkable gorge joining the Minnesota with the Red River of the North, we are brought to the largest of the glacial lakes of this class, and to the typical place in America in which to study the temporary changes of drainage produced by the ice itself daring the periods both of its advance and of its retreat.

By turning to our general map of the glaciated region of the United States,[CO] one can readily see the relation of the valley between Lake Traverse and Big Stone Lake to an area marked as the bed of what is called Lake Agassiz. During the Glacial period Brown's Valley, the depression joining these two lakes, was the outlet of an immense body of water to the north, whose natural drainage was towards Hudson Bay or the Arctic Ocean, but which was cut off, by the advancing ice, from access to the ocean-level in that direction, and was compelled to seek an exit to the south.

[Footnote CO: See page 66.]

Thus for a long period the present Minnesota River Valley was occupied by a stream of enormous dimensions, and this accounts for the great size of the trough--the present Minnesota being but an insignificant stream winding about in this deserted channel of the old "Father of Waters," and having as much room as a child of tender age would have in his parent's cast-off garments. This glacial stream has been fittingly named River Warren, after General Warren, who first suggested and proved its existence, and so we have designated it on the accompanying map of Minnesota.

Lake Traverse is fifteen miles long, and the water is nowhere more than twenty feet deep. Big Stone Lake is twenty-six miles long, and of about the same depth. Brown's Valley, which connects the two, is five miles long, and the lakes are so nearly on a level that during floods the water from Lake Traverse sometimes overflows and runs to the south as well as to the north.

The trough occupied by these lakes and valley is from one mile to one mile and a half in width and about 120 feet in depth. If we had been permitted to stand upon the bluffs overlooking it during the latter part of the Glacial period, we should have seen the whole drainage of the north passing by our feet on its way to the Gulf of Mexico. As lie follows down the valley of the Minnesota River, the observant traveller, even now, cannot fail to see in the numerous well-preserved gravel terraces the high-water marks of that stream when flooded with the joint product of the annual precipitation over the vast area to the north, and of the still more enormous quantities set free by the melting of the western part of the great Laurentide Glacier.

Numerous other deserted water-ways in the northwestern part of the valley of the Mississippi have been brought to light in the more recent geological surveys, both in the United States and in Canada. During a considerable portion of the Glacial period the Saskatchewan, the Assiniboine, the Pembina, and the Cheyenne Rivers, whose present drainage is into the Red River of the North, were all turned to the south, and their temporary channels can be distinctly traced by deserted water-courses marked by lines of gravel deposits.[CP]

[Footnote CP: For further particulars, see Ice Age, pp. 293 _et seq._]

In Dakota, Professor J. E. Todd has discovered large deserted channels on the southwestern border of the glaciated region near the Missouri River, where evidently streams must have flowed for a long distance in ice-channels when the ice still continued to occupy the valley of the James River. From these channels of ice in which the water was held up to the level of the Missouri Coteau the water debouched directly into channels with sides and bottom of earthy material, which still show every mark of their former occupation by great streams.[CQ]

[Footnote CQ: For particulars, see Ice Age, p. 292.]

In Minnesota, also, there is abundant evidence that while the northeastern part of the valley from Mankato to St. Paul was occupied by ice, the drainage was temporarily turned directly southward across the country through Union Slough and Blue Earth River into the head-waters of the Des Moines River in Iowa.

_Ancient River Terraces._

The interest of the whole inquiry respecting the relation of man to the Glacial period in America concentrates upon these temporary lines of southern drainage. Wherever they existed, the swollen floods of the Glacial period have left their permanent marks in the deposition of extensive gravel terraces. The material thus distributed is derived largely from the glacial deposits through which they run and out of which they emerge. While the height of the terraces depended upon various conditions which must be studied in detail, in general it may be said that it corresponds pretty closely with the extent of the area whose drainage was turned through the channel during the prevalence of the ice. The height of the terraces and the coarseness of the material seem also to have been somewhat dependent upon the proximity of their valleys to the areas of most vigorous ice-action, and this, in turn, seems to lie in the rear of the moraines which President Chamberlin has attributed to the second Glacial epoch. Southward from this belt of moraines the terraces uniformly and gradually diminish both in height and in the coarseness of their gravel, until they finally disappear in the present flood-plain of the Mississippi River.

An interesting illustration of this principle is to be observed in the continuous valley of the Alleghany and Ohio Rivers. The trough of this valley was reached by the continental glacier at only a few points, the ice barely touching it at Salamanca, N. Y., Franklin, Pa., and Cincinnati, Ohio. But throughout its whole length the ice-front was approximately parallel to the valley, and occupied the head-waters of nearly all its tributaries. Now, wherever tributaries which could be fed by glacial floods, enter the trough of the main stream, they brought down an excessive amount of gravel, and greatly increased the size of the terrace in the trough itself, and from the mouth of each such tributary to that of the next one below there is a gradual decrease in the height of the terrace and in the coarseness of the material.

This law is illustrated with special clearness in Pennsylvania between Franklin and Beaver. Franklin is upon the Alleghany River, at the last point where it was reached directly by the ice. Below this point no tributary reaches it from the glaciated region, and none such reaches the Ohio after its junction with the Alleghany until we come to the mouth of Beaver Creek, about twenty-five miles below Pittsburg.

But at this point the Ohio is joined by a line of drainage which emerges from the glaciated area only ten or twelve miles to the north, and whose branches occupy an exceptionally large glaciated area. Accordingly, there is at Beaver a remarkable increase in the size of the glacial terrace on the Ohio. In the angle down-stream between the Beaver and the Ohio there is an enormous accumulation of granitic pebbles, many of them almost large enough to be called boulders, forming the delta terrace, upon which the city is built and rising to a height of 135 feet above the low-water mark in the Ohio. In striking confirmation of our theory, also, the terrace in the Ohio Valley upon the upper side of Beaver Creek is composed of fine material, largely derived from local rocks and containing but few granitic pebbles.

From the mouth of Beaver Creek, down the Ohio, the terrace is constant (sometimes upon one side of the river and sometimes upon the other), but, according to rule, the material of which it is composed gradually grows finer, and the elevation of the terrace decreases. According to rule, also, there is a notable increase in the height of the terrace below each affluent which enters the river from the glaciated region. This is specially noticeable below Marietta, at the mouth of the Muskingum, whose head-waters drain an extensive portion of the glaciated area. From the mouth of the Little Beaver to this point the tributaries of the Ohio are all small, and none of them rise within the glacial limit. Hence they could contribute nothing of the granitic material which enters so largely into the formation of the river terrace; but below the mouth of the Muskingum the terrace suddenly ascends to a height of nearly one hundred feet above low-water mark.

Again, at the mouth of the Scioto at Portsmouth, there is a marked increase in the size of the terrace, which is readily accounted for by the floods which came down the Scioto Valley from the glaciated region. The next marked increase is at Cincinnati, just below the mouth of the Little Miami, whose whole course lay in the glaciated region, and whose margin is lined by very pronounced terraces. At Cincinnati the upper terrace upon which the city is built is 120 feet above the flood-plain.

Twenty-five miles farther down the river, near Lawrenceburg, these glacial terraces are even more extensive, the valley being there between three and four miles wide, and being nearly filled with gravel deposits to a height of 112 feet above the flood-plain. Below this point the terraces gradually diminish in height, and the material becomes finer and more water-worn, until it merges at last in the flood-plain of the Mississippi. The course of the Wabash River is too long to permit it to add materially to the size of the terraces which characterise the broader valley of the Ohio below the Illinois line.

It is in terraces such as these just described that we find the imbedded relics of man which definitely connect him with the great Ice age. These have now been found in the glacial terraces of the Delaware River at Trenton, N. J.; in similar terraces in the valley of the Tuscarawas River at New Comerstown, and in the valley of the Little Miami at Loveland and Madisonville, in Ohio; on the East Fork of White River, at Medora, Ind.; and still, again, at Little Falls, in the trough of the Mississippi, some distance above Minneapolis, Minn.

I append a list of the points at which various streams from the Atlantic Ocean to the Mississippi River emerge from the glacial boundary, and below which the terraces are specially prominent. Of course, with the retreat of the ice, the formation of the terraces continued northward in the glaciated area to a greater or less distance, according to the extent of the valley or to the length of time during which the drainage was temporarily turned into it. These points of emergence are: In the Delaware Valley, at Belvidere, N. J.; in the Susquehanna, at Beach Haven, Pa.; in the Conewango, at Ackley, Warren County; in Oil Creek, above Titusville: in French Creek, a little above Franklin; in Beaver Creek, at Chewtown, Lawrence County; on the Middle Fork of Little Beaver, near New Lisbon, Ohio; on the east branch of Sandy Creek, at East Rochester, Columbiana County; on the Nimishillin, at Canton, Stark County; on the Tuscarawas, at Bolivar; on Sugar Creek, at Beech City; on the Killbuck, at Millersburg, Holmes County; on the Mohican, near the northeast corner of Knox County; on the Licking River, at Newark; on Jonathan Creek, Perry County; on the Hocking, at Lancaster; on the Scioto, at Hopetown, just above Chillicothe; on Paint Creek, and its various tributaries, between Chillicothe and Bainbridge; and on the Wabash, above New Harmony, Ind.; to which may be added the Ohio River itself, at its junction with the Miami, near Lawrenceburg.

Another class of terraces having most interesting connection with the Glacial period is found in the arid basins west of the Rocky Mountains. Over wide areas in Utah and Nevada the evaporation now just balances the precipitation, and all the streams disappear in shallow bodies of salt water of moderate dimensions, of which Great Salt Lake in Utah, and Mono, Pyramid, and North Carson Lakes in Nevada, are the most familiar examples. These occupy the lowest sinks of enclosed basins of great depth.

But there is abundant evidence that in consequence of the increased precipitation and diminished evaporation of the Glacial period one of these basins was filled to the brim and the other to a depth of several hundred feet. These former enlargements have been named after the first explorers of the region, Captains Lahontan and Bonneville, and are shown on the accompanying sketch map by the shading surrounding the existing lakes.

Lake Lahontan has been carefully studied by Mr. I. C. Russell, and has been found to extend from the boundary of Oregon to latitude 38° 30' south, a distance of two hundred and sixty miles. The Central Pacific Railroad runs through its dried-up bed from Golconda to Wadsworth, a distance of one hundred and sixty-five miles. The terraces of the former lake are distinctly traceable at a height of 700 feet above the present level of Lake Mono.

Lake Bonneville, whose present representative is Great Salt Lake, is the subject of a recent monograph by Mr. G. K. Gilbert, from which it appears that this ancient body of water occupied 19,750 square miles--an area about ten times that of the present lake. At the time of its maximum extension its depth was 1,000 feet, while Great Salt Lake ranges only from fifteen to fifty feet in depth.

The pass through which the discharge finally took place is at Red Rock, on the Utah and Northern Railroad, at the head of Cache Valley on the south and the lower part of Marsh Creek Valley on the north. During the long period preceding and accompanying the gradual rise of water in the Utah basin to the level of the highest terrace, Marsh Creek (the upper portion of which comes from the mountains on the east and turns at right angles) had been at work depositing a delta of loose material in the col which separates the two valleys. This deposit rested upon a stratum of limestone at the bottom of the pass, and covered it with sand, clay, and gravel to a depth of 375 feet. Thus, when the water was approaching its upper level, the only barrier to prevent its escape was this unstable accumulation of loose material upon top of the rock. It would have required, therefore, no prophet's eye to predict that the way was preparing for a tremendous _débâcle_.

The critical point at length was reached. After remaining nearly at the elevation of the pass for a considerable period, during which the 1,000-foot shore-line was formed, the crisis came when the water began to flow northward towards Snake River. Once begun in such loose material, the channel rapidly enlarged until soon a stream equal to Niagara, and at times probably much larger, was pouring northward through the valley heretofore occupied by the insignificant rivulets of Marsh Creek and the Port Neuf. It is impossible to tell how rapidly the loose barrier wore away, but there is abundant evidence in the valley below that not only the present channel of the lower part of Marsh Creek, but the whole bottom of the valley for a mile or more in width, was for a considerable time covered by a rapid stream from ten to twenty feet in depth, and descending at the rate of thirteen feet to the mile.

The continuance of this flood was dependent upon the amount of water to be discharged, which, as we have seen, was that contained in an area of 20,000 square miles, with a depth of 375 feet. A stream of the size of Niagara would occupy about twenty-five years in the discharge of such a mass, and this may fairly be taken as a measure of the time through which it lasted. When the loose material lying above the strata of limestone in Red Rock Pass had been washed away, the lake then continued at that level for an indefinite period, with an overflow regulated by the annual precipitation of the drainage basin. This stage of the lake, during which it occupied 13,000 square miles and was 625 feet above its present level, is also marked by an extensive and persistent shore-line all around the basin. But, finally, the balance again turned when the evaporation exceeded the precipitation, and the vast body of water has since dwindled to its present insignificant dimensions.

My own interest in this discovery of Mr. Gilbert is enhanced by the explanation it gives of a phenomenon in the Snake River Valley which I was unable to solve when on the ground in 1890. The present railroad town of Pocatello is situated just where this flood emerged from the narrower valley of Marsh Creek and the Port Neuf, and spread itself out upon the broad plain of the Snake River basin. The southern edge of the plain upon which the city is built is a vast boulder-bed covered with a thin stratum of sand and gravel. Everywhere, in sinking wells and digging ditches on the vacant lots and in the streets of the city, water-worn boulders of a great variety of material and sometimes three or four feet in diameter are encountered. I was debarred from regarding this as a terminal moraine, both by the water-worn character of the boulders and by the absence of any sign of ice-action in the surrounding mountains, and I was equally debarred from attributing it to any ordinary stream of water, both by the size of the boulders and the fact that for a mile or more up the Port Neuf Valley there is an intervale, forty or fifty feet below the surface at Pocatello, and occupying the whole width of the valley, in which there is only gravel and fine sand, through which the present Port Neuf pursues a meandering course. The upper end of this short intervale is bounded by the terminus of a basaltic stream which had flowed down the valley and filled it to a considerable depth, but had subsequently been much eroded by violent water-action.

In the light of Mr. Gilbert's discoveries, however, everything is clear. The tremendous _débâcle_ which he has brought within the range of scientific vision would naturally produce just the condition of things which is so puzzling at Pocatello. Coming down through the restricted channel with sufficient force to roll along boulders of great size and to clear them all out from the upper portion of the valley, the torrent would naturally deposit them where the current was first checked, a mile below the lava cliffs. The plunge of the water over these cliffs would keep a short space below clear from boulders, and the more moderate stream of subsequent times would fill in the depression with the sand and gravel now occupying it.

What other effects of this remarkable outburst may be traced farther down in the Snake River Valley I cannot say, but it will be surprising if they do not come to light and help to solve some of the many geological problems yet awaiting us in this interesting region.

It should have been said that during the formation of the 625-foot, or so-called Provo shore-line, glaciers descended from the cañons on the west flank of the Wahsatch Mountains, and left terminal moraines to mark the coincidence of the Glacial period with that stage of the enlargement of the lake. Evidences of a similar coincidence are to be found on the high-level terraces surrounding Lake Mono, to which glaciers formerly descended from the western flanks of the Sierra Nevada.

The ancient shore-lines surrounding Lakes Bonneville and Lahontan bear evidence also of various other episodes in the Glacial period. Evidently there were two periods of marked increase in the size of the lakes, with an arid period intervening. During the first rise the level of Bonneville attained to within ninety feet of the second, and numerous beaches were formed, and a large amount of yellow clay deposited. Then it seems to have been wholly evaporated, while its soluble mineral matter was precipitated, and so mingled with silt that it did not readily redissolve during the second great rise of water. Partly on this account, and partly through the influence of the outlet into the Snake River, the lake was nearly fresh during its second enlargement.

_European Facts._

In Chapter VI it came in place to mention many of the facts connected with the influence of the Glacial period upon the drainage systems of Europe. We there discussed briefly the probable influence of the ice-obstructions that extended across the mouths of the Dwina, the Vistula, the Oder, the Elbe, the Weser, and the Rhine. The drainage of the obstructed rivers in Russia was perhaps turned southward into the Caspian and Black Seas, and then assisted in forming the fertile soil of the plains in the southern part of that empire.

The obstructed drainage of the German rivers was probably turned westward in front of the ice through the Straits of Dover or across the southern part of England. This was during the climax of the Glacial period; but later, according to Dawkins, during a period in which the land of the British Isles stood about 600 feet above its present level, the streams of the eastern coast--namely, "the Thames, Medway, Humber, Tyne, and others, joined the Rhine, the Weser, and the Elbe, to form a river flowing through the valley of the ocean. In like manner, the rivers of the south of England and of the north of France formed a great river flowing past the Channel Islands due west into the Atlantic, and the Severn united with the rivers of the south of Ireland; while those to the east of Ireland joined the Dee, Mersey Ribble, and Lune, as well as those of western Scotland, ultimately reaching the Atlantic to the west of the Hebrides. The water-shed between the valleys of the British Channel and the North Sea is represented by a ridge passing due south from Folkestone to Dieppe, and that between the drainage area and the Severn and its tributaries on the one hand, and of the Irish Channel on the other, by a ridge from Holyhead westward to Dublin.

"This tract of low, undulating land which surrounded Britain and Ireland on every side consisted not merely of rich hill, valley, and plain, but also of marsh-land studded with lakes, like the meres of Norfolk, now indicated by the deeper soundings. These lakes were very numerous to the south of the Isle of Wight and off the coast of Norfolk and Suffolk."[CR]

[Footnote CR: Early Man in Britain, p. 151.]

The evidence first regarded by scientific men to be demonstrative of the formation of extensive lakes during the Glacial period by the direct influence of ice-dams exists in the Parallel Roads of Glen Roy in Scotland.

According to the description of Sir Charles Lyell, "Glen Roy is situated in the western Highlands, about ten miles north of Fort William, near the western end of the great glen of Scotland, or Caledonian Canal, and near the foot of the highest of the Grampians, Ben Nevis. Throughout nearly its whole length, a distance of more than ten miles, three parallel roads or shelves are traced along the steep sides of the mountains, each maintaining a perfect horizontality, and continuing at exactly the same level on the opposite sides of the glen. Seen at a distance they appear like ledges, or roads, cut artificially out of the sides of the hills; but when we are upon them, we can scarcely recognize their existence, so uneven is their surface and so covered with boulders. They are from ten to sixty feet broad, and merely differ from the side of the mountain by being somewhat less steep.

"On closer inspection, we find that these terraces are stratified in the ordinary manner of alluvial or littoral deposits, as may be seen at those points where ravines have been excavated by torrents. The parallel shelves, therefore, have not been caused by denudation, but by the deposition of detritus, precisely similar to that which is dispersed in smaller quantities over the declivities of the hills above. These hills consist of clay-slate, mica-schist, and granite, which rocks have been worn away and laid bare at a few points immediately above the parallel roads. The lowest of these roads is about 850 feet above the level of the sea, and the next about 212 feet higher, and the third 82 feet above the second. There is a fourth shelf, which occurs only in a contiguous valley called Glen Gluoy, which is twelve feet above the highest of all the Glen Roy roads, and consequently about 1,156 feet above the level of the sea. One only, the lowest of the three roads of Glen Roy, is continued through Glen Spean, a large valley with which Glen Roy unites. As the shelves, having no slope towards the sea like ordinary river terraces, are always at the same absolute height, they become continually more elevated above the river in proportion as we descend each valley; and they at length terminate very abruptly, without any obvious cause, or any change either in the shape of the ground or in the composition or hardness of the rocks."[CS]

[Footnote CS: Antiquity of Man, pp. 252, 253.]

Early in his career Charles Darwin studied these ancient beaches, and ascribed them to the action of the sea during a period of continental subsidence. In this view he was supported by the majority of geologists until the region was visited by Agassiz, who saw at once the true explanation. If these were really sea-beaches, similar deposits should be found at the same elevation on other mountains than those surrounding Glen Roy. Their absence elsewhere points, therefore, to some local cause, which was readily suggested to the trained eye of one like Agassiz, then fresh from the study of Alpine glaciers, who saw that these beaches were formed upon the margin of temporary lakes, held back during the Glacial period (as the Merjelen See now is) by a glacier which came out of one glen and projected itself directly across the course of another, and thus obstructed its drainage. The glacier of Glen Spean had pushed itself across Glen Roy, as the great Aletsch Glacier in Switzerland now pushes itself across the little valley behind the Eggishorn.