Popular Science Monthly/Volume 70/February 1907/Glacial Erosion in Alaska






By Professor RALPH S. TARR


WHEN Henry Gannett made the statement that "thousands of cubic miles" of rock had been removed from the fiords of southeastern Alaska by glacial erosion, and that "the relief features of this region, its mountains and its gorges partly filled by the sea, are all of glacial origin,[2] it is probable that many readers had the feeling that he had greatly exaggerated the case of glacial erosion. For my own part, I distinctly remember reading this with the feeling that, although glaciers are unquestionably capable of doing great work of erosion, it would require the most convincing evidence to satisfy me of even the approximate accuracy of this statement. Having now made four trips over a part of the route upon which Mr. Gannett based his statements, and having examined the phenomena attentively, there and elsewhere, I have the conviction that in reality his statement of the case is in close harmony with the truth. It is the purpose of this paper to state the argument upon which this conclusion is based.

It is a well-known fact that it is possible to go from Seattle to Sitka, along a series of 'Channels' 'Canals' and 'Reaches' without once entering the open ocean. In addition to this unique 'Inside Passage' of upwards of 1,000 miles, there is a maze of branches of such enormous extent that the whole system of channels has not yet been charted. Everywhere these arms of the sea are enclosed between
PSM V70 D104 Inland passage alaska result of glacial erosion.jpg

Fig. 1. Aligned Spurs, Inside Passage. Three such spurs seen on the right, the most distant one showing the change in slope. Two shown on the left, with the change in slope plainly visible in the more distant one. From such a condition as this there is every gradation to straight walled 'canals' Photograph by O. von Engeln.

mountain walls, and in many places they have the characteristics of grand fiords.

Such a topography as this has, until recently, been quite generally explained as a result of subsidence of the land, by which the lower ends of the land valleys have been drowned by the admission of the sea water into them. In this way the irregular coast of Patagonia, the fiords of Norway, and other similar coast lines have been explained.

Under ordinary conditions, the development of valleys by stream erosion produces certain characteristic features which are easily recognizable. These features are well understood by physiographers and have been fully stated on many occasions, and especially by Professor Davis, to whom, more than to any other, we owe our clear recognition of them and their application to the problems of glacial erosion.

One of these features is the cross-section of the valley, which varies in width and steepness according to the stage of its development. A young stream valley is steep-sided and gorge-like. Its width is narrow in proportion to its depth. A mature valley, having long been exposed to action of the weather, has been broadened out by the weathering back of the valley walls so that its width is great as compared with its depth. For a stream valley to pass from youth to maturity, even under the most favorable conditions, requires a great lapse of time.

The form of river valley to be expected in such a mountainous country as the coast of British Columbia and Alaska, would therefore depend largely upon the length of time that the streams had been working to cut the valleys. Had the stream action been brief, we should expect to find profound gorges; had it been long, broader valleys and the more gentle slopes of maturity. If, as is the case in Alaska, the same valleys have some of the characteristics of youth and some of maturity, a special explanation must be sought.

A second characteristic which results from the normal development of stream valleys is the accordance in grade between main and tributary streams. No matter how fast the main stream may be lowering its valley, even though it be a Colorado River, the side streams, including even weak tributaries, lower their mouths at approximately the same rate that the main stream deepens its valley. This feature is so well established as a normal condition of valley development, that it may be stated as a law that, under normal conditions of stream development, tributary valleys enter main valleys approximately at grade. That this is not the case in many instances in Alaska will be shown below.

A third feature normally developed during the formation of stream valleys is that of a somewhat winding course with overlapping spurs, alternating first on one side then on the other. Because of this

PSM V70 D105 Grenville channel through a large tributary valley.jpg

Fig. 2. Large Tributary Valley entering Grenville Channel from the East, below the Sea Level. Note steepened lower slope on left side of tributary valley. Photograph by Lawrence Martin.

PSM V70 D106 Hanging valley grenville channel inside passage.jpg

Fig. 3. Hanging Valley in Grenville Channel, Inside Passage. Waterfall of stream draining the broad, U-shaped valley seen near water in right-hand half of picture. Photograph by Lawrence Martin.

feature a view up or down such a valley is not usually very extensive, being cut off by the projecting spurs around which the stream swings. The absence of this feature in those Alaskan valleys where there is a

PSM V70 D106 Hanging valley waterfall sara island inside passage.jpg

Fig. 4 Waterfall on the Very Face of the Rock Lip of a Hanging Valley behind Sara Island, Inside Passage. Photograph by Lawrence Martin.

discordance in the other directions mentioned above, calls for explanation.

The partial submergence of a region traversed by a series of valleys with the characteristics just stated, would produce results which can be readily and accurately predicted. The line up to which the new sea level reached would be rendered irregular for two reasons. In the first place, the overlapping spurs would introduce a winding coast line in the fiords, with capes on one side opposite reentrants on the other. In the second place, since the tributary valleys joined the main valleys at grade, the sea water would enter their mouths and thus transform their lower portions to bays.

Examining the actual conditions along the Inside Passage to Alaska, we find very wide departures from this postulated result of a drowning of normal land valleys. Many of the passages are in the form of long straight 'Peaches' and 'Canals,' up and down which one can look for miles without obstruction to the view. In other cases the 'Reaches,' though not perfectly straight, have alternating projections and reentrants (Fig. 1). These, however, depart from typical overlapping spurs in two important respects. In the first place, they are much less pronounced. In the second place, instead of having a uniform slope from the crest to the tip of the spurs, they have a moderate slope above, like that of ordinary valley spurs, but terminate on the water side in a steep and even precipitous slope. They have the appearance, therefore, of being truncated valley spurs; and a view through such a channel often shows a succession of these partial spurs with the truncated faces in alignment. The general appearance of these aligned spurs suggests that some powerful rasping agent has moved through the fiord and truncated the overlapping spurs back to a fairly uniform distance.

The fiords of the Inside Passage furnish all gradations from typical overlapping spurs to aligned spurs, and to straight, smoothed 'Canals' from which all semblance of spurs has been erased. In the latter case the valley walls themselves often possess a double slope, steep and even precipitous below, more gentle above. The steepened lower slope has the appearance of having been incised in a valley whose remnant is represented by the upper more gentle slope.

In those 'Peaches' which are long and straight, and in those with aligned spurs, the tributary valleys enter the main valley at very different levels. Some, especially the larger, enter below the level of the sea, and in these cases there are bays in their mouths (Fig. 2); many others have their mouths high above the fiord level (Figs. 3, 4 and 5). Although there is no uniform height at which these side valleys enter the main trough, in general it is true that, the smaller the tributary valley, the higher its mouth lies above the main valley bottom. These are called hanging valleys because their mouths hang above the bottom
PSM V70 D108 Hanging valley grenville channel inside passage.jpg

Fig. 5. Hanging Valley, Grenville Channel, Inside Passage. The forest-covered lip is solid rock, but it looks like a dam. Doubtless if one went to the crest of this lip he would find the broad valley extending, with moderate grade up the distant mountain. Note the waterfall near center of picture. Photograph by Lawrence Martin.

of the main valley to which they are tributary, instead of entering at grade, as is normal.

Where these Alaskan hanging valleys are most typically developed, the appearance is quite remarkable. The valley wall of the long, straight 'Beach' or 'Canal' is broken by a broad U-shaped tributary valley, whose cross section, if explained by ordinary methods of valley formation, would require a long period of time for its formation. The stream occupying the hanging valley flows with moderate grade up to the point where the tributary valley is intersected by the straight wall of the main 'Reach.' Then, instead of continuing into the main valley with the same grade, it tumbles over the lip of the hanging valley and descends to the fiord in a succession of leaps, sometimes on the very face of the main valley wall (Fig. 4), sometimes in a shallow gorge (Figs. 3 and 5).

In these most typical cases, there is such an absolute discordance of conditions as to cause comment from even the most casual observers, as I had occasion to observe in many instances in sailing through the Inside Passage. The first feature to attract attention was the waterfall. It was then noticed that the stream emerged from a broad valley, far up which one could look, though without seeing its bottom (Figs. 3-5). This produced the impression that the lip of the hanging valley was really a dam across the mouth of a broad tributary valley. This deceptive appearance was so striking that, on asking fellow voyagers for an explanation of the hanging valleys, I have again and again received the answer that the mouth of the valley has been dammed and a lake formed behind it. So apparent is this explanation that the captain of the steamer stated positively that there are always lakes behind these lips.

Thus the hanging valley is so abnormal a feature that even to ordinary observers it seems to demand some special explanation. That there are lakes in some of the hanging valleys is probable; but it is not a necessary condition. The lip is not a dam; it is unconsumed rock in a valley bottom that has been left high above the main valley by exceptional conditions which have deepened the main trough. It was of course impossible to stop and go into the many hanging valleys which we passed in the Inside Passage, but farther up the coast I was able to enter such valleys and prove, what I was well aware of before, that the lip is not a dam and that lakes form no necessary part of the hanging valley condition (see Figs. 7 and 8).

Two other features of the valleys in the Inside Passage are noteworthy. One is the fact that in both the main and tributary valleys the rock walls have been smoothed and rounded by glacial action, proving the former extension of glaciers through this series of 'Reaches.' The other is the remarkably uniform cross-section of both the main and tributary valleys, as is so well illustrated in many of the accompanying

PSM V70 D109 Hanging valley on the south side of nunatak fjord.jpg

Fig. 6. A Hanging Valley on the South Side of Nunatak Fiord. This valley lies on the same side, but about a mile west of the succeeding pictures (Figs. 7, 8 and 9). The floating ice is from Nunatak glacier about four miles distant. Photograph by Lawrence Martin.

photographs. They are distinctly U-shaped with smooth and regular walls. In spite of their breadth, which is a normal characteristic of mature valleys, the enclosing walls, especially in the lower portions, are oftentimes exceedingly steep and even precipitous, a characteristic of young, not of mature, stream valleys. Thus the same valley has the characteristics of two stages of development, the breadth of maturity and the steep-sidedness of youth.

It is evident that such conditions as those which characterize so many of the valleys of the Inside Passage can not be due to normal conditions of stream valley development. The discrepancies and anomalies are altogether too numerous and striking for such an explanation.

PSM V70 D110 Rock lip of a hanging valley in nunatak fjord.jpg

Fig. 7. The Rock Lip of a Hanging Valley Just West of the Nunatak in Nunatak Fiord. There is a vertical difference of 700 feet between the camera site and the lip of the valley. Photographs 9 and 10 were taken from this lip. Photograph by O. von Engeln.

If this is true of the origin of the valley forms, it follows that the present outline of the intricate maze of channels on this coast cannot be explained as a result of the drowning of normal stream-made valleys, as has been so universally believed to be the case.

It is now quite generally admitted that some of the features which characterize the 'Reaches' of the Inside Passage do not admit of explanation as a result of normal stream work. The feature that has been most uniformly admitted, to be abnormal is that of discordance of tributary and main valleys. The explanation of this hanging valley condition as a result of glacial erosion, which this paper is supporting, is not, however, so uniformly accepted; the chief objection of those who have not yet accepted it being their belief that glaciers are incompetent to perform such great work as would be required if hanging valleys are explained in this way. In consequence of this inability to accept the conclusion that glaciers are powerful agents of erosion, a number of alternate hypotheses have been suggested, of which the following are some of the most prominent.

One of these special explanations is based upon the conception that glaciers act to protect rather than to erode. This explanation assumes that glaciers occupied and protected the tributary valleys while the main valleys were free from ice, and that, while this condition lasted, the main valleys were so deepened that, when the ice finally melted from the protected tributary valleys, they were hanging well above the overdeepened main troughs. When it is considered that thousands of hanging valleys are already known, and that in each case it was necessary for a small glacier to linger with its terminus at the very lip

PSM V70 D111 Hanging valley from rock lip at 700 feet.jpg

Fig. 8. Looking into Hanging Valley (Fig. 7) from Rock Lip at Elevation of 700 Feet. The stream flows in a small gorge at the right. The elevation in the middle background of the valley is the moraine-covered terminus of a dwindling glacier. The valley floor is all rock, and rock extends continuously across its mouth. Photograph by O. von Engeln.

of the hanging valley throughout the long period of time required to deepen the main channel, this explanation seems almost too absurd to consider. It furthermore fails to account for the aligned spurs, and, above all, for the great breadth and U-shape of the main troughs. While one might admit this as a possible cause for individual cases, it fails utterly as a general explanation.

A second hypothesis proposed, is that glacial erosion is lateral rather than vertical, and that the hanging valleys are due to the wearing back of the tributary mouths so that they are left hanging. That
PSM V70 D112 View from hanging valley towards nanatak fjord.jpg
Fig.9. From Same Site as Fig. 8, Looking Out into the Main Trough of Nunatak Fiord to which the Valley (Figs. 7 and 8) is Tributary. Gorge cut in rock lip on left, bottom being just out of sight. Walls of gorge is rock, like rest of lip. Elevation 700 feet; depth of fiord unknown; but large icebergs float freely in it. Photograph by O. von Engeln.

there is marked lateral erosion is generally admitted by all believers in glacial erosion; but that this is the dominant form of glacial erosion would require for its acceptance much better evidence than has been presented. It may fairly be asked, if there is such pronounced lateral erosion, why should there not also be vertical erosion of equal or greater amount? Even if excessive lateral erosion should be granted as a possibility, of which there is no proof, it alone would fail to account for all the conditions observed. It would fail to explain why remnants of valley spurs are left side by side with pronounced hanging valleys; for in such cases the spurs should certainly be rubbed completely away. But, even more fatal than this is the fact that if the grade of the hanging valley is projected out into the main valley, it will, in a vast number of cases, fall far short of meeting the main valley at grade. Consequently, if glacial erosion is admitted at all, the element of vertical erosion must be granted as a prominent part of the process.

A third explanation proposed for the hanging-valley condition is that of capture and diversion of tributary streams. No one would deny that the diversion of a stream by capture might leave it hanging above the valley to which it was originally tributary. But to attempt to apply such an explanation to the multitude of known cases of hanging valleys would not be so generally accepted. It would require a marvelous development of stream capture in special localities and, strangely enough, almost entirely in regions of former glaciation. Before this hypothesis could be seriously considered as a general explanation of hanging valleys, it would be necessary to account for the fact that this process has operated so extensively in glaciated regions, whereas it so rarely operates in unglaciated countries. But even if this explanation were otherwise probable for hanging valleys, it still leaves unexplained the associated phenomena of aligned spurs, steepened lower slopes and general U-shape of the main troughs.

A fourth hypothesis proposed is that of rejuvenation. By this it is assumed that the main and lateral valleys had an accordance of grade during an earlier cycle of development, but that recent uplift, or other cause, gave to the streams a new power of cutting, making them young again, or rejuvenating them. As a result of this there was rapid cutting, the main streams working much faster than the laterals and leaving them hanging. This explanation is totally inadequate for the Alaskan conditions. It fails to account for the truncated spurs; it gives no explanation of the difference in level at which the laterals are hanging; and, moreover, even if it operated, it could not possibly produce the other results observed. Such rejuvenation would not develop a broad main valley, but a narrow gorge. But, even if we were to admit, which physiographers would not, that such deepening and broadening of the main valley would be possible without corresponding deepening at the mouths of the laterals, it is inconceivable that, during all the time required for the deepening and broadening of the main trough, the lateral stream was scarcely able to even scratch the lip of the hanging valley (Fig. 4). This point may be illustrated by a specific case, taken not from the Inside Passage, but from Nunatak Fiord, a branch of the Yakutat Bay Inlet which lies about midway between Sitka and Controller Bay just southeast of Mount St. Elias. This fiord has been so recently occupied by ice that vegetation, excepting scattered annual plants, has not yet been able to take hold on the soil. The Nunatak Glacier (Fig. II) has receded up this fiord more than a mile in ten years. Unquestionably there has been powerful glacial erosion here, for the walls of the fiord are smoothed and grooved by glacial grinding, and there are no valley spurs left. Several of the valleys tributary to the fiord are hanging high above it (Figs. 6

PSM V70 D114 View across disenchantment bay from russell valley.jpg
Fig. 10. Looking Across the Mouth of Disenchantment Bay, Russell Valley (Fig. 11) on Left. This valley is hanging at about sea level. A small valley to the right of this hangs fully 1.000 feet above sea level. A somewhat larger valley in the extreme right of the picture is hanging at a level intermediate between these two. To account for such discordance by faulting would demand very complex block faulting. But the rock walls of the fiord are plainly exposed and there is no evidence of it. Photograph by O. von Engeln.

and 7), and in all the larger of these small glaciers are still present. The entire absence of forest exposes the conditions here far more clearly than is the case along the forest-clothed Inside Passage.

Viewed from the fiord, the hanging valley selected for this illustration is plainly seen to be a broad, U-shaped trough heading well back in the mountains and with a small glacier at its head. The wide open mouth of this broad valley is truncated by the straight, steep rock wall of Nunatak Fiord and left perched high above even its water surface. This valley wall extends completely across the mouth of the hanging valley, forming a rock lip seven hundred feet high (Fig. 7). Climbing to the crest of this lip, one is able to look up the hanging valley to its mountain-walled head (Fig 8). It is found to be a broad, U-shaped valley with a flat floor and moderate grade.

The ice-born stream which flows along the bottom of this valley has cut only a shallow trench in the rock floor, through which it flows with moderate grade until the lip of the hanging valley is reached, when its grade abruptly increases and it tumbles down the main valley wall, as a succession of waterfalls, in the bottom of a gorge so shallow that the entire series of cascades, from the crest of the lip to its bottom, is plainly visible from the fiord. The stream has begun to lower its grade to harmonize with the main valley; but it has not had time yet to carry the process very far. That there is no possibility of the presence of a drift-filled valley of earlier date is proved by the fact that bed rock outcrops across the entire lip.

On any assumption of stream rejuvenation, it is utterly incredible that all the time required to deepen the main trough of Nunatak Fiord, and to broaden it into the form of maturity which it possesses (Figs. 9 and 14), should have been too short to have permitted the stream in the hanging valley to cut a more profound gorge, on such a steep slope, and to attain a better approximation to that accordance of grades toward which all tributaries tend in their relation to the main streams. Wherever one critically examines a hanging valley in its relation to the main trough, the same conclusion is necessitated.

A fifth explanation that has been proposed is faulting. It is of course admitted that a block fault, by dropping down the bottom of a main valley, would leave the tributary valleys hanging. Although admitted as a possibility for individual cases, the application of such an explanation to Alaskan conditions in general, fails utterly to account for the facts. It would not explain the truncated spurs on both sides, nor the U-shape of the main and lateral valleys. Furthermore, without the introduction of complicated secondary faulting, it would not account for the difference in level at which the valleys hang above the main trough to which they are tributary (Fig. 10). Another fact which ordinary block faulting would fail to explain is the frequent presence of a condition of double hanging valleys,—a lateral hanging above the main valley, and a tributary of this lateral hanging above it.

Such a condition of double hanging valleys may be illustrated by the case of Russell Valley (Fig. 10) which enters the lower end of Disenchantment Bay, a part of the Yakutat Bay inlet. This valley has a moderate slope and a remarkably well-developed U-shape (Fig. 11). Where it joins the fiord it has-built a gravel delta, so that there the actual rock bottom is not visible; but about a mile back from the fiord, bed rock occurs in the valley bottom near its center. Extending
PSM V70 D116 View of russell valley from across disenchantment bay.jpg

Fig 11. Photograph of Russell Valley with Long Focus Lens from West Side of Disenchantment Bay. Mouth of valley about three miles distant: length of valley up to glacier (where the snow line ends) about five miles. Note U-shape, steep, smoothed walls, absence of spurs, and flat floor. The hanging valley shown in Fig. 12. is the first upland valley on the left. Photograph by O. von Engeln.

the grade of this valley out into the main trough of Disenchantment Bay, where the nearest soundings show a depth of from 600 to 1,000 feet, the profile falls far short of reaching the bed of the bay. It is assumed, therefore, to be a hanging valley with the lip at or just below the surface of the fiord water. If we grant that this particular hanging valley may be due to faulting, which can not be disproved, we are still left with the necessity of assuming block faulting along the axis of the Russell Valley to account for the hanging condition of its own tributaries whose lips lie fully a thousand feet above the Russell Valley bottom (Fig. 12). In some cases even a third series of laterals have been seen hanging above a tributary, which itself hangs above another, which is hanging above a main trough.

To propose faulting as an explanation for such a complex system of hanging valleys does not seem rational without definite evidence of the faulting, and without some explanation of why the results of such recent faulting are so common in glaciated regions and so rare in unglaciated areas. Moreover, in some of the cases mentioned, for example, the Russell Valley itself, if there had been such faulting, it would be easily detected in the sedimentary rocks which form the walls of the valley. Since a search for evidence of recent faulting in this valley failed to find it, I feel warranted in asserting that there has been no such faulting as the theory demands. A glance at the photographs (Figs. 10 and 11) is sufficient to show that the form of this valley could not be accounted for on the basis of block faulting. Its flaring, curving, U-shaped sides are not the forms characteristic of cliffs due to faulting. Should it be stated that block faulting occurred at a date sufficiently remote to permit the weathering back of the valley walls to the present curve, it is sufficient to answer that in all the time required for this, the lateral streams must of necessity have trenched the bottoms of the hanging valleys and reduced them to an accordant grade with the Russell Valley stream. As Fig. 12 clearly shows, this is far from being the case.

From the above statement of hypotheses it will be seen that it is generally admitted that hanging valleys are a peculiar phenomenon calling for special explanation. It is also true that this phenomenon is practically confined to regions of former glaciation. Together with the U-shaped valley, truncated spurs, and steepened main valley slopes, the condition of hanging valleys is reported not only from a wide area in Alaska and British Columbia, but in such other regions of former glaciation as the Sierra Nevada, the Rocky Mountains, the Finger Lake Valleys of central New York, the coast of Norway, the Alps, the Himalayas and New Zealand. While exceptional instances of hanging valleys, which are readily explained in other ways, have been reported from unglaciated regions, these are so few and scattered, and
PSM V70 D118 Hanging valley tributary flowing to russell valley.jpg
Fig. 12. Hanging Valley Tributary to Russell Valley (Fig. 11). The first tributary from the mouth on the north side. The lip of this valley lies about 1,000 feet above the main valley floor, and the stream flows over it on the very surface of the rock, forming a gorge, below which it crosses moraine. A small glacier lies at the head of this hanging valley. Photograph by Lawrence Martin.

so unlike their abundant and striking development in glaciated regions, that they are hardly to be considered as bearing upon the problem.

The facts discovered in reading the literature and in field investigation, point to glacial erosion as the cause of the hanging valleys and associated phenomena, while no facts are found that are vitally opposed to it. Of no other hypothesis proposed may the same be said; on the contrary, all other explanations are open to fatal objections. . The great majority of students of glacial action are now in accord with the belief in profound glacial erosion in favorable situations. Even those opposed to the explanation by glacial erosion admit that the forms under discussion are what would be expected if it were possible for glaciers to perform such great erosive work.

The few who are opposed to this explanation have been able to offer no better argument against it than their failure to believe in the ability of ice to do erosive work in great amount. Some of this opposition is based upon observations at the margins of small glaciers. But all such observations have little value; for, as has been well stated by another, if an observer could have been where ice was really capable of profoundly eroding, he would not have been able to come back and talk about it. The weak, retreating margin of a small valley glacier gives no better basis for understanding profound glacial erosion than a small meadow brook gives for a conception of the mode of formation of a Colorado Canyon. The objections to ice as an agent of profound erosion remind one very much of the objections which, in the early days, were urged against water as an agent of erosion. In this connection reference may be made to a short note, signed H. G., on page 249 of the National Geographic Magazine, Vol. 16, 1905. This little squib, which we may fairly safely ascribe to Henry Gannett, although written in a humorous and somewhat sarcastic vein, is really a noteworthy contribution to the discussion on glacial erosion. In it, as a sort of reply to a recent arraignment of glacial erosion, he applies to the now accepted belief in river erosion some of the same class of arguments as those which have been urged against glacial erosion, and with telling effect.

Since the establishment of the theory of profound glacial erosion is the work of the last fifteen years, and since the full force of the evidence has only recently been accepted by some of our leading physiographers, it is natural that as yet there should not be universal acceptance of so new an idea, carrying with it such tremendous consequences. But the fact that some workers have not yet accepted the doctrine does not necessarily constitute a strong argument against it, and certainly not enough to counterbalance the overwhelming evidence in its favor. When a large number of people are involved, ultraconservatism is always to be expected among some of them. There are, for example, even at the present day, some highly intelligent men who are writing

PSM V70 D119 Hidden glacier valley a tributary to the yakutat bay inlet.jpg
Fig. 13. The North Wall of Hidden Glacier Valley, a Tributary to the Yakutat Bay Inlet, the Glacier Terminus Showing in the Midground. Note the smoothed, striated lower walls due to glacial erosion as contrasted with the irregular topography of the higher slopes due to ordinary weathering and stream erosion. A hanging valley enter-at about the level between these two classes of slopes about a third of the way from the right margin above the glacier. Photograph by R. S. Tarr.

polemics in opposition to the belief in former continental glaciation, which almost every one now considers definitely established, though after a hard fight.

It does not seem necessary at the present time to undertake to show how the glaciers did this, nor to prove that they could do it when the evidence is so clear that they actually did do it. Suffice it to say, that if glaciers smooth, scratch and pluck the rocks over which they pass, as every one knows they do (Fig. 13), it requires only a sufficiently long continuation of this action to lower valleys to any extent up to the time when they cease to further smooth, scratch and pluck. A century ago it seemed to many observers that at the slow observed rate of recession of Niagara Falls it was impossible to explain

PSM V70 D120 East view of nunatak fjord.jpg
Fig. 14 Looking Up (East) Nunatak Fiord. The rock knoll, or Nunatak, in the middle of the picture, 1,400 feet high, splits the Nunatak glacier one arm, on the left, descending to the sea through the broader valley, the other occupying a smaller U-shaped valley on the right side of the Nunatak. but not upon reaching the sea. When first seen by Prof. Russell in 1891 these two arms nearly enclosed the Nunatak. The site of the hanging (Fig. 7) valley is on the right side of the picture. Photograph by Lawrence Martin.

the seven miles of gorge as a result of this process. No one now doubts this explanation of the Niagara gorge; and it is not doubted that the Colorado Canyon has been formed by slow sawing into the strata, like that which the river is now engaged in, but continued through a long period of time. An application of the same principle—a slow rate of erosion working for a long period of time—is all that is necessary to understand profound glacial erosion, once it is granted that glaciers do scour their beds at all, as every one admits, and that there is plenty of time available, as is well known to be the case.

Accepting ice erosion as a doctrine now established, as it seems to me we must, we will briefly examine some of the consequences of such erosion. Hanging valleys, U-shaped valleys, aligned spurs, and steepened valley slopes are among the more prominent of these consequences. From their existence we must of necessity infer enormous vertical as well as lateral erosion, such erosion occurring in places where actively moving streams of ice were concentrated in valleys along relatively narrow lines. Along the Inside Passage, and in Yakutat Bay, the two sections immediately under consideration in this paper, the amount of erosion which must be deduced from the evidence is in places not less than two thousand feet vertically; and erosion of this magnitude has occurred along hundreds of miles of fiords.

In discussions of the significance of hanging valleys, it has been rather common to speak as if the main valleys were eroded while the tributaries were left undeepened. This has been done here, as doubtless in other writings, in order not to introduce an unnecessary complexity into the discussion. It would, however, be entirely erroneous to suppose that the lateral valleys were not eroded also. It requires only an examination of the photographs accompanying this paper to see that the normal cross-section of the hanging tributary valleys has the same curve as that of the main valleys; that is, the curve which glacial erosion produces.

From the statement just made, it follows that the level at which a lateral valley now hangs above the main trough is not to be taken as the full measure of vertical erosion along the main valley. That this is true is indicated by the fact that of several valleys tributary to a main trough, no two usually hang at exactly the same level. There may be, and in many cases are, wide differences in the hanging levels of neighboring valleys (Fig. 10); some being perched far up on the mountain side, others so far lowered that the sea water enters and drowns their mouths (compare Figs. 2 and 5), which, however, are still hanging above the bottom of the fiord. Such differences in the hanging level are, in the main, a measure of the difference in amount of erosive work performed by glaciers in the several hanging laterals.

In general, those valleys occupied by the largest glaciers have been lowered most; and it may be stated as a law that, other conditions being equal, the height of a hanging valley above the bottom of the main trough varies inversely with the size of the glacier. The operation of this law is, of course, modified by the influence of varying rock texture, slope and other causes which tend to modify the rate of ice erosion. We are not yet in full enough possession of the facts relating to the process of glacial erosion to warrant an attempt at a full statement of the nature and result of the various influences which tend to modify the rate of erosion. There can be no question, however, that the nature of the valley rock is of profound importance, some weak rocks being eroded with relative rapidity by small glaciers, other rocks resisting the erosion of even large, powerful glaciers. Two causes, the size of the glacier and the nature of the enclosing rock, are, in all probability, of most importance in the modification of the height of valleys left hanging by more rapid erosion along the main trough.

An argument which has been advanced against the power of glaciers to erode, is the fact that rock islands sometimes rise from the floor of valleys through which powerful glaciers have passed. It has been claimed that such protuberances should have been erased if the glaciers were really eroding greatly. When the operation of glaciers as agents of erosion is truly understood, however, this argument seems to favor rather than to oppose glacial erosion. It is not to be supposed that glaciers would erode everywhere at the same rate. There is naturally a variation in the rate of erosion of a valley bottom dependent upon at least two important influences—nature of rock and rapidity of ice currents—both of which are liable to vary in any valley and thus necessarily give rise to irregularities in the ice-eroded valley bottom. Once an obstacle arose in the path of a powerfully moving glacier, it would have the tendency to split the ice current around itself, much as a sand bar spilts the current of a river. By interfering with the ice current in line with the obstacle, and by causing a concentration of movement on either side of it, the size of the obstacle would naturally increase. Rock knolls, islands and nunataks (Fig. 14) are such characteristic features in glacially eroded valleys that, when the full significance of glacial erosion is understood, I believe they will be found to constitute one of-the distinctive evidences of glacial erosion, to be-classed with hanging valleys, truncated spurs, steepened slopes and U-shaped profiles.

In discussions on glacial erosion much attention has been paid to rock basins,—basins with rock rims in the bottoms of glaciated velleys, and oftentimes holding lakes. Such basins also occur on the fiord floors of the Inside Passage. Irregularities in erosion, due to differences in rock resistance and in ice currents, readily account for these. As Andrews has shown in his remarkable papers on glacial erosion in the New Zealand fiords, one important cause for such basins, and other forms of vigorous erosion, is the convergence of ice currents in a valley of smaller cross section, causing acceleration of motion. Rock basins must be added to the land forms resulting from and hence indicative of profound glacial erosion.

Another feature at first apparently opposing glacial erosion is that hanging valleys, truncated spurs, and steepened slopes are at times well developed on one side of a main trough and either absent or poorly developed on the other. This, however, seems a perfectly normal result of ice erosion, for, as in a river, the current naturally at time impinges upon one side with greater force than on the other, as, for example, when by the entrance of a tributary the ice current is pushed against the opposite side of the valley.

A prominent feature in regions of former glaciation, both of continental glaciers and mountain-valley glaciers, is the presence of through valleys, that is, valleys in which there is now no pronounced divide. Such valleys abound in the Finger Lake region of central New York, and they are common also in Alaska, and, as Penck has shown, in the Alps. The evidence points to the conclusion that many of these through valleys owe their characteristics to the passage of ice across divides, and the consequent lowering of the divides by glacial erosion. In some places in Alaska, as in the Yakutat Bay region, the ice is still pouring across such divides; in other cases, owing to the shrunken state of present-day glaciers, the through valleys are now occupied by glaciers which flow both ways from a low, flat divide area across which, at a higher stage of the ice, through glaciers once passed. So far as seen in the Yakutat Bay region, none of the through valleys are entirely free from ice; but in many cases the glaciers are so shrunken as to expose the valley form, which is distinctly that characteristic of glacial erosion. In central New York, where the work was performed by continental glaciers instead of valley tongues, and where the ice is entirely gone, the character of these through valleys is easily observed. They are often U-shaped, steep-sided, straight-walled, and possess hanging valleys.

The acceptance of the conclusion that glaciers have been powerful agents of erosion, and doubtless still are where now in active operation, seems a necessary result of a candid consideration of the evidence. Once this conclusion is reached, a number of remarkable phenomena, otherwise not satisfactorily explained, find ready explanation. The belief in glacial erosion carries with it stupendous consequences, for it assigns to glacial action some of the most striking topographical features of regions formerly occupied by actively moving ice. Nowhere is the evidence clearer, or the results more striking, than along the Inside Passage to Alaska, and in the fiords northwest of this, such as Yakutat Bay. For those who still doubt the effectiveness of ice erosion, a trip through these fiords is strongly recommended instead of a study of the weak termini of small, dwindling Alpine glaciers.

  1. Published by permission of the Director of the U. S. Geological Survey. I am indebted to Lawrence Martin and O. von Engeln, members of my expeditions, for photographic work, as indicated under the illustrations, and to Mr. Martin and B. S. Butler for valuable assistance in my field investigations.
  2. Harriman Alaska Expedition, Vol. II., History, Geography, Resources, 1902, pp. 258-259.