USGS Logo Geological Survey Professional Paper 160
Geologic History of the Yosemite Valley




Several widely different hypotheses have been advanced in explanation of the hanging valleys of the Yosemite region. Professor Whitney36 supposed them to have been left hanging by the sinking of the floor of the main chasm, a great block of the earth's crust underlying the floor and bounded by fault fractures having dropped to a depth of several thousand feet owing to disturbances in the interior of the earth. Clarence King37 assumed that the two sides of the Yosemite chasm were torn asunder by some catastrophic earth movement, the bottom of the abyss thus formed having been filled later with débris up to the level of its present floor. Among those who believed the chasm to have been hewn to its present depth mainly by the glaciers of the ice age, Henry Gannett38 and Douglas W. Johnson39 regarded the hanging valleys as a characteristic and necessary product of glacial action; the powerful main glacier, in their opinion, having been able to excavate its bed to much greater depth than the relatively feeble branch glaciers had been able to excavate theirs. Prof. Andrew C. Lawson,40 the latest scientist to offer an explanation, has suggested that some of the hanging valleys might not have come into existence until after the ice age, when the waters began to channel the glacially smoothed surface of the uplands. These hanging valleys, according to his view, are shallow because they are of recent origin, the streams not yet having had time to cut them to any great depth.

36Whitney, J. D., Geology of the Sierra Nevada: Geological Survey of California, Geology, vol. 1, pp. 421-422, 1865; The Yosemite guidebook, pp. 84-87, 1870.

37King, Clarence, Mountaineering in the Sierra Nevada, 4th ed., pp.7, 134, 1874.

38Gannett, Henry, The origin of the Yosemite Valley: Nat. Geog. Mag., vol. 12, pp. 86-87, 1901.

39Johnson, D. W., The origin of the Yosemite Valley: Appalachia, vol. 12, pp. 138-146, 1910; Hanging valleys of the Yosemite: Am. Geog. Soc. Bull., vol. 43, pp. 1-25, 1911.

40Lawson, A. C., Geology, in Handbook of Yosemite National Park, compiled by Ansel F. Hall, pp. 117-120, 1921.

Whitney's dropped-block hypothesis seemed to have support in the fact that there are in and adjoining the Sierra Nevada several well-authenticated examples of deep basins produced by the subsidence of crustal blocks. The best known is the basin that holds Lake Tahoe. However, careful search has failed to disclose any proof that the Yosemite is an abyss produced by subsidence: the great fault fractures which the mechanics of such a dislocation demand have not been discovered. It is true that such fractures would not be traceable along the bases of the walls on both sides, on account of the great thicknesses of gravel and rock waste that hide the rock floor of the chasm; but if they did exist they would necessarily extend some distance beyond the confines of the chasm and cut through its walls. However, the walls reveal no such fractures, although they are bare and afford the best possible chances for the detection of structural features of that sort.

The argument of the glacialists was undeniably a strong one, for hanging side valleys are notably characteristic features of profoundly ice-worn canyons in many parts of the world. Nor can it be doubted that a trunk glacier several thousand feet thick has inherently much greater excavating power than a glacier only a few hundred feet thick. Prolonged glaciation is therefore almost certain to result in deepening the main canyon below the levels of its tributary valleys, so that those valleys will be left hanging.

The history of the Yosemite, however, is much more complicated than the glacialists suspected, nor does their theory account for all the facts that are now known. It does not satisfactorily account for the great disparities in height between the different hanging valleys of the Yosemite region, nor does it explain the presence of hanging valleys on the sides of the lower Merced Canyon, which, it now appears, has never been glaciated.

Professor Lawson's surmise that some of the hanging valleys of the Yosemite region may be products of postglacial stream erosion, carved in the glacially smoothed surface of the uplands, is, in the first place, based on a misapprehension as to the magnitude of the glaciers that have lain on the uplands and as to the amount of erosional work they have performed, and, in the second place, it fails to consider the brevity of postglacial time and the exceptionally resistant nature of the rocks of which the uplands are made. It is now definitely known that the uplands have been covered with ice only in part and nowhere to any great depth, so that their original features could not possibly have been smoothed away by glaciation. And it is manifest that the little upland streams could not have carved, in the few thousand years that have elapsed since the departure of the glaciers, valleys from several hundred to a thousand feet in depth, especially when it is considered that in the same interval the Merced itself, with its superior volume, has been able to deepen its channel but a few feet, in rock very similar to that which prevails on the uplands. All the facts ascertained by the recent investigations, indeed, go to prove that the upland valleys of the Yosemite region, instead of being among its youngest features, are among its most ancient and date far back into the Tertiary period.


The key to the mystery is to be found in the lower Merced Canyon, more especially in the part below El Portal. This part of the canyon also has several hanging side valleys, although it has never been glaciated. Good examples are Saxon Gulch and the valleys of Feliciana Creek and its branches. (See pl. 2.) To be sure, these are not hanging valleys of the clean-cut type prevalent in the Yosemite region. They do not end abruptly at the brink of the canyon, nor do their waters pour from their lips in spectacular falls. Instead, their lips are cut by narrow gulches through which the waters cascade steeply down to he river. Nevertheless, these valleys are properly classed as hanging valleys, for their gently sloping floors lie at considerable heights above the bottom of the Merced Canyon, as will be evident from a glance at Figures 5 and 6. Comparison with Figures 7, 8, 9, and 10, furthermore, will show that, except for the gulches at their mouths, these valleys do not differ materially from the best-developed hanging upland valleys of the Yosemite region, and they are quite comparable to notched hanging valleys such as those of the two forks of Indian Creek.

FIGURE 5.—Longitudinal profile of Saxon Gulch, a hanging side valley of the unglaciated Merced canyon below El Portal. The curve of the profile, when extended forward to the axi, of the canyon, indicates at A the approximate level at which the Merced flowed before it intrenched itself—that is, prior to the last great uplift. B marks the original mouth of the hanging valley. C marks the point to which the hanging valley has been trenched by a new gulch. This profile and those in Figures 6-12 and Plate 27 are drawn strictly to scale—that is, without vertical exaggeration.

FIGURE 6.—Longitudinal profile of hanging valley of Feliciana Creek. Although this valley is trenched by a gulch for more than a mile up from its mouth, its profile indicates for the Merced a former level that is closely accordant with that indicated by the profile of Saxon Gulch.

That the Merced Canyon below El Portal has remained in all probability unglaciated was pointed out in 1911 by Prof. Douglas W. Johnson,41 who based his conclusions mainly on the general shape and character of the inner gorge. That gorge is V-shaped in cross section, no wider at the bottom than the channel of the river, and it winds in strongly serpentine curves, sharp spurs projecting into it from both sides. It has thus all the marks of a stream-cut gorge and none of the marks of a typical glacier channel, which is generally U-shaped in cross section, broad-bottomed, and fairly smooth-sided. However, there is now more direct proof at hand. The systematic search for glacial deposits which was a part of the recent investigations (see pp. 67-68) has shown that the moraines of the Yosemite Glacier end in the vicinity of El Portal, and hence it is reasonably certain that the glacier never extended much beyond that point.

41Johnson, D. W., Hanging valleys of the Yosemite: Am. Geog. Soc. Bull., vol. 43, pp. 1-25, 1911.

There being, then, no possibility that the side valleys of the lower Merced Canyon were left hanging as a result of glacial action, the question is, To what other process can they owe their hanging character? The most obvious explanation that suggests itself is that when the Merced cut its inner gorge, upon acceleration by the tilting of the Sierra block, its feeble side streams were unable to keep up with it. They were handicapped not only by reason of their lesser volume but also by reason of their northwesterly courses, arranged at right angles to the direction of the Sierra slope. The side streams thus remained unsteepened and their flow unaccelerated.

Indirect proof of the soundness of this explanation is to be found in the fact that those side streams which did not labor under these handicaps have no hanging valleys. For instance, the stream in Ned Gulch, which is of small volume but has a southwesterly course, substantially in the direction of the tilting, descends steeply all the way from its head to the bottom of the canyon, and the South Fork of the Merced, which runs northwestward for many miles but which, next to the Merced, is the largest stream of the region, has cut just as deeply as the master stream. Probably neither of these side streams was able at first to cut as fast as the master stream, but evidently they were able to catch up with it later when the cutting power of the Merced had declined owing to the reduction of its slope in its lower course.

It might be suggested, perhaps, that the hanging valleys are held up by bodies of extremely resistant rock that have retarded the cutting action of their streams. But the reverse is actually true. These valleys are underlain by relatively unresistant rocks. As is explained on page 25 (see also fig. 2), the Merced Canyon from El Portal to the foothills traverses a broad belt of upturned, nearly vertical strata of sedimentary rocks—the worn-down remnants of the great rock folds that formed the northwestward-trending mountain ridges of Cretaceous time. The differences in resistance to stream wear of these upturned rock beds have largely determined the positions and trend of most of the valleys and ridges of the present landscape; the valleys follow the belts of weaker rocks, the ridges are composed of the more resistant rocks. Valleys such as those of Feliciana Creek and Saxon Gulch, therefore, are hanging not because of the resistant nature but in spite of the unresistant nature of their floors.

It may seem surprising, in view of this curious fact, that these valleys should be hanging at all. It is to be remembered, however, that only a relatively brief time has elapsed since the last Sierra uplift took place and the Merced began to cut its inner gorge. The time has been too short to permit feeble streamlets unaccelerated by tilting to do much cutting. The headway they have made may be gaged by the length of the gulches they have cut at the mouths of their hanging valleys. The streamlet in Saxon Gulch, which is among the smallest, has cut its gulch back a distance of only half a mile; Feliciana Creek, which is somewhat larger, has cut back a distance of 1-1/2 miles; and Bear Creek, which is larger still, has destroyed the greater part of its hanging valley.

The Merced Canyon, it should be added, is not the only canyon in the Sierra Nevada that has hanging side valleys in its unglaciated lower course. The canyons of the Stanislaus and San Joaquin Rivers notably have hanging side valleys down to points within a few miles of the foothills. Almost every one of these hanging side valleys has a northwesterly or southeasterly course. Those of the lower Stanislaus, being underlain by unresistant sedimentary rocks, are in process of being trenched by headward-growing gulches. (See topographic map of Copperopolis quadrangle.) Those of the lower San Joaquin, on the other hand, being carved in prevailingly massive granite, are as a rule well preserved. Their lips are but slightly notched, and their waters still leap down abruptly in spectacular cascades and falls. (See topographic map of Kaiser quadrangle.)


In Figure 5, if the curve of the longitudinal profile of a hanging valley such as Saxon Gulch is prolonged forward to the axis of the Merced Canyon, the destroyed lower part of the valley may be "restored" with a fair degree of accuracy, and the level at which its streamlet formerly joined the master stream may be determined. The method assumes, of course, that there was originally no break in the profile, but this assumption is entirely justified, in view of the mature form of the valley and the smoothly concave curve of its profile, as far as it remains preserved. These two characteristics, which all the hanging valleys here concerned have in common, show that they were developed in a protracted epoch of undisturbed stream erosion—a "cycle of erosion," as it is termed—during which the side streams evolved smoothly graded courses down to the level of the master stream.

Applying this method of restoration to each of the hanging side valleys of the lower Merced Canyon yields a series of points indicative of the former level of the river. These points when plotted on the longitudinal profile of the Merced Canyon, as in Plate 27, B, are remarkably accordant. They lie on a smooth, flat curve that extends nearly parallel to the present profile of the river, though at a height of 1,400 to 1,500 feet above it. Doubtless this line represents the profile of the Merced at the stage which immediately preceded the cutting of the inner gorge—in other words, it represents the profile of the Merced in the mountain-valley stage, which has been tentatively assigned to the end of the Pliocene epoch. It will be referred to hereafter, for the sake of brevity, as the mountain-valley profile.

PLATE 27.—A (top), Yosemite Valley, showing three former profiles of the Merced determined by three sets of hanging valleys. A—A' is the profile of the earliest or broad-valley stage; B—B' is that of the second or mountain-valley stage; C-C' is that of the third or canyon stage, which immediately preceded the first ice invasion. The depth of the Yosemite then was 2,400 feet, measured from the brow of El Capitan. D—D' is the profile of the rock floor of ancient Lake Yosemite, which occupied a basin scooped out by the glaciers. The space between C—C' and D—D' affords a measure of the depth of glacial and interglacial excavation. This depth ranges from 500 feet at the lower end of the valley to 1,500 feet near the upper end. The space between D—D' and the present valley profile indicates the depth of the sand and gravel that now fill the basin of ancient Lake Yosemite.

B (bottom), Yosemite Valley and part of lower Merced Canyon. The broken line indicates the old profile of the mountain-vally stage, which antedated the last great uplift of the Sierra Nevada and the cutting of the inner gorge. It is determined by the profiles of a number of hanging side valleys, some of which lie far below El Portal and the extreme limit reached by the glaciers of the ice age.

(click on image for an enlargement in a new window)

The correctness of this interpretation being granted, it follows that not only the hanging valleys but the spurs and ridges between them must be remnants of the landscape of the mountain-valley stage—somewhat eroded, of course, during Quaternary time. The Quaternary erosion, however, has been in all probability slight, not more than 100 feet in the bottoms of the valleys. Allowance for this erosion has been made in plotting the profiles, but on the small scale on which they are here reproduced it is scarcely noticeable.

The question now arises, naturally, whether any of the hanging valleys in the Yosemite region belong to the same category. Are any of them remnants of the landscape of the mountain-valley stage? If so, their restored profiles should give points accordant with the old profile of the Merced established by the hanging side valleys of the lower Merced Canyon. Actual trial shows that many of them do give accordant points.

The first hanging valley above El Portal is that of Crane Creek. It forms part of the Big Meadow flat, which evidently is closely related to the broad older valley floor into which the inner gorge is cut. The flat, it is true, was invaded by a lobe of the Yosemite Glacier that spilled over the low divide between the gorge and the flat, but that circumstance does not introduce any serious complication, for the lobe spread out with only moderate thickness and had therefore but slight excavating power. Besides, the flat is underlain by fairly resistant granite; hence it was presumably but very slightly deepened by the ice. The profile of the hanging valley, unfortunately, is rather short, owing to the rapid gulch cutting done by Crane Creek in its lower course, which is on the unresistant rocks that mark the zone of contact between the granite and the sedimentary rocks near El Portal. Nevertheless, the profile, when carefully plotted, indicates a point on the mountain-valley profile of the Merced that accords well with the points already obtained. (See pl. 27, B.)

About 3-1/2 miles farther up is the hanging valley of Grouse Creek. This also presents certain complications, and its profile could scarcely be used in this connection without an intimate knowledge of local conditions. Grouse Creek, it would appear, has been deflected from its original lower course by a heavy embankment of glacial débris—the left lateral moraine of the Yosemite Glacier—and it now follows a new course, having broken through the embankment at a point half a mile farther south. Clearly the profile of the old course, which is still recognizable, is the one to be used, and when this is extended forward, it furnishes an accordant point.

It is scarcely necessary here to enter into the details of each individual valley: suffice it to say that throughout the Yosemite region there are hanging valleys that furnish accordant points for the extension of the mountain-valley profile of the Merced. Among them are the valley of Tamarack Creek, which has remained unglaciated and presents a simple case; the hanging valley of Fireplace Creek, which is likewise unglaciated (fig. 7); several small hanging valleys or ravines on the south side of the Yosemite chasm, opposite Fireplace Bluff; and similar hanging ravines on the north side, opposite the Pohono Bridge. The profiles of these little valleys and ravines are short, but the accurate contouring of the topographic map (pl. 7) permits them to be extended forward with a fair degree of confidence.

FIGURE 7.—Longitudinal profiles of hanging valley of Fireplace Creek and of ravine on opposite side of Yosemite Valley. These profiles indicate a common point on the mountain-valley profile of the Merced about 1,700 feet above the present level of the river.

Farther up in the Yosemite chasm are the hanging valleys of the two branches of Indian Creek. (See fig. 8.) These valleys, though repeatedly buried under ice to a depth of fully 600 feet, have suffered but slight excavation, for the ice came into them through the passes to the north and northeast and spread out as a partly inert mass that was held back by the powerful current of the passing Yosemite Glacier, much as backwater is held stagnant in a tributary channel by a swollen river. The profiles of these two hanging valleys should therefore be particularly valuable for an accurate determination of the mountain-valley profile of the Merced. They give, in fact, a common point for the extension of the profile.

FIGURE 8.—Longitudinal profile of hanging valley of Indian Creek and section across Yosemite Valley to Sentinel Dome. The profile, extended forward, indicates the depth to which the Yosemite had been cut in the mountain-valley stage.

At the extreme head of the Yosemite chasm is the hanging valley of Illilouette Creek. This valley has always seemed puzzling to students of the Yosemite region, as it hangs considerably lower than most of the other side valleys. The real difficulty is that the glacial history of this valley has hitherto been misunderstood. It has commonly been supposed that the Illilouette Valley was deepened by a powerful tributary glacier that came down from its own headwater basin, but it is now definitely established that most of the deepening was done by a massive lobe of ice that split from the Merced Glacier below the mouth of the Little Yosemite and that forced its way a short distance up the valley, thereby excavating the lower end with particular vigor. The profile of the Illilouette Valley, when properly corrected for this local glacial deepening, as well as for the considerable cutting done by Illilouette Creek, gives a point on the mountain-valley profile that accords well with the other points.

As will be clear from Plate 27, B, the mountain-valley profile of the Merced can be extended up into the Little Yosemite by means of several points determined from short side valleys. The total distance over which this ancient profile of the river can be reconstructed with reasonable certainty is thus about 40 miles. Future research, aided by accurate mapping, probably will permit its extension still farther, both toward the foothills of the Sierra Nevada and toward the crest.

It is possible likewise to reconstruct the mountain-valley profile of Tenaya Creek as far up as the head of Tenaya Canyon. The hanging valley of Snow Creek furnishes an accordant and valuable point, and the profile of the upper course of Tenaya Creek, which lies above the Tenaya Cascade, when duly corrected for glacial deepening, appears itself to lie directly in the line of continuation of the mountain-valley profile.


The reader may wonder why the hanging valleys just enumerated do not include those of Yosemite Creek and Bridalveil Creek. These two certainly stand preeminent among all the hanging valleys of the Yosemite region, being associated with its most famous waterfalls. The reason is that they do not accord with the other hanging valleys but belong to a separate category.

If the profile of the hanging valley of Yosemite Creek (fig. 9) is extended forward over the main chasm, duly corrected for the effects of stream erosion and glacial erosion, it is found to indicate for the Merced a former level about 700 feet higher than that of the mountain-valley profile. Similarly the extended profile of the upper part of the hanging valley of Bridalveil Creek (fig. 10) indicates for the river a former level nearly 900 feet above the mountain-valley profile.

FIGURE 9.—Longitudinal profile of hanging valley of Yosemite Creek. Extended forward to the axis of the Yosemite Valley it indicates the level at which the Merced flowed in the broad-valley stage, about 700 feet above the level of the mountain-valley stage. The lower part of the hanging valley, from A to B, was trenched and steepened by the stream presumably during the mountain-valley stage.

FIGURE 10.—Longitudinal profile of banging valley of Bridalveil Creek. The upper part, as far down as A, belongs to the broad-valley stage; the middle part, from A to B, was trenched and steepened during the mountain-valley stage; the lower part, from B to C, was cut during the canyon stage.

These large disparities can not be due merely to inaccurate plotting of the profiles. These profiles have been plotted on a much larger scale than that of the published diagrams and with considerable care, in strict accordance with the contouring of the topographic maps. The probable error to be expected from that source is in the neighborhood of 50 feet for the detailed map of the valley (pl. 7) and in the neighborhood of 100 feet for the map of the park (pl. 2). Neither can the disparities be due to erroneous estimates of the depth of stream erosion and glacial erosion, for which allowance must be made in the restoration of the hanging valleys. Such estimates are necessarily approximate, it is true, but the observational data now at hand concerning each individual valley are so plentiful that there is scarcely any chance of errors as great as 900 feet, or even half as great, creeping in. Indeed, careful consideration of all the facts involved makes it entirely clear that the disparities are not errors susceptible of elimination by adjustment or compromise. The high levels indicated by the valleys of Yosemite Creek and Bridalveil Creek can not in any legitimate way be brought down, nor can the low levels indicated by the other set of valleys be raised so as to make all accordant.

It is highly significant, furthermore, that several other hanging valleys, those of Ribbon Creek, Cascade Creek, Meadow Brook, and Sentinel Creek, indicate for the Merced high levels closely accordant with those indicated by the valleys of Yosemite Creek and Bridalveil Creek. When plotted on the longitudinal profile of the Yosemite Valley, the points obtained from these hanging valleys are found to lie on a smooth, unbroken curve (A—A, pl. 27, A) analogous to that which was previously interpreted as the Merced's profile of the mountain-valley stage but considerably higher and somewhat flatter. Evidently this curve represents a still older profile—namely, that of the broad-valley stage, which antedated the mountain-valley stage and has been referred to late Miocene time.

The valleys of Yosemite Creek and Bridalveil Creek present certain peculiarities that would seem to strengthen this conclusion. The profile of the valley of Yosemite Creek (fig. 9) describes on the whole a remarkably smooth curve, yet it is broken at a point about 1 mile back from the brink and there becomes appreciably steeper. This lower part of the valley is quite unlike the prevailingly open upper part, being trenched by a narrow, rugged gorge through which the waters make a boisterous descent. Now the profile of this gorge, when extended forward, satisfactorily meets the mountain-valley profile of the Merced. It may be inferred, then, that while the Merced was carving its mountain valley Yosemite Creek was cutting a gulch at the mouth of its hanging valley in precisely the same fashion in which Feliciana Creek is cutting a gulch at the present time. The head of the gulch has reached a point about a mile back of the brink.

The profile of the valley of Bridalveil Creek is similarly broken and seems to indicate a similar history. As far down as the point marked "A" in Figure 10 the valley is broadly open and characterized in places by gently sloping meadows. But from point A to the brink (B) it is trenched by a steep-walled gorge that increases gradually in depth downstream and has a distinctly steeper gradient. The profile of this gorge, when extended forward, also meets the mountain-valley profile of the Merced, thus showing that Bridalveil Creek, like Yosemite Creek, had made some headway in gulch cutting during the mountain-valley stage.

The profile of Bridalveil Creek suffers a second break at B (fig. 10), the part BC being much steeper than the rest. This lower part is the profile of the V-shaped gulch between the Cathedral Rocks and the Leaning Tower though which the stream rushes to the precipice of the Bridalveil Fall. As the profile clearly shows, this gulch is not part of the upland valley properly so called but lies wholly below the brink. It is a feature produced during the canyon stage of the Yosemite Valley, which will be considered further on.

The establishment of the old profile of the Merced (AA, pl. 27, A) as the profile of the broad-valley stage leads to some interesting inferences. In the first place, it follows that the upland valleys from which its level has been determined—the upland valleys of Yosemite Creek, upper Bridalveil Creek, Sentinel Creek, Meadow Brook, Cascade Creek, and Ribbon Creek—are themselves remnants, only slightly modified, of the landscape of that early stage. That being granted, it follows further that the rounded hills and low mountains that form the divides between the valleys are also parts, somewhat modified, of that ancient landscape. In short, it appears that the configuration of the uplands is to-day, save for minor changes, still representative of the country as it was in that remote epoch (late Miocene) when the Yosemite region was a land of moderate altitude above the sea and when the Merced, as yet unaccelerated by any major uplifts of the Sierra block, wandered sluggishly in a broad-floored valley cut only a few hundred feet below the tops of the flanking ridges and hills.

This may seem a daring statement to make, for the time that has elapsed since those early days amounts probably to about 8,000,000 years, and the erosive processes have been active throughout that period. However, it is to be borne in mind that the upland streams have small volume and for the most part gentle gradients and therefore slight cutting power; that the massive granitic rocks from which both hills and valleys are carved are exceptionally resistant; and that throughout most of the long interval, until the latest great uplift, the land was covered with luxuriant, dense vegetation promoted by a humid climate, so that the effectiveness of the erosional processes was minimized by a protective mat. Thus, it will be seen, several circumstances have combined to preserve the Yosemite upland in nearly its original state during those very epochs when the canyon of the Merced was being actively cut.

Probably the most conspicuous changes in the configuration of the upland were wrought by the glaciers of the ice age, which broadened and deepened some of the valleys, scraped and rounded the divides which they overrode, and hewed out the valley heads on some of the higher ridges into amphitheaterlike hollows. But those changes were confined only to certain parts of the upland. Large areas have remained uninvaded by the ice and therefore retain their preglacial configuration but slightly changed.

The vertical measure of the wearing down which the Yosemite upland has suffered, apart from the changes wrought by the ice, is in any event small—presumably less than 200 feet on an average—and accordingly it may be properly said that the present billowy topography of the upland is in a general way still representative of that which was evolved before the land acquired its present altitude.


Of the various circumstances to which the Yosemite upland owes its preservation, the most influential by far is the exceeding durability of its massive granitic rocks. The massive structure of these rocks is, indeed, peculiar to the Yosemite region and is in large measure the cause of its distinctive sculpture. Only in a few other localities in the Sierra Nevada and in others widely scattered over the earth is there granite of comparable structure. It seems appropriate, therefore, to set forth the characteristics of this unusual material, in order that the reader may fully appreciate the part which it has played in the development of the Yosemite landscape.

All igneous rocks are characteristically traversed at intervals by straight or nearly straight fractures, commonly termed "joints" because they resemble the joints between stones in masonry. (See pl. 41, B.) Such fractures occur as a rule in parallel sets, three or more sets crossing one another in such a way as to divide the rock into plane-sided, sharp-edged blocks and slabs. Ordinarily the joints are spaced about a foot or a few feet apart, and a mountain of granite so jointed may be conceived as being composed of a great multitude of blocks and slabs accurately fitted together. The granitic rocks of the Yosemite region, however, are for the most part sparsely jointed, the intervals between fractures measuring tens or hundreds of feet, in some places even thousands of feet. (See pls. 35, B, 36, and 17.) The uplands, accordingly, are to be conceived as being made up in part of large blocks and sheets of granite, in part of huge monoliths42 measuring hundreds or thousands of feet in horizontal and vertical extent.

42The term monolith was applied by the ancient Greeks to a column or monument composed of a single large block.

Now the rate at which a mass of igneous rock yields to the agents of decomposition and disintegration depends largely on the spacing of its joints, for every joint constitutes a plane of weakness. Through the joints water penetrates to the interior, carrying with it carbon dioxide and acids derived from decomposing vegetal matter, which dissolve the weaker minerals. In the joints also water congeals in freezing weather, exerting its well-known expansive force. Thus a closely jointed mass of rock suffers attack from within as well as from without and tends to be converted into an aggregate of loose blocks. Manifestly, the more closely jointed the rock the more rapidly will it break up; and conversely the more sparsely jointed it is, the longer will it hold out. A huge monolith, being wholly devoid of fractures and vulnerable only at its surface, will endure for a very long time.

The rate at which streams cut their channels in igneous rocks also depends in large measure upon the spacing of the joints. For in hard rocks of this kind streams accomplish little by abrading with the sand and gravel they carry; they erode most effectively by plucking and removing entire joint blocks and joint slabs that have been partly loosened by solution and frost. This is evident from the hackled, angular configuration of their beds. It follows that in the Yosemite region, whose granitic rocks are very hard, channel erosion can proceed with some rapidity only where the rock is divided into small, light blocks or slabs. Wherever the rock is only sparsely fractured or wholly massive—wherever, in other words, the individual blocks and slabs are too large and too heavy to be dislodged by the current—erosion is limited to the abrasive processes only and works at an extremely slow rate.

How nearly impotent even the most powerful stream of the Yosemite region is to erode massive granite may be readily seen at various points along the course of the Merced, notably above the Vernal and Nevada Falls. There in all the 10,000 years or more that has elapsed since the glaciers of the ice age melted away the river has worn its bed only a foot or two, in places only a few inches, below the polished surface left by the ice.

Thus the Yosemite upland and its hanging valleys have remained well preserved, in spite of their great age. The upland is limited, of course, to the area of the prevailingly massive granitic rocks. Its western border coincides in general with the western limit of those rocks, which follows an irregular line passing from north-northwest to south-southeast through The Gateway, at the elbow bend of the Merced Gorge. West of this line the granitic rocks are prevailingly fractured, and in the vicinity of El Portal they give place to the thin-bedded and generally shattered sedimentary rocks of the lower Sierra slope. There the country is deeply and intricately dissected, and only scattered remnants of the ancient upland surface remain preserved on narrow skeleton ridges and isolated peaks. Hanging upland valleys dating back to the broad-valley stage are wholly absent, and there are consequently no data for the extension of the broad-valley profile of the Merced. A tentative extension of that profile in a downstream direction (pl. 27, A) shows that it must have passed but little below the higher summits, such as Pinoche Peak and Trumbull Peak.

It may be asked, perhaps, why Indian Creek and Illilouette Creek were able to cut their valleys down to the level of the mountain-valley stage. The reason is that each of these streams flowed over more or less regularly jointed rock and thus had a decided advantage over those which flowed over prevailingly massive rock. Even in the relatively short interval that has elapsed since the second Sierra uplift (which initiated the canyon stage) these two streams have cut deep gulches at the mouths of their hanging valleys. Indian Creek, in spite of its small volume, has carved a gulch 1 mile in length (known as Indian Canyon), thereby depriving the Yosemite landscape of an additional waterfall. Illilouette Creek, though a much larger stream, has carved a gulch only half a mile in length, having to deal with more sparsely jointed rock. It still makes a waterfall 370 feet in height, its gulch cutting having been arrested by a body of massive granite.

Upstream the broad-valley profile of the Merced is readily extended into the Little Yosemite, for the upland south of that valley is also a large remnant of the broad-valley landscape, preserved on massive granite, and the hanging valley of the brook that runs through the Starr King Meadows furnishes a closely accordant point for the profile. (See pl. 27, A.) Still farther upstream the broad-valley profile of the Merced is more difficult to reconstruct, as the hanging side valleys have been deeply excavated by the glaciers of the ice age. Nevertheless, close study of these hanging valleys and of the high shoulders that extend on both sides of the main canyon trough leaves little doubt that these features, all of which are carved from prevailingly massive granite, are remnants, more or less modified by glaciation, of the ancient landscape of the broad-valley stage. That landscape can thus be traced all the way to the head of the drainage basin of the Merced, and it would appear that even there, in the heart of the High Sierra, the hanging valleys are not products of glacial action alone but are at least in part of preglacial origin, having been left hanging in the first instance through the rapid trenching of the Merced after the first great Sierra uplift.


Hanging valleys and upland tracts that have remained preserved since remote geologic time as a result of the exceeding durability of massive granite are common throughout the southern half of the Sierra Nevada. Examples might be cited by the score, but one seems particularly deserving of mention—that which is associated with the valley of the South Fork of the Merced near Wawona. (See pl. 2.)

This valley forms a wide basin in an otherwise narrow canyon and is in this respect analogous to the Yosemite Valley. But it is only what may be called a half-yosemite, for though it is flanked on the north by imposing cliffs, on the south it is flanked merely by low mountains and hills whose forested slopes descend at moderate angles. The cliffs on the north side rise abruptly 3,000 feet to the brink of a billowy upland and are adorned by a beautiful cascade, the Chilnualna Falls, which pours from the mouth of a typical hanging valley. (See fig. 11.) From the mountains and hills on the south side, by contrast, the streamlets descend through deeply cut valleys, without any cascades or falls. Indeed, the great height of the hanging valley of Chilnualna Creek and the upland of which it forms a part is rendered the more impressive because of the absence of an upland of equal height on the other side of the South Fork.

FIGURE 11.—Longitudinal profile of hanging valley of Chilnualna Creek and section across the "half-yosemite" of the South Fork of the Merced at Wawona. The hanging valley belongs to the broad-valley stage. Only its lower portion appears to have been cut down to the level of the mountain-valley stage.

The explanation of this remarkable contrast in land sculpture on the two sides of the valley at Wawona is that on the north side the progress of erosion has been greatly retarded by the presence of massive granite comparable to that of the Yosemite region, whereas on the south side erosion has proceeded at a relatively rapid rate in prevailingly well-jointed rocks, granitic and sedimentary. Particularly impressive is the fact that the valley of Chilnualna Creek remains in part untrenched, although it has a southwesterly course and consequently must have been steepened by each tilting of the Sierra block. Its general level corresponds to that of the upland valleys of Bridalveil and Yosemite Creeks, and it therefore forms part of the ancient landscape of the broad valley stage. Only its lower portion appears to have been trenched and steepened during the mountain valley stage.

It is highly significant, further, that the valley of the South Fork at and below Wawona is wholly a product of stream erosion and not in any measure of glacial erosion, for the boulder deposits that mark the farthest limits reached by the South Fork Glacier lie at and above Wawona. It follows that the hanging valley of Chilnualna Creek lies 2,000 feet above the valley of the South Fork solely because the creek has been unable to trench the massive granite of the upland as rapidly as the master stream has trenched the jointed rocks along its course.


Besides the two sets of hanging valleys from which the profiles of the broad-valley stage and the mountain-valley stage have been determined, there is still another set, situated at a lower level. These are really gulches rather than valleys, for their gradients are very steep. The largest is the hanging gulch from whose mouth the Bridalveil Fall leaps. (See pl. 3.) It terminates at a height of only 850 feet above the floor of the Yosemite Valley, or fully 1,600 feet below the mouth of the upland valley of Bridalveil Creek. Another example is the hanging gulch of Royal Arch Creek, which terminates at the shoulder above the Royal Arches, at a height of 1,400 feet above the main valley. To the same set belong also the gulches of Cascade Creek and Tamarack Creek, which unite just above the brink of the Merced Gorge below the valley, at a height of 650 feet above the river; and the gulch of Wildcat Creek, which terminates at the brink of the gorge at a height of 700 feet.

None of these hanging gulches have been given prominence in the literature on the Yosemite Valley, probably because they are only minor features in the landscape and because their streams and falls have but small volume. Even the gulch of Bridalveil Creek has received but scant attention, although it is associated with one of the world's most famous waterfalls and is itself a decidedly anomalous feature, projecting on a boldly sculptured promontory far out into the Yosemite chasm. Yet these hanging gulches hold the key to the Yosemite's most significant secret: they indicate the level to which the Merced had cut the valley immediately prior to the advent of the glaciers.

The depth to which any valley or canyon was cut before being remodeled by glacial action is as a rule impossible to ascertain with any degree of precision. In most localities only the roughest sort of approximation can be made, the glaciers having blurred or destroyed all vestiges of the preglacial topography. The Yosemite is exceptional in this respect: several features of its preglacial topography remain preserved, and these are of such a kind as to afford a clue to the depth to which the chasm had been cut by the Merced before the glaciers began their work. This is truly a fortunate circumstance, for the Yosemite problem centers about the question, How much excavating was done by the river, and how much by the glaciers?

The profile in Figure 12 shows that the gulch of Bridalveil Creek lies wholly below the level to which the Yosemite had been cut in the mountain-valley stage. What is equally significant, the lower part of the flanking spur surmounted by the Cathedral Rocks, including the lowest of their three summits, also lies below that level. It is clear, then, that all these features have been carved since the mountain-valley stage during that cycle of vigorous erosion which produced the canyon stage. To some extent, of course, they have suffered erosion also during glacial and postglacial time, but their configuration is such as to show beyond a doubt that they are primarily stream-cut features of the canyon stage.

The gulch of Bridalveil Creek is sharply V-shaped and in general exhibits the characteristic forms of stream erosion but hardly any marks of glacial erosion. True, it has been completely submerged by the highest ice floods, for it lies below the level of the highest moraines (see pp. 64-65), but evidently it has not been exposed to the full force of the Yosemite Glacier, being protected by the mighty bulwark of the Cathedral Rocks. Even when that bulwark was overwhelmed by the ice, as must have happened at least twice during the glacial history of the valley, the gulch was filled mostly with stagnant ice that was held imprisoned by the main current of the Yosemite Glacier, as backwater is held imprisoned in a side slough by a powerful river. Nor is there any evidence that the gulch has ever been traversed lengthwise by an actively excavating tributary glacier. On the contrary, the moraines on the upland show that the feeble Bridalveil Glacier failed to reach the brink of the chasm, save at the time of maximum glaciation, when the chasm was filled with ice literally to the brim.

Stream erosion effected during and since glacial time accounts probably for only a small part of the total depth of the gulch. The clean-cut form of the lip of the gulch, which is not marred by any stream carved notch or recess, shows in itself that stream erosion proceeds but very slowly here owing to the resistant nature of the granite. The effects produced by stream erosion since the glacial epoch are in fact negligible, and the cutting accomplished by it during glacial and interglacial time aggregates probably but little over 100 feet. In the profile (fig. 12), therefore, an allowance of that amount has been made for stream erosion since the canyon stage.

FIGURE 12.—Longitudinal profiles of hanging gulch of Bridalveil Creek and of hanging upland valley of Ribbon Creek. The former indicates the depth to which the Yosemite had been cut in the canyon stage, before it was invaded by any glaciers of the ice age. The latter indicates its depth in the broad-valley stage. The difference in level, 2,100 feet, is the total depth of cutting done by the Merced during the mountain-valley and canyon stages. From this diagram it is manifest that the Yosemite prior to the ice age had a depth of 2,400 feet, measured from the brow of El Capitan.

It can be demonstrated, similarly, that none of the other hanging gulches mentioned have been deepened appreciably by glacial erosion, in spite of the fact that they have all been submerged by the highest ice flood. The spurs flanking them have been in part planed away by the ice, but the gulches themselves evidently were sufficiently recessed in the sides of the chasm to escape the eroding force of the glacier's current. Probably they were filled by feeble eddies at the margin of the ice stream. Neither is there any indication that any of the gulches have been exposed to the eroding action of tributary glaciers. The cutting done by their respective streamlets during and after the ice age is probably even less than that done by Bridalveil Creek. It may be properly concluded, then, in view of these considerations, that the hanging gulches all are primarily features left over from the canyon stage.

Quite different is the story told by the cliffs at the mouths of the hanging gulches. The abruptness with which these cliffs cut off the gulches as well as the flanking spurs shows that they were produced later than those features of the canyon stage and by an agency that worked in a different way and with much greater vigor than the streams. Their sheer profiles, their prevailingly straight or smoothly curved courses, and their alinement essentially parallel to the axis of the main chasm, moreover, show plainly that that agency was a glacier—a mighty glacier that forced its way lengthwise through the chasm. (The processes whereby glaciers operate to produce such cliffs are explained on pp. 89-91.) It may be concluded, then, that whereas the gulches are essentially preglacial features, the cliffs in which they terminate are glacial features produced during the ice age.

It is not desired to imply, however, that all the excavating done in the main chasm below the lips of the hanging gulches was performed by the ice. The glaciers did not work continuously throughout the ice age but were interrupted by long intervals of relatively mild climate, when they melted back to their sources, or nearly so, and the streams resumed their normal activities. Doubtless, therefore, that part of the cross section of the Yosemite Valley which lies below the lips of the hanging gulches represents stream work as well as glacier work accomplished during the ice age. But the glacier work clearly was greatly preponderant, for the cliffs are characteristic glacial features, and the configuration of the bottom of the valley, in so far as it is not modified by postglacial stream work, has the characteristic form of a glacier bed. It is permissible, then, to speak of the lower part of the valley, below the lips of the hanging gulches, as the glacially excavated part.

It follows from all this that the profiles of the hanging gulches, if extended forward by the method applied to the profiles of the hanging valleys above them, should indicate very closely the depth to which the Yosemite Valley had been cut by the beginning of the ice age.


The altitude of the preglacial floor of the Yosemite Valley, indicated by the profile of Bridalveil Creek in Figure 12, appears to be about 4,600 feet—that is, 700 feet higher than the present floor. The levels indicated by the profiles of the other gulches are entirely consistent with it, and when plotted on the longitudinal profile of the Yosemite Valley (pl. 27, A) are found to lie on a smooth curve analogous to the two older profiles of the Merced but more strongly concave. The very smoothness and regularity of this curve would seem to confirm the belief that it truly represents the profile of the preglacial canyon floor. It can be drawn from the elbow bend of the Merced Gorge to the mouth of the Little Yosemite Valley.

Of peculiar interest is the rapidly increasing steepness of the profile toward the head of the Yosemite Valley. It shows that during the canyon stage the Merced was still cutting its inner gorge headward and had not yet carried it up to the Little Yosemite. The abrupt headward termination of the gorge was due in part, doubtless, to the fact that above the junction of Tenaya Creek and Illilouette Creek the river had less volume and consequently less cutting power than below; but of much greater influence, probably, was the fact that at the head of the chasm the relatively well jointed Sentinel granodiorite is replaced by the prevailingly massive Half Dome quartz monzonite, which is exceedingly resistant to stream erosion. The boundary between these two types of rocks crosses the head of the Yosemite Valley, as may be seen on the map in Plate 51.

Up along the course of Tenaya Creek, however, a branch of the inner gorge extended probably for a distance of several miles, for, though Tenaya Creek has small volume, its cutting action was favored by the presence of a zone of fractures in the granite, the same zone which even now enables Tenaya Creek to trench more effectively than the Merced River above the Yosemite Valley.

The profile of the canyon stage, as drawn in Plate 27, A, permits fairly definite answers to be given to the crucial questions. To what depth was the Yosemite cut by the Merced prior to the ice age, and how much deeper was it cut by the glaciers? The space between the profile of the broad-valley stage (A—A') and the profile of the canyon stage (C—C') gives the measure of the stream cutting, and the space between the canyon profile (C—C') and the profile of the glacially carved rock bottom of the valley (D—D') gives the measure of the glacial cutting, including whatever stream cutting was accomplished during interglacial time.

The depth of the preglacial stream cutting ranges from 1,600 feet at the head of the chasm to 2,100 feet at its lower end; the glacial cutting, on the other hand, increases from 500 feet at the lower end to fully 1,500 feet at the head. A better appreciation of the great depth to which the Yosemite had been cut by the beginning of the ice age may be gained from the statement that its depth measured from the brow of El Capitan was then fully 2,400 feet (compared with 3,000 feet at the present time), and its depth below the promontory of Glacier Point was about 2,000 feet (compared with 3,200 feet at the present time).


The two older profiles of the Merced (pl. 27, A) still require a bit of scrutiny. To the layman they may seem to be mere lines in a diagram, but to the geologist they convey a wealth of information concerning the configuration of the Yosemite region at each of the early stages which they represent. What is more, they afford a basis for estimates of the magnitude of the two great Sierra uplifts that caused the Merced to trench so deeply. Indeed, most of the foregoing story of successive uplifts and periods of valley and canyon cutting and all the figures that have been given for the height of the Sierra Nevada and the depth of the Yosemite Valley at each stage are based largely on these profiles.

In Plate 27, A, both profiles appear as they now are, steepened by the tiltings of the Sierra block. The mountain-valley profile has been steepened by one tilting, the broad-valley profile by two tiltings. It is thus clear that originally both had much flatter slopes. By what method can these original slopes be ascertained?

Professor Lindgren,43 in his studies on the gold-bearing gravel in the ancient stream beds of the northern Sierra Nevada, found that the profiles of these ancient stream beds, when plotted with care, exhibit steep stretches alternating with relatively flat stretches. The slope of each stretch depends upon its trend with respect to the general slope of the range. Naturally those stretches which trend west-southwestward, directly down the slope of the range, are steepest, having been most strongly affected by the tilting; those stretches which trend at considerable angles to the direction of the tilting are less steep; and those few stretches which trend east-northeastward, in a direction opposite to the tilting, are flattest or even slightly reversed. Most instructive, however, are those stretches that have north-northwesterly or south-southeasterly trends, at right angles to the direction of the tilting, for they have been neither steepened nor flattened by the tilting and still have their original slopes. There are a sufficient number of such stretches to afford a fair conception of the prevailing slopes of the ancient rivers in the northern part of the Sierra Nevada.

43Lindgren, Waldemar, The Tertiary gravels of the Sierra Nevada of California U. S. Geol. Survey Prof. Paper 73, pp. 46-48, pl. 10, 1911.

The Merced in the Yosemite Valley has but one stretch with northwesterly trend that might possibly afford such indication of its former slope—the stretch from the mouth of Illilouette Creek to the mouth of Indian Creek. This stretch is only 2-1/2 miles long, scarcely long enough to serve as a basis for definite figures, yet it deserves to be considered in this connection. In the mountain-valley profile in Plate 27, A, this stretch has a slope of only 60 feet to the mile, whereas the southwesterly stretch immediately below averages 140 feet to the mile, and the southwesterly stretch immediately above rises to about 175 feet to the mile. From these figures it may be inferred that this part of the Merced's course was steepened by the last tilting from an original slope of about 60 feet to the mile to the present average slope of 150 feet—that is, by about 90 feet.

However rough these data may be, they are in any event consistent and of the kind to be expected. Moreover, they harmonize remarkably well with those which Lindgren obtained for a number of northern rivers and also with those which the present writer has obtained for the San Joaquin. They may, therefore, be tentatively adopted as representative in a general way. A slope of 60 feet, it may be added, is entirely compatible with the volume and character of the Merced and with the conditions of flow that must have prevailed during the mountain-valley stage.

The original slope of the broad-valley stage is less readily determined, for the profile of that stage, as it appears in Plate 27, A, has been steepened by two uplifts, and, besides, it is not controlled by a sufficient number of points to afford any definite data for the northwesterly stretch from Illilouette Creek to Indian Creek. However, a rough approximation can be made for it. The last tilting, as has just been shown, steepened the slope of the mountain valley by about 90 feet, and it must therefore have steepened the slope of the broad valley by the same amount. Deducting 90 from 114, the slope indicated in Plate 27, A, leaves 24 feet to the mile as the slope prior to the last tilting. The slope prior to the first tilting must have been still less, presumably in the neighborhood of 12 feet to the mile. The remarkable smoothness of the broad-valley profile and its near approach to a straight line up to a point within a short distance of the sources of the river show that the river's course was "well graded"—that is, had reached a stage in which, as a result of long-continued erosion, a flat, smooth slope had been evolved.

To explain in detail how, with the aid of these slopes of the Merced River in the broad-valley and mountain-valley stages, the increase in the height of the range with each uplift may be computed would lead beyond the bounds of this paper. Suffice it to say that the extension of each of the profiles toward the foothills and the similar extension of the old profiles of the San Joaquin and the northern rivers investigated by Lindgren permit the hinge line of the Sierra block—that is, the line east of which the block was elevated and west of which the block was depressed—to be determined approximately for each stage. From this zero line as a base it is possible to calculate within reasonable limits the altitude of the crest of the range and likewise the altitude of the Yosemite for each stage.

Mount Lyell, which at present stands 13,090 feet above the sea, accordingly, appears to have had an altitude of about 4,000 feet in the broad-valley stage (at the end of the Miocene epoch) and of about 7,000 feet in the mountain-valley stage (at the end of the Pliocene epoch). The first uplift, in other words, raised the crest of the Sierra Nevada at the head of the Merced River about 3,000 feet; and the second uplift raised it about 6,000 feet more. The Yosemite Valley, or rather, that part of the Merced's valley which ultimately became the Yosemite, had an altitude of only about 800 feet in the broad-valley stage and of about 1,800 feet in the mountain-valley stage.

Many other deductions of interest may be based upon the two old profiles of the Merced. To a geologist familiar with the laws that govern the eroding action of streams and with the laws of valley development in general, these profiles, together with their original slopes, as determined approximately, are suggestive of the character and aspect which the valley of the Merced in general, and the Yosemite Valley in particular, must have had at each of those two remote stages.

Thus the flatness and straightness of the broad valley profile indicate for the valley of late Miocene time a very mature stage of development, such as could have been attained only by erosion during a long period of prevailing stability of the land. They show that the river was cutting but feebly and in the lower part of its course was gradually broadening its valley and creating what is termed a "flood plain." The steeper slope and the more pronouncedly concave curvature of the mountain-valley profile, on the other hand, indicate for the valley of late Pliocene time a much less mature stage of development. Evidently the interval of time between the first uplift and the second, though of long duration, was much shorter than that during which the broad-valley profile had been evolved. It was not long enough, in any event, to enable the Merced to produce a flat, even slope more than halfway up to its source. Within the length of the Yosemite alone the slope increased headward from 50 feet to 90 feet to the mile. In this part of its course doubtless the Merced was still a swift and actively cutting mountain stream, and its valley narrowed gradually headward to a simple V shape.

The inferences drawn from the two old profiles of the Merced accord closely with the indications as to the character of the two older valleys that are afforded by the features of the landscape of to-day and also with the story told by the hanging valleys. There is thus available from these three sources a considerable mass of data, consistent among themselves and in large part of a quantitative nature, from which the landscape of the Yosemite at each of these two stages in its evolution can be reconstructed in imagination.

<<< Previous <<< Contents >>> Next >>>

Last Updated: 28-Nov-2006