Chapter II: Part 2

Let AB, fig. 9, be a stream whose initial consequent course led it down the gently sloping axial trough of a syncline. The constructional surface of the syncline is shown by contours. Let the succession of beds to be discovered by erosion be indicated in a section, laid in proper position on the several diagrams, but revolved into the horizontal plane, the harder beds being dotted and the baselevel standing at OO. Small side streams will soon be developed on the slopes of the syncline, in positions determined by cross-fractures or more often by what we call accident; the action of streams in similar synclines on the outside of the enclosing anticlines will be omitted for the sake of simplicity. In time, the side streams will cut through the harder upper bed M and enter the softer bed N, on which longitudinal channels, indicated by hachures, will be extended along the strike, fig. 10 (La Noë and Margerie). Let these be called "subsequent" streams. Consider two side streams of this kind, C and D, heading against each other at E, one joining the main stream lower down the axis of the syncline than the other. The headwaters of C will rob the headwaters of D, because the deepening of the channel of D is retarded by its having to join the main stream at a point where the hard bed in the axis of the fold holds the main channel well above baselevel. The notch cut by D will then be changed from a water-gap to a wind-gap and the upper portion of D will find exit through the notch cut by C, as in fig. 11. As other subsequent headwaters make capture of C, the greater depth to which the lateral valley is cut on the soft rock causes a slow migration of the divides in the abandoned gaps towards the main stream, and before long the upper part of the main stream itself will be led out of the synclinal axis to follow the monoclinal valley at one side for a distance, fig. 12, until the axis can be rejoined through the gap where the axial portion of the controlling hard bed is near or at baselevel. The upper part of the synclinal trough will then be attacked by undercutting on the slope of the quickly deepened channels of the lateral streams, and the hard bed will be worn away in the higher part of the axis before it is consumed in the lower part. The location of the successful lateral stream on one or the other side of the syncline may be determined by the dip of the beds, gaps being cut quicker on steep than on gentle dips. If another hard bed is encountered below the soft one, the process will be repeated; and the mature arrangement of the streams will be as in fig. 13 (on a smaller scale than the preceding), running obliquely off the axis of the fold where a hard bed of the syncline rises above baselevel, and returning to the axis where the hard bed is below or at baselevel; a monoclinal stream wandering gradually from the axis along the strike of the soft bed, AE, by which the side-valley is located and returning abruptly to the axis by a cataclinal[17] stream in a transverse gap, EB, in the next higher hard bed, and there rejoining the diminished representative or survivor of the original axial or synclinal stream, GB.

[Footnote 17: See the terminology suggested by Powell. Expl. Col. R. of the West, 1875, 160. This terminology is applicable only to the most detailed study of our rivers, by reason of their crossing so many folds, and changing so often from longitudinal to transverse courses.]

17. _Terminology of rivers changed by adjustment_.--A special terminology is needed for easy reference to the several parts of the streams concerned in such an adjustment. Let AB and CD, fig. 14, be streams of unequal size cutting gaps, H and G, in a ridge that lies transverse to their course. CD being larger than AB will deepen its gap faster. Of two subsequent streams, JE and JF, growing on the up-stream side of the ridge, JE will have the steeper slope, because it joins the deeper master-stream. The divide, J, will therefore be driven towards AB, and if all the conditions concerned conspire favorably, JE will at last tap AB at F, and lead the upper part, AF, out by the line FEGD, fig. 15, through the deeper gap, G. We may then say that JE becomes the _divertor_ of AF, which is _diverted_; and when the process is completed, by the transfer of the divide from J, on the soft rocks, to a stable location, H, on the hard rocks, there will be a short _inverted_ stream, HF; while HB is the remaining _beheaded_ portion of the original stream, AB, and the water-gap of AB becomes a wind-gap, H. It is very desirable that geographic exploration should discover examples of the process of adjustment in its several stages. The preparatory stage is easily recognized by the difference in the size of the two main streams, the difference in the depth of their gaps, and the unsymmetrical position of the divide, J. The very brief stage of transition gives us the rare examples of bifurcating streams. For a short time after capture of the diverted stream by the divertor, the new divide will lie between F and H, in an unstable position, the duration of this time depending on the energy of the process of capture.

The consequences resulting from readjustments of this kind by which their recent occurrence can be detected are: a relatively sudden increase of volume of the divertor and hence a rapid deepening of the course of the diverting stream, FE, and of the diverted, AF, near the point of capture; small side-streams of these two being unable to keep pace with this change will join their masters in local rapids, which work up stream gradually and fade away (Löwl, Penck, McGee). The expanded portion, ED, of the larger stream, CD, already of faint slope, may be locally overcome for a time with the increase of detritus that will be thus delivered to it at the entrance, E, of the divertor; while the beheaded stream, HB, will find itself embarrassed to live up to the habits of its large valley [Heim]. Geographic exploration with these matters in mind offers opportunity for the most attractive discoveries.

18. _Examples of adjustment_.--Another case is roughly figured in the next three diagrams, figs. 16, 17, 18. Two adjacent synclinal streams, EA and HB, join a transverse master stream, C, but the synclines are of different forms; the surface axis of one, EA, stands at some altitude above baselevel until it nearly reaches the place of the transverse stream; while the axis of the other, HB, descends near baselevel at a considerable distance from the transverse stream. As lateral valleys, E and D, are opened on the anticline between the synclines by a process similar to that already described, the divide separating them will shift towards the stream of fainter slope, that is, towards the syncline, EA, whose axis holds its hard beds above baselevel; and in time the upper part of the main stream will be withdrawn from this syncline to follow an easier course by crossing to the other, as in fig. 17. If the elevation of the synclinal axis, AES, take the shape of a long flat arch, descending at the further end into a synclinal lake basin, S, whose outlet is along the arching axis, SA, then the mature arrangement of stream courses will lead the lake outlet away from the axis by some gap in the nearer ascending part of the arch where the controlling hard bed falls near to baselevel, as at F, fig. 18,[18] and will take it by some subsequent course, FD, across the lowland that is opened on the soft beds between the synclines, and carry it into the lower syncline, HB, at D where the hard beds descend below baselevel.

[Footnote 18: This figure would be improved if a greater amount of wasting around the margin of the hard bed were indicated in comparison with the preceding figure.]

The variety of adjustments following the general principle here indicated is infinite. Changes of greater or less value are thus introduced in the initial drainage areas, until, after attaining an attitude of equilibrium, further change is arrested, or if occurring, is relatively insignificant. It should be noticed that the new stream courses thus chosen are not named by any of the terms now current to express the relation of stream and land history; they are neither consequent, antecedent nor superimposed. The stream is truly still an original stream, although no longer young; but its channel is not in all parts strictly consequent on the initial constructional form of the land that it drains. Streams thus re-arranged may therefore be named original streams of mature adjustment.

It should be clearly recognized that the process of adjustment is a very slow one, unless measured in the extremely long units of a river's life. It progresses no faster than the weathering away of the slopes of a divide, and here as a rule weathering is deliberate to say the least, unless accelerated by a fortunate combination of favoring conditions. Among these conditions, great altitude of the mass exposed to erosion stands first, and deep channeling of streams below the surface--that is, the adolescent stage of drainage development--stands second. The opportunity for the lateral migration of a divide will depend on the inequality of the slopes on its two sides, and here the most important factors are length of the two opposite stream courses from the water parting to the common baselevel of the two, and inequality of structure by which one stream may have an easy course and the other a hard one. It is manifest that all these conditions for active shifting of divides are best united in young and high mountain ranges, and hence it is that river adjustments have been found and studied more in the Alps than elsewhere.

19. _Revival of rivers by elevation and drowning by depression_.--I make no contention that any river in the world ever passed through a simple uninterrupted cycle of the orderly kind here described. But by examining many rivers, some young and some old, I do not doubt that this portrayal of the ideal would be found to be fairly correct if opportunity were offered for its development. The intention of the sketch is simply to prepare the way for the better understanding of our actual rivers of more complicated history.

At the close or at any time during the passage of an initial cycle such as the one just considered, the drainage area of a river system may be bodily elevated. The river is then turned back to a new youth and enters a new cycle of development. This is an extremely common occurrence with rivers, whose life is so long that they commonly outlive the duration of a quiescent stage in the history of the land. Such rivers may be called revived. Examples may be given in which streams are now in their second or third period of revival, the elevations that separate their cycles following so soon that but little work was accomplished in the quiescent intervals.

The antithesis of this is the effect of depression, by which the lower course may be drowned, flooded or fjorded. This change is, if slow, favorable to the development of flood-plains in the lower course; but it is not essential to their production. If the change is more rapid, open estuaries are formed, to be transformed to delta-lowlands later on.

20. _Opportunity for new adjustments with revival_.--One of the most common effects of the revival of a river by general elevation is a new adjustment of its course to a greater or less extent, as a result of the new relation of baselevel to the hard and soft beds on which the streams had adjusted themselves in the previous cycle. Synclinal mountains are most easily explained as results of drainage changes of this kind [Science, Dec. 21st, 1888]. Streams thus rearranged may be said to be adjusted through elevation or revival. It is to be hoped that, as our study advances, single names of brief and appropriate form may replace these paraphrases; but at present it seems advisable to keep the desired idea before the mind by a descriptive phrase, even at the sacrifice of brevity. A significant example may be described.

Let it be supposed that an originally consequent river system has lived into advanced maturity on a surface whose structure is, like that of Pennsylvania, composed of closely adjacent anticlinal and synclinal folds with rising and falling axes, and that a series of particularly resistant beds composes the upper members of the folded mass. The master stream, A, fig. 19, at maturity still resides where the original folds were lowest, but the side streams have departed more less from the axes of the synclinals that they first followed, in accordance with the principles of adjustment presented above. The relief of the surface is moderate, except around the synclinal troughs, where the rising margins of the hard beds still appear as ridges of more or less prominence. The minute hachures in figure 19 are drawn on the outcrop side of these ridges. Now suppose a general elevation of the region, lifting the synclinal troughs of the hard beds up to baselevel or even somewhat above it. The deepening of the revived master-stream will be greatly retarded by reason of its having to cross so many outcrops of the hard beds, and thus excellent opportunity will be given for readjustment by the growth of some diverting stream, B, whose beginning on adjacent softer rocks was already made in the previous cycle. This will capture the main river at some up-stream point, and draw it nearly all away from its hard path across the synclinal troughs to an easier path across the lowlands that had been opened on the underlying softer beds, leaving only a small beheaded remnant in the lower course. The final re-arrangement may be indicated in fig. 20. It should be noted that every capture of branches of the initial main stream made by the diverting stream adds to its ability for further encroachments, for with increase of volume the channel is deepened and a flatter slope is assumed, and the whole process of pushing away the divides is thereby accelerated. In general it may be said that the larger the stream and the less its elevation above baselevel, the less likely is it to be diverted, for with large volume and small elevation it will early cut down its channel so close to baselevel that no other stream can offer it a better course to the sea; it may also be said that, as a rule, of two equal streams, the headwaters of the one having a longer or a harder course will be diverted by a branch of the stream on the shorter or easier course. Every case must therefore be examined for itself before the kind of re-arrangement that may be expected or that may have already taken place can be discovered.

21. _Antecedent and superimposed rivers_.--It not infrequently happens that the surface, on which a drainage system is more or less fully developed, suffers deformation by tilting, folding or faulting. Then, in accordance with the rate of disturbance, and dependent on the size and slope of the streams and the resistance of the rocks, the streams will be more or less re-arranged, some of the larger ones persisting in their courses and cutting their channels down almost as fast as the mass below them is raised and offered to their action. It is manifest that streams of large volume and considerable slope are the ones most likely to persevere in this way, while small streams and large ones of moderate slope may be turned from their former courses to new courses consequent on the new constructional form of the land. Hence, after a disturbance, we may expect to find the smaller streams of the former cycle pretty completely destroyed, while some of the larger ones may still persist; these would then be called antecedent streams in accordance with the nomenclature introduced by Powell.[19] A fuller acquaintance with the development of our rivers will probably give us examples of river systems of all degrees of extinction or persistence at times of disturbance.

[Footnote 19: Exploration of the Colorada River of the West, 1875, 153, 163-166.]

Since Powell introduced the idea of antecedent valleys and Tietze, Medlicott and others showed the validity of the explanation in other regions than the one for which it was first proposed, it has found much acceptance. Löwl's objection to it does not seem to me to be nearly so well founded as his suggestion of an additional method of river development by means of backward headwater erosion and subsequent capture of other streams, as already described. And yet I cannot help thinking that the explanation of transverse valleys as antecedent courses savors of the Gordian method of explaining a difficult matter. The case of the Green river, to which Powell first gave this explanation, seems well supported; the examples given by Medlicott in the Himalayas are as good: but still it does not seem advisable to explain all transverse streams in this way, merely because they are transverse. Perhaps one reason why the explanation has become so popular is that it furnishes an escape from the old catastrophic idea that fractures control the location of valleys, and is at the same time fully accordant with the ideas of the uniformitarian school that have become current in this half of our century. But when it is remembered that most of the streams of a region are extinguished at the time of mountain growth, that only a few of the larger ones can survive, and that there are other ways in which transverse streams may originate,[20] it is evident that the possibility of any given transverse stream being antecedent must be regarded only as a suggestion, until some independent evidence is introduced in its favor. This may be difficult to find, but it certainly must be searched for; if not then forthcoming, the best conclusion may be to leave the case open until the evidence appears. Certainly, if we find a river course that is accordant in its location with the complicated results of other methods of origin, then the burden of proof may be said to lie with those who would maintain that an antecedent origin would locate the river in so specialized a manner. Even if a river persist for a time in an antecedent course, this may not prevent its being afterwards affected by the various adjustments and revivals that have been explained above: rivers so distinctly antecedent as the Green and the Sutlej may hereafter be more or less affected by processes of adjustment, which they are not yet old enough to experience. Hence in mountains as old as the Appalachians the courses of the present rivers need not coincide with the location of the pre-Permian rivers, even if the latter persisted in their courses through the growth of the Permian folding; subsequent elevations and adjustments to hard beds, at first buried and unseen, may have greatly displaced them, in accordance with Löwl's principle.

[Footnote 20: Hilber, Pet. Mitth., xxxv, 1889, 13.]

When the deeper channelling of a stream discovers an unconformable subjacent terrane, the streams persist at least for a time in the courses that were determined in the overlying mass; they are then called superimposed (Powell), inherited (Shaler), or epigenetic (Richthofen). Such streams are particularly liable to readjustment by transfer of channels from courses that lead them over hard beds to others on which the hard beds are avoided; for the first choice of channels, when the unconformable cover was still present, was made without any knowledge of the buried rock structure or of the difficulties in which the streams would be involved when they encountered it. The examples of falls produced when streams terrace their flood-plains and run on buried spurs has already been referred to as superimposed; and the rivers of Minnesota now disclosing half-buried ledges here and there may be instanced as illustrating the transition stage between simple consequent courses, determined by the form of the drift sheet on which their flow began, and the fully inconsequent courses that will be developed there in the future.

22. _Simple, compound, composite and complex rivers_.--We have thus far considered an ideal river. It now seems advisable to introduce a few terms with which to indicate concisely certain well marked peculiarities in the history of actual rivers.

An original river has already been defined as one which first takes possession of a land area, or which replaces a completely extinguished river on a surface of rapid deformation.

A river may be simple, if its drainage area is of practically one kind of structure and of one age; like the rivers of southern New Jersey. Such rivers are generally small. It may be composite, when drainage areas of different structure are included in the basin of a single stream. This is the usual case.

A compound river is one which is of different ages in its different parts; as certain rivers of North Carolina, which have old headwaters rising in the mountains, and young lower courses traversing the coastal plain.

A river is complex when it has entered a second or later cycle of development; the headwaters of a compound river are therefore complex, while the lower course may be simple, in its first cycle. The degree of complexity measures the number of cycles that the river has entered.

When the study of rivers is thus attempted, its necessary complications may at first seem so great as to render it of no value; but in answer to this I believe that it may be fairly urged that, although complicated, the results are true to nature, and if so, we can have no ground of complaint against them. Moreover, while it is desirable to reduce the study of the development of rivers to its simplest form, in order to make it available for instruction and investigation, it must be remembered that this cannot be done by neglecting to investigate the whole truth in the hope of avoiding too great complexity, but that simplicity can be reached safely only through fullness of knowledge, if at all.

It is with these points in mind that I have attempted to decipher the history of the rivers of Pennsylvania. We find in the Susquehanna, which drains a great area in the central part of the state, an example of a river which is at once composite, compound and highly complex. It drains districts of divers structure; it traverses districts of different ages; and it is at present in its fourth or fifth degree of complexity, its fourth or fifth cycle of development at least. In unravelling its history and searching out the earlier courses of streams which may have long since been abandoned in the processes of mature adjustment, it will be seen that the size of the present streams is not always a measure of their previous importance, and to this we may ascribe the difficulty that attends the attempt to decipher a river's history from general maps of its stream lines. Nothing but a detailed examination of geological structure and history suffices to detect facts and conditions that are essential to the understanding of the result.

If the postulates that I shall use seem unsound and the arguments seem overdrawn, error may at least be avoided by not holding fast to the conclusions that are presented, for they are presented only tentatively. I do not feel by any means absolutely persuaded of the correctness of the results, but at the same time deem them worth giving out for discussion. The whole investigation was undertaken as an experiment to see where it might lead, and with the hope that it might lead at least to a serious study of our river problems.

PART FOURTH. _The development of the rivers of Pennsylvania_.

23. _Means of distinguishing between antecedent and adjusted consequent rivers_.--The outline of the geological history of Pennsylvania given above affords means of dividing the long progress of the development of our rivers into the several cycles which make up their complete life. We must go far back into the past and imagine ancient streams flowing down from the Archean land towards the paleozoic sea; gaining length by addition to their lower portions as the land grew with the building on of successive mountain ranges; for example, if there were a Cambro-Silurian deformation, a continuation of the Green Mountains into Pennsylvania, we suppose that the pre-existent streams must in some manner have found their way westward to the new coastline; and from the date of this mountain growth, it is apparent that any streams then born must have advanced far in their history before the greater Appalachian disturbance began. At the beginning of the latter, as of the former, there must have been streams running from the land into the sea, and at times of temporary elevation of the broad sand-flats of the coal measures, such streams must have had considerable additions to their lower length; rising in long-growing Archean highlands or mountains, snow-capped and drained by glaciers for all we can say to the contrary, descending across the Green Mountain belt, by that time worn to moderate relief in the far advanced stage of its topographic development, and finally flowing across the coal-measure lowlands of recent appearance. It was across the lower courses of such rivers that the Appalachian folds were formed, and the first step in our problem consists in deciding if possible whether the streams held their courses after the antecedent fashion, or whether they were thrown into new courses by the growing folds, so that a new drainage system would be formed. Possibly both conditions prevailed; the larger streams holding their courses little disturbed, and the smaller ones disappearing, to be replaced by others as the slopes of the growing surface should demand. It is not easy to make choice in this matter. To decide that the larger streams persisted and are still to be seen in the greater rivers of to-day, only reversed in direction of flow, is certainly a simple method of treating the problem, but unless some independent reasons are found for this choice, it savors of assumption. Moreover, it is difficult to believe that any streams, even if antecedent and more or less persistent for a time during the mountain growth, could preserve till now their pre-Appalachian courses through all the varying conditions presented by the alternations of hard and soft rocks through which they have had to cut, and at all the different altitudes above baselevel in which they have stood. A better means of deciding the question will be to admit provisionally the occurrence of a completely original system of consequent drainage, located in perfect accord with the slopes of the growing mountains; to study out the changes of stream-courses that would result from later disturbances and from the mutual adjustments of the several members of such a system in the different cycles of its history; and finally to compare the courses thus deduced with those now seen. If there be no accord, either the method is wrong or the streams are not consequent but of some other origin, such as antecedent; if the accord between deduction and fact be well marked, varying only where no definite location can be given to the deduced streams, but agreeing where they can be located more precisely, then it seems to me that the best conclusion is distinctly in favor of the correctness of the deductions. For it is not likely, even if it be possible, that antecedent streams should have accidentally taken, before the mountains were formed, just such locations as would have resulted from the subsequent growth of the mountains and from the complex changes in the initial river courses due to later adjustments. I shall therefore follow the deductive method thus indicated and attempt to trace out the history of a completely original, consequent system of drainage accordant with the growth of the central mountain district.

In doing this, it is first necessary to restore the constructional topography of the region; that is, the form that the surface would have had if no erosion had accompanied its deformation. This involves certain postulates which must be clearly conceived if any measure of confidence is to be gained in the results based upon them.

24. _Postulates of the argument_.--In the first place, I assume an essential constancy in the thickness of the paleozoic sediments over the entire area in question. This is warranted here because the known variations of thickness are relatively of a second order, and will not affect the distribution of high and low ground as produced by the intense Permian folding. The reasons for maintaining that the whole series had a considerable extension southeast of the present margin of the Medina sandstone have already been presented.

In the second place, I shall assume that the dips and folds of the beds now exposed at the surface of the ground may be projected upwards into the air in order to restore the form of the eroded beds. This is certainly inadmissible in detail, for it cannot be assumed that the folded slates and limestones of the Nittany valley, for instance, give any close indication of the form that the coal measures would have taken, had they extended over this district, unworn. But in a general way, the Nittany massif was a complex arch in the coal measures as well as in the Cambrian beds; for our purpose and in view of the moderate relief of the existing topography, it suffices to say that wherever the lower rocks are now revealed in anticlinal structure, there was a great upfolding and elevation of the original surface; and wherever the higher rocks are still preserved, there was a relatively small elevation.

In the third place, I assume that by reconstructing from the completed folds the form which the country would have had if unworn, we gain a sufficiently definite picture of the form through which it actually passed at the time of initial and progressive folding. The difference between the form of the folds completely restored and the form that the surface actually reached is rather one of degree than of kind; the two must correspond in the general distribution of high and low ground and this is the chief consideration in our problem. When we remember how accurately water finds its level, it will be clearer that what is needed in the discussion is the location of the regions that were relatively raised and lowered, as we shall then have marked out the general course of the consequent water ways and the trend of the intervening constructional ridges.

Accepting these postulates, it may be said in brief that the outlines of the formations as at present exposed are in effect so many contour lines of the old constructional surface, on which the Permian rivers took their consequent courses. Where the Trenton limestone is now seen, the greatest amount of overlying strata must have been removed; hence the outline of the Trenton formation is our highest contour line. Where the Helderberg limestone appears, there has been a less amount of material removed; hence the Helderberg outcrop is a contour of less elevation. Where the coal beds still are preserved, there has been least wasting, and these beds therefore mark the lowest contour of the early surface. It is manifest that this method assumes that the present outcrops are on a level surface; this is not true, for the ridges through the State rise a thousand feet more or less over the intervening valley lowlands, and yet the existing relief does not count for much in discussing the enormous relief of the Permian surface that must have been measured in tens of thousands of feet at the time of its greatest strength.

25. _Constructional Permian topography and consequent drainage_.--A rough restoration of the early constructional topography is given in fig. 21 for the central part of the State, the closest shading being the area of the Trenton limestone, indicating the highest ground, or better, the places of greatest elevation, while the Carboniferous area is unshaded, indicating the early lowlands. The prevalence of northeast and southwest trends was then even more pronounced than now. Several of the stronger elements of form deserve names, for convenient reference. Thus we have the great Kittatinny or Cumberland highland, C, C, on the southeast, backed by the older mountains of Cambrian and Archean rocks, falling by the Kittatinny slope to the synclinal lowland troughs of the central district. In this lower ground lay the synclinal troughs of the eastern coal regions, and the more local Broad Top basin, BT, on the southwest, then better than now deserving the name of basins. Beyond the corrugated area that connected the coal basins rose the great Nittany highland, N, and its southwest extension in the Bedford range, with the less conspicuous Kishicoquilas highland, K, in the foreground. Beyond all stretched the great Alleghany lowland plains. The names thus suggested are compounded of the local names of to-day and the morphological names of Permian time.

What would be the drainage of such a country? Deductively we are led to believe that it consisted of numerous streams as marked in full lines on the figure, following synclinal axes until some master streams led them across the intervening anticlinal ridges at the lowest points of their crests and away into the open country to the northwest. All the enclosed basins would hold lakes, overflowing at the lowest part of the rim. The general discharge of the whole system would be to the northwest. Here again we must resort to special names for the easy indication of these well-marked features of the ancient and now apparently lost drainage system. The master stream of the region is the great Anthracite river, carrying the overflow of the Anthracite lakes off to the northwest and there perhaps turning along one of the faintly marked synclines of the plateau and joining the original Ohio, which was thus confirmed in its previous location across the Carboniferous marshes. The synclinal streams that entered the Anthracite lakes from the southwest may be named, beginning on the south, the Swatara, S, fig. 21, the Wiconisco, Wo, the Tuscarora-Mahanoy, M, the Juniata-Catawissa, C, and the Wyoming, Wy. One of these, probably the fourth, led the overflow from the Broad Top lake into the Catawissa lake on the middle Anthracite river. The Nittany highland formed a strong divide between the central and northwestern rivers, and on its outer slope there must have been streams descending to the Alleghany lowlands; and some of these may be regarded as the lower courses of Carboniferous rivers, that once rose in the Archean mountains, now beheaded by the growth of mountain ranges across their middle.

26. _The Jura mountains homologous with the Permian Alleghanies_.--However willing one may be to grant the former existence of such a drainage system as the above, an example of a similar one still in existence would be acceptable as a witness to the possibilities of the past. Therefore we turn for a moment to the Jura mountains, always compared to the Appalachians on account of the regular series of folds by which the two are characterized. But while the initial topography is long lost in our old mountains, it is still clearly perceptible in the young Jura, where the anticlines are still ridges and the longitudinal streams still follow the synclinal troughs; while the transverse streams cross from one synclinal valley to another at points where the intervening anticlinal arches are lowest.[21] We could hardly ask for better illustration of the deductive drainage system of our early Appalachians than is here presented.

[Footnote 21: This is beautifully illustrated in the recent monograph by La Noë and Margerie on "Les Formes du Terrain."]

27. _Development and adjustment of the Permian drainage_.--The problem is now before us. Can the normal sequence of changes in the regular course of river development, aided by the post-Permian deformations and elevations, evolve the existing rivers out of the ancient ones?

In order to note the degree of comparison that exists between the two, several of the larger rivers of to-day are dotted on the figure. The points of agreement are indeed few and small. Perhaps the most important ones are that the Broad Top region is drained by a stream, the Juniata, which for a short distance follows near the course predicted for it; and that the Nittany district, then a highland, is still a well-marked divide although now a lowland. But there is no Anthracite river, and the region of the ancient coal-basin lakes is now avoided by large streams; conversely, a great river--the Susquehanna--appears where no consequent river ran in Permian time, and the early synclinal streams frequently turn from the structural troughs to valleys located on the structural arches.

28. _Lateral water-gaps near the apex of synclinal ridges_.--One of the most frequent discrepancies between the hypothetical and actual streams is that the latter never follow the axis of a descending syncline along its whole length, as the original streams must have done, but depart for a time from the axis and then return to it, notching the ridge formed on any hard bed at the side instead of at the apex of its curve across the axis of the syncline. There is not a single case in the state of a stream cutting a gap at the apex of such a synclinal curve, but there are perhaps hundreds of cases where the streams notch the curve to one side of the apex. This, however, is precisely the arrangement attained by spontaneous adjustment from an initial axial course, as indicated in figure 13. The gaps may be located on small transverse faults, but as a rule they seem to have no such guidance. It is true that most of our streams now run out of and not into the synclinal basins, but a reason for this will be found later; for the present we look only at the location of the streams, not at their direction of flow. As far as this illustration goes, it gives evidence that the smaller streams at least possess certain peculiarities that could not be derived from persistence in a previous accidental location, but which would be necessarily derived from a process of adjustment following the original establishment of strictly consequent streams. Hence the hypothesis that these smaller streams were long ago consequent on the Permian folding receives confirmation; but this says nothing as to the origin of the larger rivers, which might at the same time be antecedent.

29. _Departure of the Juniata from the Juniata-Catawissa syncline_.--It may be next noted that the drainage of the Broad Top region does not follow a single syncline to the Anthracite region, as it should have in the initial stage of the consequent Permian drainage, but soon turns aside from the syncline in which it starts and runs across country to the Susquehanna. It is true that in its upper course the Juniata departs from the Broad Top region by one of the two synclines that were indicated as the probable line of discharge of the ancient Broad Top lake in our restoration of the constructional topography of the State; there does not appear to be any significant difference between the summit altitudes of the Tuscarora-Mahanoy and the Juniata-Catawissa synclinal axes and hence the choice must have been made for reasons that cannot be detected; or it may be that the syncline lying more to the northwest was raised last, and for this reason was taken as the line of overflow. The beginning of the river is therefore not discordant with the hypothesis of consequent drainage, but the southward departure from the Catawissa syncline at Lewistown remains to be explained. It seems to me that some reason for the departure may be found by likening it to the case already given in figs. 16-18. The several synclines with which the Juniata is concerned have precisely the relative attitudes that are there discussed. The Juniata-Catawissa syncline has parallel sides for many miles about its middle, and hence must have long maintained the initial Juniata well above baselevel over all this distance; the progress of cutting down a channel through all the hard Carboniferous sandstones for so great a distance along the axis must have been exceedingly slow. But the synclines next south, the Tuscarora-Mahanoy and the Wiconisco, plunge to the northeast more rapidly, as the rapid divergence of their margins demonstrates, and must for this reason have carried the hard sandstones below baselevel in a shorter distance and on a steeper slope than in the Catawissa syncline. The further southwestward extension of the Pocono sandstone ridges in the southern than in the northern syncline gives further illustration of this peculiarity of form. Lateral capture of the Juniata by a branch of the initial Tuscarora, and of the latter by a branch of the Wiconisco therefore seems possible, and the accordance of the facts with so highly specialized an arrangement is certainly again indicative of the correctness of the hypothesis of consequent drainage, and this time in a larger stream than before. At first sight, it appears that an easier lateral capture might have been made by some of the streams flowing from the outer slope of the Nittany highland; but this becomes improbable when it is perceived that the heavy Medina sandstone would here have to be worn through as well as the repeated arches of the Carboniferous beds in the many high folds of the Seven Mountains. Again, as far as present appearances go, we can give no sufficient reason to explain why possession of the headwaters of the Juniata was not gained by some subsequent stream of its own, such as G, fig. 18, instead of by a side-stream of the river in the neighboring syncline; but it may be admitted, on the other hand, that as far as we can estimate the chances for conquest, there was nothing distinctly in favor of one or the other of the side-streams concerned; and as long as the problem is solved indifferently in favor of one or the other, we may accept the lead of the facts and say that some control not now apparent determined that the diversion should be, as drawn, through D and not through G. The detailed location of the Juniata in its middle course below Lewistown will be considered in a later section.

30. _Avoidance of the Broad Top basin by the Juniata headwaters_.--Another highly characteristic change that the Juniata has suffered is revealed by examining the adjustments that would have taken place in the general topography of the Broad Top district during the Perm-Triassic cycle of erosion. When the basin, BT, fig. 22, was first outlined, centripetal streams descended its slopes from all sides and their waters accumulated as a lake in the center, overflowing to the east into the subordinate basin, A, in the Juniata syncline along side of the larger basin, and thence escaping northeast. In due time, the breaching of the slopes opened the softer Devonian rocks beneath and peripheral lowlands were opened on them. The process by which the Juniata departed from its original axial location, J, fig. 22, to a parallel course on the southeastern side of the syncline, J, fig. 23, has been described (fig. 18). The subsequent changes are manifest. Some lateral branch of the Juniata, like N, fig. 23, would work its way around the northern end of the Broad Top canoe on the soft underlying rocks and capture the axial stream, C, that came from the depression between Nittany and Kishicoquillas highlands; thus reënforced, capture would be made of a radial stream from the west, Tn, the existing Tyrone branch of the Juniata; in a later stage the other streams of the western side of the basin would be acquired, their divertor constituting the Little Juniata of to-day; and the end would be when the original Juniata, A, fig. 22, that once issued from the subordinate synclinal as a large stream, had lost all its western tributaries, and was but a shrunken beheaded remnant of a river, now seen in Aughwick creek, A, fig. 24. In the meantime, the former lake basin was fast becoming a synclinal mountain of diminishing perimeter. The only really mysterious courses of the present streams are where the Little Juniata runs in and out of the western border of the Broad Top synclinal, and where the Frankstown (FT) branch of the Juniata maintains its independent gap across Tussey's mountain (Medina), although diverted to the Tyrone or main Juniata (Tn) by Warrior's ridge (Oriskany) just below. At the time of the early predatory growth of the initial divertor, N, its course lay by the very conditions of its growth on only the weakest rocks; but after this little stream had grown to a good-sized river, further rising of the land, probably in the time of the Jurassic elevation, allowed the river to sink its channel to a greater depth, and in doing so, it encountered the hard Medina anticline of Jack's mountain; here it has since persisted, because, as we may suppose, there has been no stream able to divert the course of so large a river from its crossing of a single hard anticlinal.

The doubt that one must feel as to the possibility of the processes just outlined arises, if I may gauge it by my own feeling, rather from incredulity than from direct objections. It seems incredible that the waste of the valley slopes should allow the backward growth of N at such a rate as to enable it to capture the heads of C, Tn, F, and so on, before they had cut their beds down close enough to the baselevel of the time to be safe from capture. But it is difficult to urge explicit objections against the process or to show its quantitative insufficiency. It must be remembered that when these adjustments were going on, the region was one of great altitude, its rocks then had the same strong contrasts of strength and weakness that are so apparent in the present relief of the surface and the streams concerned were of moderate size; less than now, for at the time, the Tyrone, Frankstown and Bedford head branches of the Juniata had not acquired drainage west of the great Nittany-Bedford anticlinal axis, but were supplied only by the rainfall on its eastern slope (see section 39)--and all these conditions conspired to favor the adjustment. Finally, while apparently extraordinary and difficult of demonstration, the explanation if applicable at all certainly gives rational correlation to a number of peculiar and special stream courses in the upper Juniata district that are meaningless under any other theory that has come to my notice. It is chiefly for this reason that I am inclined to accept the explanation.

31. _Reversal of larger rivers to southeast courses_.--Our large rivers at present flow to the southeast, not to the northwest. It is difficult to find any precise date for this reversal of flow from the initial hypothetical direction, but it may be suggested that it occurred about the time of the Triassic depression of the Newark belt. We have been persuaded that much time elapsed between the Permian folding and the Newark deposition, even under the most liberal allowance for pre-Permian erosion in the Newark belt; hence when the depression began, the rivers must have had but moderate northwestward declivity. The depression and submergence of the broad Newark belt may at this time have broken the continuity of the streams that once flowed across it. The headwater streams from the ancient Archean country maintained their courses to the depression; the lower portions of the rivers may also have gone on as before; but the middle courses were perhaps turned from the central part of the state back of the Newark belt. No change of attitude gives so fitting a cause of the southeastward flow of our rivers as this. The only test that I have been able to devise for the suggestion is one that is derived from the relation that exists between the location of the Newark belt along the Atlantic slope and the course of the neighboring transverse rivers. In Pennsylvania, where the belt reaches somewhat beyond the northwestern margin of the crystalline rocks in South mountain, the streams are reversed, as above stated; but in the Carolinas where the Newark belt lies far to the east of the boundary between the Cambrian and crystalline rocks, the Tennessee streams persevere in what we suppose to have been their original direction of flow. This may be interpreted as meaning that in the latter region, the Newark depression was not felt distinctly enough, if at all, within the Alleghany belt to reverse the flow of the streams; while in the former region, it was nearer to these streams and determined a change in their courses. The original Anthracite river ran to the northwest, but its middle course was afterwards turned to the southeast.

I am free to allow that this has the appearance of heaping hypothesis on hypothesis; but in no other way does the analysis of the history of our streams seem possible, and the success of the experiment can be judged only after making it. At the same time, I am constrained to admit that this is to my own view the least satisfactory of the suggestions here presented. It may be correct, but there seems to be no sufficient exclusion of other possibilities. For example, it must not be overlooked that, if the Anthracite river ran southeast during Newark deposition, the formation of the Newark northwestward monocline by the Jurassic tilting would have had a tendency to turn the river back again to its northwest flow. But as the drainage of the region is still southeastward, I am tempted to think that the Jurassic tilting was not here strong enough to reverse the flow of so strong and mature a river as the Anthracite had by that time come to be; and that the elevation that accompanied the tilting was not so powerful in reversing the river to a northwest course as the previous depression of the Newark basin had been in turning it to the southeast. If the Anthracite did continue to flow to the southeast, it may be added that the down-cutting of its upper branches was greatly retarded by the decrease of slope in its lower course when the monocline was formed.

The only other method of reversing the original northwestward flow of the streams that I have imagined is by capture of their headwaters by Atlantic rivers. This seems to me less effective than the method just considered; but they are not mutually exclusive and the actual result may be the sum of the two processes. The outline of the idea is as follows. The long continued supply of sedimentary material from the Archean land on the southeast implies that it was as continually elevated. But there came a time when there is no record of further supply of material, and when we may therefore suppose the elevation was no longer maintained. From that time onward, the Archean range must have dwindled away, what with the encroachment of the Atlantic on its eastern shore and the general action of denuding forces on its surface. The Newark depression was an effective aid to the same end, as has been stated above, and for a moderate distance westward of the depressed belt, the former direction of the streams must certainly have been reversed; but the question remains whether this reversal extended as far as the Wyoming basin, and whether the subsequent formation of the Newark monocline did not undo the effect of the Newark depression. It is manifest that as far as our limited knowledge goes, it is impossible to estimate these matters quantitatively, and hence the importance of looking for additional processes that may supplement the effect of the Newark depression and counteract the effect of the Newark uplift in changing the course of the rivers. Let it be supposed for the moment that at the end of the Jurassic uplift by which the Newark monocline was formed, the divide between the Ohio and the Atlantic drainage lay about the middle of the Newark belt. There was a long gentle descent westward from this watershed and a shorter and hence steeper descent eastward. Under such conditions, the divide must have been pushed westward, and as long as the rocks were so exposed as to open areas of weak sediments on which capture by the Atlantic streams could go on with relative rapidity, the westward migration of the divide would be important. For this reason, it might be carried from the Newark belt as far as the present Alleghany front, beyond which further pushing would be slow, on account of the broad stretch of country there covered by hard horizontal beds.

The end of this is that, under any of the circumstances here detailed, there would be early in the Jurassic-Cretaceous cycle a distinct tendency to a westward migration of the Atlantic-Ohio divide; it is the consequences of this that have now to be examined.