Outline.—Introduction.—Topography of Connecticut: The upland plateau, its origin, date, elevation, valleys sunk beneath its surface.—Lowland on the Triassic area.—Later oscillations.—RÉsumÉ of the topography.—Early drainage.—Re-adjusted streams.—Revived streams.—Unconformable rivers, consequent or superimposed.—Pleistocene changes; the Farmington, Quinnipiac, Scantic.—Abandoned gaps.

Introduction. In order to study intelligently the history of a river, one must first become acquainted with the present physical geography of the region in which the river lies, and know the stages of its development. Therefore, before classifying the rivers of Connecticut, I shall consider the topography of the state, and in a few paragraphs outline the successive cycles in the history of its growth. The scope of this article will not permit a discussion or even a full statement of the evidence on which these conclusions are based. They have been stated at considerable length by Professor W.M. Davis,34 and the reader is referred to his papers for the complete discussion. His conclusions in respect to the physical geography are accepted here without question, and form the basis for the discussion on the rivers of the state.

Topography of Connecticut. Connecticut can be said to consist of two great areas quite distinct in topography and geologic structure.35 On the east and on the west are the crystalline uplands which rise from sea level along the Sound to 1,700 and 1,800 feet in the northwestern part of the state, and to 600 and 700 feet in the northeastern. These uplands consist chiefly of gneiss and granite, probably of pre-Paleozoic age, which are now much folded, faulted and crumpled. Between these two areas of crystallines is a lowland belt of Triassic sandstone and shale, twenty to twenty-five miles wide, extending from New Haven north through the center of the state and including in its borders New Haven, Meriden, Hartford, New Britain and many towns of lesser note. These sandstones form a monocline with an eastward dip of 10° to 30°, and in addition to being tilted they have been faulted since their deposition in a shallow, slowly-subsiding trough of crystallines. Their thickness is variously estimated—3,000 to 5,000 feet, Dana; 10,000 or more, Davis. This lowland is interrupted by a series of trap ridges, which in general present steep faces toward the west, whereas their eastward slope is gradual, less than the dip of the sandstones.

The upland plateau. Suppose we ascend the highest point of these trap ridges, the old tower on Talcott Mt., nine miles west of Hartford; we are 900 feet above the sea level and more than 600 above the plain at our feet. A few miles to the west across the sandstone valley, rise the crystalline uplands, which extend far to the north and to the south. On the east across the Connecticut we see the eastern uplands. The first impression, which comes to one as he gazes upon these uplands and which is strengthened with each view, is that few hills rise above the general level of the plateau; the crest line is nearly horizontal, declining gently to Long Island Sound. Above this general level are a few rounded domes, but no sharp, towering peaks. Below it valleys have been cut, but they do not destroy the plateau-like appearance. A view from the western plateau across the sandstone valley shows the remarkably even crest line of the trap ridges, a crest line which approximates in height the uplands on the east and west. A nearer view of the upland corroborates our first impressions of the gently rolling character of the inter-stream surfaces, but we have a better view of the valleys which have been sunk beneath the general level and of the low rounded hills which rise above it. In popular parlance the country is “hilly.” It is uneven, not because there are high hills, but rather because there are deep valleys. If in imagination we fill up these valleys and the wide Triassic lowland to the general level of the broad inter-stream surfaces, we shall have constructed a gently undulating plateau, dipping to the south and east—a peneplain.36

Origin of the peneplain. This is not a constructional surface, for the rocks are greatly tilted, folded and faulted, so that the surface consequent upon such disturbance must have been complex and mountainous. Long subaËrial denudation upon a folded and faulted mass when the land stood much lower than at present produced this plateau. Evidently it could be produced by denudation only at or near baselevel, for the effect of erosion upon a mass high above baselevel is to accentuate its topographic relief, not to reduce it. We naturally ask ourselves, “At what stage in geologic history did this denudation occur?”

Date of the peneplain. The erosion which accomplished this great work must have commenced after the formation and dislocation of the Triassic beds, for the even crest line of the trap ridges, a part of which—perhaps all—were contemporaneous with the sandstones, is a part of the dissected peneplain; but to fix the date of the completion of the peneplain, we must turn to evidence presented in New Jersey.37 There we learn that by the close of Cretaceous times, the country was eroded nearly to baselevel, and we may therefore speak of the relative position of the land and sea, to which the land was at this time reduced, as the Cretaceous baselevel, and this land surface as the Cretaceous peneplain.

Elevation of the peneplain. In post-Cretaceous, presumably early Tertiary38 times, the land was elevated to nearly its present height and remained at that altitude, so far as topographic evidence shows, during Tertiary times. The proofs of this elevation are the valleys which the streams have sunk below the general level. That this was not a simple uplift, but was accompanied with tilting and warping, is clear from the following considerations. The depth to which a stream can cut its valley depends directly upon its height above baselevel. If the present surface were a peneplain uniformly elevated, the head waters and middle courses of a river would not be cut so deep in the surrounding plain as its lower course. But the reverse is true of the rivers of Connecticut. The depth of the valley increases inland, being greater in those regions where the peneplain was raised the highest. A comparison of the upper and lower valleys of the Housatonic, Naugatuck, Quinnebaug, and of the Connecticut at Middletown, where it enters the plateau, and at its mouth, will give some idea of the amount of the warping. It will not give an exact measure of it for several reasons: first, the upper courses of the rivers have not yet reached the present baselevel; second, the present altitude of the uplands is the result of the post-Cretaceous uplift and warping, plus a probable later post-Tertiary uplift (to be mentioned later), besides several minor oscillations, the last of which was downward, and is recorded near the coast in the drowned condition of the rivers. As has been already said, the peneplain is highest in the northwest, and gradually declines to sea level toward the south and east.

Consequences of the uplift. The consequences of this uplift are seen in the valleys, which are cut into the peneplain, and which have destroyed the level character of the country. In the hard crystalline rocks the valleys are generally narrow and deep, with bold slopes;39 where they are cut in the crystalline limestone, they are wider and more open. In marked contrast, however, is the lowland on the Triassic area in which only the trap ridges remain to tell of the former altitude of the general surface, and the immense amount of erosion which has taken place on the soft sandstones and shales. Indeed erosion has progressed so rapidly on these soft rocks, that they have been worn down almost to a new baselevel in the same length of time in which the hard crystallines have been only trenched. This fact cannot be too strongly emphasized. The broad sandstone lowland from New Haven north into Massachusetts has been carved out of the uplifted peneplain in soft rocks, during the same time in which the Connecticut has excavated its gorge in the crystallines below Middletown, and the Housatonic has opened its upland valley on the limestones. The difference in results is due not to a difference of time, but to the difference in the relative hardness of the rocks.

On the basis of this principle the age of certain river gorges to which reference will be made later can be fixed. The narrow passage of the Quinnipiac through a sandstone ridge southwest of Meriden cannot belong to the same cycle of erosion as the broad sandstone lowland on either side of it, but manifestly must be much younger. So, also, the narrow passage of the Farmington at Tariffville, where it crosses the trap ridge through a gorge free from drift, is of much later date than the broader valley more or less encumbered with drift which the upper part of the same river has cut in the hard crystalline schists. Cook’s Gap in the trap sheet west of New Britain is much broader than either of the above, and belongs to the Tertiary cycle of erosion, although as I shall endeavor to show later, it was probably not occupied by a stream during the whole cycle. In marked contrast, also, with the Tariffville gorge is the gap by which the Westfield river in Massachusetts cuts the trap ridge. This gap was formerly broad and open—the result of Tertiary erosion—but is now filled with drift, in which the river is at present working. Since these two rivers are essentially the same in size, are now at the same level, and the rock is the same in both cases, the only explanation for the difference in the two passages is that they belong to different cycles.

To recapitulate, the results of the post-Cretaceous uplift are seen in the valleys which have been cut in the peneplain. The narrow valleys in the gneisses and schists, the upland valleys in the limestones, the wide open, drift encumbered gaps in the trap ridge,—Cook’s and the Westfield river gaps,—the broad open lowland on the sandstones, are all the result of erosion in this cycle. The Quinnipiac gorge in the sandstone, and the Tariffville gorge in the trap are just as surely of a later date. They do not at all accord with the work of the earlier cycle either in size, angle of slope, or depth.

This conclusion is somewhat at variance with an opinion expressed by Professor J.D. Dana,40 but it seems justifiable in view of the successive cycles in the physical development of the region. In another part of this article I shall consider these gaps again in connection with their river histories, and shall give additional reasons why I venture to differ from so eminent an authority.

Length of this cycle. This cycle of erosion beginning with the post-Cretaceous uplift was not so long as the preceding cycle. In the earlier one the whole state was reduced to a peneplain; in the later cycle only the soft Triassic sandstones were brought near to baselevel. It probably lasted through Tertiary times, and was brought to a close by a slight uplift. The result of this uplift is well shown in Pennsylvania41 and New Jersey.42 It is not well shown in Connecticut, but there seem to be some traces of it in the trenches the rivers have cut below the level of the sandstone peneplain. However, these trenches are so much obscured by drift that a positive statement is not warranted. It may, however, be spoken of provisionally as the post-Tertiary uplift. There may have been later oscillations of small amount, probably were; here and there are shreds of evidence which point to such oscillations, but only one movement has had an effect upon the topography, which can be recognized. The fjorded condition of all the rivers along the Sound—the Norwalk, Saugatuck, New Haven bay, Niantic and Thames are the best examples—shows that within comparatively recent time there has been a slight subsidence of the land. But this movement is not to be compared in amount with those of the earlier cycles.

The drift. Over all the state in varying thickness lies the glacial drift, either in its typical unmodified development as till, or in its modified form, as river terraces, kames, eskers and sand-plains. It is of importance in this connection only as it has affected the topography of the country and so modified the drainage. Examples of these modifications will be mentioned later.

RÉsumÉ. There was first a long cycle of denudation in pre-Triassic times, during which the contorted crystallines were worn down to a comparative level; second, a cycle of subsidence, deposition and volcanic outburst, during which the sea entered the crystalline trough, and the Triassic conglomerates, sandstones and shales were deposited with the intercalated layers of lava; third, a long cycle of elevation, folding, faulting and erosion, during which the sedimentary beds were elevated—tilted into the present faulted monocline, and this constructional surface worn down to a baselevel of erosion in late Cretaceous times. Each of these cycles probably represents the sum total of several subordinate cycles. There was, fourth, a post-Cretaceous uplift inaugurating a period of erosion lasting through Tertiary times and resulting in the formation of valleys in the hardest rocks, and a lowland approaching baselevel on the Triassic sandstones and shales; fifth, a probable late or post-Tertiary uplift, when the valleys were deepened and the lowlands trenched—obscure in Connecticut, but well shown farther south; sixth, the land, near the coast at least, is now slightly lower than it has been in the not remote past, as is shown by the fjords.

With the changes of the physical geography clearly in mind, the rivers of Connecticut may now be examined in respect to their conditions of origin, the number of cycles through which they have lived, and the approach they have made to mature old age. But at the very outset a serious difficulty is encountered, for the geological structure of the state is nowhere well described, nor have topographic maps of all the districts yet been issued. Since the structural details are to some extent unknown it is unwise in many cases to attempt more than tentative conclusions. Several of the problems to be presented cannot be considered as settled. Considerable progress toward a final settlement will have been made, however, if the conditions of the problems are made clear, various hypotheses suggested, and the attention of workers in this field called to these questions.

Early drainage. Of the drainage of Connecticut during Jurassic and Cretaceous times very little can be said. It is not even known whether it was consequent upon the Jurassic tilting and faulting, or whether these deformations were so slow in their movement that the rivers persisted in spite of them. It may have been that the larger rivers were victorious, while the smaller were conquered and compelled to assume new consequent courses. Whatever was their origin there must have been abundant opportunities during the long erosion which resulted in the Cretaceous baselevel, and again in the period of revived and quickened degradation succeeding the post-Cretaceous uplift, for the streams to adjust themselves in a large degree to the geological structure. The contrast of hard and soft beds and the great elevation must have been potent factors in bringing to pass such a result. We expect to find the streams so far re-adjusted as to render improbable the discovery of their manner of origin.

The Housatonic, a re-adjusted stream. The best example of re-adjustment is found in the northwestern part of the state where the Housatonic and some of its branches follow well adjusted courses. From its headwaters, near Pittsfield, Mass., to New Milford, Conn., it has nearly all the way chosen its course along the Cambrian crystalline limestones in preference to the harder granites and gneisses on either side. The stratigraphical relationships of the limestone are not fully understood, but they seem to be deeply eroded anticlines and synclines, whose axes plunge north or south at various angles. The course of the river, if the drainage was consequent, was at first along the synclinal valleys, passing from one to another across the lowest points in the anticlinal ridge between them. But by a series of changes43, resulting from the differential rates of erosion as hard or soft beds became exposed, the river previously to the Cretaceous baseleveling, seems to have re-adjusted its course to the softer limestones. However, there are several places where this conformity to structure does not seem to be the law; where the river departs from a limestone valley to flow for a time in the crystallines, only to return to the limestone again. The most marked instance of this is in the towns of Sharon and Cornwall, where the river leaves the limestone valley, which continues to the southwest, and flows for ten miles in a narrow gorge in the gneiss, only to again enter at its northern end a long narrow bed of limestone. The following seems to be the probable explanation. When the land stood at the elevation represented by the Cretaceous peneplain, these hard beds were below or but very slightly above baselevel, and were therefore undiscovered by the stream or had just begun to make themselves known late in the cycle. Had they been reached early in the cycle, when the stream was far above baselevel and presumably before many of its tributaries had been developed, and when it was therefore a smaller river, it is quite probable that further re-adjustments would have occurred, and the stream been led away from the hard rocks onto the softer beds to the west; but when they were reached the stream had cut so deeply and so nearly to baselevel that it was safe from capture. After the elevation of the peneplain the stream was revived and disclosed more and more of these hard beds, but was then, owing to the development and head-water growth of its tributaries, too important a river to be diverted by any rival. A river of this kind may be said to be “conformably superimposed” in distinction to one which is superimposed from an unconformable cover.

Revived streams. It is important to recognize the effect of the post-Cretaceous uplift upon the rivers at that time established. As the land was baseleveled and the velocity of the streams decreased, they lost in large degree their cutting power and sluggishly meandered more or less in broad flood-plains. During and for a period after the uplift, their cutting power was restored to them by virtue of their increased velocity and they excavated the deep narrow valleys which we find in the crystalline highlands. The upper course of the Housatonic is a good example of a river re-adjusted to the structure during one cycle, revived by uplift to a second cycle of erosion, and in places “conformably superimposed” upon structures from which it would have been led away in the ordinary course of re-adjustment. Its tributaries, the East Aspetuck, Still, Shepaug, and Pomeraug follow courses re-adjusted in one cycle and revived in a later uplift.

We can assert with the more confidence that such was the history of the upper Housatonic, because we find in other states, in regions whose history has been the same, similar examples of “conformably superimposed” and “revived” streams. The Musconetcong and Pequest, highland rivers of New Jersey, are streams “revived” from mature old age to vigorous youth and “conformably superimposed” upon saddles of gneiss between two limestone valleys.44

Unconformable rivers. In considering the course of the lower Housatonic we meet with some difficulty at the outset. In the southern part of the town of New Milford the river leaves the limestone belt which continues with some slight interruptions to the Hudson, and swings sharply into the crystalline plateau in a southeasterly course until it is joined by the Naugatuck, when their united waters flow south for a few miles to the sound. The course of the lower Connecticut is even more surprising. At Middletown it leaves the broad open Triassic sandstone lowland, and through a gorge enters the plateau, which has an average elevation of 600 to 700 feet. In this plateau of crystallines the river has sunk its valley nearly to sea-level. The slopes are steep compared to the lines in the sandstone lowland, and the contrast between the two parts of the river is one of the striking features of Connecticut scenery. Several theories may be framed to account for the curious behavior of these two rivers, but none of them are free from all difficulty.

As a consequent river. The lower Connecticut has been thought45 to be a revived river, whose course was consequent upon the post-Triassic tilting and faulting. The faulted monocline seems to have had the shape of a half-boat, ends to the north and south, and one gunwale rising toward the west, the combined effect of the tilting and faulting being to swing the river to the southeast, where the keel of the boat was lowest. The probable existence of faults, with upthrow on the east, along the eastern margin of the Triassic rocks, is a difficulty in the way of the complete acceptance of this theory. Unfortunately too little is known about the structure of the western plateau to say whether the course of the lower Housatonic could be accounted for on such an hypothesis. On this theory the Connecticut would be consequent upon the Jurassic deformation, and revived by the post-Cretaceous uplift.

It may be suggested that the southeast courses are due to the tilting of the peneplain at the time of elevation, the plateau now being, as we have seen, much higher in the northwestern part of the state than elsewhere. But the acceptance of this theory necessitates a degree of smoothness and absence of even mild relief in the peneplain, which is hardly possible. The present average slope of the plateau is but a few feet per mile, and it seems incredible that so gentle a tilting could force rivers as large as these to take new courses. Besides, if the Housatonic and Connecticut were deflected, why were not the smaller streams—the Naugatuck and Quinnebaug—also given southeastern deflections? Clearly, this explanation is not the correct one.

Superimposition. It has been suggested that these courses may be inherited from a Cretaceous cover, which formerly stretched over Connecticut for a considerable distance, but of which no traces now remain in the state. On parts of Long Island the Cretaceous deposits are found, and it is not inherently impossible nor improbable that they once stretched far over the main land. In New Jersey46 several lines of evidence seem to show that the Cretaceous beds formerly extended across the Triassic, probably to the margin of the highland plateau. The curious drainage of the Watchung Crescent is one evidence of this, but the other proofs are along entirely different lines, so that there is apparently good evidence that the Cretaceous beds extended twenty-five miles or more farther inland. If, in the time which has elapsed since the deposition of these beds, there has been erosion sufficient to strip them off from such a broad area in New Jersey, may they not, in Connecticut, under presumably similar conditions, have been equally eroded?

There is much which makes this hypothesis attractive, and, as the facts were first studied, it seemed the most likely one. It affords a good explanation, not only for the courses of the Housatonic and Connecticut, but also for other rivers along the sound. It seems, also, at first thought, to be well supported by analogy from New Jersey. But a closer study of the situation in that state reveals marked differences in the attendant circumstances. There the soft Triassic sandstone must have been worn down to a lowland early in the Cretaceous cycle, perhaps by the close of Jurassic time or thereabouts, while the harder crystallines retained a strong relief. The slight subsidence, which marked the beginning of marine Cretaceous in New Jersey, allowed the Cretaceous sea to transgress rapidly the baseleveled sandstones to the foot of the crystalline hills, but not to cover them to any extent. It is not probable that the crystallines in Connecticut had been brought nearer to baselevel than those in New Jersey at the time of the Cretaceous deposits. There is no evidence to show that the subsidence was greater in Connecticut than in New Jersey, and, therefore, from a priori considerations, the conclusion would seem to follow that the subsidence, which permitted the Cretaceous sea to cover the Triassic sandstone area of New Jersey, was not sufficient to permit the sea to cover the then unsubdued crystalline hills of Connecticut. Although this hypothesis is not to be hastily thrown aside, for theoretical reasons, yet it would seem necessary to hold it very lightly, at least until some positive proof is found of the former existence of the Cretaceous or some later formation in that region. The first suggestion, that the lower Connecticut was a consequent river in the Cretaceous cycle and was revived by the post-Cretaceous uplift, is, at the present state of knowledge, the most probable.

The Farmington. The roundabout course of this river presents another interesting problem, which is not free from difficulties. From its source in Massachusetts it flows southeast across the crystallines to the village from which it takes its name, where it turns abruptly north along the Triassic sandstones for ten or twelve miles, when with another wide sweep it crosses the trap ridges at Tariffville by a deep gorge, and resumes its southeasterly course to the Connecticut. Of this latter part I will speak later, but now arise the questions, “what has been the history of this river,” and “why does it turn north at Farmington?”

The Farmington in the Tertiary cycle. A course more accordant with the structure would seem to be south along the Quinnipiac and Mill river valleys to the sound at New Haven. As has been said before (page 376), Prof. Dana has expressed the opinion that the gorge at Tariffville was occupied by the Farmington in Tertiary times, and that the Westfield river gap further north and the gorge of the Quinnipiac southwest of Meriden are also of earlier date than the glacial epoch. One reason has also been given why I differ from him in regard to the Quinnipiac and Tariffville gorges—they are narrower and steeper than those made in similar rocks during the Tertiary cycle. But more than this, the constructional topography, resulting from the tilting and faulting of the region, could not, it would seem, have caused the Farmington to take its present course. Even if it had taken this roundabout course during the baseleveling of the country, it must, since it would have had to cross three trap sheets, have been captured and led to the sea by the shorter and easier way along the sandstone area. The fact that the Connecticut probably persisted in its consequent course is no argument for similar conditions for the Farmington, because the latter is much the smaller stream, and so more easily captured. Nor could the river have been forced into this course during or after the post-Cretaceous uplift, for the land was then raised more at the north than at the south, and any changes from this cause would have been to confirm the river in its southward course. It is very probable, therefore, that in at least the latter part of the Tertiary cycle, the Farmington did not have its present course, but followed the open sandstone valley, along the course of the Quinnipiac and Mill rivers of to-day. The earlier history of this river is purely conjectural; one fact may shed a little light upon it, a fact which may indicate that this course was an adjusted one taken during Tertiary times.

In pre-Tertiary times. Origin of Cook’s Gap. A few miles southeast of where the river emerges from the crystallines, the trap ridge is cut by a deep wind notch—Cook’s Gap—through which the New York and New England Railroad passes west from New Britain. As was pointed out some time ago by Prof. Davis,47 this is not a fault gap, because the alignment of the ridge is not broken, but it is probably an abandoned water gap, the head-waters of the stream which formerly occupied it having been abstracted by a rival, which did not have to cross a hard trap ridge. Perhaps this river was the ancestor of the present Farmington, and in that case its history would seem to have been as follows. A stream consequent upon the constructional topography after the faulting and tilting at the close of the Triassic, it had its upper course on the crystallines, its lower on the sandstones and buried trap sheets. In its old age it crossed by a shallow gap the trap sheet, which had been uncovered by erosion. In the second or Tertiary cycle it was simply a revived stream quickened to a new life by the post-Cretaceous uplift of the peneplain. This uplift gave opportunity to a rival stream, which did not have to cross the hard trap beds to intercept the waters of the old Farmington, and lead them out by a shorter, easier path, probably down the sandstone valley west of the trap ridge. The path across the trap was abandoned, and the notch became a wind gap; the river following its new course, until the incursion of the ice-sheet interrupted its normal development. This is of course almost entirely speculative. Cook’s Gap is best explained as an abandoned river gap; the Farmington is the nearest river of a size proportional to the size of the gap, and the hypothesis is a rational one. There is, however, no direct evidence that the Farmington once occupied Cook’s Gap.

The Tariffville cut. Before attempting to answer the second question, “why the river flows north at Farmington?” let us consider for a moment the history of the Tariffville cut. The river occupies a gorge whose sides are steep and talus covered, but which is not at all clogged with drift. There is naturally no room at or near the water level, even for the wagon road, place for which has been blasted near the top of the gorge. The profile of the gap shows a gentle ascent from the top of the gorge, up to the nearly level crest line of the ridge. That is to say, the recent gorge has been cut in the bottom of a sag in the ridge. We have already given our reasons for believing that the gorge here is much younger than the Westfield river gap; that it is a part of the work of the next cycle; that it is post-Tertiary. The sag, however, in the bottom of which the gorge is cut, is clearly of the earlier cycle. The bottom of the sag is much above the level to which the rivers had cut their valleys in the late Tertiary, and, therefore, it is certain that a river could not have occupied it at the close of that cycle. It was probably an abandoned water-gap whose stream had been captured in the same way and in the same cycle as the river, which formerly occupied Cook’s Gap.

The fact that the sag and gorge, although located very near a fault line, do not correspond to it, but are transverse and independent of it, is instructive and needs a moment’s attention. It seems probable that the stream consequent upon the faulted blocks would have flowed down the slope of the tilted block and then along the fault line at the foot of the fault cliff and would have held this course during the baseleveling of the country. When the area was baseleveled the stream must have swung from side to side in its broad flood plain, and thus departed from the fault line. When it was revived by the post-Cretaceous uplift, it was confined to the course it had unwittingly taken on the sandstones just above the hard ridge, and it was forced to cut down through the trap. Subsequently a rival, which did not have to work against this obstacle, abstracted its head waters and the gap was abandoned. The accompanying diagrams may make this easier to understand. Figure3 is a cross-section of the faulted monocline, R showing the position of the river along the foot of the fault cliff. The line B L represents the surface of the country after baseleveling, the trap outcrops forming low hills (much exaggerated in the diagram). Figure4 shows the dislocated trap sheets, the fault line and the winding course of the river, which has abandoned the fault line except where it passes between the low trap hills. Here the country is at baselevel. Figure5 represents the region after the elevation and resulting erosion. The trap ridges have become more pronounced, and have migrated eastward in the direction of the dip. The river has been slowly let down upon the northern one from the sandstone at point G and has there cut into the solid trap.

The transverse notch of Cook’s Gap, already described, was probably located in a somewhat similar manner, but the case is not so clear as at Tariffville.

Gravel terraces of the Farmington. A consideration of some facts concerning the height and slope of the terraces along this part of the river may give a clue to the answer to our question. One-half a mile east of Tariffville and east of the trap ridge, the highest terrace is 210 to 215 feet. Half a mile south of the same place but west of the ridge the height is 275 feet.48 The top of the gorge at Tariffville is about 190 feet above the sea-level. It does not seem probable that these highest terraces were ever continuous over all the Farmington valley. But if they represent the level reached by the maximum flood accompanying the melting of the glacier, the great difference in their height on the two sides of the trap ridge, in connection with the other evidence already noted, gives strong reason for believing that the gorge as it exists to-day had not then been cut. A mile and a half east of Tariffville there is a lower terrace which is wide-spread. Its general height is about 190 feet, in places a little more. In this terrace the lower part of the Farmington has cut a trench 90 to 100 feet deep. The shape of the valley makes clear the fact that before this trench was cut the river flowed at about the 190 foot level, which is the height of the bottom of the sag at Tariffville. On the west side of the trap ridge there is also a more or less wide-spread terrace at about the same height. It seems very probable therefore that the river was raised to the level of the old sag in the trap ridge by the building of these terraces.

The present average southward slope of the highest terraces west of the trap ridge from Northampton, Mass., to Farmington, Conn., forty-four miles, is seven inches per mile,49 and Professor Dana is inclined to believe that this is approximately the slope at the time the terraces were built. The character of the deposits shows that the current which formed these deposits flowed south. The present river, flowing north, falls twenty feet between Farmington and Tariffville, or 1? feet per mile. The reversal of the river was probably determined by two factors. Near the village of Farmington, the waters of 200 square miles of territory are poured into the valley by the upper Farmington and its tributary, the Pequabuck. During the terrace building stage the great mass of dÉbris contributed by these streams was deposited where the steep gradient of the highlands was exchanged for the gentle slope of the lowland. The main north and south valley was thus choked by the dÉbris of its tributaries and a long stretch of comparatively still water extended north from Farmington, in which nearly horizontal deposits were made. South of Farmington the terrace deposits are much coarser than to the north, and the face of the terraces is much greater. It is not impossible that, as the deposits between Farmington and the Massachusetts state line approached nearer and nearer to horizontality, the waters of the upper Farmington began to divide, part flowing north and part south, the northward flowing portion finding an outlet at the sag at Tariffville. If this was the case, the terraces between Farmington and Tariffville must have had a slight slope to the north. Their present southward slope could readily be accounted for by the re-elevation of the land after the disappearance of the ice. This explanation rests upon the ability of the upper Farmington and the Pequabuck to have completely dammed the southward flowing current and turned it northward by the great mass of their deposits. If this was not the case, and there may be some doubt on the matter, the subsidence which accompanied the later stages of the ice-retreat is the other factor in the problem. It is estimated that an average depression of 1.25 feet50 through the Connecticut valley would restore it to an altitude approximating that at the close of glaciation. It seems highly probable that these terrace-deposits were built before the maximum depression was reached. If this was the case, the depression would be efficient in reversing the Farmington, and this factor would supplement the first. It is impossible at present to say to what extent these two factors enter into the problem. That they are not mutually exclusive is evident, and that they are together quantitatively competent seems certain. Among the several hypotheses which have been considered, this seems the most probable, and in the light of the present evidence the most rational.

At first thought it might seem that if the Farmington was reversed by the differential subsidence of the land, the Connecticut ought to have suffered a similar fate, and since it did not, the explanation cannot apply to the Farmington. But the terraces of the Connecticut have a much greater southward slope than those on the smaller river, and the depression was not sufficient to reverse the stream. The conditions on the two sides of the trap ridge were not the same.

To sum up, then, the history of the Farmington seems to have been as follows: Its original consequent course was southeast on the crystallines and perhaps across the trap ridge at Cook’s Gap, from which course it was turned in the Tertiary cycle by a stream whose course was approximately that of the Mill river of to-day. The damming of the valley by the deposits of the Upper Farmington, and the depression in the north accompanying the ice retreat, reversed the river at Farmington, and it took a new course on the terrace deposits, escaping by the sag in the trap at Tariffville into the Connecticut valley.

The Quinnipiac. The gorge of the Quinnipiac, already mentioned several times, seems closely comparable to the gorge of the Farmington. It is not of the Tertiary cycle, and is best referred to the inter-glacial or post-glacial epochs. We should expect the Quinnipiac, instead of turning eastward, to cut through this sandstone ridge, to continue southward along the Mill river valley. Dana51 finds from the heights of the terraces that the drainage of the terrace-building period was not along the Quinnipiac, but along the Mill river, and concludes that the Quinnipiac gorge was obstructed by an ice-dam. I have not as yet studied it enough in detail to do more than express the opinion here reiterated, that this gorge is later than the cycle in which the open sandstone lowland on either side of it was excavated. Its topographic form would put it in the cycle which has been called post-Tertiary.

The Scantic. In the Scantic we have a typical example of a river whose lower course is manifestly of a later date than the upper. In this it is similar to several of our Atlantic rivers, notably those of North Carolina, whose upper courses are on the Piedmont crystallines, being probably established previous to the Cretaceous baseleveling, and whose lower courses stretch seaward over the unconsolidated Tertiary deposits of the coastal plain. As the plain of these recent deposits emerged from the sea, the rivers were forced to extend their courses eastward over the freshly raised surface to the retreating shore line. The Scantic river has a similar history. Its upper course in southern Massachusetts on the crystalline plateau is a remnant of the drainage established before Cretaceous baseleveling and revived by the subsequent uplift. How much that revived drainage has been modified by drift can only be determined by long field study, but the topography, as read from the topographical atlas would seem to indicate, that it has not been much. The valleys were undoubtedly clogged with drift, and the drainage area may be somewhat modified, but the drainage seems to be substantially along the same lines.

Just below the village of Hampden, the Scantic leaves the plateau and enters the Triassic lowland. From this point to its mouth at the Connecticut, opposite Windsor, a distance of twenty miles, it flows nearly all the way through the gravel, sand and clay deposits of the period of ice-retreat. The topography of the lower course of the river is entirely characteristic of a stream which has recently attacked a level, easily eroded district. The inter-stream surfaces are broad and flat; the descent to the stream bed which is sunk seventy or eighty feet below the general surface is exceedingly steep. These two lines, that of the inter-stream surface and that of the valley side, meet at a sharp angle. The side streams are as yet very short, and have cut narrow gorges down to the main river. Tributary to them are deep side ravines, whose bottoms ascend rapidly to the inter-stream surface, the whole making a dendritic system of drainage in its earlier stages. The Scantic, having reached base level in its lower course, has developed a narrow flood-plain.

Manifestly this part of the river valley is of much later date than the upper part. If, during the period of ice-retreat, the lower Connecticut valley was an estuary, the Scantic was a much shorter river than at present. Its mouth could not have been far from the point where now it leave the crystallines, but as the land was elevated and estuarine conditions gave place to fluviatile, the Scantic lengthened “mouthward,” consequent upon the minor inequalities of the newly made beds. The effect would be substantially the same if the terraces were built by great valley floods, as Dana supposes. In pre-glacial times this river, in common with several other rivers rising on the crystallines and flowing into the Connecticut, had courses of various lengths over the Triassic sandstones, but these old valleys are lost entirely, the later trenches in the terrace deposits being altogether independent of them.

Other examples. The lower Hockanum, Farmington, Park, and the entire length of many short streams are similar to the lower Scantic, and originated under similar conditions. Stony Brook, a little stream north of Windsor Locks, presents the same features, but with this variation: It is superimposed through a thin layer of drift upon the sandstone, into which it has cut a deep, picturesque gorge. The Hockanum and Farmington are also “locally superimposed” in a few places. The Connecticut, also, north of Middletown, although following its pre-glacial valley, has departed in numerous places from its former bed, and has cut down through the valley-filling onto ledges of rock beneath. The water-power at Enfield, Conn., and at Turner’s Falls and Bellows Falls, Mass., is the result of this superimposed position.

Abandoned gaps. Many abandoned water-gaps must exist among the hills of the state. Cook’s Gap, through which the New York and New England Railroad crosses the trap ridge, three miles west of New Britain, has already been discussed. It must not be confounded with the majority of the other gaps in the trap ridge, which are oblique, break the alignment of the ridge, and are due to faults.

The New York and New England Railroad in ascending to the eastern plateau passes through Bolton Notch, a few miles east of Manchester. This notch, also, is an abandoned river bed but, as it seems, abandoned at a later date and for another reason than that assigned for Cook’s Gap. The drift is very heavy in this region, and the most probable explanation is that the post-glacial streams do not altogether follow pre-glacial valleys. This gap, used by turnpike and railroad, testifies of another and older drainage system.

That in this brief article all the problems connected with the Connecticut rivers have been solved, or even noted, is not to be expected. It is hoped, however, that the work done may prove a help to further study of the same regions, and that the tentative conclusions advanced may be substantiated by further investigation.

Henry B. Kummel.

33The author desires to express his obligation to Professor W.M. Davis for aid in the preparation of this article. It was first written under his direction and with the help of his suggestions when the author was in the graduate school of Harvard University. Prof. Davis is not responsible, however, for the statement of the views herein advanced, although in general it is believed that he is in accord with them.