When subglacial streams were confined to definite channels, the same may have been constant in position, or may have shifted more or less from side to side. Where the latter happened there was a tendency to the development of a belt or strip of stratified drift having a width equal to the extent of the lateral migrations of the under-ice stream. Where the channel of the subglacial stream remained fixed in position, the deposition was more concentrated, and the bed was built up. If the stream held its course for a long period of time, the measure of building may have been considerable. In so far as these channel deposits were made near the edge of the ice, during the time of its maximum extension or retreat, they were likely to remain undisturbed during its melting. The aggraded channels then came to stand out as ridges. These ridges of gravel and sand are known as osars or eskers. It is not to be inferred that eskers never originated in other ways, but it seems clear that this is one method, and probably the principal one, by which they came into existence. Eskers early attracted attention, partly because they are relatively rare, and partly because they are often rather striking topographic features. The essential conditions, therefore, for their formations, so far as they are the product of subglacial drainage, are (1) the confining of the subglacial streams to definite channels; and (2) a sufficient supply of detritus. One esker only has been found in the region under consideration. It is located at the point marked j, Plate [II], seven and one-half miles northeast of Merrimac and one and one-half miles south of Alloa (g, Plate [II]). The esker is fully a quarter of a mile long, about thirty feet high, and four rods wide at its base.

Subglacial deposits of stratified drift were sometimes made on unstratified drift (till) already deposited by the ice before the location of the stream, and sometimes on the rock surfaces on which no covering of glacier drift had been spread.

It is to be kept in mind that subglacial drainage was operative during the advance of an ice sheet, during its maximum extension, and during its retreat, and that during all these stages it was effecting its appropriate results. It will be readily seen, however, that all deposits made by subglacial waters, were subject to modification or destruction or burial, through the agency of the ice, and that those made during the advance of the ice were less likely to escape than those made during its maximum extension or retreat.

RELATIONS OF STRATIFIED TO UNSTRATIFIED DRIFT.

When it is remembered that extraglacial and subglacial waters were active at all stages of an ice sheet's history, giving rise, or tending to give rise to all the phases of stratified drift enumerated above; when it is remembered that the ice of several epochs affected much of the drift-covered country; and when it is remembered further that the edge of the ice both during advance and retreat was subject to oscillation, and that each advance was likely to bury the stratified drift last deposited, beneath unstratified, it will be seen that the stratified drift and the unstratified had abundant opportunity to be associated in all relationships and in all degrees of intimacy, and that the relations of the one class of drift to the other may come to be very complex.

As a result of edge oscillation, it is evident that stratified drift may alternate with unstratified many times in a formation of drift deposited during a single ice epoch, and that two beds of till, separated by a bed of stratified drift, do not necessarily represent two distinct glacial epochs. The extent of individual beds of stratified drift, either beneath the till or inter-bedded with it, may not be great, though their aggregate area and their aggregate volume is very considerable. It is to be borne in mind that the ice, in many places, doubtless destroyed all the stratified drift deposited in advance on the territory which it occupied later, and that in others it may have left only patches of once extensive sheets. This may help to explain why it so frequently happens that a section of drift at one point shows many layers of stratified drift, while another section close by, of equal depth, and in similar relationships, shows no stratified material whatsoever.

Such deposits as were made by superglacial streams during the advance of the ice must likewise have been delivered on the land surface, but would have been subsequently destroyed or buried, becoming in the latter case, submorainic. This would be likely to be the fate of all such superglacial gravels as reached the edge of the ice up to the time of its maximum advance.

Streams descending from the surface of the ice into crevasses also must have carried down sand and gravel where such materials existed on the ice. These deposits may have been made on the rock which underlies the drift, or they may have been made on stratified or unstratified drift already deposited. In either case they were liable to be covered by till, thus reaching an inter-till or sub-till position.

Englacial streams probably do little depositing, but it is altogether conceivable that they might accumulate such trivial pockets of sand and gravel as are found not infrequently in the midst of till. The inter-till position would be the result of subsequent burial after the stratified material reached a resting place.

Complexity of relations.—From the foregoing it becomes clear that there are diverse ways by which stratified drift, arising in connection with an ice sheet, may come to be interbedded with till, when due recognition is made of all the halts and oscillations to which the edge of a continental glacier may have been subject during both its advance and retreat.