(9) _Different Directions of Movement._ If, after its maximum advance, the ice suffered merely a minor recession and then remained stationary, or nearly so, for a time, the general direction of its movement in a subsequent advance would probably be essentially the same as in the earlier. But if, after its maximum advance, the ice receded to a great distance, and especially if it entirely disappeared, a subsequent ice-sheet might have a very different direction of movement, since its center of accumulation and dispersion might be very different. It is conceivable that this center might shift during the history of a single ice-sheet. In this case there should be a gradual change in the direction of ice movement, not an abrupt one. If, therefore, there be found one sheet of drift made by an ice movement in one direction, overlaid by another sheet of drift deposited by ice moving in a very different direction, with an abrupt transition between them, such drift sheets would be presumptive evidence of distinct ice epochs. An exception would need to be made in the case of drift sheets along the margins of confluent or proximate ice lobes. In such cases, if the one lobe temporarily secured the advantage of the other, drift beds formed by movements from opposite directions might be found in vertical succession, without being evidence of separate ice epochs. It is no part of the purpose of this essay to point out the difficulties which might arise in the application of this criterion of diverse directions of ice movements. It is possible that gradual changes in the direction of movement might leave records which would seem to indicate abrupt changes instead. This possibility makes care necessary in the application of the criterion, but does not destroy its value. When not itself conclusive, this criterion may be so associated with differential weathering, differential erosion, forest beds, etc., that their combined testimony makes but one conclusion possible. The absence of evidence of radically diverse directions of movement during the time of deposition of the various sheets of drift, would be no proof that there were not distinct epochs. In the first place, the movements of different epochs might be harmonious--a condition of things more probable than any other if the more common views of the causes of glaciation be correct. In the second place, if the movements were diverse, the deposits might still be so similar that their differentiation, when the one is buried, might not be easily made. In the third place, the later ice might have so far incorporated the older drift material with that which belonged more properly to it, as to have destroyed all definition between them.