their margins, and deposited on their seaward slopes. The shelves are thus built outward and are largely constructional topographic forms. Their outer slopes, where best defined, represent about the "angle of repose" in water of the fine material of which they are composed. These slopes are in several regions so precipitous that they probably would not retain their present forms, but descend in landslides, should the restraining pressure of the sea-water be removed.

In certain favoured regions, as about the southern extremity of Florida, over an extensive area in the West Indies, and on both sides of Central America, the conditions favour the growth of reef-building coral-polyps, and portions of the continental shelf in that region are covered with an irregular layer of living coral and dead coral rock. The importance of this resistant superficial layer on the minor features of the relief of the submarine banks, etc., needs to be considered in studying the secondary topographic features of many portions of the floor of shallow tropical seas.

In addition to the débris from the land and the rain of the hard parts of organisms from the water covering the continental shelf there is in northern regions a third but less important source of material furnished by floating ice. About the northern shores of America sea ice forms in winter, some of which is frozen fast to boulders and stones in shallow water, and when this ice-foot, as it is termed, is adjacent to steep cliffs, rock débris falls upon it. When

the ice becomes broken into cakes in the spring-time or during storms, it floats away, under the influence of the winds and currents, and as it melts drops its freight on the floor of the sea. This shore ice seldom travels far, and is probably not an important factor in the building of continental shelves. Of greater interest are the bergs derived from glaciers, especially in Greenland, many of which contain hundreds of thousands and even millions of cubic feet of ice and travel hundreds of miles before melting. In some instances these bergs carry with them rock masses, mud, etc., derived from the land over which their parent glaciers flowed, and as they melt, distribute this material over the sea-floor. The greater portion of this ice-carried freight derived from Greenland is dropped on the continental shelf, and not infrequently reaches the latitude of Halifax, and even journeys farther south. This berg-carried débris is mainly deposited on the continental shelf, for the reason that the cold currents which bring the bergs southward follow the coast in a general way, and are bordered on their seaward margins by warmer currents flowing northward. To the north of Nova Scotia the additions of material to the continental shelf through the agency of bergs is considerable in the aggregate, and as the process has been in operation for thousands of years, the banks or shoals in the sea off the Newfoundland coast are due in part to this cause.

Ice-carried débris forms an important source of material for the building of the continental shelf from New England northward and westward about the shores of North America, including Greenland, to Bering Sea, and to a less extent on the south coast of Alaska, where many comparatively small bergs are set afloat by glaciers which reach tide-water. Supplementing the distribution of débris over the continental shelf by shore ice and bergs, is the similar work carried on by the ice discharged into the sea by northern rivers, such as the St. Lawrence, Mackenzie, and the Yukon.

During the glacial epoch great ice-sheets like those now discharging bergs along the Greenland coast, but vastly

larger, entered the Atlantic all the way from New York to the Arctic Ocean, and along the Pacific coast from the Aleutian Islands to the State of Washington. During certain periods of this time of intense glaciation great additions of ice-borne débris must have been made to the continental shelf. The banks to the east of Newfoundland and other similar shoals as far south as Nantucket are probably due in large part to the débris deposited by the glaciers which formerly entered the sea in that region. It is of interest in this connection to note that the glaciers, even at the time of their greatest expansion, could not have extended beyond the seaward margin of the continental shelf, for the reason that on passing that boundary and entering deep water they must have broken off and given origin to bergs.

Submerged River Channels.—One of the most interesting features in connection with the continental shelf bordering North America is that its generally plane surface is trenched in several places by cañon-like depressions similar to the narrow steep-sided valleys which streams sometimes cut in the surfaces of plateaus. This suggestion that the surface of the continental shelf is crossed by stream-cut channels is supported by the fact that several such depressions, leading seaward from the present mouths of large rivers, have been discovered by the sounding-line. The best known example occurs off the mouth of the Hudson and has been traced from New York Bay about 120 miles seaward to the edge of the continental shelf. It is deepest and best defined on the outer portion of the submerged plateau, where for a distance of 23 miles, beginning 97 miles from Sandy Hook, it has an average width of 3 miles and a maximum depth of about 2,500 feet below the surface of the bordering submarine plain, which has 20 fathoms of water over it. This cañon opens out in the seaward face of the plateau and forms a deep notch in the generally uniform crest-line of that escarpment. Farther "up-stream," so to speak, the channel narrows to a mile and a quarter, with some irregularities in depth, and near Sandy Hook it is not apparent, owing to the amount of débris, largely sand, swept

into it by shore currents. This evidence, strengthened by the fact that the true rock-cut valley of the Hudson as far as Troy is filled with clay and sand to a considerable but unknown depth, is abundant proof that the land was formerly higher than at present by at least 3,000 feet, and that the now submerged continental shelf off Long Island was then a plain above water, across which the ancient Hudson was extended. The river flowed across this plain for a sufficient length of time to excavate a cañon over 2,500 feet deep and 3 miles wide from crest to crest of its walls in its seaward portion. This submerged channel has the characteristics of a young, stream-cut valley and suggests that the plain across which it flowed to the eastward of Long Island was a submerged continental shelf previous to being upraised so as to be trenched by the Hudson.

The evidence as to changes in the elevation of the Atlantic coast furnished by the submerged valley of the Hudson does not stand alone. Similar but less well-defined channels have been discovered by soundings off the mouths of the Delaware and the Susquehanna, while the most remarkable instance of all is furnished by the submerged valley of the St. Lawrence, which has been traced through the Gulf of St. Lawrence and out to the brink of the submerged continental escarpment some 200 miles eastward of Nova Scotia. The tide now rises and falls in the St. Lawrence to within a few miles of Montreal; that is, the "Greater St. Lawrence" has lost about 1,000 miles of its length owing to a downward movement of the land.

Evidence of the nature just considered is lacking, or, more correctly, surveys and soundings which would perhaps reveal the presence of submerged river channels have not been made about the shores of the more northerly portion of the continent, but instructive results in this connection are to be expected when that region is thoroughly studied.

On the Pacific coast several transverse channels in the