The two volumes of Gulf Stream water in the Fram's southern section of 1910 are divided by a volume of water, which lies over the Porcupine Bank, and has a lower salinity and also a somewhat lower average temperature. On the bank to the south of Ireland (Stations 1 and 2) the salinity and average temperature are also comparatively low. The fact that the water on the banks off the coast has lower salinities, and in part lower temperatures, than the water outside in the deep sea, has usually been explained by its being mixed with the coast water, which is diluted with river water from the land. This explanation may be correct in a great measure; but, of course, it will not apply to the water over banks that lie out in the sea, far from any land. It appears, nevertheless, on the Porcupine Bank, for instance, and, as we shall see later, on the Rockall Bank, that the water on these ocean banks is -- in any case in early summer -- colder and less salt than the surrounding water of the sea. It appears from the Frithjof section across the Rockall Bank, as well as from the two Fram sections, that this must be due to precipitation combined with the vertical currents near the surface, which are produced by the cooling of the surface of the sea in the course of the winter. For, as the surface water cools, it becomes heavier than the water immediately below, and must then sink, while it is replaced by water from below. These vertical currents extend deeper and deeper as the cooling proceeds in the course of the winter, and bring about an almost equal temperature and salinity in the upper waters of the sea during the winter, as far down as this vertical circulation reaches. But as the precipitation in these regions is constantly decreasing the salinity of the surface water, this vertical circulation must bring about a diminution of salinity in the underlying waters, with which the sinking surface water is mixed into a homogeneous volume of water. The Frithjof section in particular seems to show that the vertical circulation in these regions reaches to a depth of 500 or 600 metres at the close of the winter. If we consider, then, what must happen over a bank in the ocean, where the depth is less than this, it is obvious that the vertical circulation will here be prevented by the bottom from reaching the depth it otherwise would, and there will be a smaller volume of water to take part in this circulation and to be mixed with the cooled and diluted surface water. But as the cooling of the surface and the precipitation are the same there as in the surrounding regions, the consequence must be that the whole of this volume of water over the bank will be colder and less salt than the surrounding waters. And as this bank water, on account of its lower temperature, is heavier than the water of the surrounding sea, it will have a tendency to spread itself outwards along the bottom, and to sink down along the slopes from the sides of the bank. This obviously contributes to increase the opposition that such banks offer to the advance of ocean currents, even when they lie fairly deep.