An undercurrent flows out from the Red sea through the Strait of Bab-el-Mandeb, and from the Mediterranean through the Strait of Gibraltar, raising the salinity as well as the temperature of the part of the ocean outside the gates of the respective seas. The action of the Red sea water affects the whole of the Gulf of Aden and Arabian sea, raising the temperature at the depth of I,000 metres to 52° or 53° F or 9° F higher than the water of the Bay of Bengal at the same depth. The effect of the Mediter ranean water in the North Atlantic does not require such large figures to express it, but is extraordinarily far-reaching. There are clear traces of the warm and salty Mediterranean water northward as far as Ireland, westward as far as the Azores and southward as far as the Canaries, at a depth of from i,000 to 1,600 metres. In shallow seas as the North sea and the British fringing seas, where tidal currents run strong, there is a general mixing together of the surface and deeper water, thus making the arrangement of vertical temperature anathermic in summer and katathermic in winter, while at the transitional periods in spring and autumn it is practically homothermic. Thus at Station E2 of the international series at the mouth of the English channel in
N.,
42' W., the following distribution of temperature F has been observed by Matthews It is noticeable that there is a marked vertical temperature gradient only at the end of summer when a warm surface layer is formed, though in Aug. 19°4 that was only 14 metres thick. The heat content of the water column, at Station
and other places where investigations have occurred in great detail, has been shown by Harvey to vary greatly from year to year.
The freezing-point of sea water is lower as the salinity increases and normal sea water of 35 per mille salinity freezes at 28.6° F. Experience shows that sea water can be cooled considerably below the freezing-point without freezing if there is no ice or snow in contact with it. The dissolved salts are ex cluded in the process in a regular order according to temperature. As very low air temperatures seldom occur, freshly formed sea ice always has a salt content of from 12 to 18%0 if the original salt content of the water was
and the sea ice is therefore not suitable as a source of drinking water. In general the salt is returned to the lower unfrozen water, but during a continued hard frost it may appear as crystals on the surface ; these crystals are known to polar explorers by the Siberian name of rassol. Ice is a very poor conductor of heat and accordingly protects the sur face of the water beneath from rapid cooling; hence new-formed pancake ice does not increase excessively in thickness in one win ter, and even in the centre of the Arctic basin the ice-covering only amounts to 2 or at most 3 metres in the course of a year, while in the Antarctic regions the season's growth is only half as great; in the latter also the accumulated snow is an important factor in the thickness of the ice, and snow is an even worse conductor of heat. The influence of wind and tide breaks up the frozen surface of the sea, and sheets yielding to the pressures slide over or under one another and are worked together into a hummocky ice pack, the irregularities on the surface of which, caused by re peated fractures and collisions, may be from 3 to 7 metres high. Such formations termed toross by the Russians, may extend under water, according to Makaroff's investigations, to at least an equal depth. Such old sea ice, when prevented from escaping, forms the palaeocrystic sea of Nares; but, as a rule, it is carried south ward in the East Greenland and Labrador currents, and melted in the warmer seas of lower latitudes. In the Southern Hemi
sphere the ice-pack forms a nearly continuous fence around the Antarctic continent. Pack ice forms regularly in the inner part of the Baltic every winter, but not in the Norwegian fjords. Even in the Mediterranean, sea ice is formed annually, in the northern part of the Black sea, and more rarely in the Gulf of Salonika and at the head of the Adriatic off Trieste. Hudson bay is blocked by ice for the greater part of the year, and the Gulf of St. Lawrence is blocked every winter. Ice also clothes the conti nental shores of the northern fringing seas of eastern Asia. In addition to sea ice, icebergs which are of land origin occur at sea. In the north, icebergs break off, as a rule, from the ends of the great glaciers of Greenland, and in the far south from the edge of the great Antarctic ice-barriers. The latter often give birth to prodigious icebergs and ice islands, which are carried north ward by ocean currents nearly as far as the tropical zone, before they melt. The Antarctic icebergs are of tabular form and much larger than those of Greenland, but in either case an iceberg rising to 7o metres above sea-level is uncommon and one exceeding Ioo metres is very rare. The Greenland icebergs are carried by the Labrador current across the great banks of Newfoundland, where they are often very numerous in the months from March to August, when they constitute a danger to shipping as far south as 40° N. No icebergs occur in the North Pacific and none has ever been reported nearer the coasts of Europe than off the Orkney islands, and there only once in 1836.
Although observations on marine cur rents were made near land or between islands even in antiquity, accurate observations on the high seas have only been possible since chronometers furnished a practicable method of determin ing longitude, i.e., from the time of Cook, the circumnavigator. The difference between the position as determined astronomically and by dead-reckoning gives an excellent idea of the general direc tion and velocity of the surface-currents. The first compre hensive study of the currents of the Atlantic was that carried out by James Rennell (1790-1830), and since that time Findlay in his Directories, Heinrich Berghaus, Maury and the officials of the various hydrographic departments have produced increasingly accurate descriptions of the currents of the whole ocean. Direct observations of currents in the open deep sea are difficult, but Americans have measured directly from anchored ships the Gulf Stream currents in the Caribbean sea. Similarly, for the whole South Atlantic region, the German survey ship "Meteor" (1925 27) was anchored to measure the currents at all depths with Ekman's current-meter. One of the indirect methods of investi gating currents is by taking account of the initial temperature of the current and following it by the thermometer throughout its course. Benjamin Franklin (1775) and Charles Blagden (1781), by means of numerous observations of temperature made on board the packets plying on the Atlantic passage, determined the boundaries and seasonal variations of Gulf Stream and Labrador current. In the northern European seas, salinity often proves a sharper criterion of the boundaries than temperature. Evidence drawn from driftwood, wrecks or special drift bottles is less dis tinct but still interesting and often useful. Icebergs and plankton also serve as indicators of the trend of currents.