influence by the land, and the tides and currents of the ocean affect them to a greater or less extent. They owe their origin to depressions of the earth's crust of no very wide extent and not running very far into the continental mass, and geologically they are of recent age and still subject to change. In these respects they contrast with the great oceans which owe their origin to the most extensive and the profoundest depressions of the crust, date back at least to Mesozoic times, and have perhaps remained permanently in their present position from still remoter ages.
Seas may be classified according to their form either as “enclosed” or as “partially enclosed” (or “fringing”). Enclosed seas extend deeply into the land and originate either by the breaking through of the ocean or by the overflowing of a subsiding area. They are connected with the ocean by narrow straits, the salinity of the water contained in them differs in a marked degree from that of the ocean, and the tidal waves are of small amplitude. Four great intercontinental enclosed seas are included between adjacent continents—the Arctic Sea, the Central American or West Indian Sea, the Australo-Asiatic or Malay Sea and the Mediterranean Sea. There are also four smaller continental enclosed seas each with a single channel of communication with the ocean, viz. the Baltic Sea and Hudson Bay with very low salinity, the Red Sea and Persian Gulf with very high salinity.
The fringing or partially enclosed seas adjoin the great land masses and are only separated from the oceans by islands or peninsulas. Hence their tidal conditions are quite oceanic, though their salinity is usually rather lower than that of ocean water. The four fringing seas of eastern Asia, those of Bering, Okhotsk, Japan and East China, are arranged parallel to the main lines of dislocation in the neighbouring land-masses, and so are the Andaman Sea and the Gulf of California. On the contrary, the North Sea, the British fringing seas (English Channel, Irish Sea and Minch), and the Gulf of St Lawrence cross the main lines of dislocation.
In addition to these seas notice must be taken of the subordinate marginal features, such as gulfs and straits. Gulfs may be classified according tc their origin as due to fractures of the crust or overflowing of depressed lands. The former are either the extensions of oceanic depressions, e.g. the Arabian Sea, Bay of Bengal and Gulf of Arica, or such caldron-depressions as the Gulfs of Genoa and Taranto, or rift-depressions like the Gulfs of Aden and Akaba. Compound gulfs are formed seawards by fracture and landwards by the overflowing of depressed land, e.g. the Bay of Biscay, Gulf of Alaska and Gulf of the Lion. Gulfs formed by the overflowing of depressed lands lie upon the continental shelf, e.g. the Gulf of Maine, Bay of Fundy, Bay of Odessa, Gulf of Martaban.
Straits have been formed (1) by fracture across isthmuses, and such may be by longitudinal fracture as in the Strait of Bab-el-Mandeb, or transverse fracture as in the Strait of Gibraltar or Cook Strait; (2) by erosion, e.g. the Strait of Dover, the Dardanelles and Bosporus; (3) by overflowing through the subsidence of the land, as in the straits of Bering, Torres and Formosa.
Surface of the Ocean.—If the whole globe were covered with a uniformly deep ocean, and if there were no difference of density between one part and another, the surface would form a perfect ellipsoid of revolution, that is to say, all the meridians would be exactly equal ellipses and all parallels perfect circles. At any point a sounding line would hang in the line of the radius of curvature of the water surface. But as things are the water-surface is broken by land, and the mean density of the substance of the land is 2.6 times as great as that of sea-water, so that the gravitational attraction of the land must necessarily cause a heaping up of the sea around the coasts, forming what has been called the continental wave, and leaving the sea-level lower in mid-ocean. Hence the geoid or figure of the sea-surface is not part of an ellipsoid of rotation but is irregular. The differences of level between different parts of the geoid have been greatly overestimated in the past; F. G. Helmert has shown that they cannot exceed 650 ft. and are probably much less. Recent pendulum observations have shown that it is incorrect to assume a uniform density of 2.6 in the elevated part of the earth's crust, that on the contrary there are great local differences in density, the most important being a confirmation of Airy's discovery that there is a marked deficiency of mass under high mountains and a marked excess under the bed of the ocean. The intensity of gravity at the surface of the sea far from land has been measured on several occasions. During Nansen's expedition on the “Fram” in 1894-1895, Scott Hansen made observations with a Sterneck's half-seconds pendulum on the ice where the sea was more than 1600 fathoms deep and found only an insignificant deviation from the number of swings corresponding to a normal ellipsoid. In 1901 O. Hecker took the opportunity of a voyage from Hamburg to La Plata, and in 1904 and 1905 of voyages in the Indian and Pacific Oceans to determine the local attraction over the ocean by comparing the atmospheric pressure measured by means of a mercurial barometer and a boiling-point thermometer, and obtained results similar to Scott Hansen's. The inequalities of the geoid in no case exceed 300 ft. Distortion of the ocean surface may also arise from meteorological causes, and be periodic or unperiodic in its occurrence, but it does not amount to more than a few feet at the utmost. Solar radiation warms the tropical more than the polar waters, but, assuming equal salinity, this cause would not account for a difference of level of more than 20 ft. between tropical and polar seas. The annual range of temperature between summer and winter of a surface layer of water about 25 fathoms thick in the Baltic is as much as 20° F., but this only corresponds to a difference of level of 1¼ in. due to expansion or contraction.
Atmospheric precipitation poured into the sea by the great rivers must necessarily create a permanent rise of the sea-level at their mouths, and from this cause the level round the coasts of rainy lands must be greater than in mid-ocean. H. Mohn has shown how the inequalities of what he terms the density-surface can be found from the salinity and temperature; and he calculates that the level of the Skagerrak should be about 2 ft. higher than that of the open Norwegian Sea between Jan Mayen and the Lofoten Islands. The level of the Gulf of Finland at Kronstadt and of the Gulf of Bothnia at Haparanda should similarly be 15 in. higher than that of the Skagerrak. Recent levellings along the Swedish and Danish coasts have confirmed the higher level of the Baltic; and the level of the Mediterranean has also been determined by exact measurements to be from 15 to 24 in. lower than that of the Atlantic on account of evaporation. Apart from the effects of varying precipitation and evaporation the atmosphere affects sea-level also by its varying pressure, the difference in level of the sea-surface from this cause between two given points being thirteen times as great as the difference between the corresponding readings of the mercurial barometer. In the north tropical belt of high pressure south of the Azores the atmospheric pressure in January is 0.87 in. higher than in the Irminger Sea; hence the sea-level near the Azores is almost 1 ft. lower than in the northern sea. In the monsoon region, where the barometer rises 0.38 in. between July and January, the level of the sea falls in consequence by 5 in. Wind also gives rise to differences of level by driving the water before it, and the prevailing westerly wind of the southern Baltic is the chief cause of the sea-level at Kiel being 5½ in. lower than at Arkona on Rügen. Periodic variations of level due to meteorological causes account for the Baltic being fuller in the time of the summer rains than in winter, when the rivers and lakes are frozen and most of the precipitation on the land is in the form of snow. The range on the Arkona gauge is from 3.5 in. below mean level in April to 2.75 in. above the mean level in August. A similar range occurs on the Dutch coast in the North Sea, where the maximum level is reached in October, the month of highest rainfall, and there is a range of 8 in. to the minimum level at the time of least rainfall in early spring. In the monsoon regions the half-yearly change from on-shore to off-shore winds produces noticeable differences in
