It was formerly believed that while the surface temperature--which depended upon direct solar radiation, the direction of currents, the temperature of winds, and other temporary causes--might vary to any amount, at a certain depth the temperature was permanent at 4° C., the temperature of the greatest density of fresh water. Late investigations, however, have led to the conclusion that instead of there being a permanent deep layer of water at 4° C., the average temperature of the deep sea in temperate and tropical regions is about 0° C., the freezing point of fresh water. In the atmospheric ocean, aëronauts not seldom meet with warm air currents flowing above others of a colder temperature; while, according to a general law, the warmth of the air constantly diminishes as its elevation above the surface of the sea increases. Similar exceptions to the general rule are met with in the ocean. In moderate depths sometimes the whole mass of water from the surface to the bottom is abnormally warm, owing to the movement in a certain direction of a great body of warm water, as in the "warm area" to the north-west of the Hebrides, where, at a depth of 500 fathoms, the minimum temperature was found to be 6° C. On the other hand, the whole body of water is sometimes abnormally cold, as in the "cold area," between Scotland and Faeroe, where, at a depth of 500 fathoms, the bottom temperature is found to average -1° C.[C] The only feasible explanation of these enormous differences of temperature, amounting to nearly 13° F. in two areas freely communicating with one another, and in close proximity, is that in the area to the north-west of the Hebrides a body of water warmed even above the normal temperature of the latitude flows northwards from some southern source, and occupies the whole depth of that comparatively shallow portion of the Atlantic, while an arctic stream of frigid water creeps from the north-eastward into the trough between Faeroe and the Shetland Islands, and fills its deeper part in consequence of its higher specific gravity. There can be no doubt that similar phenomena occur in various parts of the ocean, and that the deep seas are frequently intersected by streams differing in temperature from the surrounding waters. [Footnote C: "The Depths of the Sea," by Professor Wyville Thomson, p. 307.] In some places, owing to the conformation of the neighbouring land or of the sea-bottom, superficial warm and cold currents are circumscribed and localised, thereby occasioning the singular phenomenon of a patch or stripe of warm and a patch of cold sea meeting in an invisible but well-defined line. The temperature of the sea apparently never sinks at any depth below -3·5° C. This is about the temperature of the maximum density of sea water, which contracts steadily till just above its freezing point (-3·67° C.), when kept perfectly still.