The motion of the sun in the ecliptic occasions perpetual variations in the length of the day, and in the direction of his rays with regard to the earth; yet, as the cause is periodic, the mean annual temperature from the sun’s motion alone must be constant in each parallel of latitude. For it is evident that the accumulation of heat in the long days in summer, which is but little diminished by radiation during the short nights, is balanced by the small quantity of heat received during the short days of winter and its radiation in the long frosty and clear nights. Were the globe everywhere on a level with the surface of the sea, and of uniform substance, so as to absorb and radiate heat equally, the mean heat of the sun would be regularly distributed over its surface in zones of equal annual temperature parallel to the equator, and would decrease regularly to each pole. The distribution of heat, however, in the same parallel is very irregular in all latitudes, except between the tropics, from the inequalities in the level and nature of the surface of the earth, so that lines drawn on a map through all places having the same mean annual temperature are nearly parallel to the equator only between the tropics; in all other latitudes they deviate greatly from it, and from one another.[129] Radiation is the principal cause of temperature; hence, the heat of the air is most powerfully modified by the ocean, which occupies three times as much of the surface of the globe as the land, and is more uniform in its surface, and also in its radiating power. On the land the difference in the radiating force of the mountains and table-lands from that of the plains—of deserts from grounds covered with rich vegetation—of wet land from dry, are the most general causes of variation; the local causes of irregularity are beyond enumeration. There are two points in the northern hemisphere, both in the 80th parallel of latitude, where the cold is more intense than in any other part of the globe with which we are acquainted. One north of Canada in 100° W. long. has a temperature of -3°·5 of Fahrenheit; while, at the Siberian point, in 95° E. long., the temperature of the air is +1°; consequently it is four and a half degrees warmer than that north of Canada—a difference that has an influence even to the equator, where the mean temperature of the air is different in different longitudes.