The geographical theory, which states that the Ice Age was brought about by elevation in high latitudes, and by changes in the land and sea distribution, though never seriously challenged, has suffered until recently from a lack of precision. The present author attempted to remedy this by a close mathematical study of the relation of temperature to land and sea distribution at the present day. The method at attack was as follows: from the best available isothermal charts of all countries the mean temperature reduced to sea-level was read off for each intersection of a ten-degree square of latitude and longitude, for January and July, from 70° N. to 60° S. latitude; this gave 504 values of temperature for each of these months. Round each point was next drawn a circle with an angular radius of ten degrees, divided into east and west semicircles. The area of each semicircle was taken as 100, and by means of squared paper the percentage of land to the east and land to the west were calculated; finally, in each month the percentage of the whole circle occupied by land, ice, or frozen sea was calculated, this figure naturally being greater in winter than in summer. The projection used was that of the “octagonal globe,” published by the Meteorological Office, which shows the world in five sections, the error nowhere exceeding six per cent. These figures were then analysed mathematically, and from them the effects on temperature of land to the east, land to the west, and ice were calculated. The detailed numerical results are set out in an Appendix; it is sufficient here to give the following general conclusions: (1) In winter the effect of land to the west is always to lower temperature. (2) In winter the effect of land to the east is almost negligible, that is to say, the eastern shore of a continent is almost as cold as the centre of the continent. The only important exception to this rule is 70° N., which may be considered as coming within a belt of polar east winds. (3) In summer the general effect of land, whether to the east or west, is to raise temperature, but the effect is nowhere anything like so marked as the opposite effect in winter. (4) The effect of ice is always to lower temperature. (5) For every latitude a “basal temperature” can be found. This is the temperature found near the centre of an ocean in that latitude. This “basal temperature” is a function of the amount of land in the belt of latitude. Poleward of latitude 20° an increase of land in the belt lowers the winter basal temperatures very rapidly and raises the summer basal temperature to a less extent. The “basal temperature” is important, since it is the datum line from which we set out to calculate the winter and summer temperatures of any point, by the addition or subtraction of figures representing the local effect of land in a neighbouring 10° circle.