In the first part of the paper effort is made to discover to what extent periods of abnormally high or abnormally low temperature in the United States synchronize and also as to whether or not there is evidence of a periodicity in the occurrence and recurrence of these phenomena. The basic material for the study was 12-month consecutive or overlapping monthly means of the temperature for fairly large geographic districts. The districts used are (1) the New England States, (2) Minnesota, (3) Colorado, (4) Washington, and (5) Louisiana. The monthly mean temperature for each of these districts was originally computed from the means of all of the individual stations therein. The period 1888–1919 furnished the data for the study.

As was to have been expected, the control of the changes in temperature of the various parts of the United States is clearly that of the general circulation of the atmosphere, as modified, of course, by the secondary circulation due to the movement of cyclones and anticyclones. In order to get beyond the influence of the latter the study was extended to include certain tropical stations, viz, Batavia, Habana, Honolulu, and Arequipa. Consecutive means were also computed for these stations.

Both tropical and temperate zone stations show very clearly the persistence of short-period variations of about 40 months in length; occasionally, for reasons not understood, some of these short-period maxima and minima are greatly intensified and consequently appear as primary maxima or minima in the series. The length of the interval between these so-called primary maxima and minima is greater than and probably some multiple of the 40-month period.

One of the chief characteristics of the data is the tendency of any marked variation in the temperature to be followed by another one of opposite phase almost immediately. While this tendency amounts to almost certainty it is useless for forecast purposes because there is no means of discovering the precise duration of any existing phase. A comparison of the temperature variations in the border region between tropics and subtropics led to the conclusion that the influence of the general circulation at times extends to the northern portion of tropical areas. There also seems to be in operation at times a common temperature control for both Tropics and temperate zones. Extracts from the Reseau Mondial for 1910, 1911, and 1912 are quoted in support of this view.

The literature of the sun spot-terrestrial temperature relation is very briefly touched upon and the difficulty of separating the terrestrial from the extra-terrestrial influences is discussed. The annual temperature variations of the United States as a single geographic unit is compared with the sun-spot curve and the resemblances and differences are discussed.