610 PROCEEDINGS OP THE AMERICAN ACADEMY 



orology, I adopted the general belief of meteorologists that, as there was 

 uo rational reason vvliy the moon shonld influence the weather, it prob- 

 ably did not. I had, however, in my investigation found evidence of 

 j)eriodic changes in the weather independent of the annual and diurnal 

 periods. I decided to lay aside any theory as to their cause, and to 

 determine as definitely as possible the lengths, ranges, and methods of 

 oscillation of the periodic changes. My investigations led me to the 

 following conclusions which 1 believe are important : — (1) That every 

 weather period is leiidered complex by the existence of periods which 

 bear the relation of harmonics to the primary, that is, their lengths are 

 twice, one half, one third, one fourth, etc., the length of the primary. 

 (2) The periods in different parts of the world have different phases, as 

 for example in the annual period it is cold in the northern hemisphere 

 when it is warm in the southern, and in the sunspot period it is dry in 

 Russia when it is wet in India. (3) At any given place on the earth's 

 surface the harmonics, and in some cases the primaries, reverse in j^hase. 

 In the case of some of the longer periods this has been traced to a move- 

 ment of the centre of oscillation from place to place. (4) At any given 

 place the periods and their harmonics do not vary synchronously. Some- 

 times the primary period is weak, while one or more of the harmonics 

 are strong, and the reverse. 



One of my earliest investigations, made in 1882-84, was concerning 

 a period of about two years wliicli I now believe is a periodic cliange 

 in the weather, arising from the annual period, and having twice its 

 leno'th. The results were published in the American INIeteorological 

 Journal for August, 1884, and April, 1885. Some of the diagrams are 

 here reproduced. The contiimous curve in Plate V. was obtained by 

 takino- the average of the monthly barometric means for each twelve 

 consecutive months from 1874 to 1881. The dotted curve was obtained 

 by taking the mean of each consecutive twenty-five months. The dotted 

 and broken curves in Plate VI. were obtained in the same way for the 

 rainfall. In each case the annual period is eliminated, and the unbroken 

 curves show secondary oscillations in pressure and rainfall with about 

 two years between each maximum and minimum. 



Tliat the same oscillation prevaHed in the temperature is shown by 

 Table II., page HI 3, giving the departures from the average temperature. 



On the eastern coast of the United States the oscillations in pressure 

 were found to be opposite in phase to those over the interior. A table 

 was made showing the departures from the average of twenty-five 

 months at the times of the maxima and minima of the waves. These 



