616 PROCEEDINGS OF THE AMERICAN ACADEMY. 



would fulfil the condition as well as the period I had supposed, and I 

 was led to adopt the length of the lunar periods in my investigations as 

 a working hypothesis. This seems justified by the fact that other inves- 

 tigators have found certain constant relations between the positions of 

 the moon and weather phenomena as described in the foregoing pages. 

 The average changes in the weather conditions as related to the moon 

 appear to be very small, and meteorologists . have heretofore assumed 

 that, if such a relation existed, it was so small as to be negligible. How- 

 ever, since certain of the weather periods have been found to reverse in 

 phase, is it not possible that periodic motions in the atmosphere set up in 

 some way by the motions of the moon may be much larger than shown 

 by the average for a long time at any given place ? 



On Plate IX. are plotted the mean daily departures from the normal 

 temperatures at Blue Hill from October, 1898, to February, 1899. It is 

 seen that during October and November there was a rise and fall in tem- 

 perature about every three and a half days, which is about one eighth of 

 a lunar period. A dotted curve representing a harmonic exactly one 

 eighth of tlie synodic period of the moon is plotted for comparison with 

 the observed temperatures. The curves show that for every maximum 

 of the dotted curve there was a corresponding maximum of the observed 

 temperature, though the two were not always synchronous, and near the 

 end of November the phenomenon of reversal of phase appears. Fur- 

 thermore, it is seen that the minimum temperatures of October, December, 

 January, and February occurred very near the times indicated by the 

 round black circles. These circles indicate the time of new moon, and 

 show that the intervals between the minima approximate the length of a 

 synodic period of the moon. Moreover these minima were from 10° to 

 30° F. below the normal of the time of year, and, if periodic, were the 

 result of some strongly acting cause. Whatever the cause of these oscil- 

 lations may be, a knowledge of the laws controlling their action would 

 be of vast importance for weather forecasting. An urgent present need 

 is to find the reason for, or the law of, their reversal of phase. Such 

 researches deserve financial aid and sympathy. The problems to be solved 

 are fascinating and important, and in the time which can be spared from 

 other pursuits I am making an earnest effort to find a clue to the cause 

 or the method of the reversals. One cause appears to be a movement 

 of the centres of oscillation. In other cases there is evidence of an annual 

 period. Taking, for example, Luke Howard's tables, which show for 

 the eighteen years 1815-32 the mean temperatures for the weeks when 

 the moon was in south and in north declination respectively, I found the 



