NO. 4 SUX'S VARIATION AND TEMPPZRATURES ABKOT 33 



them to tell in advance when such changes of phase and of amplitude 

 will occur. If such changes could be forecasted, then, as the dates of 

 recurrence of the sun's changes can be predicted indefinitely, very 

 long-range weather forecasts might perhaps be made. 



It is pointed out that temperature forecasts of the order of a month 

 in ad\ance may perhaps even now be practicable with the aid of this 

 new cycle. For, as stated, the phases of recurrence of temperature 

 efifects almost always remain nearly unchanged in any one month, and 

 the variation of amplitude is not often excessive within that interval. 

 Hence with a continuously kept running curve of the temperature 

 effect, similar to figures 2 and 3, predictions might l)e hazarded with 

 considerable confidence for a month in advance. .\ preliminary trial 

 forecast for 36 days was made, and gives an average deviation of 

 6? 5 F. from the event. Omitting 7 days of sporadically wide dis- 

 crepancies and of obviously increased amplitude of the temperature 

 periodicity, 2^ days after the forecast began, the reinaining 29 days 

 give an average deviation from the event of 4?2 F. 



