NO. 7 THE ATMOSPHERE AND THE SUN CLAYTON 45 



Hence, to follow the sequences of weather resulting from increased 

 solar activity it is necessary to consider the month or seasons sepa- 

 rately and to work out expected conditions for different intensities of 

 solar activity. 



In regard to the use of the second method, forecasting weather as 

 ordinarily practiced at the present time depends on anticipating for 

 a day or two at a time the drift of weather conditions. Such fore- 

 casts can be improved and extended in time by analyzing weather into 

 waves of different lengths and forecasting the progress of the 

 stronger waves. Even long range forecasts can be made on this basis, 

 as I have demonstrated by actual tests. 



The third method of forecasting is by means of the periodic 

 vibrations in the sun and atmosphere. Any pulsation in solar condi- 

 tion will be attended by similar pulsations in the earth's atmosphere. 

 The shorter pulsations will be felt relatively more in high latitudes 

 of the earth and the longer pulsations relatively more at low lati- 

 tudes, but all will be repeated to some extent in every part of the 

 atmosphere. An analysis of the periodic terms in the weather at any 

 point on the earth would make it possible to project the periodic 

 terms ahead to^ any length of time desired, were there no variations 

 in the amplitude and phase of the periods. But there are variations 

 and for this reason it is necessary to redetermine the periodic terms 

 at short intervals and to limit the time in advance which they are 

 made to cover. When these variations in the periodic terms become 

 calculable, this method of forecasting will probably replace all others. 

 Already considerable progress has been made along this line. 



In practical forecasting at present it is desirable to consider all of 

 the three methods mentioned and to use them as checks on each 

 other. Forecasting in words has but little meaning to the average 

 expert, because the meanings of words can be interpreted in various 

 senses and there are no accepted rules for verifying such cases. 

 Quantitative forecasts can, however, be verified by accepted stand- 

 ards ; so that from the beginning of my experiments in forecasting, 

 both verbal and quantitative forecasts were made. These quantitative 

 forecasts were made first for about a week in advance, then for 

 longer intervals up to a month. Figure 31 gives one of the more 

 recent of these forecasts of pressure made on November 24, 1929, for 

 2^ days in advance beginning on November 26 and ending on 

 December 21. The forecast was made up from a combination of 

 cycles varying in length from 3 days to 13 days. The correlation of 

 the forecasted with the observed pressure is 0.64 ±0.06. 



