378 



A STUDY OK ('ORIiKLATIONS AMON(i TERRESTRIAL TEMPERATURES. 



of the sun's synodic rotation. It is true that there is a minute increase of 0.004 

 between the 4th and 5th terms of the set. But an examination of the several separate 

 sums through which this is formed shows that the increase is too small and uncertain 

 to be regarded as the effect of periodicity. 



But a quasi-periodicity is still possible, the persistently positive sign of x indicating 

 a tendency of the departures to persist through a period of more than 25 days. The 

 exact general fact brought out by the correlation is as follows : 



Whatever be the departure of temperature at San Diego during any 5-day term 

 we may expect the subsequent departures to lie in the general average in the same 

 direction for more than a month, the ultimate amount at the end of the month being 

 about one fifth that of the departure taken as the initial one. This persistence cer- 

 tainly seems singular, and it may be that had the correlation period been extended, 

 periodicity would have been brought out. 



As a further illustration of the method, without expecting to reach definitive 

 results, I have made a similar time-correlation of the general mean temperatures for 

 each decade as given in Table XL preceding. The correlation-products were carried 

 through periods of four terms, or 40 days each, counting from the middle of the 

 initial to the middle of the last term. The actual period included is 50 days between 

 extremes. The result, summed by terms of three years, is as follows : 



A general tendency is here shown in the departures of temperature to continue in 

 the same direction for a period of at least 50 days. The time required for them to 

 disappear entirely can be determined only by continuing the products through a 

 longer period, which requires little more than a work of routine computation. 



What is striking in the present case is the small increase of the fourth sum, 

 following the rapid diminution of the first three sums. This is what we should expect 

 fi'om temporary inequalities in the temperature of the two solar photospheres. If 



