350 



A STUDY OF CORRFXATIONS AMONO TERRESTRIAL TEMPERATURES. 



Table VIII. 

 Time Correlations Through Nine-year Terms. 



The reality of the periodicity can be established only by carrying the investi- 

 gation back through the years preceding 1871. I have done this with Kcipping's 

 table of annual departures already cited, after correction for the sun-spot inequality. 

 The result is found in the second part of Table IX, preceding. There is here not 

 only no periodicity, but, on the contrary, a tendency toward a persistence of the 

 departure in the same direction for as much as six years. The products are, in gen- 

 eral, several times larger than those for the modern period, showing wider accidental 

 deviations. We may attribute both this and the systematic character of the correla- 

 tion products to the imperfections of the older instruments and observations. But 

 this would not be likely to mask entirely a six-year periodicity, if any such existed. 

 We must, therefore, regard the seeming period as unreal, or at least open to serious 

 doubt, notwithstanding the plausibility of the statistical evidence in its favor. 



CHAPTER IV. 

 Discussion of Monthly Departures. 

 Since the only period exceeding a month that we can assign a 'priori as probable, 

 that of the sun-spots, has already been investigated in the preceding chapter, the pur- 

 pose of the present chapter is to determine whether the monthly departures of world- 

 temperature show any systematic character not found in the results of the annual 

 departures. If this result were the only one aimed at, ideal simplicity and perfection 

 would require that we first correct the normal temperatures from which the departures 

 are computed for the fluctuations already derived from the annual means. In other 

 words, our normal temperature should include at least the sun-spot fluctuation. But 

 this has not been done. Consequently, the general departures t^, affecting all parts 



