A STUDY OF CORRELATIONS AMONG TERRESTRIAL TEMPERATT^RES. 3H1 



the total sum comes tmt with a positive value markedly exceeding its probable value, 

 we may infer with a corresponding degree of probability that some at least of the. 

 departures are real. If the excess is not great, then what we should conclude is that 

 there is a greater or less pi-obabilit}', in a general way, that the sun's radiation is 

 variable, but not that it had a definite variation at a definite time. To draw the latter 

 conclusion from the data would be fallacious, not from any defect in the method but 

 from the very nature of the case. But, if the well marked excess were a general rule, 

 then we could fairly infer that, as a general rule, the fluctuations of temperature indi- 

 cated corresponding fluctuations in the solar energy. For example, referring to the col- 

 umn T in Table VI, which shows the residual departure of the annual temperatures after 

 eliminating the effect of the sun-spot period, we may say that the temperature appears 

 to have been above the normal in 1871, again in the years 1881-83 and againin 1896- 

 97. It seems to have been below the normal in 1874-84, 1887 and 1892, and 1893. 



Although these fluctuations, even if real, are so small that we cannot expect to 

 trace them in any other meteorological phenomena than the temperature, the question 

 of their reality is of scientific interest. This can be determined only by more 

 extended researches. 



To state the limitation in a more condensed form, the proof of general invari- 

 ability does not positively establish the negative proposition that the sun's heat has 

 never, on an}' one occasion whatever, undergone a perturl^ation during the period 

 covered by our researches. In the absence of better positive evidence than is yet 

 available, the assumption of such a perturbation would be a purely gratuitous one, to 

 be refuted by a consideration of its improbability rather than by positive evidence. 



§ 17. Relation Between the Solar Radiation and. Meteorological Processes. 



The preceding studies being primarily of fluctuations in the temperature of the 

 air at the earth's surface, the question arises how far, from the steadiness of tempera- 

 tui-e we have established, we are justified in affirming that the sun's thermal radiation 

 is steady in a corresponding degree. The consideration of this question will be facil- 

 itated by calling to mind certain points bearing upon it. A general proposition 

 which, the writer conceives, needs no enforcement, is that so far as our science can 

 show, the earth receives an appreciable supply of heat only from the sun. We may 

 safely assume that the minute amount of heat reaching the earth's surface from the 

 stars or other bodies in the celestial spaces, or by conduction from the earth's interior, is 

 too minute to materially affect the temperature around us. This temperature is 

 determined in a general way by the condition that it is such that the earth shall 

 radiate into space as much heat as it absorbs from the sun's rays. 



A. p. S.— XXI. ZZ. 15, 1, '08. 



