332 A STUDY OF CORRELATIONS AMONG TERRESTRIAL TEMPERATURES. 



stations. But there is always a possibility at any one station of varying systematic 

 errors from one cause or another. Hence, it has received no greater weight than the 

 best of the remaining stations. 



§ 7. Arrangement of the work. 



Owing to the complexity of the conditions which have determined the final form 

 of the work, the task of studying its processes will be facilitated by a condensed state- 

 ment of its arrangements. The main features to be borne in mind are the following :■ 



Firstly, as regards geographic distribution ; that portion of the earth best avail- 



"able for the purpose is divided into regions within each of which the fluctuations of 



temperature are prima facie independent of those in other regions. The question 



whether this independence is real is regarded as open to question and therefore has 



been investigated in special cases where a correlation is possible. 



Then, within each region as many stations of observation as practicable are to be 

 selected in order that the accidental fluctuations of the regions may be reduced. Fre- 

 quently there is but one station in a region. 



Secondly, as to the time; the whole period included in each special branch of the 

 discussion is divided up into time-terms. The time- terms actually used are three, (1) 

 the year; (2) the calendar month; (3) the decade. 



Were the work ideally complete in every particular, we should logically begin 

 with the decade, then pass to the month, and then to the yearly terms, because this is 

 the order in which the observations are made and the work has to be done. But, for 

 reasons not necessary to set forth at length, the difiei'ent series of time-terms are pre- 

 sented in reverse order, beginning with the year. 



The material used is different for the three classes of terms. In discussing the 

 ten-day terms it was found that, quite apart from the labor of forming ten-day means, 

 the available material in the form of daily observations was comparatively limited. 

 But monthly and annual means are found iu so many publications that the data 

 available for this branch of the research is great. This additional wealth of material 

 has permitted the use of a much greater number of regions than are available for the 

 ten-day means. 



CHAPTER III. 



Discussion of Annual Deviations of Temperature. 

 § 8. F lad tuitions Having the Period of the Sun-spots. 



Proceeding according to the plan mapped out, our first step will be to determine 

 the fluctuations in temperature corresponding to the 11 -year inequality in the solar 



