330 A STUDY OF CORRELATIONS AMONG TERRESTRIAL TEMPERATURES. 



region. Having done this, we are not confined to a single observing station in each 

 region. In fact the more observing stations used in each, and the more widely they 

 are separated, the greater will be the weight to which the mean result for the region 

 is entitled. 



We shall now review the material made use of, and the method of handling it, so 

 far as seems necessary to enable critical investigators to examine and test the processes 

 in detail, and to form a judgment as to the reliability of the result. An entirely syste- 

 matic statement of the plans and methods is difficult from the fact that they had to be 

 changed in detail from time to time as the work progressed, owing to the unexpected 

 cases of incompleteness and other imperfections which showed themselves here and 

 there as the compilation went on. Lack of uniformity in treatment has also arisen 

 from the discovery from time to time of new material which it was thought advisable 

 to use in the work. Moreover a certain perfection of method originally aimed at, 

 involving a rigorous reduction to a 24-hour mean for every day, was found impractic- 

 able, and such means as chanced to be available had very generally to be used. The 

 effect of this drawback upon the results of the work itself is practicall}' quite unim- 

 portant ; but it prevents the material made use of from having the completeness and 

 certainty that it otherwise might have claimed as a basis for more extended meteoro- 

 logical researches. 



It may be added that the conclusions reached in the research can be judged 

 without any reference to the original materials on which the work is based ; but, as 

 already intimated, a knowledge of this material will not only facilitate the judgment 

 of any details but will be of assistance to any one desiring to review and extend the 

 work. The following are the sources from which the data were mostly derived. 



Records of the United States Weather Bureau. — The original plan was to choose a 

 number of widely separated stations in the United States and, from the manuscript 

 records of the Bureau, to reduce the recorded mean temperature of each day to the 

 rigorous 24-hour mean, and then obtain a daily deviation from the normal during the 

 entire period. But the discussion of the entire 35 years of records on this plan was 

 found to be too laborious, especially as the hours and system of observation had been 

 changed twice during the history of the bureau. It was therefore deemed sufficient 

 to take as the standard temperature for each day the mean of the maximum and 

 minimum temperatures. This was, for the most part, reduced to tlie 24-hour mean 

 when data for doing so could be readily found. 



The Argentine Republic. — The main source for this region has been the publica- 

 tions of the Officina Meteorologica Argentina. I am also indebted to Dr. Davis, Direc- 

 tor of the Meteorological Office, for the communication of observations additional to 



