OF THE ATMOSPHERIC TEMPERATURE. 229 



method of interpolation is quite perfect, the second is less satisfactory, yet it is not 

 apprehended that the annual mean could in the worst case be vitiated or in general 

 rendered uncertain by more than i 0°.5. In all cases where such limited interpola- 

 tion had to be resorted to the fact is indicated in the tables by an asterisk affixed. 



It should also be understood that all tabular annual means were found by dividing 

 by 12, the sum of the monthly means belonging to the calendar months ; the small 

 correction for inequality of months (previously referred to) is nearly constant, and 

 would not affect any conclusions we may deduce from the tables ; of the same 

 nature are index errors to the thermometers and reductions for difference of eleva- 

 tion or different exposures of stations at no great distance apart, as for instance 

 within the limits of a city. 



The bottom line of the tables contains the resulting mean temperatures for the 

 respective stations; they are in general the mean of all the annual means in their 

 respective columns, but they are made up from the separate monthly means, and 

 include consequently all monthly means whether they belong to complete or 

 incomplete years, in fact we might have a resulting annual mean from observations 

 scattered over all the months but in different years and yet no single year complete. 

 This explains the occasional differences of the resultant temperature from the 

 simple mean of the individual complete years, and has nothing to do with interpo- 

 lation. 



In conformity with custom the mean temperatures are given to two places of 

 decimals, but the hundredths of a degree have very little real value, and that only 

 differentially. 



TABLES OF THE MEA¥ ANNUAL TEMPERATURE IN THE UNITED STATES 

 AND BRITISH NORTH AMERICA 



FOR A SUCCESSION OF TEARS. 



ALL NXraiBEKS AEE EXPRESSED IN DEOEEES AND FEAOTIONS OF THE FAHEENHEIT SCALE. 



