ABSOLUTE RANGES. 



137 



exceeded 50y but did not exceed lOOy, the last cases in which it exceeded 250y. Of the 317 days 

 included, only 5 had a range as small as 25y, while 101, or 32 per cent., had a range exceeding lOOy. The 

 latter figure is rather smaller than the corresponding percentage in the case of H. 



^1 -2. In addition to the results for individual days, Table XLVIII gives means for individual months, 

 varying from 234y for January, 1903, to 50y for July, 1902. The monthly means, however, fluctuate 

 somewhat irregularly, and can hardly claim to closely represent average conditions. The January mean, 

 for example, appears abnormally large. This is partly accounted for by the exceptionally large ranges on 

 January 11 and 12. It is by no means impossible that some of the irregularities may be due to errors in 

 the scale values, or in the temperature corrections. The former source of error is most to be feared in the 

 months of July, August, and September, 1902, the latter in the Midsummer months of 1902-3, when the 

 temperature range was especially large. 



If we combine corresponding months from the two years, allowing equal weight to each day, we obtain 

 the following somewhat more regular mean monthly values : 



TABLE LII. Mean Absolute Ranges of V. (Unit ly.) 



43. Table LIII gives the largest and smallest absolute ranges recorded in each individual month. 



TABLE LIII. Absolute Ranges. 



Even in the case of D there are 10 months in which the largest range of the month is underestimated 

 owing to the limit of registration being exceeded. In the case of H every month, except July, 1902, 



T 



