459 



case, but in a great many days, especially in winter, it was otherwise. As the 8 a.m. reading applied to a 

 period of which 8 hours belonged to clay n, as against 4 belonging to day n- 1, the reading was assigned 

 to day n when the general trend of temperature was not sufficiently definite to settle the question. When 

 temperature showed a general trend downwards after 8 p.m. on day n, the 8 a.m. reading on dy n+ 1 

 was assigned to day n; in such a case, however, the maximum at 8 p.m. was practically certain to be tho 

 real maximum for day n. When temperature showed a general trend upwards after 8 p.m. on day n, the 

 8 a.m. reading next morning was assigned to day m + 1, and the ordinary reading at 10 p.m. or midnight 

 on day n, if greater than the 8 p.m. maximum, was accepted as the maximum for day n. In March, 1902, 

 maximum thermometer readings were lacking in about half the days of the month ; on these days the 

 highest of the ordinary thermometer readings was taken as the maximum. There were very few other 

 occasions throughout the two years when maximum readings were lacking, and the days when this 

 occurred were left out of consideration. 



When ordinary thermometer readings were accepted as maxima the consequence was, no doubt, as a 

 rule, an underestimate of the maximum, and thence of the daily range (maximum less minimum) and 

 the figures for March, 1902, probably are slightly affected by this cause. With the exception of this one 

 month, whilst individual daily ranges were doubtless occasionally in appreciable error, there can be but 

 little doubt of the substantial accuracy of the mean data given in Table VII. 



It shows for each month of the two years the means of the daily maxima and of the daily minima, and 

 the mean of the daily ranges (maximum less minimum). 



Mean results are also given for the two years combined. 



The seasons are taken as in the previous tables. 



Data are omitted for the whole year and for midsummer in 1902, because January results were lacking 

 for that year; those entered in brackets really belong to 1904. 



Even at a temperate station the daily range from the maximum and minimum thermometers is 

 considerably in excess of that derived from the highest and lowest of the hourly readings, and increasing 

 the interval between readings from one hour to two naturally reduces the range. The differences, 

 however, between the ranges in Tables VII. and I. are on an enormously enhanced scale as compared to 

 the corresponding differences in a temperate climate. 



The five months November to March in all of which the mean maximum temperature for the day 

 exceeds F. show ranges which differ comparatively little amongst themselves, but which are consider- 

 ably less than those in the other months. On the average of the two years the range is largest in June, 

 and only slightly less in July and August. 



7. In some places the mean of the daily maximum and minimum is on the average of the days of a 

 month, a very close approach to the mean temperature of the 24 hours. Table VIII. gives, for each 



TABLE VIII. 



* The values entered under January, 1902, belong really to January, 1904. 



3 N 2 



