460 



month of the two years, the mean derived from the daily maxima and minima, and the amount by which 

 it falls short of the corresponding mean derived from the 2-hourly readings. In two months, December, 

 1902, and January, 1904, the mean from the 2-hourly readings was the smaller. In the latter month the 

 night readings were curtailed, and allowance had to be made for the fact, so that the result is exposed to 

 greater uncertainty than in the other months. It is clear, however, from the table that, while the mean 

 from the 2-hourly readings is in general decidedly the higher, its excess is decidedly less in the summer, 

 and even in the equinoctial months, than in winter. Taking the two years, the mean from the daily 

 maximum and minimum is on the average only 1 less than that from the 2-hourly readings. It would, 

 in fact, form a not at all bad approximation to the true mean temperature, though somewhat over- 

 estimating the cold of winter. 



8. The amplitude of the daily range as derived from the daily maximum and minimum is a quantity 

 varying immensely from day to day. Table IX. is intended to give an idea of the extent of this variation. 

 It records the greatest and the least daily range actually observed during each month of the two years. 



TABLE IX. Greatest and Least Daily Ranges. 



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



Very small ranges are occasionally met with at all seasons. On the whole, the smallest ranges of the 

 year are encountered in the summer months. The largest ranges in winter are very considerably larger 

 than those in summer. In five separate months there were daily ranges exceeding 40 F. 



9. Table X. does for individual hours what Table III. does for individual days. It gives the highest 

 maximum and the lowest minimum readings recorded in each individual month and the difference 

 between the two, which constitutes the extreme range of temperature for the month in question. 



The highest temperature recorded throughout the month varied from +42'0 in December, 1903, 

 to +2-0 in April, 1903, a range of 40, there being no single month in which the temperature failed to 

 reach F. 



The lowest temperature of the month varied from -58 '5 in September, 1903, to + 9'0 in January, 

 1903, the range being 67 '5, and so 69 per cent, larger than that of the highest temperature of the month. 



The monthly range, i.e., the difference between the highest and thu lowest temperature of the month, 

 shows a fairly regular annual variation, and is notably least at midsummer. 



10. The particular hour at which the highest maximum or the lowest minimum of a particular month 

 occurs must be regarded as largely a matter of accident, and it cannot throw any certain light on the 

 question whether there is a decided tendency for exceptionally high or exceptionally low temperatures to 

 occur at one hour of the day rather than another. 



To throw further light on this question, means were derived from the six lowest minima readings 

 corresponding to each of the 2-hour intervals of which the day was composed. If in any particular month 



