457 



regular annual variation, with a minimum in the hottest months, December and January, and a maximum 

 in July and June. The minimum thus occurs at a season when the periodic diurnal changes of temjwra- 

 ture are greatest, the maximum when they arc least ; moreover, the maximum is fully double the 

 minimum. The hours at which the difference between the ordinary and minimum readings is greatest 

 and least vary a good deal from month to month, especially in winter. In summer the phenomena are 

 fairly regular. From the three midsummer months, November to January, we get a very regular diurnal 

 inequality in the difference, having a maximum for the 2-hour period 2 to 4 a.m., and a minimum for the 

 period 2 to 4 p.m., the "range" amounting to 0'99. These two periods immediately follow the hours 

 of minimum and maximum respectively in the diurnal inequality of ordinary temperature for the season. 

 This shows that if a diurnal inequality were formed from the minimum thermometer readings of the 

 month, it would resemble that deduced from the ordinary thermometer ; the two inequalities would show 

 maxima and minima about the s une hours, but the range from the minimum readings would be about 1" 

 the larger. 



In equinox the difference between the ordinary and minimum readings shows a less regular diurnal 

 inequality, but there is a distinct maximum for the period 6 to 8 a.m., and a distinct minimum for the 

 period 2 to 4 p.m. These periods again follow immediately the minimum and the maximum of the 

 ordinary temperature. In these two seasons, then, the difference between the ordinary and minimum 

 readings has a diurnal inequality resembling that of ordinary temperature, but having a smaller range. 

 In the case of the ordinary temperature the diurnal inequality for midwinter differed from that for the 

 two other seasons in showing very distinctly two maxima, the larger at from 2 to 4 p.m., the smaller 

 from 4 to 6 a.m., with intervening minima. It is thus with considerable interest that one turns to the 

 diurnal inequality obtained in Table V. for midwinter. Clearly a marked double period in that inequality 

 would be evidence in favour of the reality of the diurnal inequality of ordinary temperature. On 

 examining ths figures assigned to midwinter in Table V. one sees traces of apparently three maxima ; the 

 two largest and best defined answer to the periods 2 to 4 p.m. and 6 to 8 a.m. 



5. The twelve 2-hourly readings of the minimum thermometer supply information as to the time at 

 which the lowest temperature for the day is reached. Supposing that in a month of 30 days the lowest 

 temperature occurred on 6 days between and 2 a.m., then the percentage of occurrence of the minimum 

 in this 2-hour period for the month in question is (6/30) x 100, or 20 '0. 



Table VI. gives the percentage occurrence of the minimum throughout the 24 hours for each month of 

 the year, for the year as a whole, and for three seasons made up as explained in describing Table I. The 

 results, except in November, December, and January, are means from two years. 



TABLE VI. Percentage Occurrence of Minimum for the Day. 



3 N 



