358 SIR DAVID BREWSTER ON THE HOURLY 



II. On the Arrangement of the Monthly Curves into three separate Groups. 



By examining the daily curve for each month, it will be seen that it preserves 

 the general form of the daily annual curve, occasionally deviating into salient 

 and re-entering portions ; but were we to delineate the individual daily curves, we 

 should, in most cases, find the very form of a curve obliterated, and a capricious 

 succession of elevations and depressions substituted in its place. 



The most remarkable result, however, is the distribution of the monthly curves 

 into three separate groups, namely, curves of high temperature, such as those of 

 June, July, August, and September; curves of low temperature, such as those of 

 November, December, January, February, and March ; and curves of moderate 

 temperature, such as those of April, May, and October. 



This distinct separation of the monthly group is well seen in the Plates XV. 

 and XVI. of Volume X., which represents them as in 1824 and 1825. In that for 

 1824 there is a very slight encroachment of the April curve upon that of January, 

 but in that for 1825 the separation is complete. 



In 1826 and 1827 (See Plates XXIV. and XXV.) these curves are grouped, 

 though less distinctly, according to the same law; but, what is very remarkable, 

 the curve for January 1826 is entirely thrown out of the cold group, and in con- 

 sequence of the extraordinary cold which prevailed in that month, its curve is 

 as far separated from those of the winter group, as any one of the groups are 

 separated from each other* 



III. On the Determination of the Tivo Hours of the Bay when the Mean Temperature 



occurs. 



Previous to the establishment of the hourly Register at Leith Fort, nothing 

 was known respecting the times of the day when the mean annual temperature 

 occurs. It was generally supposed to be about 8 o'clock in the morning, and 

 Professor Playfair adopted this as the most probable result. With regard to 

 the time when the annual mean occurred in the evening, I am not aware that 

 even a conjecture had been formed. 



* The extraordinary character of the October curve in 1827 requires to be explained. When the 

 daily schedules for that month were sent to me from Leith Fort, I was surprised to find two for the 

 same day of October with very different numbers. Upon inquiring into the cause, I found that 

 some of the non-commissioned officers who had voluntarily undertaken the duty of observing the 

 thermometer, and for doing which they were liberally paid, had neglected to make the observations, 

 and had filled up the daily schedules with false numbers. It is obvious from the curve that this 

 fi'aud was committed by the person who made the afternoon and evening observations. 



It is interesting to observe how little effect these erroneous observations have had upon the 

 general results for 1827, when compared with those of other years. 



