MR W. H. DINES ON THE VERTICAL TEMPERATURE 



Obviously the figures may be smoothed with advantage, and I consider that the 

 best way is to form the first term of a Fourier series and use the values that it gives. 

 The figures would doubtless give an appreciable coefficient to the second, and perhaps 



TABLE II. Mean Monthly Temperatures Corrected for Barometric Height. 



to higher terms, but, in my opinion, the observations are too few to allow us to attach 

 any importance to such coefficients. The annual variation is a certainty, but the six 

 months' variation seems to me very probably due to pure chance. Taking a sufficient 

 number of terms of a Fourier series we, of course, approximate to the actual figures 

 given by the observations, but we do not thereby obtain a smooth curve. 

 The temperatures thus obtained are shown in Table III. 



TABLE III. Mean Monthly Temperatures at each Height, Smoothed. 



; is of interest to compare these figures with similar results that have been 

 pre' iously given. On p. 42 of Messrs. GOLD and HARWOOD'S report to the British 

 Association, Section A, Winnipeg, 1909, a table is given showing the mean value and 



