790 MR J. Y. BUCHANAN ON THE 



the maximum to the minimum value in one montli, from January to February, is very 

 remarkable, but the recovery in March to a much higher value, and indeed to a second 

 maximum, suggests that there are special reasons for this irregularity which may be 

 accidental to the particular years under consideration. It is certainly difficult to 

 imagine, though it is well worth while to try to find out, what conditions vary so 

 much and so rapidly between January and April that the excess of pressure in 

 question should show the variations which we observe in the table. 



The curves of yearly march of pressure, Plate VIIL, show a remarkable rise in 

 February in foggy weather and a sharp foil in clear weather, so that the curves are 

 brought much nearer each other. In both weathers there is a minimum in March and 

 a maximum at the beginning of June. In foggy weather this is the principal maximum. 

 In clear weather the principal maximum is in September. 



Turning now to the distribution of pressure over the day in each month, Tables XV. 

 to XVI II., Plates I. and II., we find considerable differences. In Plate I. we have the 

 months of the winter half-year, and in Plate II. those of the summer half-year, both on 

 a scale of 20 : 1. It is seen at once that the curves for foggy weather are smoother 

 than those for clear weather, although sharp irregularities are not wanting, as, for 

 instance, in the month of April. Looking at the clear weather curves, especially those 

 of the winter half-year, it would be imagined that they were curves of a single day's 

 observations, but they are the mean curves for the clear days of the month during 

 thirteen years. 



It will be observed that in all the clear weather curves there is a maximum in the 

 evening between eighteen hours and midnight, this is particularly developed in winter. 

 In these months, especially December, January, and February, the curves are most 

 irregular. In summer they are a good deal smoothed out, and run down to the 

 early morning minimum, which is strongly marked in nearly all the curves. The 

 midsummer months, June and July, show curves which, without straining, could he 

 smoothed into one having a single minimum between three and four a.m., and a single 

 maximum about six p.m. In all the others a double period is perceptible, and in the 

 winter months the serrated character of the curves is too remarkable to be accidental. 



In foggy weather we have, as has been pointed out, greater regularity. The con- 

 tinual change of water from the liquid to the gaseous state and vice versd has a regulating 

 influence on the changes of pressure as it undoubtedly has on the changes of temperature. 

 In comparing the curves for clear and for foggy weather, it has to be remembered that 

 the mean pressures, the hourly deviations from which are shown in the curves, are 

 always much lower than the corresponding ones for clear weather. 



The curves for May and June ver}^ much resemble those for June and July in clear 

 weather. They give at once a smooth curve with single minimum and single maximum, 

 but these points fall rather later, the minimum at 4 to 5 a.m., and the maximum at 9 P.M. 

 A second minimum at 6 p.m. appears in July, providing a second maximum at 2 p.m., 

 and the curve preserves this form in August and September. In October the second 



