527 



I. FROM Forenoon Observations. 



II. FROM Means for each Day. 



III. FROM all available Observations. 



The general character of the annual variation as deduced from any of the three tables is the same ; 

 there is a maximum in December, a minimum in June or July, and apparently a secondary minimum in 

 March or April. 



Table I. is unsatisfactory on account of the very small number of the forenoon observations in the 

 winter months. To the use of the methods of Tables II. or III. in determining the annual variation it 

 might be objected that when the observations are distributed irregularly over the 24 hours, there is a 

 danger of an apparent annual variation arising from a variation in the hours of most frequent observation 

 at different seasons of the year. In the summer months, for which alone the data are in any way sufficient 

 to test the existence of a daily variation, the mean of the forenoon observations for each month does not 

 differ to any serious extent from the means derived from all the observations as in Tables II. or III. 

 Whether the method of Table II. or that of Table III. is the better it is not easy to decide. Table II. is 

 not really derived from the values of the true mean potential for each day ; the values used for the mean 

 potential for each day being the mean of the results of the undisturbed observations for that day, the 

 number of such observations varying for different days from 1 to 24. It would seem natural to attach 

 more weight to the mean for a day the greater the number of observations from which such mean has 

 been deduced. If we were to weight the means for the various days simply in proportion to the number 

 of observations, we should arrive at the numbers given in Table III. On the other hand, it might be held 



