482 Dr. C. Chree : Results obtained at Kew Observatory with 



the results refer to only a small fraction of the day, 

 viz., 2 to 4 p.m., and that if we were dealing with mean 

 results from the whole 24 hours the annual variation in a + 

 and a_ and the relation to P might be different, 



We know that in the case of P the character of the regular 

 diurnal variation varies with the season of the year. In 

 summer, the element is between 2 and 4 p.m. notably below 

 its mean value for the day, but at midwinter at this hour it 

 is if anything above the mean. Thus the annual variation of 

 P shown by Table I. differs from that shown by the 24-hour 

 mean, in the direction of increasing the difference between 

 summer and winter. In the case of a + and a_ it is at least 

 as likely as not that the opposite is true. 



In the case of q and q', annual variation if existent 

 appears to be small, but whether accidentally or otherwise 

 the four equinoctial months all give values above the 

 mean. 



§ 7. In order to see whether a + or a_ exhibited any 

 parallelism to P or to any meteorological element, the days 

 of each month of the three years were arranged in two 

 equal groups, according to the value of the element under 

 consideration. When temperature, for instance, was being 

 considered, the two groups were composed respectively 

 of the warmest and coldest days. If there happened 

 to be an odd number of clays of observation, the day which 

 was central as regards temperature was omitted. Mean 

 values were found for each of the groups, for tempera- 

 ture, a + , a_, a_/a + , and P, as well as the corresponding- 

 values of (%a_l%a + ). Calling these two mean values for 

 any element from any one month of the .year m and m' ', and 

 distinguishing the three years by the suffixes 1, 2, 3, means 



M = (mj + m 2 + m 3 )/3, and M' = (m^ + m 2 ' + m 3 ')/3 



were then calculated. The means of the 12 M's and the 

 12 M"s were ascribed to the year as a whole, the means of 

 the M's and M"s for November, December, January, and 

 February were ascribed to winter, and so on. The difference 

 between the final means M and M' for the year, or for any 

 season, was taken, and in the case of a + , a_, and P, this 

 difference was expressed as a percentage of the mean value 

 of the element for the season. The results thus found are 

 given in Tables III. to VI. 



