[ 408 ] 



XXXIX. The Annual March of Temperature. 

 By R. J. A. Barnard, ALA., Melbourne*. 



THE accompanying curves give the results of an examina- 

 tion of the temperature for forty years at Melbourne, 

 and were suggested by Rijkevorsel , s work on European 

 meteorological elements. 



The process adopted was as follows : — The forty years' 

 observations were divided into two groups, 1859-78 and 

 1879-98. In each group the average for each day is obtained 

 and smoothed twice by replacing the temperature of each day 

 by the mean of the five consecutive days of which the par- 

 ticular day is the middle. The results of the second smooth- 

 ing are given. 



An examination of the curves does not lead to such definite 

 results as given by Rijkevorsel for Europe, but some striking 

 correspondences are noticeable, the most marked being that 

 at the end of March. In the second week of March it will be 

 found that the temperature begins to drop rapidly, reaching a 

 secondary minimum for each curve on the 19th. It then 

 rises again to the extent of 2°\5 during the next week, reaching 

 a maximum on the 25th and 2(ith, the date being again the 

 same for both. I have examined the four groups of ten years, 

 and all the curves show the same minimum and maximum in 

 March. The curves for the first three weeks of January, 

 for the whole of April, and some other parts also show great 

 resemblances, but in many other cases where the curves appear 

 very similar the times of the secondary maxima and minima 

 do not coincide sufficiently closely in time, and in some parts 

 the maxima of one occur with the minima of the other. 



The conclusions I am inclined to draw from my results are 

 the following : — First, that the spells of weather are in all 

 probability not so marked here as in Europe. If this is so 

 the probable cause lies in the more uniform conditions of the 

 southern hemisphere, tending to make the annual march more 

 uniform and free from spells. Secondly, I do not think that 

 a period of less than 40 years can give any real information 

 about spells. • For example, if one had only the first curve 

 of the figure, the spells in June would seem to be quite as 

 striking as that in March, yet the second curve is exactly 

 opposite in the former case. Thirdly, I have no doubt that 

 to establish the existence of spells the most satisfactory method 

 is to take the observations of a particular place and divide 

 them into groups and compare them as I have done, and that 

 these groups should not be less than 20 years each. 



* Communicated by the Author. 



