517 



In Table IV. the moan proportion of clouded sky is given for each hour of observation for each of the 

 twenty-font' months (lining which the "Discovery" remained at Boss Island. The figures show for CVITV 

 month a very uniform distribution throughout the day, and, speaking generally, it may be said that the 

 sky was as a nuV about half covered with cloud. The winter months, however, were the least cloudy, and 

 in July, 190.3, there was a particularly cloudless sky, the average daily amount of cloud being only 

 two-tenths. 



In Table V. the result of a different treatment of the observations is shown, the cloud amounts having 

 been grouped to represent the sky conditions indicated by the letters of the Beaufort weather scale. In this 

 si, ile "b" indicates blue sky, or that not more than one-tenth of the sky expanse was covered with 

 cloud ; " be" denotes blue sky and detached cloud, not more than four-tenths of the sky being clouded ; 



TABLE V. Cloud Distribution as Indicated by the Beaufort Weather Scale. 

 (The values given are percentages of the total number of observations made each month.) 



"c" signifies a cloudy sky, where the expanse of sky covered may amount to seven-tenths; "o," or 

 overcast, implies that from eight-tenths to the whole of the sky is clouded. The number of observations 

 falling under each of these divisions in each month is expressed as a percentage of the total number for the 

 month. 



The unexpected result which the table exhibits is that in every month the great majority of the 

 observations fall under one or other of the two extremes "b" or "o," and that as a rule, to which there 

 are but few exceptions, the percentage frequency either of " blue sky " or of " overcast " sky exceeds that 

 of the two intermediate stages combined. Put in another way, this apparently means that small quantities 

 of detached clouds were relatively infrequent, and that when clouds were present they were usually present 

 in quantity large enough to practically cover the entire sky expanse. 



Comparing these results with the observed percentages of the total possible amount of sunshine given in 

 Table I., a fairly close relation is seen to exist between the latter values and the combined percentages 

 of " 1) " and " be " sky in the months of October to February ; the values are as follows : 



