166 PRESSURE. 



1911, however, barometer observations were commenced at intervals of four hours. The.se 

 were continued ahnost without break until the end of October 1912. In November 1912, 

 owing to nearly all the members of the exjDedition taking part in the different search parties, 

 the barometer observations became irregular, but on every day the barometer was read at 

 least once and on most days three or four times. 



The eye readings of the barometers were reduced to sea-level and constant gravity. 

 The barograph traces were measured at each hour and the readings tabulated. Against 

 the corresponding barograph measurement each reduced barometer reading was entered, and 

 the difference taken. Ihese differences were then plotted on the actual barograph sheet to a 

 scale 10 times larger than the barograph scale. As most of the changes in the difference 

 were due to lag in the barograph, it was found that the curve of differences was very similar 

 to the actual barograph curve. This similarity made it very easy to draw the curve of 

 differences and is a great advantage of this method of reducing a barograph trace. When 

 the curve of dift'erences had been drawn, it was read off at every hour and the value entered 

 against the corresponding value from the barograph curve. The barograph and difference 

 values were then added together, giving the correct barometer reading at each hour. These 

 measurements were all made to jjjyth of an inch, which is the greatest accuracy possible 

 with a barograph. The mean values of the month from the barograph reductions for each 

 of the hours at which the barometer was read were found to agree to -fj^jjy-th of an inch 

 with the mean of the barometer readings. 



During the period when barometer observations were not taken every four hours the 

 reduced barograph records are correspondingly less accurate, but there is no doubt that even 

 in these periods the mean monthly value is as accurate as the barometer correction, and 

 the mean daily variation for the month is probably as accurate as could be obtained with the 

 full set of observations, for this element is more affected by the non-periodic pressure changes 

 than by the method of reduction. 



A similar method was used in reducing the Cupe Adare barograph record, but as during 

 the greater part of the year eye observations were not taken during the night, the reduction 

 is not so perfect. It was therefore considered sufficient to reduce the Cape Adare barograph 

 trace only at every second hour. During a few periods all the members of the Cape Adare 

 Party were away from the station for several days. The values of the pressure for these 

 days have been obtained whenever possible from the barometer observations taken in the field- 



Mean pressure and annual variation. 



McMurdo Sound* 



The following table contains the monthly values of pressure at McMurdo Sound for 

 the four years for which data are now available. • 



* Since this chapter was finished ii,nd ready for the press, Mr. Mossman has very kindly drawn my attention 

 to the fact that while the pressure data contained in tables 1 and 3 of chapter XI, pp. 476-482 of the volume 

 of results of the Discovery Exi)edition have been reduced to gravity at latitude 45°, those in table 2 have not. 

 As there is nothing to point this out in the chapter itself, I had used the mean monthly pressux'es given in 

 table 2 under the impression that they were fully reduced. Thus half the data used in this chapter was wrong 

 by -070 inch. This necessitated recalculating most of the tables of pressure and also altering many of the 

 diagrams. All the tables have been corrected, but it was too late to alter all the diagrams. Tho.se diagrams which 

 were materially affected have been corrected, but the remainder have been left unchanged. Thus in a few 

 diagrams the plotted points do not agree with the figures in the tables, but in none of these cases is the character 

 of the diagram affected- One cannot too strongly deprecate the use of pressure data, some of which are fully 

 and the remainder only partially corrected. 



