200 



PRESSURE. 



Table 119. 

 Correlation coefficients after deducting the McMurdo Snuiid yearly variation. 



These numbers leave little doubt that for the year under investigation the monthly pres- 

 sures at these stations varied in a similar way. In spite of the large margin between tht 

 correlation coefficients and the probable error, it would not be right to conclude that all 

 years would show the same high relationship. The observations at Snow Hill continued for 

 another eight months after the twelve considered above, and during this period the relationship 

 was much less marked. It must however be pointed out that the added months were re- 

 markably free from large changes and therefore were unsuited for bringing out the relationship. 



Returning now to the data for Kerguelen we see from the curves in figure 60 that the 

 Kerguelen pressure is not very intimately related with the Antarctic pressure, although the 

 changes are somewhat similar. It is remarkable that the Kerguelen curve is more like the 

 curves for Hut Point and Snow Hill than for the Gauss Station, although the latter is so 

 much nearer. This is shown in the correlation coefficients which are +'37, -f-'22 and +'08 

 respectively. We cannot correct the Kerguelen observations for yearly variation, for there is 

 much evidence to show that the yearly variation over the Southern Ocean is quite different 

 from the variation over or near the continent. Considering the small amount of data and the 

 small correlation coefficients, it would be unmse to draw any other conclusion than that the 

 correlation between Kerguelen and the Antarctic is small but positive. 



Observations in the Antarctic region simultaneously with some of those in McMurdo 

 Sound are also available for the South Orkneys and South Georgia. At the South Orkneys 

 simultaneous observations were made from April 1903 to January 1904 and during the 

 whole of 1911 and 1912, i.e., 34 months in all. At South Georgia only 17 months' simulta- 

 neous observations are available from June 1911 to December 1912. 



These observations have been treated as follows : The ten years' observations which are 

 available for the South Orkneys * were used for obtaining the annual variation. The variation 

 so obtained was deducted from each of the 34 months mentioned above for which simul- 

 taneous observations are available in McMurdo Sound.' The annual variation was also deducted 

 from the corresponding monthly pressure in McMurdo Sound. We have then two sets of monthly 

 values which have been freed from periodic change ; the variations of the individual months from 

 the mean of the set were then calculated, and the correlation coefficients determined. 



Exactly the same procedure was employed with the data for South Georgia. In this 

 case nearly nine years' observations are available for determining the yearly variation. 



It will be noticed that this procedure is practically the same as that used for the Antarctic 

 stations except that in the latter case only the mean of 12 months was used as the basis 

 from which to calculate the departures. The aim in both cases is to find departures from 

 the mean pressure of the whole period for which simultaneous observations are available. 



* Mv beat thanks are due to Mr. R. C. Mossnian for providing me with the.«e and other valuable <lata,. 



