148 Transactions of the South African Philosoi^hical Society. 



they imply in Table 23, but rather because of the lower temperature. 

 But the question arises whether the test is a fair one. And if we 

 examine the almanac distribution of the dew-points used here we- 

 shall soon see that it is anything but fair. This distribution is dis- 

 played in Table 25, intended to show the number of days in any 

 month upon which the mean hourly dew-point of the day had any 

 assigned degree. Evidently the low dew-points are characteristic of 

 the winter months, and the higher ones of the summer. So that 

 after all we have merely epitomised, and rather obscurely at the 

 best, the facts of Table 17. The classification, in fact, has failed to 

 isolate the moisture conditions from the equally potent temperatures 

 and lengths of daylight ; and is no more solid argument than the 

 Indian case of Buchan's already cited. It is, therefore, up to this 

 point, and so far as Table 24 can help us, still an open question 

 whether the cloud effect is one of temperature or of moisture. 



Now Tables 26 and 27 are constructed upon a different plan. 

 Instead of attempting to determine directly the ranges of pressure 

 corresponding to the quantities of aqueous vapour present, the dew- 

 points under clear skies have been roughly divided into two sets, one 

 in which they are less, and the other in which they are greater, than 

 the mean of all the clear days of the month ; the concurrent baro- 

 metric pressures being then averaged in monthly columns. On the- 

 face of it the inference is that the barometer rises secularly when 

 the dew-point is low, but displays some disposition to fall when it is 

 high. Nevertheless, considering that the rule, when the monthly 

 numbers are submitted to individual scrutiny, is not universal, this 

 is possibly a fortuitous result. 



As they stand they give these mean ranges : — 



High Dew-points. Low Dew-points. 

 Inch. Inch. 



A -008 -004 



B -047 -055 



-091 -083 



D -047 -052 



But if we eliminate the secular variation they become : — 



A -008 -004 



B -049 -049 



C -089 -090 



D -048 -045 



The first of these results would seem to support that obtained 

 when clear and cloudy days of pressure are arrayed over against. 



