1889.] Africa from Barometric Observations. 319 



Even one month's continuous observations with a similar correction 

 ..should give results sufficiently accurate to be of considerable practical 

 value. As a justification for the abandonment of the theoretical 

 formula in the construction of this table, it may l^e observed that this 

 formula is based on certain hypothetical conditions which can seldom 

 or never exist. ^ 



Thus it is assumed that the difference of pressure at two levels is 

 due to the weight of the intermediate vertical column of air, whereas 

 it is obvious that this will only be the case when the atmosphere is in 

 equilibrium. If there is disturbance of equilibrium, whether of rest or 

 motion, this will clearly affect the tension of the air. The error in 

 this case enters principally in the correction for temperature. Tem- 

 peratures registered by thermometers near the earth's surface are local 

 and do not represent the temperature of the bulk of the super- 

 incumbent air stratum. Thus for example the mean of the tempera- 

 tures registered by two thermometers, one, say at the Observatory 

 and one on the top of Table Mountain, will not, except by mere 

 accident, give the mean temperature of the stratum of air between 

 these levels. This stratum, coming as it does over vast expanses of 

 ocean, has attained a temperature which can be very little affected by 

 radiation from a few miles of terrestrial surface, whereas the tempera- 

 ture registered by thermometers within a few feet of the surface 

 must be greatly affected by such radiation. 



When the points of observation are many miles apart, as for 

 instance in the case of two thermometers, the one at Cape Town and 

 the other at Kimberley, the ordinary correction for temperature at the 

 time of observation is utterly fallacious. Whenever such observa- 

 tions have to be used, better results could probably be obtained by 

 correcting for 'the mean temperature of the month. Barometric 

 pressure, as is well known, does not in any way obey the daily 

 changes of temperature. Temperature has generally but one 

 maximum and one minimum in the twenty-four hours, while 

 barometric pressure has two maxima and two minima. On the other 

 hand variations of average monthly barometric pressure ali seem con- 

 nected with variations of average temperature. 



It has been stated that what is at present required for meteoro- 

 logical purposes is not so much the collection of additional observa- 

 tions as the careful digestion of those Avhich have been already 

 accumulated. There are, no doubt, some good grounds for this 

 observation, but the experience obtained in drawing up this tables 



