part at any rate, the surface was rising slightly from the sea-level at Winter ',)ii;n NTS, but how much is 

 very doubtful. Captain HKI-NVOUTII (p. 439), using an average height for the barrier, is of opinion that 

 the observations give no conclusive indication of an increase of pressure southward, and conjectures an 

 isobar running nearly north and south. Mr. CURTIS shows that the rise of surface necessary to account 

 for the observed diminution of pressure southward is so slight that more might well lie allowed, and infers 

 that the sea-level pressure increased to the southward. 



The Antarctic Anticyclone and Local Wiwls. 



In this connexion it may be desirable to reconsider the general aspect of the question. First of all it is 

 necessary to distinguish the average conditions and the instantaneous conditions on any particular occasion. 

 No doubt the average is made up of the aggregate of individual occasions, and no process of averaging 

 can convert a persistent south-westerly wind into an easterly one, or vice, versd. Actual averages are only 

 available for a few points, and the true relation of even a considerable number of isolated observations in 

 different localities to the average conditions is not easily estimated. The Antarctic anticyclone, if it exists, 

 is a comparatively superficial effect attributable to the surface cold. But to give an easterly wind there 

 must be sufficient thickness of cold air to reverse the gradient of the upper air, which, as shown by the 

 smoke of Erebus, by cloud observations, and by the observations at high levels, is poleward ; 5000 feet is 

 probably an ample allowance for the thickness of the cold surface cap which has an east to west rotation. 

 In order to give a resultant gradient for easterly winds, the gradient of the cold surface layer must exceed 

 that of the westward moving layer up above. One requires, therefore, a rapid change of pressure in the 

 surface layer, and with that we should expect to find a rapid temperature gradient poleward.* 



There seems no reason to suppose that when once the permanent ice is reached there is very rapid 

 diminution of temperature further to the south. We may suppose, on the contrary, that when the tran- 

 sition from open water to ice takes place there will be a transition of air temperature, and a pressure 

 gradient associated therewith, but once upon the permanent ice continent the changes of general 

 temperature and their influence upon surface pressure will be slight and possibly local. 



The pressure conditions to be anticipated over the Antarctic may, perhaps, be described rather as a 

 plateau with a steep fringe along the line of transition from ice to water than as a region of continually 

 increasing elevation. The shape of the plateau will follow, roughly speaking, the shape of the permanent 

 ice continent, and the average isobars will therefore roughly follow the coast line. Captain HEPWORTH, 

 on p. 440, has drawn attention to the tendency of isobars to conform to the coast line in other regions. 

 What local variations there may be within the area of the plateau is altogether undetermined. 



The number of calms observed in the Antarctic shows that the surface gradients are often very slight. 

 The rapid fluctuations of temperature which must be associated with change of surface pressure, extending 

 perhaps to no great height and therefore not recorded on the barograph, show that changes must be local 

 and spasmodic, and we get the idea of the polar region within the line of minimum pressure as a region 

 with a fringe of gradient for easterly winds where the rapid transition of temperature takes place from 

 sea to ice, and within that belt little regular change of pressure, frequent calms with spasmodic changes 

 of temperature and wind not closely associated with the general surges shown on the barograph. 



Attention may also be called here to Mr. DINES' conclusion (p. 475) that the air of the Antarctic is 

 exceptionally dry, and that the dryness may be attributed to the descending air of an anticyclonic region. 

 This suggestion obviates to some extent the difficulty of the divergence of observation as regards wind 



* It may be noted that, in order to account for the transition from the surface gradient for easterly winds through an 

 intermediate stage of southerly airs to a gradient for westerly winds at, the level of the upper clouds, we require a temperature 

 distribution with a fall of temperature towards the east of south in the first stage and towards the west of south in the second 

 stage. Regarding these distributions as a fall of temperature towards the eouth in both stages, with a west to east gradient 

 of temperature in the lower stage and one from cast to west in the upper stage, we get a combination of conditions which 

 is by no means improbable, in view of the fact that the temperature fell to the south along the surface, and in the upper 

 layer the high land surface to the west would imply lower temperature there than at the same level in the free air over the 

 Bea of cold air lying on the barrier surface. 



