138 WIND. 



blizzard conditions. In other woids the blizzards at the surface do not as a rule affect the 

 normal conditions at the height of Erebus smoke, but every now and then the blizzards 

 reach as high as Erebus and then give south-easterly winds there. 



The number of observations of high clouds is too small to insist on small differences 

 in the two diagrams, but the greatly increased frequence of motion from the E.8.E. during 

 blizzards at the surface almost certainly points to the fact that occasionally during blizzards 

 the south-easterly motion does extend right uji to the cirrus cloud layer. 



It must be remembered that the upper aii' motion could not be observed owing to low 

 clouds when the most violent blizzards occurred. The occasional observations of cloud and 

 smoke motion with high southerly winds were made when the sky was more or less free 

 from lower clouds, hence the diagrams do not represent true blizzard conditions. We may 

 therefore conclude that the increase of southerly motion at the different cloud levels would 

 have been more marked if the motion could have been observed during the more violent 

 blizzards. 



It will be noticed that there is still a considerable amount of wind from the south-east 

 quadrant throughout the upper air when -the blizzards were not blowing at the surface. This 

 liowever is not surprising, for blizzard conditions in the upper air may start before and remain 

 after they are active at the surface. With the data available it is impossible to prove that 

 all the winds from the south-east quadrant in the upper air are connected with blizzards, 

 but we have shown a sufficiently close relationship to make this probable. 



We can now sum up the above discussion and state the general outlines of the mean 

 air motion over Cape Evans from the ground upwards. 



Ground level. — The air motion is entirely backwards and forwards through McMurdo Soimd 

 and the motion through the Sound from the south far outbalances the motion from the noith, 

 the relative frequency in the two cases being 64 per cent, and 12 per cent. (24 per cent. 

 being accounted for by calms, variable winds and no observations). 



Height of the medium clouds (10,000 to 13,000 Jeel and therefore below the level of the 

 toj) oj Erehiis). — At this height the motion is still backwards and forwards through the .Sound, 

 but the motion from the north is now more frequent than the motion from the south, the 

 relative frequencies being 55 per cent, and 37 per coit. (8 per cent, being accounted for by 

 no motion of the clouds). 



HeigJit of Erebus smoke (13,000 feet to 16,000 feet).— At tliis level the motion of the 

 atmosphere is no longer affected by the surrounding land masses. There is now considerable 

 motion from the west and south-west — directions which were not possible below the siuumit 

 of Erebus. If we assume that the wi-nds from the south-east are mainly due to blizzards 

 and therefore of local origin, the observations of Erebus smoke show that Ijetween 13,000 

 and 16,000 feet the general motion of the atmosp^here is almost entirely from the western 

 half of the horizon with north and south added. The mean direction from all the observa- 

 tions corrected by observations made on the Discovery Expedition is probably from the 

 west-south-west, and if the local blizzard winds are neglected the mean motion is practically 

 from due west. 



Height of the high clouds {above 16,000 feet). — The high clouds show a certani amount 

 of motion from the south-east quadrant which in all probability is due to the local blizzards. 

 If these winds are neglected practically all the motion at this height is from the northern 

 half of the horizon with east and west motion added. The mean motion excluding the winds 

 in the south-east quadrant is slightly from the west of north and including these winds 

 N. 27° E. 



