SUEFACE WATERS OF THE NORTH ATLANTIC OCEAN. 
Ill 
liuring the midsummer period, pressure was ou the whole above the average ; but 
the dilFereuces did uot lead to any definite disturbance of the normal gradients, and 
so far as the winds are concerned the conditions are to be regarded as normal. 
The months of September and October are characterised by deficient pressures in 
the low latitudes, due to the passage of cyclones from the south-west to the Bay of 
Biscay and the British Isles, and by relatively high pressures in Greenland and 
Iceland. This distribution would give an easterly tendency to the winds in the 
north, and the slowness of the easterly movement of the fresh surface waters derived 
from the ice is therefore probably abnormal. In November the cyclone track moved 
northward, and an anticyclone developed over South-western Europe, conditions 
which would increase the easterly movements in the higher latitudes in the 
Atlantic. 
In December 1896 we find the beginning of a different distribution of pressure, 
which continues, somewhat irregularly, but with little interruption, till August 1897. 
The characteristic feature is pressure above the average in the north and west, and 
below it in the south and south-east, the region of deficient pressure being chiefly 
south and south-west of the British Isles. The Atlantic anticyclone does not therefore 
expand north-eastwards as far as it did in 1896, and the track of cyclones skirting it 
is further south and more directly eastward ; it appears also that the cyclones were 
shallower or less numerous than usual. 
Hence the main easterlv drift is weaker on its northern maroin, and the direction 
of movement is more to the southward, tlie chief region of banked-up ^rater, the 
source of the northward-moving currents, is further south, and the currents receive 
less direct aid from the surface drifts. The relatively high pressure in tlie higher 
latitudes would give weaker westerly winds in the Atlantic, and therefore a weaker 
drift circulation, and less spreading of tlie Labrador stream water eastward. The 
drift delivery northward on the eastern side would lie less, the winds being weaker 
and more westerly ; but, on the other hand, the southward deflection of the main 
cyclone track would increase the easterly component of the winds between Iceland 
and Spitsbergen. The water sent northward by the current from the coast of Europe 
would therefore tend to mix less with the water underlying it, and on reaching the 
Spitsbergen region to drift westwards. We know, as a matter of fact, that an 
unusually large area west and south-west of Spitsbergen was open during 1897 — 
probably the result of the enormous amount of warm water sent up in the preceding 
year, and that the open area was covered to an unusual extent by Atlantic 
water (47). 
The supplies of Atlantic water being smaller, and the ice more remote, in 1897 
than in 1896, the increase of the Polar streams in autumn is much less marked ; hence 
an unusually large area is then occupied by the warmer and salter s-urfajce-waters, 
and this is maintained by the peculiar atmospheric changes which take place in the 
latter part of the year. 
