period of 1896, and in the far north the whole circulation of hoth warm and cold 
streams at the surface seems to have been slower. 
In September, 1897, the general direction of the drift loses its southerly component, 
and in October it becomes due east. Hence there is at hrst “ banking-up ” of water 
against the European coast and escaj^e noidhward, causing the nortliAvard stream west 
of the British Isles to run stronger, and salter water to spread over the central area 
generally. After October the supply from the lower latitudes diminishes, hut the 
drift eastevarel from the Labrador coast continues into November. It is to he 
remarked that the spreading of the Polar water eastwarel from the soutli-east of 
Greenland, so strongly marked in 1896, is scarcely noticeable in 1897. 
Towards the end of the vear the drift from the Labrador current, which has not 
contracted to its usual size after the expansion in August, expands southward again, 
and the “north and south” distribution becomes more marked than in 1896, partly 
because of tills, and partly l)ecause of tiie iielated strengthening of the northward 
streams on tiie eastern side. 
It is a matter of some difficulty to ascertain with any detail hmv far the features 
common to tlie circulation in tlie two vears descrilied are reproduced every year. Tlie 
most relialile means of comparison is proliaiily the series of temperature and current 
charts published by the Meteorological Gftice (lO), 1)ut these do not contain much 
information about the circulation north of 40° N. during the winter months. In the 
lower latitudes tiiev show, liowever, vliat is important to our purpose, that the cir¬ 
culation round the Atlantic anticyclone is more active and definite in summer tlian 
in winter (January and August). 
The current chart, of course, cannot define the source of the surface waters off the 
Newfoundland Banks, lint in spring and early summer (April and June charts) the 
apparent direct continuation of tlie Gulf Stream becomes shorter, while there is 
increased eastward movement from the east of Newfoundland. 
In August the easterly drift shows a tendency to divide near the south-west ot 
Ireland, the greater part appearing to go southv’ard, while a narrow stream moves to 
the north. Further north the general direction of movement is more to the south. 
The charts for October and November show tliat the head of the Gulf Stream 
broadens and retreats, while in the north the easterly component lieconies well marked ; 
note the easterly direction of the arrows south-east of Greenland. The dividing line 
formed by a hand of “ no current ” between Nevdbundland and the British Isles is an 
important feature : it moves northward in winter, southward in summer. 
It appears, therefore, that there are certain important seasonal changes in the 
surface circulation which occur in the two years 1896 and 1897, and which can, to 
some extent at least, be traced in the less definite outlines of the composite pictures 
obtained by the method of averages. These changes may be summarised as 
follows :— 
The surface circulation in the North Atlantic between the parallels of 40° and 
VOL. CXCVI.—A. P 
