SUEFACE WATERS OF THE NORTH ATLANTIC OCEAN. 
67 
during the “ intervals,” we are justified in assuming that we have made a legitimate 
and adequate application of the method, and ultimately that the supposed changes 
have actually taken place. A more limited application of the method can be made 
where sets of observations are separated from others by intervals too long for the 
assumption of continuous change. Under this heading comes, for example, a com¬ 
parison of the state of affairs at the same season in different years; wdiich is valuable 
in that strictly seasonal variations are eliminated, and irregular or long-period changes 
can be studied by themselves. 
The joint observations of 1893-94 were, for the most part, made wfithin a period of 
a week or ten days, and except in certain localities, or under unusual weather 
conditions, it was found that, at least in so far as temperature and salinity were 
concerned, the observations of each set could be treated as simultaneous, and plotted 
together on curves or charts. Again, the interval of three months was, for the most 
part, found to be sufficiently short to allow of comparison of each set of observations 
with its successor, at least in the main outlines (21) ; and the general nature of the 
changes in temperature and salinity could be traced. 
The net result of these comparisons was to establish, beyond all reasonable doul_)t, 
that the variations in the circulation already knowai to exist in more or less enclosed 
areas like the Baltic, occurred not only in the comparatively open North Sea, ljut 
even in the open channels connecting it with the Atlantic (22); that not only d() 
seasonal variations of wdde amplitude take place from month to month, hut tliat 
irregular variations of probably equal magnitude render the type of circulation 
markedly different at the same period of different years. These latter variations, 
about which no chart of average conditions can give any infoiination, certainly 
originate in the waters at or near the surface, and prol^alily involve remote regions 
of the ocean. Althoim'li tlie existence of irreo'ular variations in the surface conditions 
O O 
of the North Atlantic was recognised by Petermann, and probably accounts for the 
widely divergent views held about the general circulation by scientific men almost up 
to the present day, as well as for the persistent scepticism of a large number of 
efiicient navigators, the large proportion wdiich they bear to the wdiole mean move¬ 
ment in circulation has not been realised, and no systematic attempt has hitherto 
been made to ascertain their nature and extent. The reason for this is perhaps to be 
looked for in the tenacious hold Avhicli the idea of “rivers in the ocean ” still retains 
in the minds of many with regard to currents. 
The importance of a knowledge of the changes just referred to wns strongly 
impressed upon me in the course of the work done under my direction on board 
H.M.S. “ Jackal ” in 1893 and 1894. Believing that, at least in so far as the open 
ocean is concerned, the greater part of the information required could be obtained 
from surface observations, I determined to see if the observations ordinarily recorded 
in the log-lx)oks of sea-going vessels w’ould provide sufficient material for the constinic- 
tion of synoptic charts of temperature and salinity, wdien, as seemed nec&ssary (partly 
K 2 
