'9^ DISCOVERY REPORTS 



isotherms for each month (bound separately in an ' Atlas '), and although these no doubt give a correct 

 picture of the general distribution of Atlantic surface temperature, some features of the isotherms 

 south of about 50 S (e.g. o and ^ 1° in March) are certainly not compatible with the Discovery data 

 In his later paper (1938) Bohnecke gives a table of surface temperatures in 2° squares 



The Norwegian Meteorological Institute published a series of records (1935) collected by whaling 

 factories for the International Polar Year, 1932^3. These include readings of sea temperature which 

 are no doubt accurate, but the great majority were taken near the ice-edge and for our present purpose 

 do httle more than confirm that temperatures here are about or below -- iC. Some temperatures 

 recorded in the first part of the homeward voyages, however, are helpful, though these again do not 

 fill any important gaps. A limited number of observations between the ice-edge and the convergence 

 are to be found also in results of the ' Norvegia ' expeditions (see Mosby, 1933 1934) 



The Australasian expeditions under Sir Douglas Mawson provide material in the Australian Sector 

 in some months in which the Discovery Committee's ships collected no data in that region Some of 

 the temperatures recorded in the earlier expedition (1911^:4) seem very high, and here again it is 

 TaZVI^T 1"!''^'''^'^ -^ f""y ^-1-ble; but those recorded in the Station List of the 

 B.A.N.Z.A.R. Expedition of 1929-31 are undoubtedly reliable (see Mawson, 1940; and Johnson, 1937). 



METHOD OF DRAWING THE ISOTHERMS 

 The distribution of surface temperature is of course subject not only to annual variations, but also to 

 innumerable local and temporary variations and complexities. The object here is to draw isotherms 

 to show as nearly as possible the average distribution of temperature for each month. Thus Plates 

 II-XII show the temperatures which .^ould probably be found in certain positions at different 

 times of year. For example, in January the 2° isotherm crosses longitude 30° W in c^° 30' S This 

 means that according to past experience a ship arriving at that position about the middle of January 

 IS more hkely to find that sea temperature than any other. If a higher or lower temperature is found 

 It IS probable but not certain, that in that region the surface water is warmer or colder than usual, and 

 the degree of probability will depend mainly on the number of different years from which data near 

 to that position are available. Data from the preceding and succeeding months will be relevant as 

 well as from the month in question. 



Since in some regions data are lacking, and in some other regions insufficient to indicate the extent 

 . of annual variations. Plates II-XII can only be a first approximation to the average distribution of 

 temperature The data are, however, very extensive, and the charts should give a more correct view 

 of the distribution ot temperature in these waters than any that have been published before 



It IS obvious that the basic pattern of the isotherms must be a system of concentric rings, generally 

 representing a falling gradient from north to south. The rings, however, are distorted in various ways 

 by water movements and the configuration of the land and the sea floor. The gradient is not usually 

 a steady one, but apart from small fluctuations it is not often inverted on a substantial scale. Inspection 

 of the temperature charts ^ indicates that the highest temperatures generally prevail in February and 

 suggests that the lowest occur about September and October (see also Fig. 6, p. 190). In drawing the 

 isotherm charts I have assumed that any isotherm should be placed progressively farther south in 

 each month from October to February, and similarly farther north from February to September 

 A break in the average seasonal rise and fall of temperature is not perhaps impossible, but the assump- 

 tion that there is no such break seems justified in the absence of evidence to the contrary. There are 



are rlftVero'l^Ts^thr "chti " ""' '°^ '""^ '^"^^'"^' ^'^"^^ °" ^^^'^'^ '^^ ^^^^ temperatures were plotted. Plates II-XII 



