52 



BJ0RN HELLAND-HANSEN 



[rep. of the "MICHAEL SARS" NORTH 



Figs. 12 and 13 demonstrate the difference in the re- 

 sults which one obtains by taking the actual temperatures 

 and the "anomalies" of temperature. The graphs refer 

 to 100 m. below the surface for the two areas B and C. 

 The marks in the upper part of each of these figures 

 show the variations in the temperatures directly observed 

 while the marks in the lower part represent the anomalies 

 mentioned. It appears quite clearly from these figures 

 that the large dispersion which the direct observations of 

 temperature exhibits is greatly reduced by the new method. 

 The dispersion is, in fact, reduced mostly to one half 

 or one third, in some cases even more. By help of the 

 "temperature anomalies" one can, with a rather high de- 

 gree of accuracy, draw mean curves whereby seasonal 

 variations stand out clearly. Beside the seasonal variations 

 some annual changes may occur, a point which we shall 

 return to later on. 



Most of the observations have been taken in spring 

 and summer, and it is to be regretted that so few obser- 

 vations are available from September to April. The re- 

 sults with regard to the latter period are, therefore, rather 

 uncertain. In the northern area A, there are no observa- 

 tions at all from the beginning of September to the end 

 of May. 



Figs. 14, 15 and 16 show the results from each of 

 the 3 areas mentioned, for the surface, 25, 50, 100 and 

 200 metres. Even though, due to absence of adequate 

 material of observations, many details may be uncertain 

 the chief features of the seasonal variations in temperature 

 seem to appear fairly clearly. The curves present the fol- 

 lowing results with regard to the upper 100 metres in the 

 middle and the southern areas, the Roman ciphers indi- 

 cating the months when minimum and maximum of tem- 

 perature occur and a r the annual range of temperature: 



The annual range found for the surface seems to 

 correspond as nearly as may be expected to the values 

 represented in Schott's chart. Fig. 9. In the region corre- 

 sponding to our area B the annual range of surface tem- 

 perature varies from 6 to 8° C. according to Schott. In 

 the southern area the annual range found from the chart 

 amounts to between 4 and 6° C. for most of the stations 

 here dealt with. The coincidence is as good as may be 

 expected and speaks in favour of the method here used. 



The amplitudes are still large at 25 metres and not 

 much less than at the surface. They decrease rapidly 

 downwards. At 200 metres the amplitude as found from 

 the curve is practically nought in the southern area. In 

 the area off the Bay of Biscay the curve for 200 metres 

 shows some greater variations, especially a narrow maxi- 

 mum in July, but these variations are rather doubtful 

 (see further down). 



In Fig. 17 the curves from the 3 areas from 0, 25, 

 50 and 100 metres are grouped together. The curves 

 from one depth show great similarity in the general features, 

 but some variations appear from one area to another which 

 have a certain significance if they be real. These varia- 

 tions probably chiefly depend upon the different origin 

 and the "history" of the water masses which dominate 

 in the area. 



The shapes of the curves do not correspond to simple 

 curves of sine. At all depths the curves show a relatively 

 flat course for the winter months. The physical expla- 

 nation is evidently the following: When the surface water 

 is cooled in late autumn and winter, the vertical convec- 

 tion makes the deeper water layers take part in the cooling. 

 A neutral equilibrium or even a state of instability develops 

 to greater and greater depths as the winter cooling goes 

 on. In this way increasing quantities of water have to 

 be cooled with the effect that the fall of temperature 



becomes less and less. In the later part of the 



(-r;) 



winter the water layers down to 100 metres or more give 

 off some heat by being carried to the surface where the 

 radiation outward and the convection of heat to the 

 atmosphere exceed the absorption of solar radiation. Later 

 on the heating of the surface starts and gives a relatively 

 quick rise of temperature in the uppermost layers where 

 the maximum of temperature is reached in August or 

 September. The fail of temperature takes place compara- 

 tively rapidly at the surface and 25 metres in September, 

 October and November. When the temperature at the 

 surface rises, the stability just below the surface increases 

 and reduces the virtual conductivity of temperature down- 

 wards. The deeper layers are thus comparatively little 

 heated in summer (section 31) while the winter cooling 

 in these layers is quite effective. This means that the 

 mean annual temperature at, for instance, 50 and 100 

 metres is much lower than it would have been if the 

 vertical conductivity of heat had been constant all the 

 year round. The difference amounts to more than 2° C. 

 The winter minimum at 50 and 100 metres shows much 

 the same temperature as tiie minimum at the surface but 

 the summer maximum at those depths is very low com- 

 pared with the maximum at the surface. 



The curves for 50 metres show a deformation in 



