60 



BJ0RN HELLAND-HANSEN 



(rep. of the "MICHAEL SARS" NORTH 



when they may also be expected to represent the water 

 at considerable depths below the surface, in consequence 

 of the vertical convection in the cold season. Our exa- 

 mination embraced observations taken between February 

 3rd and March 4th and partly also between March 15th 

 and April 15th in each of the years 1898 — 1910 [Helland- 

 Hansen and Hansen, 1917, 1920]. 



Fig. 20 illustrates some of the results of this investiga- 

 tion. The curves show the variations of surface tempera- 



10 - 20° W FEBRUARY 



1900 1905 



1910 



Fig. 20. Vari.itions in the surface temperature for February in the 

 eastern North Atlantic between 10° and 20° W. and between 37° 



and 59° N. 



ture for the first of the two epochs mentioned (February) 

 in the North Atlantic off Europe between 10° and 20° W. 

 The temperatures at the base-lines are the mean tempera- 

 tures for February referred to the 11-year period 1900 — 

 1910. The uppermost curve belongs to the area called 

 "A" in section 32 (see Fig. 11); it represents the condi- 

 tions within a bell which goes NE— SW and is about 2 

 degrees of latitude in width. The other curves represent 

 the conditions in an E— W direction, the second curve 

 froin above referring to a field which is 1 degree of 

 latitude in width, the others to belts of 2 degrees. Curves 

 2 and 3 fall well within the area "B", and the lowermost 

 curve within "C". 



The difference between the highest and lowest tempe- 

 rature for February recorded during the years 1898 — 1910 

 amounts to about 1-5° C (taken in the same succession as 

 the curves in Fig. 20, beginning from above, the figures 

 for the separate fields are: 1-4, l-:3, 1-4, 11, 15 and 

 1-9° C). There is sometimes a difference of 1° C or 

 more between February in one year and February in the 



following year. We observe that taken as a whole the 

 curves change from one field to another, which means 

 that the annual variations in the temperature of the upper 

 water-layers are not the same within different areas even 

 if these are fairly small and not far apart. The transi- 

 tions are, however, mostly gradual, representing a certain 

 progression in the appearance of the anomalies. Some 

 features are common to all the curves, as, for instance, 

 a fall in the February temperature from 1903 to 1904, 

 a rise to 1905 and then a fall again to 1906, but the 

 magnitude of the variations is not the same everywhere. 



As a rule the perfect mixing of the upper water- 

 layers due to the vertical convection in winter has not 

 reached its maximum depth as early as February; — 

 generally it is not completed until the middle of March 

 or later. The surface temperature in February may, 

 however, be regarded as being approximately representa- 

 tive of, say, the upper 100 metres, especially when the 

 temperature is below the mean average for the month. 

 The seasonal variations at 100 metres below the surface 

 are relatively small according to the results set forth in 

 the preceding section, and from February to June or even 

 August— September they are smaller than many of the 

 annual variations exhibited by the curves in Fig. 20. 



We have only a very scanty number of observations 

 for the direct study of possible annual variations at 100 

 metres or any other depth below the surface. As men- 

 tioned above, a departure of a "temperature-anomaly" 

 (in the sense described in section 32) from the average 

 of the date, may to some e.xtent indicate annual varia- 

 tions. It must, however, be borne in mind that our 

 examination of the seasonal variations of temperature in 

 the upper water-strata is based upon very heterogeneous 

 material, the observations having been collected from 

 water-masses (currents) of different origin with unequal 

 salinities. We have tried to eliminate the local changes 

 and find the average seasonal temperature-variations, of 

 what may be called a purely thermal character, for fairly 

 large areas. Our method of elimination is not, and cannot 

 be, so perfect that the irregular distribution of the points 

 in graphs representing the actual temperatures altogether 

 disappears; although it is much reduced. We have just 

 drawn attention to the fact that the annual variations in 

 the upper water-strata do not, as a rule, run parallel over 

 very wide areas of the ocean, but alter from one field to 

 another. Some of the water-masses may, therefore, have 

 arrived on the scene with an original temperature which 

 deviates, to a larger or smaller extent, from the temperature 

 of the other water-masses in the same area as regards the 

 annual variations Thus we cannot expect to find exactly 

 similar anomalies everywhere within even a limited field 

 where the annual variations may show appreciable differ- 



