ATI.ANT. DEEP-SEA EXPED. 1910. VOL. i) PHYSICAL OCEANOGRAPHY AND METEOROLOGY 



93 



Morocco westwards towards the Sargasso Sea and north- 

 wards towards the Bay of Biscay and still farther north. 

 This is due to the propagation of water from the Medi- 

 terranean. The salinity-anomalies reach a maximum off 

 Southern Portugal, which indicates that the water from 

 the Mediterranean at this depth chiefly moves along the 

 south-western coast of Spain and Portugal, probably as 

 an effect of the rotation of the earth. The high values 

 of the salinity-anomalies in these regions are not due to 

 variations in the vertical distribution of the water-masses, as 

 a similar distribution is also found at 800 and 1200 metres. 

 At 2000 metres we also find comparatively high po- 

 sitive values of salinity-anomalies in a region e.xtending 

 from Southern Spain and Portugal to the west and north. 

 Relatively salt water appears at this depth off Southern 

 Europe and in the Bay of Biscay, and in the sea around 

 Aladeira and south and south-west of the Azo'es. 



The small-scale charts on p. 97* are rather schematic. 

 Figs. 35 and 36 show the distribution of salinity-anomalies 

 in more detail at 400 and 1000 metres in the eastern 

 part of the North Atlantic. These charts correspond to 

 the charts of temperature and salinity reproduced in Figs. 

 28 and 29. 



Some sections are reproduced on pp. 87*, 89* and 

 93*, illustrating the vertical distribution of salinity-ano- 

 malies. A comparison of these sections with the cor- 

 responding sections showing the distribution of tempe- 

 rature and salinity gives an impression of the advantages 

 afforded by the method of salinity-anomalies. By this 

 method the admi.xture with the Atlantic water of Medi- 

 terranean water on the one hand, and of Polar water on 

 the other, seems to come out more clearly than by the 

 usual study of the distribution of isotherms and isohalines 

 alone. 



VIII. STABILITY. 



40. Calculation of the Stability. 



We have mentioned above (section 34) that a perfect 

 mixture of water-masses creates a uniform salinity while 

 the temperature rises slightly downwards owing to the 

 adiabatic effect. In these circumstances we have a state 

 of neutral equilibrium and the stability is =; 0. The 

 potential density is then the same everywhere. In by far 

 most cases the water-masses are not thoroughly mixed, 

 the salinity showing vertical variations and the temperature 

 a vertical gradient different from the adiabatic, with the 

 potential densities increasing downwards. The greater the 

 increase of the potential densities in vertical direction, 

 the greater is the stability. In hydrographical tables the 

 values of at are, as a rule, published, and we may obtain 

 an approximate value of the stability simply by comput- 

 ing the vertical variation per metre of <rt. In so doing 

 we neglect the adiabatic influence. In the deep water, 

 for instance, we may find a decrease of at downwards 

 and, seemingly, a state of instability, while neutral equili- 

 brium or even a state of positive stability is found when 

 the adiabatic variations of temperature are taken into 

 account. In such cases a more exact calculation of tiie 

 stability is needed. 



Hesselberg and Sverdrup [1915] have published tables 

 for the computation of the stability, E. They have ex- 

 pressed the stability in the following way: 



Az 



where o means the density of the water at the depth z 

 and {>' the density of a water particle which has been 

 moved to the same depth from a depth z + A z. In 

 a series of tables Hesselberg and Sverdrup have given 

 values for the effects of variations in salinity, temperature 

 and depth upon the stability, so the final value of E can 

 be found by addition. The observations from the "Mi- 

 chael Sars" Expedition have been manipulated by means 

 of these tables, and the results are given in the 5th column 

 of Table III in Part II. In a later publication |1929] Hes- 

 selberg has shown that the values found in this way are 

 not quite exact, but the errors are of no consequence 

 in the present connection. 



As mentioned on p. 21, \i)* E - 1000 corresponds 

 nearly to a vertical variation of 0-01 of at per metre. 



41. Horizontal and Vertical Variations of 

 Stability. 



The stability varies in" the upper water-strata in the 

 course of the year on account of variations in heating 

 and cooling. The cooling at the surface in winter tends 

 to create a state of instability, and sometimes the water 

 at the very surface may be a little heavier than the water 

 below until the vertical convection brings about a thorough 

 mixing. Such instability may also be observed as the 

 result of cooling during the night or an increase of the 

 surface salinity on account of evaporation. E.xamples of 



