ATLANT. DEEP-SEA EXPED. 1910. VOL. i] PHYSICAL OCEANOGRAPHY AND METEOROLOGY 



43 



the uppermost layer of water and cause a greater rise of 

 the temperature at, for instance, 5 or 10 metres than 

 would have been found when no waves stirred the water. 

 In .section 32 we shall, discuss the seasonal variations of 

 temperature. It may, now, just be mentioned that the 

 maan rise of temperature at the greater part of the sur- 

 face of the North Atlantic between 25" and 50 ' N. Lat. 

 from February to August amounts to 5°- 8 ' C. or around 

 0-04° C per 24 hours as an average for the whole epoch. 



Our computations slioxc that the temperature at 50 

 metres in spring and summer on an average rises about 

 0004" C per day as a result of the absorption at this 

 depth of radiation from sun and sky. It corresponds to 

 an increase of about 0-7° C from April to September. 

 The corresponding rise of temperature at 100 metres be- 

 loic the surface is on an average about 0-0013^ C per 

 day or 0-2^ C from April to September. 



Absorption of heat radiation from sun and sky goes 

 on in the great oceans every day all the year round. 

 The visible rays cause a heating (however slow) quite 

 far down below the surface. The water itself only radiates 

 dark heat-rays with a very great value of t. This radia- 

 tion is included in the molecular conductivity of heat 

 which has been determined by experiments and found 

 to be exceedingly small in water. Practically speaking, 

 the heat once given to the water at, say, 100 metres' 

 depth by radiation from above will not escape again in 

 the form of radiation unless the water comes to the sur- 

 face. It would be retained and the effect be accumulated 

 so that the temperature always would be on the increase 

 if the heat were not taken away by convective processes 

 such as turbulence and currents. 



31. Conduction of Heat. 



We shall here only discuss the vertical conduction of 

 heat and not the transport of heat by currents. 



In water which is perfectly motionless heat from above 

 will propagate downwards at an exceedingly slow rate. 

 The coefficient of molecular heat conduction is so small 

 that temperature variations due to such conduction may 

 be perfectly ignored in most cases as far as the ocean 

 is concerned. 



If the water is in motion (waves and currents) the 

 water particles generally will acquire disordered move- 

 ments in many directions. This phenomenon is called 

 turbulence. We shall, for the sake of argument, assume 

 that we have a horizontal current with velocities that 

 decrease from level to level downwards. Many water par- 

 ticles which at a given moment are found at a certain 

 level will shoot away from it in various directions, up and 



down, and not move parallel to the main direction of the 

 current. They may return to the first level again, and 

 perhaps continue further to the opposite side. The average 

 length of the vertical distances traversed by all the par- 

 ticles depends chiefly upon the velocity of the current and 

 the vertical stability of the water layers. On the assump- 

 tion made, the water particles will, when shooting up- 

 wards, upon the whole get a slightly increased velocity 

 in horizontal direction corresponding to the difference in 

 current velocity between the two levels, but at the same 

 time reduce the current velocity at the upper level. The 

 opposite effects are caused by the water particles shooting 

 downwards. These variations in horizontal velocity are 

 obviously effected momentarily. The process is of great 

 dynamic significance; it leads to the notion of "turbulent 

 friction" or "turbulent viscosity", and a "virtual coefficient 

 of friction" which is different from the ordinary coefficient 

 of molecular viscosity. 



Now the water particles take their physical and chemi- 

 cal properties with them when moving from one level to 

 another, so also their contents of heat. By the turbulent 

 motion an exchange of heat between the different levels 

 must take place, but we may a priori presume that this 

 effect requires a longer time than the corresponding dyna- 

 mic effect to be fully established [cf. J. P. J.\cobsi:n, 

 1913, p. 71). We may speak of a "virtual coefficient of 

 heat conductivity" which is much larger than the coeffi- 

 cient of molecular conductivity of heat, but probably nu- 

 merically smaller than the corresponding coefficient of 

 friction. 



Processes such as these take place nearly always when 

 a current appears in the sea. The turbulent motion gene- 

 rally becomes more and more vivid the quicker the current 

 flows. In homogeneous water disordered motion may 

 easily arise and the mean vertical transfer of the water 

 particles from their original place be very appreciable. 

 The latter is reduced in water where a stable equilibrium 

 exists and the more so the greater the stability is. In 

 marked discontinuity layers (cf. section 20) the single water 

 particles will not move up and down in turbulent motion 

 to any noteworthy extent. Such a layer forms a very 

 great obstacle to vertical displacements by turbulence and, 

 therefore, prevents or at any rate very materially reduces 

 an interchange of water from both sides. 



Thus, the virtual conduction of heat is also relatively 

 great in water with small vertical gradients of density. 

 Supposing that the density chiefly depends upon the tem- 

 perature (the salinity nearly constant) we must find less 

 turbulence by great vertical gradients of temperature than 

 by small. But at the same time a small vertical motion 

 of the particles in water with great vertical variations in 

 temperature may have just as great or greater effect on 



