280 MALKus [chap. 4 



would lead in fifty years to a cooling of the column by nearly 3^C! Thus, even 

 if all other budget terms were surely unaltered by less than this margin (which 

 is, of course, preposterous) and evaporation were determined regularly over 

 wide regions to the best accuracy foreseeable methods permit, it still is in- 

 herently impossible to pin down this long trend in Qe from its shorter and larger 

 fluctuations. It would be well for the reader to contemplate how differences of 

 this size in Qvo or in radiation could be estimated. Furthermore, when we come 

 to the point where energy fluxes of 1 cal per cm^ per day are important, the 

 terms left out of equation (1) are no longer negligible. In particular, the dif- 

 ference in heating produced by kinetic energy dissipation amounts to this much 

 if winds are changed from only 6 to 8 m/sec. 



No climatological pictures of exchange like those we have presented in 

 Sections 4 and 6 of this chapter existed fifty years ago and, although it is to be 

 hoped that fifty years hence much better determinations will be available, it 

 should be clear from the discussions in this chapter that the uncertainties in 

 the present computations alone must preclude any chance of relating the 

 ensuing climatic or oceanic structural changes in any rigorous, formal, or even 

 demonstrably sound physical manner. 



Nevertheless, we have enough data today to infer the likelihood of secular 

 changes in the decades since the turn of the century, particularly in atmospheric 

 circulations and sea-temperature patterns. A brave paper by Bjerknes (1959) 

 has attempted to describe these in the North Atlantic and to relate them 

 temporally and spatially by mechanistic hypotheses. 



Test periods of eight years each were selected on the basis of data coverage, 

 namely 1890-97 and 1926-33; this interval averages out short period fluctua- 

 tions with which the Bjerknes study was not concerned. His analysis of changes 

 in sea-surface temperature between these intervals is shown in Fig. 88. The 

 significant features are a marked warming in the Gulf Stream region off Grand 

 Banks, a slight cooling centered around 53°N south of Iceland, and a slight 

 warming in the Bermuda-Sargasso Sea area. We are plagued here with the 

 problem of inadequate and uneven sampling and perhaps also poor representa- 

 tiveness of data, particularly in the 5°-latitude square of the intense warming, 

 through which the shipping lanes' were shifted between the sample periods (due 

 to the Titanic disaster). Nevertheless, more recent weather-ship data studied 

 by Rodewald (1956) confirms a similar general pattern in the post-war 

 years. 



The pressure change map at sea-level for the same periods, superimposed on 

 average sea-surface isotherms, is shown in Fig. 89. Bjerknes used this, some- 

 what in the manner of Namias, to deduce changes in the prevailing geostrophic 

 wind between periods, and thence to infer what might be the alterations in 

 surface-wind stress upon the sea via equation (17). Therefore, the geostrophic 

 wind speed squared for each period is shown in Table XXV, computed from 

 Fig. 89. 



The profile from Bermuda to Port-au-Prince is intended to give a measure 

 of change in trade-wind strength, while that from Bermuda to Eastport 



