dont In recent oceanographic literature Is that, when analyses 

 are made either of time-series or synoptic data, the ocean is 

 found to be much less invariant ^^fith time than was heretofore 

 assumed. No better Illustration of this thesis could be given 

 than that of Defant (1950) who demonstrated that, when one elim- 

 inates the effect of each observation being taken at a time and 

 place different from every other observation, an entirely dif- 

 ferent and more credible pattern of the oceanographic variables 

 results. In this case Defant eliminated the effect of the tidal 

 internal wave, but undoubtedly a thorough investigation of other 

 factors, such as the effect of horizontal gradients in the ocean, 

 would give the same result. 



It was clear that what was needed for this investigation 

 was a series of repeated anchor stations taken at one or more 

 locations. These anchor stations initially should be in 

 locations where the variables which could not be accurately 

 measured would have little or no effect, such as an area where 

 relatively little current exists and where there are no abrupt 

 changes in bottom topography. These observations should be 

 repeated every hour or two, and frequent BT's taken at regular 

 intervals. They should also include simultaneous observations 

 of such other parameters as might be important to changes in 

 the thermal structure, such as incoming and outgoing radiation, 

 transparency, air temperature and humidity, vjind, and sea and 

 swell. It. was considered that not too much could be learned 

 from single anchor stations of one day's duration since they 

 afford little or no opportunity for testing techniques or 



69 



