646 



EIGHTH PACIFIC SCIENCE CONGRESS 



The number of points in each structure is limited by the intervals 

 of the serial observation, and could always be fitted by a plausible log- 

 plot. More data could be taken from bathythermograph observations 

 and in the majority of cases confirmed the log-plot interpretation. How- 

 ever, further examination of bathythermograph data, as summarized 

 in Table II showed that there was a small proportion of structures that 

 were not logarithmic. Generally these plots appeared as long curves, 

 somewhat similar to an inverse function of the depth. As will be dis- 

 cussed later, these are probably immature gradients which eventually 

 will become logarithmic. 



TABLE II 

 Summary of the Experience of Fitting the Log-Plot to Bathy- 

 thermograph Data 



Step Structure 



Considerably more than half the zones were defined by three or 

 more points. In the deep zone the log-plot is the best representation 

 of the data within the limits of error of the observations. The scatter 

 of points increases towards the surface, in the boundary and upper 

 zones, to something more than twice this limit of error. 



The reason for this increasing scatter towards the surface was 

 sought in an examination of continuous data records from the salinity- 

 temperature-depth recorder and the bathythermograph. These revealed 

 step-structure as illustrated in Figure 5. The structure within each zone 

 consists of a number of small, nearly homogeneous layers, separated 

 by narrow boimdary zones. The mean slope through this step structure 

 is the logarithmic gradient defining the major zone. 



The same situation observed in the usual way with sea water sam- 

 pling bottles at discrete depths w^ould indicate a random scatter along 

 the mean gradient. Such serial observations do not define the structure 

 completely, and the only possible interpretation is the logarithmic gra- 

 dient most nearly approximating the general slope in the zone, as de- 

 fined by the dashed line in the figure. 



This interpretation stresses the principal structure, ignores the small 

 variations, and accepts the error implied in the scatter of the points. 



