12 



OBSERVATIONS AND RESULTS IN PHYSICAL OCEANOGRAPHY 



The errors in depth introduced by the uncertainty in 

 q appear thus to be small but it has to be considered 

 that the pressure coefficient is not quite independent of 

 the temperature and that it also may change in course of 

 time, and the possible errors, therefore, are two or 

 three times as great as those which are stated above. 

 3. The mean density of the water from the surface and 

 to the level where the thermometers were reversed is 

 easily determined with an accuracy of 0.0005. Assuming 

 the mean value of the density to be constant and equal to 

 1.035 we find: 



The errors which are introduced on account of uncer- 

 tainty as to the density are thus always small. 



Summing up the results of this discussion, consider- 

 ing that a temperature difference of 10° roughly corre- 

 sponds to a depth of 1250 meters, we find that the errors 

 in the depth as determined by means of unprotected and 

 protected thermometers probably lie within a limit 



Depth in 

 meters 



Maximum error 

 probably within 



1000 

 2000 

 3000 

 4000 

 5000 



+ 20 

 ±21 

 ±24 

 ±28 

 ±32 



The errors of the thermometers enter here with the 

 greatest weight. 



In his discussion of the " Meteor " data, Wust (1932) 

 has shown that errors due to errors of coefficient q in- 

 crease more with increasing depth than supposed here, 

 but simultaneously he assumes the errors due to errors 

 of reading to be smaller. His estimate of the greatest 

 possible total error gives, therefore, smaller values at 

 small depths, but greater values at great depths. He ob- 

 tains, for instance, the values +14 meters and +49 me- 

 ters at 1000 and 5000 meters, respectively, whereas 

 our estimates are +20 meters and +32 meters. His 

 final conclusion is that at depth below 1000 meters the 

 mean accuracy of the thermometric determination of 

 depth is from 0.6 to 0.4 per cent, whereas our final re- 

 sults, after discussion of the actual values, gives mean 

 accuracy of about 1 per cent at 1000 meters and 0.5 per 

 cent below 3000 meters. 



In order to test this result, the cases have been ex- 

 amined in which the wire angle was equal to 5° or small- 

 er. In these cases the wire length gives an accurate 

 value of the depth and a comparison with the depths ob- 

 tained by means of pressure thermometers furnishes 

 data for an estimate of the possible errors in the ther- 

 mometric determination of the depth. The cases in which 

 the wire angle was from 6° to 10° were also studied as 

 the depth corresponds closely to the wire length even 

 when the angle is 10°. If the wire angle remained equal 

 to 10° from the surface and down to the greatest depth, 

 a wire length of 1000 meters would correspond to a depth 

 of 985 meters, but as a rule the wire is curved and the 

 difference between the wire length and the depth is 

 smaller. 



Table 1 contains the results of the comparisons be- 

 tween the depths as obtained by thermometers and the 



Table 1. Differences between wire lengths and thermometer depths 



*Two cases omitted. 



bone case omitted. 



