10 



OBSERVATIONS AND RESULTS IN PHYSICAL OCEANOGRAPHY 



Table 5. Comparisons between thermometers which 



were used during July, September, October, and 



November, 1929, and had been calibrated either 



during October, November, 1927, or January, 



1928 (old thermometers) or during March 



or August, 1929 (new thermometers) 



Old minus new 



thermometer 



PTR no. 



No. of 

 compar- 



Mean 

 differ- 

 ence 



Range 



127558-198 37 0.009 0.070 



127087-203 6 0.005 0.040 



127089-206 4 0.006 0.019 



127585-161 16 -0.004 0.026 



127586-909 5 -0.015 0.010 



502-200 15 0.013 0.056 



503-198 11 0.015 0.031 



Weighted mean 0.009 



that the differences are positive in five of the seven 

 cases, the mean weighted difference being 0.009. The 

 old thermometers thus give slightly higher temperatures 

 than the new ones, as should be expected. The conclusions 

 which were drawn from the results of Dr. BOhnecke's 

 examination of similar thermometers are thus confirmed. 



The final conclusion of this discussion is that most 

 of the thermometers give temperatures which are sys- 

 tematically too high, but that the errors are as a rule 

 smaller than 0.°02 and in no case greater than 0.°035. A 

 single temperature observation may be affected also by 

 accidental errors, which, as a rule are considerably 

 smaller than 0.°02 and in the case of the one-twentieth 

 thermometers never greater than 0.°03 nor more than 

 0.°04 in the case of the one-tenth thermometers. These 

 accidental errors could not have been avoided but the 

 systematic errors could have been reduced had it been 

 possible to re-examine the thermometers after their 

 use. The systematic errors, however, are so small that 

 in most cases they are of no significance. 



This conclusion is verified by an examination of the 

 temperatures at great depth at stations In the Peruvian 

 basin. Stations 68 to 79 are all located in this basin in 



Table 6. Temperature observations below a level of 

 2700 meters in the Peruvian basin 



which the water at great depths appears to be very uni- 

 form, since it is not in direct communication with water 

 in adjacent areas. The observations from this region 

 show that this is true because the same temperature is 

 found at all stations below a depth of 2700 meters. The 

 agreement between the individual observations is good 

 as evident from the compilation in table 6. 



From the data in table 6 we find the following mean 

 values: 



Thermometer nos. 127558 127502 127504 127506 



Mean temperature, °C 1.827 1.828 1.825 1.824 



and none of the individual values deviate as much as 

 0.°02 from these mean values. There is, thus, an excel- 

 lent agreement between the four thermometers in ques- 

 tion and the error of the individual observations appears 

 to be well within the limits which were stated above. 



LITERATURE CITED 



Bohnecke, G. 1927. Veroff. Inst. Meeresk., A, no. 17, 



p. 6. 

 Schumacher, A. 1923. Ann. Hydrog., vol. 51, pp. 273- 



280. 



Soule, F. M. 1933. Hydrog. Rev., vol. 10, pp. 126-130, 



May. 

 W0st, G. 1928. Ztschr. Gesellsch. Erdk., Erganzung- 



sheft 3, pp. 66-83. 



