a memory of thermal history to some degree but 

 these special thermometer glasses were chosen to 

 reduce this to a minimum and provide reasonable 

 stability even at high temperatures. It so hap- 

 pens that the range of temperature in the ocean 

 is relatively short, a matter of only 35 C. In 

 this short range thermal hysteresis is not appre- 

 ciable; at least I was not able to detect any 

 significant short term changes by repeatedly 

 cycling a group of good instruments during one 

 long expedition. 



I point this out in passing because it appears 

 to me that the primitive bulb-and-stem trans- 

 ducer still has certain unexploited merit for 

 measuring relatively low temperatures as well as 

 moderately high pressures. This structure appears 

 to be one preferable to some that are employed 

 in contemporary pressure transducers and it would 

 appear to be no more difficult to link it with 

 telemeters and recorders suitable for many types 

 of work. 



Limitations 



Limitations to the reversing thermometers are 

 well known. I will mention only a few. 



The CHALLENGER report emphasized the value of 

 making "serial" or multiple measurements on each 

 sounding cast. This is still true but we now do 

 not consider 8 or 10 transducers per cast suffi- 

 cient coverage for all purposes. 



Response is too slow for much of our present 

 requirements. It can be shown' that the pro- 

 tected instrument has a compound thermal lag 

 described adequately by two lag constants; one 

 of about 18 seconds and the other of about 120 

 seconds . This is a slow response even for mer- 

 curials . Speed has been compromised so as to 

 provide pressure "protection. " 



The pressure effect on "protected" instru- 

 ments has not been eliminated entirely. In 

 some modern instruments the outer bulbs are over- 

 filled with mercury and analyses of the mechani- 

 cal structure" seems to indicate that large 

 enough pressure errors would enter to be signifi- 

 cant in measuring deep water gradients. It would 

 appear that these errors might approach 0.01°C 

 per 1,000 meters change in depth. They are 

 largest in deeper water which, unfortunately, is 

 where small intervals of temperature are most 

 likely to be of interest. 



Adiabatic Errors 



A most peculiar thermometric error comes to 

 light upon inspection of the CHALLENGER'S reports, 

 one that was not at all understood until much 

 later when Professor Tait calibrated the ther- 

 mometers under pressure in his laboratory. Many 

 materials heat faster than water does when put 

 under pressure adiabatically. Hard rubber that 



was used for supporting the gLassware in the CHAL- 

 LENGER'S thermometers happens to be especially 

 sensitive to pressure in this way. This led to 

 real ambiguity in the CHALLENGER'S temperature 

 readings . Reversing and minimum-registering 

 instruments seldom agreed as would be expected 

 from the considerations below. 



The primitive reversing thermometers used on 

 these first casts had to be "tripped" by hauling 

 them up rapidly enough to make a propeller revolve. 

 Because of this requirement, and because of adia- 

 batic effects, these reversing instruments proba- 

 bly were slightly hotter than they should have 

 been. The masses of hard rubber required some 

 time to cool into equilibrium with the sea, but 

 the CHALLENGER people were not aware of this and 

 so probably hauled up too soon. 



On the other hand, the minimum-registering 

 thermometers always read the lowest temperature 

 they encountered which, on a very deep cast, 

 was not likely to be at the bottom. In fact, 

 their readings would depend a great deal upon 

 just how fast the cable was hauled in because 

 the hard rubber parts continued to cool below 

 ambient temperature as the pressure was relieved. 

 Therefore, adiabatic errors in minimum-thermometers 

 had opposite sign from those that pertained to 

 the reversing thermometers thus adding to the 

 discrepancy between instruments . 



As a result of these inaccuracies the CHAL- 

 LENGER was unable to report the important finding 

 that there was a temperature minimum at about 

 lj-OO fathoms over most of the world ocean based 

 only upon the general trend of averages of a 

 large number of deep readings . 



Why Were Other Thermometric Instruments Not 

 Perfected ? 



The question now can be raised as to why 

 oceanographers have very nearly perfected this 

 particular measuring system without devoting 

 attention to other equally promising devices . 

 This is especially interesting because there was 

 placed aboard the CH AL L E NGER, as a gift of 

 Sir William Siemens, the brother of the founder 

 of the German Siemens-Haske firm, a complete 

 electrical thermometric system which is recog- 

 nizable as very suitable. 



Only the briefest of tests of Sir Siemens ' gift 

 were carried out on the CHALLENGER even though 

 it apparently was capable of very fine measure- 

 ments. Its design was sensible and conservative. 

 It employed a simple bridge circuit with nul 

 indicated by a marine galvanometer of the type 

 invented for cable laying by Sir William Thompson 

 some twenty years earlier. The measuring trans- 

 ducer, a coil of metal wire provided with three 

 leads to compensate for cable resistance, seems 

 to have been in every way as reliable (if not as 

 convenient) as any we would choose today. Elec- 

 trical balance was attained by a method truly 



