332 DISCOVERY REPORTS 



in use, had always been kept in an upright position, over periods varying between fifteen and twenty- 

 nine years. The records of these thermometers over such exceptionally long periods are probably 

 unique. In this paper, therefore, after a review of the errors and eccentricities to which deep-sea 

 thermometers are liable, we can examine the general question whether or not they change or deterio- 

 rate with age. We hope to show that most thermometers, at least of certain British and German 

 makes, can continue to give good service for very long periods, provided they are properly cared for; 

 and that there is little change with age apart from a tendency for the zero point to rise. 



All the thirty-eight thermometers had previously been re-certified at least once, after the issue of the 

 original certificate of test. Sixteen of the thirty-six protected thermometers were made in Germany 

 prior to 1926 (by Richter and Weise, of Berlin); the remainder were made at various times between 

 1934 and 1938 by Negretti and Zambra, of London. The scale range varies considerably, but all are 

 divided to 1/10° C. In 1953-4 certificates were issued in respect of thirty-three protected thermo- 

 meters only, as the mercury column of each of the remaining three (of German origin) consistently 

 refused to break at the constriction or appendix when the thermometer was reversed unless the 

 thermometer was tapped. The action of these thermometers had, however, been satisfactory when 

 they were previously examined by the N.P.L. in 1949. It is unfortunate that only two unprotected 

 thermometers of any considerable age have survived for comparison but, in view of their extreme 

 vulnerability, this is not, perhaps, surprising. 



Errors and Eccentricities 



Although the zero point of a reversing thermometer* is dependent on the 'break-oflF' action of 

 the mercury column at the constriction or appendix peculiar to this type of thermometer, there is 

 a characteristic rise (of the order of a few hundredths of a degree) of zero with age very similar to that 

 which commonly takes place in ordinary mercury-in-glass thermometers ; except for this rise of zero 

 the majority of deep-sea thermometers appear to maintain their accuracy remarkably well for many 

 years — whether in store or frequent use — provided that they are properly cared for. That a few become 

 erratic in their habits after repeated use, is not surprising; for instance, those used for observations at 

 the greater depths may, if used in tropical waters, be cooled rapidly from 28° to 1° C. between the 

 surface and a depth of 5000 m., and as quickly heated again when returning to the surface. 



Erratic readings from deep-sea thermometers usually result from only a few basic causes, the 

 commonest of which is the presence of a minute bubble of gas in the column of mercury, somewhere 

 in the capillary of the thermometer. On reversal the mercury column breaks at the point where the 

 bubble is then situated. Rough handling of a thermometer is usually the reason for the occurrence of 

 this condition which, nevertheless can occur in normal use even when the greatest care is taken to 

 protect the instrument from shock. 



One eccentricity, however, for which it is difficult to find a satisfactory explanation is the failure 

 of the column of mercury to break at all, unless the thermometer is tapped. With very low tempera- 

 tures this is perhaps understandable, since the amount of mercury concerned is small and, if reversal 

 is unusually slow, the weight of mercury involved might not be sufficient to overcome the surface 

 tension and bring the constriction into action. (We revert to this point below, p. 238, when con- 

 sidering the histories of the various thermometers.) There is no evidence, however, that the amount 

 of mercury reversed determines whether the thread breaks or not. This eccentricity can occur at all 



* Deep-sea reversing thermometers comprise two thermometers — main and auxiUary. Since the latter is a conventional, 

 or standard type of thermometer it has not been considered in this paper, and any reference to a reversing thermometer 

 should be taken as referring to the main thermometer only. 



