DEEP-SEA REVERSING THERMOMETERS 233 



depths, and over the whole range of ocean temperatures, and it may be that with repeated use some 

 physical change of a purely temporary nature takes place in the structure of the glass ; such a change 

 may affect a particular constriction or appendix (in themselves a tribute to the glassblower's skill) 

 sufficiently to upset the normal action of the thermometer. If, however, a thermometer which behaves 

 erratically in this manner is ' rested ' for some months — in an upright position — it will usually behave 

 normally when returned to service. This recovery of normal accuracy after ' resting ', and without 

 'manipulation' of the thermometer,* seems to be strong evidence for a temporary change in the 

 structure of the glass. 



Calibration and Certification of Deep-sea Reversing Thermometers 

 The original certificates of scale correction for the thirteen German thermometers under review were 

 issued in 1924 and 1925 by the Physikalisch-Technische Reichsanstalt (P.T.R.) of Berlin, who also 

 re-examined some of them in 1933 and 1935. Unfortunately, the N.I.O. does not now possess the 

 original (1924-5) certificates for the three thermometers (nos. 1008, ion and 1257) re-examined by 

 the P.T.R. in 1933, nor for those re-examined in 1935 (nos. 863, 883 and 1019). An attempt was made 

 to obtain copies of these German certificates, but this was not possible since all pre-war thermometric 

 records of the P.T.R. were lost during the bombing of Berlin. f All checks on the German thermo- 

 meters, subsequent to 1935, have been made at the N.P.L. The original calibration of the British 

 thermometers now under review, and all subsequent checks, were also made at the N.P.L. 



The limits of accuracy of the P.T.R. calibrations of the main thermometer, as given on their 

 certificates, were io-oi° C. ; the limits quoted by the N.P.L. were, up to and including the year 1938, 

 also +o-oi° C. ; in 1949 and 1953 they were ±0-02° C. This change was actually made in 1939 but it 

 is of interest to note that in 1929, in connection with tests on a protected deep-sea thermometer (of 

 the range —2° to + 16° C.) made by Richter and Weise, the N.P.L. would only give a certificate of 

 corrections to ±0-02° C. Messrs Negretti and Zambra, in commenting on this,| wrote as follows: 

 ' . . .they (the N.P.L.) report to us that, a number of tests having been carried out, the reproductivity 

 of the readings does not admit giving the corrections closer than -02° C, and we must say that we rather 



agree with them when i ° = 8 mms ' Of the thermometers now under review those of German make, 



with one exception, differ only slightly in range and, apart from this exception, the length of i ° on 

 the scale of the main thermometers of the remainder varies between 8-i and 9-1 mm. The thickness of 

 the graduation line is approximately 0-05 mm. The range of the British thermometers is the same 

 throughout (-2° to 12° C), with the length of one scale degree varying between 9-6 and 11-3 mm. 

 The average is 10-5 mm. The thickness of the graduation line is, approximately, o-i mm. This slightly 

 more open scale is, no doubt, a help towards more accurate readings but still does not warrant correc- 

 tions being determined to a greater degree of accuracy than ±0-02°. That it is possible in field-work to 

 read these thermometers to a greater degree of accuracy cannot be gainsaid. Any experienced observer 

 expects to obtain readings accurate to ±o-oi° C. 



* 'Manipulation' is the comprehensive term used to describe all aspects of the corrective treatment used when trying to 

 restore a temporarily erratic thermometer to its normal action (Whitney, 1952). 



f Personal communication to one of the authors (H.F.P. H.). 



I Letter from Messrs Negretti and Zambra, dated 29 October 1929, to the Director of Research, Discovery Committee. 

 This extract is published by permission of Messrs Negretti and Zambra. 



