240 DISCOVERY REPORTS 



being in collision, or suffering shock of any other kind, nor have losses on station, or accidental 

 breakages, been included. When a thermometer suspected of giving anomalous readings was checked 

 several times against a reliable thermometer only one ' failure ' was recorded for the whole series of 

 comparisons. 



In view of the difficulty of making reversing thermometers, and the unavoidably rough usage which 

 they receive at times in bad weather at sea we think it is remarkable that the overall percentage of 

 failure — just over 2% — is so low. Ignoring the continued use by the 'William Scoresby', in 1926-7, 

 of one erratic thermometer, the percentage failure of German-made thermometers for the period 

 1925-51 averaged 2-25 and, in general the performance of these instruments was remarkably steady. 

 The earliest figure of 3-5% failures, in 1925-7, was on the first occasion of their use, and before we 

 had experience in spotting idiosyncrasies. Later, in 1931 in the 'WiUiam Scoresby', and in 1935-7 

 in the 'Discovery 11', almost the same set of reliable German thermometers was in use throughout 

 each period, with a consequent and marked reduction in the number of failures. For the 'William 

 Scoresby' the percentage of failure was 1-38 for 941 observations, and for 'Discovery 11' 1-54 for 

 2921 observations. 



Except for the commissions of 1933-5 ^nd 1935-7 of 'Discovery II ', when the number of failures 

 was increased considerably by the use of one rather erratic batch of thermometers (of which three, 

 nos. CD 5556, 5559 and 5563 remain in use), the British-made deep-sea reversing thermometers have 

 been equally, if not more, satisfactory. They have maintained their accuracy well and for all other 

 commissions of 'Discovery 11' they were extensively used (for 71 12 temperature observations out of 

 a total of 8142), with a percentage of failure of only 1-07. Even if the figures for the 'William 

 Scoresby's' commission of 1950 are included, this percentage failure is increased only to 1-17. 



UNPROTECTED DEEP-SEA REVERSING THERMOMETERS 

 Earlier in this paper it was mentioned that only two pre-war unprotected reversing thermometers 

 had survived for comparison with each other. Actually, four such instruments remain but of these 

 the history of two is obscure. They do not appear to have been used, nor to have been re-calibrated 

 more than once, and their characteristics have not therefore been considered. In addition, fifteen 

 post-war unprotected thermometers remain from a batch made in 1949; some of these were fairly 

 extensively used in 'Discovery 11' in 1950-1 and all have since been re-calibrated once. The scale 

 limits of both pre- and post-war instruments is —2° to +60° C. 



Zero Point 

 One of the pre-war thermometers (no. CC 14318) was calibrated by the N.P.L. in 1933, 1950 and 

 1954. A normal rise of zero point ( + 0-04° C.) was observed over this period but there is little re- 

 semblance between the trend of the scale corrections in 1933, and those of later years, though closer 

 agreement exists between these last determinations. The scale corrections for thermometer no. 

 CG 15290 (determined in 1937, 1949 and 1954) show no agreement whatsoever between the various 

 calibrations. A correction graph for each thermometer comprising the various scale corrections at each 

 calibration has been drawn. These appear as Figs. 6 and 7. On the other hand, the scale corrections 

 for the fifteen post-war instruments show a remarkable similarity in their trend at their re-calibration 

 in 1954. Only in three instances has there been no shift of zero point since the original check in 1950; 

 for the remainder there is the normal rise, which is reflected equally throughout the scale. 



