242 DISCOVERY REPORTS 



eighteen thermometers of 0-0004. After the sixth commission of the ' Discovery II ' (1950-1) nineteen 

 unprotected thermometers — all of British make — remained available for use. Four of these were of 

 pre-war origin; the remainder were made in 1949. Seventeen were re-calibrated in 1954 and, for 

 reasons which will be explained later, a more detailed check of the m.p.c. has been made in respect of 

 four of these. So that comparison can easily be made with the ' Meteor ' thermometers, the particulars 

 relating to the 'Discovery IPs' thermometers have been similarly arranged in Table 5, but without 

 the full particulars of the detailed check. 



With one exception the figures in Table 5 show a marked stability in the m.p.c. Not only is there 

 close agreement between the values at different points on the range of individual thermometers but, 

 as in the German-made thermometers of the ' Meteor ' Expedition, little change has occurred over 

 periods varying between four and seventeen years. These figures, in fact, indicate that determination 

 of the m.p.c. from observations at atmospheric pressure and at two elevated pressures is probably 

 sufficient, and that an average of the values thus obtained is sufficiently accurate for all depth calcu- 

 lations in dynamical oceanography. It is interesting to note that certificates issued by the P.T.R., 

 between 1925 and 1931, in respect of unprotected thermometers made by Richter and Wiese, and 

 held by the Discovery Committee, gave only one value for the m.p.c. to cover the whole scale range. 

 In view of the information available from Wiist's critical examination (1932) it is virtually certain 

 that this value for the m.p.c. was an average of two or more determinations at different points on 

 the scale. 



As we have already shown (Table 4) the lowest range on the scale over which the m.p.c. for the 

 N.I.O. thermometers was determined was 0° to about 13° C, for determinations at three elevated 

 pressures. For 2-point determination the lowest range was 0° to about 24°. Recently, one fishery 

 research laboratory has found it necessary to use unprotected thermometers at much shallower depths 

 than hitherto considered necessary. All thermometers of this type then held by them had been certi- 

 fied by the N.P.L. for m.p.c. at the usual three points, and they inquired whether it would be correct 

 to assume that, if these values were plotted, a value for the m.p.c. at a lower point on the scale could 

 be obtained by extrapolation. At the time of this query all the unprotected thermometers listed in 

 Table 5 were with the N.P.L. for re-certification. It was arranged that some of them, instead of 

 being checked only at the now-customary two points, should be checked over the whole range. Four 

 thermometers, nos. CC 14318, DI 18225-6 and DI 18232 were chosen and observations made at 

 seven elevated pressures. The estimated accuracy of the measurement of increase of thermometer 

 reading after the application of pressure was ±0-05° C. This uncertainty affects the computed value 

 of the m.p.c. by an amount which varies with the pressure range used. For example, at a pressure of 

 34-80 kg./cm.2, the estimated accuracy of the m.p.c. was ±0-0015 while at 536-41 kg./cm.^ the value 

 was ±0-0001. From the figures obtained it seemed at first sight that extrapolation of the m.p.c. for 

 scale points below the normal test points would be difficult but, when the increase of the thermometer 

 reading was examined arithmetically, and graphically, it was found that this increase was linear within 

 the limits imposed by the uncertainty of the actual measurements. Taking into consideration the other 

 errors which may occur in field-work, it is unlikely therefore that any additional small error due to 

 taking a mean m.p.c. over the whole pressure range will have a significant effect on the computation 

 of depth, the limit of accuracy of which is approximately ± i m. for each 1000 m. of depth. 



