SUBSURFACE TEMPERATURES 



The surface temperature was recorded continuously 

 by means of a sea-water thermograph. The instrument 

 and its operation are described in the volume dealing 

 with the meteorological data, and tables showing hourly 

 values of sea-surface temperatures are given in that 

 volume and in table 1 of Oceanography I-B. In the fol- 

 lowing, therefore, we are concerned with the subsurface 

 temperatures only. 



The subsurface temperatures were determined by 

 means of protected reversing thermometers of the well- 

 known pattern manufactured by Richter and Wiese.^ All 

 thermometers had been examined at the Physikalische 

 Technische Reichsanstalt (PTR) and will be referred to 

 by the PTR numbers. The corrections determined by the 

 Reichsanstalt will be designated the "PTR corrections." 

 Table 1 gives the range for each thermometer, the char- 

 acter of graduation, the date of the PTR certificate, and 

 the numbers of the stations at which used. 



From the footnotes in table 1 it is seen that a num- 

 ber of the thermometers were lost because of accidental 

 breaking of cable during the occupation of several sta- 

 tions. The remaining thermometers were all lost when 

 the Carnegie was destroyed. No determinations of the 

 corrections of the thermometers were undertaken at sea 

 and, since all thermometers were lost, a re-examination 

 is impossible. A large number of protected thermome- 

 ters were used in pairs, however, and from the differ- 

 ences between the corrected readings of two such ther- 

 mometers it is possible to arrive at several conclusions 

 as to the accuracy of the observed temperatures, assum- 

 ing the PTR corrections to have remained unchanged. 



Before entering on an examination of these differ- 

 ences, the possible errors of the temperature observa- 

 tions will be briefly discussed. Some of the thermometers 

 were divided to one-twentieth degree and others to one- 

 tenth. The errors of these two classes of thermometers, 

 which for sake of brevity will be referred to as the one- 

 twentieth and the one-tenth thermometers, will be treat- 

 ed separately. The following sources of error then have 

 to be considered: (1) errors of reading; (2) correction 

 errors arising from (a) reduction errors, (b) limit of ac- 

 curacy of the test, and (c) change of zero point; and (3) 

 errors of breaking-off device. 



(1) Errors of reading . All thermometers were read 

 to 0.°01 and reading was always made by means of a spe- 

 cial reading lens. The accuracy of the reading, there- 

 fore, can safely be assumed to lie within the limits 

 + 0.°01. Bohnecke (1927)statesregardingtheone-twentieth 

 thermometers that the errors of reading for such ther- 

 mometers when read to 0.°001 never exceed 0.°005 and as 

 a rule were smaller than 0f003 according to the experi- 

 ence at the Reichsanstalt. 



(2a) Correction errors arising from reduction er - 

 rors . A correction, as is well known, must be applied to 

 the reversing-thermometer reading, since as a rule it is 

 read at a temperature differing from the temperature at 

 which the column of mercury broke off. The exact for- 

 mula for this correction is 



(T + Vq) (I - t}/6100 (1) 



^For detailed description see Wissensch. Ergebn. d. 

 Deut. Atlantischen Exped. auf dem Forschungs- und 

 Vermessungsschiff Meteor 1925-27, vol. 4, pt. 1. (1932). 



where T is the temperature at which the thermometer 

 was reversed, Vq is the volume of the mercury at zero 

 degree, _t is the temperature at which the thermometer 

 was read, and 6100 is a constant depending on the quality 

 of the glass. The temperature at which the thermometer 

 was reversed, however, is unknown and in the first ap- 

 proximation this temperature, T, may be replaced by the 

 reading of the thermometer T'. As a second approxima- 

 tion, T' may be replaced by (T" + dT'), where dT" is 

 equal to the correction which is computed by means of 

 formula (1), using T' instead of T. The final formula for 

 the second approximation to the correction will thus be 



[(T'+vo) (T'-t)/6100][l + (T'+vo) + (T'-t)/6100] (2) 



This formula has been derived by Schumacher (1923) and 

 represents an improvement of formula (1) commonly 

 used. He shows that in extreme cases it may be neces- 

 sary to apply still another approximation in order to re- 

 duce the reduction error beyond the values of the errors 

 of reading, but in the case of the Carnegie observations 

 the errors in the correction, K, as computed by means of 

 formula (2) never exceeded 0.°002 and therefore may be 

 disregarded. A practical method of determining the cor- 

 rection has been described by Soule (1933). 



(2b) Correction errors arising from limit of accu - 

 racy of the test. The corrections of the thermometers 

 which were communicated by the PTR and which must be 

 applied in addition to the reduction correction, K, have 

 been rounded to OTOl. The corrections may be regarded 

 as exact within 0.°005 at the time when the thermometers 

 were tested; however, the corrections are likely to 

 change with time and, according to the experience of the 

 Meteor expedition, this change has the character of a 

 parallel displacement of the correction curve supposing 

 the breaking-off device always to function properly WOst 

 (1928). The parallel displacement of the correction 

 curve may be attributed to a change of the zero point of 

 the thermometer. 



(2c) Correction errors arising from change of zero 

 point . A change of the zero point of the thermometer 

 takes place as a rule some time after the manufacture of 

 the thermometer and in most cases may be ascribed to a 

 contraction of the bulb which causes a rise of the zero 

 point and thus a decrease in the correction which has to 

 be applied at 0°. The contraction of the bulb is hastened 

 by artificial aging of the thermometers but the process 

 usually continues for a long time afterward at a slower 

 and slower rate. During the Meteor expedition Bohnecke 

 examined the zero points of the greater number of the 

 thermometers of the expedition at intervals of about two 

 months. From this examination it appears that the zero 

 point as a rule rose during the first two to six months 

 after the manufacture and that no appreciable changes 

 took place later. In several instances a lowering of the 

 zero point occurred before the subsequent rise, this type 

 of change being characteristic of instruments of very re- 

 cent manufacture. In a few instances the variations were 

 irregular evidently because of bad functioning of the 

 break-off device. These thermometers were easily rec- 

 ognized when used together with a perfect thermometer 

 because the differences in the indications would vary ir- 

 regularly within considerable limits. Only in two cases 

 were great variations of the zero point observed (0.°6 



