I912. No. 12. THE SEA WEST OF SPITSBERGEN. 



worthy. Thus it happened in some cases that the thermometer evidently 

 gave very erroneous results, e. g. at Stat. 2, 600 m. (reading — 3.12° C), 

 Stat. 7, 600 m. ( — 2.92'^ C), and several readings at Stat. 32 (readings 

 below — 30 C, or even below — 4° C). The temperatures found at 

 Stat. 37, between 400 and 800 metres, are also wrong; they correspond 

 closely to the temperature of the air, and it seems probable, that the 

 mercury broken off in the air had not run down again and joined the 

 bulk of the mercury when the water-bottle was sent down. The tempe- 

 ratures of Stat. 34, between 200 and 400, are evidently also wrong for 

 some similar reason, though the temperature of the air was then too low 

 to explain it. But generally the observations made by means of this 

 reversing thermometer seem to be correct; it can to some extent be con- 

 trolled as far as the "bottom- water" of the Norwegian Sea is concerned. 

 The temperatures observed at a number of stations are demonstrated in 

 Figs. 2 — 10; a cross indicates that the temperature has been determined 

 by means of the Richter reversing thermometer, while a black dot on the 

 curves indicates the use of an insulated water-bottle with a Nansen-thermo- 

 meter. It is a striking fact that the differences between temperatures taken 

 with the reversing thermometer at the same levels decrease rapidly with 

 increasing depths, and in depths greater than 1200 metres they very 

 nearly coincide, even though the stations may be far apart. 13 determi- 

 nations of temperature have been made at a depth of 1200 metres, and 

 give values between — 0.75'' and - — 1-14''; 9 observations at 1500 metres 

 gave — 0.92*^ C, and — ^1.14° C. as the extreme values. The tables 

 record 7 observations from 2000 metres, all of them ranging between 

 — 1.15° C. and — 1.28" C. — a difference of 0.13° only. This proves 

 that the reversing thermometer has generally worked satisfactorily. 



The zero-correction of therm. No. 60 was found to be -|- 0.047 

 (19th July, 1909), when the thermometer was quite new from the manu- 

 facturer; another determination gave a correction of — 0.004, on the 23rd 

 March, 191 1, when the secular depression of zero had probably stopped, 

 and the correction has probably not been much greater during the expe- 

 dition. The instrumental error of this thermometer has therefore been 

 disregarded in the tables. The corrections due to the secondary tempe- 

 rature-variation of the broken-ofif mercury have been applied in the general 



1 r 1 r 1 / {n -]- T) {T — /) . . „ . 



way, by means 01 the lormula k = , ;/ bemg 79, / the 



0300 



reading of the reversing thermometer, and / the indication of the auxiliary 



thermometer. The reversing thermometer was only used in combination 



with the Ekman water-bottle. 



