I Polarstrømmen aftager Temperaturen langtsomst 
med Dybet. Her skulde Vendethermometrene, der tjene 
som Normaler for Bestemmelsen af Trykcorrectionerne, 
være bedst accommoderede, det er forholdsvis mest afkjø- 
lede, vise forholdsvis lavere, og følgelig Indexthermometre- 
nes Trykcorrection vise sig større. Som man ser, er Tryk- 
correctionen i Polarstrømmen fundet numerisk mindre end 
i den varme Strøm. Man tør altsaa, som ovenfor Side 15 
bemerket, antage, at Vendethermometrene have accommo- 
deret sig fuldstændig efter Bundtemperaturen uagtet Kas- 
sernes tabte Flydeevne. 
Den Nøjagtighed, med hvilken de for Skalafejl og 
Tryk reducerede Thermometeraflæsninger angive den rig- 
tige Temperatur 1 Dybet, kan man faa en Forestilling om 
efter Størrelsen af Middelfejlen for Trykeorrectionens Be- 
stemmelse ved de forskjellige Indexthermometre, da denne 
Bestemmelse indeslutter Skalacorrectionens Fejl. For Vende- 
thermometrenes Vedkommende haves 
Middelfejlen af Sammenligningerne 
tret. Opstilles de nævnte Værdier 
enkelt Temperaturbestemmelse, faar 
det. samme Maal i 
med Normalthermome- 
af Middelfejlen ved en 
man følgende Oversigt. 
38 
In the Polar current, the temperature diminishes 
most slowly with the depth. Here the inverting-thermometers 
serving as standards to determine the pressure-corrections, 
should prove best accomodated, i.e. comparatively most 
cooled, viz. read relatively lower, and hence the pressure- 
correction of the index-thermometers be greater. As will 
be seen, the pressure-correction in the Polar current was 
found numerically less than in the warm current. We may 
therefore assume, as previously remarked, p. 15, that the 
inverting-thermometers had become thoroughly accomodated 
in the bottom-temperature, notwithstanding the lost buoy- 
ancy of the frame. 
The accuracy with which the thermometer-readings 
reduced for errors of scale and pressure, indicate the exact 
temperature in the deep strata, can be found by the magnitude 
of the mean error of the pressure-correction for the diffe- 
rent index-thermometers, since that correction contains 
within it the error in the scale-correction. As regards the 
inverting-thermometers, we have the same standard in the 
mean error of the comparisons with the Standard-Thermometer. 
Now, if we put down the said values of the mean error 
of a single determination of temperature, the result will be 
as follows. 
Thermometer. MF 
Miller-Casella I. + 0°.061 
; Tale O .072 
Å II. O .139 
- IV. 0.020 
Å e We 0 .072 
i VIL 0 .064 
VII. 0 .050 
5 - ID, O .035 
Buch. Ky. (Merc.) Piezom. 0 .067 
Casella-Buch. 46. Q .054 
. _ 48. OP Ork 
g s 49. | 0 .070 
Negr. & Zamb. 89. 0.026 
Å 2 Sil O .047 
Middel (Mean) + 0°.061 
Middel for Indexthermometrene er 
5; »  Vendethermometrene er 
+ 00.064. 
+ 0.037: 
Da den sidste Værdi mdgaar i Bestemmelsen af den 
første bliver Middelfejlen for en 
med et Indexthermometer 
Temperaturbestemmelse 
2) 
+ V 0.0642 — © .0372 
Ogsaa paa folgende Maade kan 
peraturbestemmelsernes Nøjagtighed. I 1878 sendtes ofte 
3 Dybvandsthermometre, 2 Index- og et Vendethermometer 
(i Regelen No. 89) med Loddet til Bunds. Tager man 
Middel af disse 3 Thermometres for Skalafejl og Tryk cor- 
rigerede Angivelser, og beregner Afvigelsen mellem de en- 
+ 00.052. 
man bedømme Tem- 
Mean for the index-thermometers + 07.064. 
> i inverting-thermometers + 0 .037. 
The last of these values being contained in the deter- 
mination of the first, the mean error of a determination 
of temperature with an index-thermometer will be 
+ V 0.0642 — 0.037? = + 00.052. 
Also in another manner can the accuracy of the 
temperature-determinations be estimated. In 1878, 3 deep- 
sea thermometers, 2 index and 1 inverting-thermometer (as 
a rule No. 89), were sent to the bottom with the lead. 
Now, if we take the mean of the indications of these 3 
thermometers, corrected for errors of scale and pressure, 
6 
