Station. | Dybde. Therm. Therm. | Obs. Ber. Diff. 
| (Dopth.) | (Comp.) 
No. Hayne L N&Z | Cor. Corr, O-B. 
| (Hathoms.) | | 
| | | 
277 | 225 | iO 0.92 0".18 | — 09.04 — 09.14 
204 | 637 — 1.09 — 123 | = ©oiO | 2 os 4 
206 | Rane — 109 | — å gt — 0.22 27) = 5 
303 | 1200 — 1 .38 = i Ou — 0.23 | 22 — I 
311 898 — 1.00 — 1.31 = 0) .22) | .16 — 6 
324 233 0.090 o .89 — 0.01 | .04 + 3 
332 | AG — 1.38 == if Ali = © 3 | 21 | + 18 
342 | 523 — 0.89 — 1.05 — 0.16 oe 6 
343 743 —= 0:09 | — % 25 — 0.26 | 13 UG 
344 1017 — 1.00 || — 1.25 — 0.16 | .19 al 3 
354 1343 = 169 | = 1.25 | = 0.16 | 25 | + 9 
355 948 Se REN) i) pasa hoa =O oD 7 | SL 5 
359 416 0.85 | 0.79 — 0..06 08 | + 2 
360 42 I 0.10 — 10.08 Lise 0.13 | .08 | — 5 
367 535 NEO NEO NE Oro JO (| Oo 
368 Bris | mt oi I .61 — © 510 .06 — 4 
12043 — 20.30 Wir == op od1 
Kaldes Correctionen for Trykket ce, Dybden i Favne 
h, sætter jeg 
BEN In 
og bestemmer den sandsynligste Verdi af Coefticienten > 
af Ligningen 
Ne 
OS SJ AMES PSI (0 = 
NJ 
SS Ila 
For Thermometer No. I bliver altsaa 
2.30 
== —= = —0.000191 og 6 = —o0.000191 h. 
12043 
For 1000 Favne ¢ = —o?.191 
» 2000 så Cc = —0.382 
Man ser, at de enkelte observerede Correctionsværdier 
samtlige ere negative, saaledes som Forudsætningen er. 
I den sjette Rubrik opført de efter Formelen 
beregnede Trykcorrectioner, og 1 den sidste Rubrik For- 
Af 
Summen af disse, uden Hensyn paa Fortegn, divideret med 
Observationernes Antal (16) findes den gjennemsnitlige 
Afvigelse for en enkelt Bestemmelse lig + 09.061. 
indeslutter Fejlene i Indexthermometrets Aflæsning, i dets 
Nulpunktbestemmelse, i Skala-Correctionen, 1 Trykcorrec- 
tionen samt i Vendethermometrets Atflæsning og 1 dets Ska- 
(Ae 
skjellen mellem de observerede og beregnede Værdier. 
Denne 
lacorrection. Settes disse sidste tilsammen til + 09.05, 
bliver den midlere Fejl af en enkelt fuldstændig reduceret 
Observation med Miller-Casella No. I kun + 0903. Da 
imidlertid Vendethermometrene aflæses med samme Nøj- 
agtighed som Miller-Casella’s Indexthermometre, fremgaar 
som Resultat, at de forskjellige Correctioner ikke indføre 
nogen merkelig Usikkerhed i de reducerede Verdier for 
Temperaturen, og at Usikkerheden ene ligger i Aflæs- 
ningen. Dennes sandsynlige Fejl kan sættes til omkring 
+ 00.05. 
Now, assuming the correction for pressure to be termed 
ec, the depth in fathoms h, I put 
6 = w In 
and take the most probable value of the coefficient 3 from , 
the equation 
Ne 
NES Behi=awe ahi é= — 
NY 
SI 
Hence, for Thermometer No. I, 
Å 2.30 
fo = — —*~—~ = —o.000191 and ¢ = —0.000101 h. 
12043 
For 1000 fathoms ¢ = —o".191 
2000 å ¢ = —0.382 
We perceive that each observed value of correction is 
invariably negative, as theory assumes. 
In the sixth column are given the pressure-corrections 
computed from the formula, and in the last column, the 
difference between the observed and computed values. From 
the sum of these values, without reference to signs, divided 
by the number of observations (16), is found the average 
deviation of a single determination, which equals + 0°.061. 
This result comprises the errors in the reading of the 
index-thermometer when determining the zero-point for 
that instrument, those in the scale-correction, the pressure- 
correction, as also those in the reading of the inverting- 
thermometer and in its scale-correction. Now, if we take 
the last of these together at + 00.05, the mean error of a 
single fully reduced observation with Miller-Casella No. I 
will amount to only + 0°.03. Meanwhile, as the inverting- 
thermometers are read off with the same accuracy as the 
Miller-Casella index-thermometers, the result must be, that 
the various corrections do not occasion any sensible un- 
certainty in the reduced values for temperature, but that 
the uncertainty lies exclusively in the reading. The prob- 
able error of the latter may be put at about + 09.05. 
