Vedkommende afsattes Observationerne paa et særskilt Kart 
for hvert af de ovennævnte Expeditionsaar samt for de 
norske og danske Skibsobservationer, og paa hvert af disse 
Karter droges de til Temperatursystemet hørende Isother- 
mer. For flere Systemers Vedkommende var dette et let 
Arbejde, f. Ex. for vor Nordhavsexpedition. For andre 
Systemer, hvor Observationerne vare mindre regelmæssig 
fordelte, maatte disse først samles i Grupper, og Middel- 
tallet af disse beregnes og indføres paa den Plads, der ind- 
toges af Gruppens Tyngdepunkt. Dernæst construerede jeg 
for hver Breddegrad, fra den 80. til den 55., efter de 
nævnte Isothermer, Curver, der viste Temperaturens Stør- 
relse langs hver Breddegrad i hvert Observations- eller 
Aars-System. Efter Antallet af disse fik jeg saaledes paa 
de Partier, hvor Systemerne overskar vedkommende Bred- 
deparallel, Curver eller Stumper af Curver for Havover- 
fladens Temperatur til forskjellige Aar eller Rækker af 
Aar. Af de forskjellige Curvers Ordinater toges Middeltal, 
og gjennem de til dem svarende. Punkter droges en Me- 
diumseurve, der, saavidt det benyttede Materiale tillod, frem- 
stiller Temperaturens normale Stand langs vedkommende 
Breddeparallel å Midten af August. Beregningen af Mid- 
deltallene, der kunde uddrages af fra 2 indtil 7 Enkeltveer- 
dier, har ogsaa kunnet tjene til at finde den gjennemsnit- 
lige Afvigelse af et enkelt Aars Observation fra det nor- 
male Medium og Maximumafvigelsen. Paa de fundne nor- 
male Curver afsattes deres Skjæringspunkter med de hori- 
zontale Linier for hver hel Grad, og disse Punkter over- 
fortes paa et Kart paa de tilsvarende Længder. Gjennem 
de saaledes fundne Snitpunkter droges derpaa Havfladens 
Isothermer. 
For Østhavets og Danmarkstrædets Vedkommende 
bleve Observationerne samtlige afsatte i samme Kart, og 
inddelte i Grupper, hvis Middeltal indtegnedes i Gruppens 
Tyngdepunkt, hvorpaa Isothermerne droges efter disse Me- 
dia. Ved Beregningen af Gruppemiddeltallene er der, for det 
nordlige Østhavs Vedkommende — Øst-Spidsbergen-Nord- 
Novaja-Semlja —, taget Hensyn til, at Observationerne 
falde i gunstige Is-Aar. For hvert saadant er i Middel- 
tallets Beregning indtaget som Compensation en Temperatur 
af —19. Herved tror jeg at være kommen det normale 
Temperaturmedium nærmere. 
Resultatet af disse Constructioner og Beregninger fo- 
religger 1 Kartet Pl. XX VII. Ifølge 54 forskjellige Be- 
stemmelser, der variere mellem 0°.4 og 195, er i Gjennem- 
snit den midlere Afvigelse af en enkel (et enkelt Aars) 
Temperaturobservation paa et Sted fra den normale + 09.90, 
og man tør saaledes, da det gjennemsnitlige Antal af Ob- 
servationer for hvert Medium er 5.6, sætte Usikkerheden 
af Kartets Temperaturangivelse for et enkelt Sted til +00.38. 
De forskjellige Breddegrader udvise ikke nogen merkelig 
Forskjel i denne Henseende. Den midlere Maximumsaf- 
100 
the Norwegian Sea, the observations were put down on 
separate maps, one for each year of the above-mentioned 
Expeditions, as also for the Norwegian and Danish vessels, 
and on each of these maps were drawn the isotherms re- 
sulting from the system of observations. With several of 
the systems, this was an easy matter, e.g., with that from 
the North-Atlantic Expedition. With other of the systems, 
in which the observations occurred less regularly distributed, 
these had first to be collected into groups, and the mean 
computed and placed where the group was found to have 
its centre of gravity. I then constructed for each parallel 
of latitude, from the 80th to the 55th, by means of the 
fore-mentioned isotherms, a corresponding number of curves, 
exhibiting the course of the temperature along each parallel 
of latitude in every observing or yearly system. According 
to the number of such, I obtained, therefore, in the tracts 
where the systems cut the respective parallel of latitude, 
curves or fragments of curves for the temperature of the 
Of the 
various curve-ordinates was taken the mean, and a mean 
sea-surface in different years or series of years. 
curve was then drawn through the points corresponding 
to the said mean numbers. This curve represented, so far as 
the material made use of would admit, the normal distri- 
bution of temperature along the respective parallel of latitude 
in the middle of August. The computed mean values, based 
on from 2 to 7 single determinations, have likewise served to 
find the average deviation of a single year’s observation 
from the normal mean and the maximum deviation. On the 
normal curves thus found, the points of intersection with 
the horizontal lines for every whole degree, were marked off 
and these points transferred to a map along the cor- 
responding degrees of longitude. Through the points of 
intersection thus found, the isotherms of the sea-surface 
were then drawn. 
As regards [the Barents Sea and Denmark Strait, 
all the observations were put down on one and the same 
map, and arranged in groups, the means for which being 
placed in the centre of gravity of the group, after which 
When com- 
puting the mean number for each group, regard had to be 
taken, concerns the northern tract of the Barents 
Sea, viz., Hast Spitzbergen-North Novaja Semlja, that the 
observations are derived from favorable ice-years. For 
every such observation, a temperature of —1° has been 
introduced as compensatory in calculating the mean. Thus 
I make venture to assume I have reached a closer approx- 
imation to the normal mean temperature. 
The result of these constructions and computations 
is given in the Map, Pl. XXVII, According to 54 dif- 
ferent determinations, varying between 0°.4 and 1°.5, the 
mean deviation of a single (a single year’s) observation of 
temperature from the normal average in any given place is 
+ 00.90; and therefore, since the average number of observa- 
the isotherms were drawn from these means. 
as 
tions for each mean is 5.6, we may put the uncertainty of 
the Map’s indication of temperature for any given place at 
+ 0038. The various parallels of latitude do not exhibit 
any appreciable difference in that respect. The mean max- 
