tog jeg den 22. August 1878 i Advent Baj, Spidsbergen, 
i 15 Faynes Dyb ved 099 og i Overfladen ved 49.8. 
Den 29. November 1878 prøvedes samtlige Nulpunkter ved 
det meteorologiske Institut i Christiania i smeltende Sne. 
Denne og følgende Dag foretoges Sammenligninger ved 
højere Temperaturer mellem Dybvandsthermometrene og 
Normalthermometret. 
Ved Indexthermometrene er jeg af Erfaring bragt til 
at gjøre følgende Bemerkninger. Ved de ældste Miller- 
Casella-Thermometre (I—IX) staar Skalaen med Grad- 
inddelingen ved Siden af Thermometerrøret. Dette gjør 
Aflæsningen noget vanskelig, naar den yderste Nøjagtig- 
hed søges opnaaet. Gradskalaen fæstet til Ebonit- 
pladen med en Skrue i hver Ende. Man maa passe vel 
paa, at disse Skruer ikke ere løse, da der i dette Til- 
fælde kan opstaa en Usikkerhed ved alle Aflæsninger paa 
flere Tiendedele af en Grad. Ved Forsøg, anstillede paa 
disse Thermometre, medens de vare omgivne af smeltende 
Sne, fandt jeg den 29. November 1879, at der. 
ganske liden Forskjel paa Aflæsningen, enten jeg aflæste 
den frie Kviksølvtop eller den til samme med Magneten 
nedtrukne Index. Denne viste sig at trykke Kviksølvtoppen 
flad, og gav saaledes en højere Temperatur. Correctionen 
beløb sig i Middel for sex Thermometre til + 0003 + 
09011. Ved de Aflæsninger, der ved Sammenligningerne 
med Normalthermometret eller ved Nulpunktbestemmelsen 
ere gjorte paa Kviksølvtoppen, er denne bleven corrigeret 
med + 0903 forat faa den Aflæsning, der svarer til In- 
Det er de saaledes corrigerede Tal, 
var 
jar en 
dexens nedre Ende. 
som ere opførte i de følgende Tabeller. 
En Ulejlighed ved Indexthermometrene er, at Kvik- 
sølvtraaden undertiden splitter sig. Dette kan i Regelen 
let rettes paa ved at svinge Instrumentet i en Slynge med 
Kuglen inderst. Centrifugalkraften driver da Kviksølvet 
sammen paa dets rette Plads. Undertiden sidder Indexen 
saa fast, at en Magnet ikke formaar at flytte den. I 
saadanne Tilfælder har jeg med Held brugt to eller tre 
Magneter, enten en paa hver Side af Røret — som det 
undertiden var nødvendigt at løsne fra Indfatningen — eller 
Værre er det. naar Kviksølvet 
og helt eller delvis indeslutter 
som magnetisk Magazin. 
er kommet forbi Indexen 
denne, thi da formaa Magneterne ikke at drage den ud 
deraf. I saadanne Tilfælder, naar Centrifugalkraftens An- 
vendelse ikke har forslaaet, har jeg med Held brugt den 
Fremgangsmaade at afkjøle Thermometerkuglen, undertiden 
ved Hjelp af fordunstende Æther. Kviksølvet paa 
Minimumsiden, som jeg her altid har for Øje — trænger 
da ind i Kuglen eller den under samme værende Kaliber: 
udvidelse, samler sig i dennes nederste Del med en mindre 
Højde og en større Diameter og lader Indexen saa fri, 
at den kan extraheres med Magnet. Ved den følgende 
13 
the 
as those under which the actual observations were 
made. On the 22nd of August 1878, I made a similar 
experiment in Advent Bay Spitzbergen, at a depth of 
15 fathoms, temperature 09.9, and at the surface of the 
On the 29th of November 1878, 
zero-points of all the thermometers were tested at 
Meteorological Institute in Christiania, by immersing 
On this and the follow- 
ditions 
sea, temperature 4.8. 
the 
the 
ing 
tures between the deep-sea thermometers and the Standard- 
instruments in melting snow. 
day, comparisons were undertaken at higher tempera- 
Thermometer. i 
As to the mdex-thermometers, experience has led me 
to make the following remarks. The oldest Miller-Casella 
instruments (I—IX) have the graded scale at the side of 
the tube. This renders the reading somewhat difficult 
when great accuracy is sought to be attained. The 
scale was made fast to an ebonite plate by means 
of a screw at either end. Care must be taken 
lest these screws get loose; for if they do, an uncer- 
tainty to the extent of several tenths of a degree may 
all When experimenting with 
arise in the readings. 
these thermometers, on the 29th of November 1879, 
whilst they were immersed in melting snow, I found 
a very slight difference in the reading whether 
I read the free upper surface of the mercury or 
the index drawn down to the latter by the magnet. 
The index was found to press the surface of the mer- 
cury flat; and hence it imdicated a higher tempera- 
ture. The correction amounted in the mean for six 
thermometers to + 0°.03 + 0°.011. When reading off the 
upper surface of the mercury — for comparisons with 
the Standard-Thermometer or determining zero-points — 
a correction of + 0°.03 has been applied, in order to 
obtain the reading that corresponds to the lower end of the 
index. The figures thus corrected are those given in the 
following Tables. 
One inconvenience attaching to the index-thermometers, 
is that the thread of mercury will sometimes break. This 
can as a rule be easily rectified by swinging the instru- 
ment in a sling, with the bulb inwards. The centrifugal 
force will then force back the mercury into its right 
position. Occasionally the index gets jammed, so that one 
magnet is quite unable to move it. When such is the 
case, I have successfully made use of two or three mag- 
nets, either one on each side of the tube — which it some- 
times proved necessary to loosen from the frame — or as 
a magnetic magazine. It is a matter of far greater per- 
plexity when the mercury has passed the index and either 
wholly or in part. surrounds the latter; for the magnet is 
then powerless to draw it out. In cases the 
application of centrifugal force being found insufficient — I 
have resorted with success to the following mode of proce- 
dure, viz., that of cooling the bulb of the thermometer, 
such 
occasionally by means of evaporating ether. The mer- 
cury — that on the minimum-side, which is always to be 
understood — pushes through into the bulb or the expansion 
of the bore below it, and collects there in the lowest part with 
