Differentserne h'”—h” ere saagodtsom identiske med 
Forskjellerne mellem Lodskuddene og de af de sandsynligste 
Verdier for & og x beregnede Dybder. En constant Skraa- 
hed af Lodlinen fører saaledes til en constant Correction 
(praktisk talt) af Glassets Sammentrykkelighedscoefficient, 
som den gjør mindre end naar Linen var lodret. 
Den fundne sandsynligste Værdi for x, 0.796 Million- 
tedele, er 1 sig selv allerede paafaldende liden, sammen- 
lignet med de ellers ved Sammentrykningsundersøgelser 
fundne og benyttede Værdier (1.5 til 3). En yderligere 
Formindskelse af Værdien for x har derfor ikke Sandsyn- 
ligheden for sig, og man føres til den Slutning, at Lod- 
skuddene ere tagne uden merkelig systematiske Fejl hidrø- 
rende fra Skraahed af Linen, og saaledes angive Dybderne 
med en Nojagtighed, der væsentlig er begrændset alene ved 
tilfældige Observationsfejl. 
Disse Observationsfejls Størrelse kunne vi gjøre os 
Rede for paa følgende Maade. 
Den sandsynlige Fejl af en enkelt Aflæsning af Pie- 
zometrets Stand eller Indexens Stilling, 
+0.075 mm = dm’» Den sandsynlige Fejl af en Bestem- 
melse af Piezometerstanden m ved en bestemt Temperatur 
under almindeligt Tryk er væsentlig afhængig kun af den 
Nøjagtighed, med hvilken Temperaturen er bestemt. Ved 
de her benyttede Observationer er Temperaturen ved Hay- 
bunden bestemt hver Gang med 3 forskjellige Thermometre, 
saaledes at den sandsynlige Fejl af Middeltallet af deres 
Angivelser kan nøjagtigt beregnes. Jeg kalder den dt. 
Af Tabellen for Piezometrets Stand som Function af Tem- 
setter jeg til 
peraturen finder man Værdien af ar) Saaledes faar man | 
følgende Tabel. | 
me 
dt, 
I 20.6 0.4 Mmm. 
2 © 77 0.7 
3 — I .35 1.0 
Al — I .37 1.0 
5 42 1.0 
6 —1 .29 1.0 
7 — 1 .53 TA 
8 Sih Al] 1.1 
9 — 1 .53 oil 
10 — 1 .50 1.0 
Middel (Mean) 0.04 
Anm. - Denne Værdi er mindre end den ovenfor fundne, 
+0.155, da i denne sidste Usikkerheden i Lodskuddene 
(Trykkene) ogsaa indgaar. Den til d (m’—m) svarende 
Usikkerhed i Trykkene er i Gjennemsnit 
dp=2.239d (m'—m) 
The differences h’’”—h” are well-nigh identical with 
the differences between the sounded depths and the depths 
computed from the most probable values of zandz. Thus, 
a constant obliquity of the line implies a constant correc- 
tion (practically speaking) of the coefficient of compression 
for the glass, which it renders less than if the line were 
vertical. 
The most probable value found for zx, 0.796 millionths, 
is in itself remarkably small compared to the values (1.5 
to 3) which have been found and adopted in investigations 
of compression. Hence a further diminution of the value 
for x has not probability in its favour; and we are led to 
the conclusion that the soundings have been taken without 
appreciable systematic errors arising from obliquity of the 
line 
and will therefore indicate the depths with a precision 
chiefly limited by accidental errors of observation. 
The magnitude of these errors of observation may be 
ascertained in the following manner. 
The probable error of a single reading of the piezo- 
meter, or the position of the index, I put at + 0.075 mm. = 
dm’. he probable error of a determination of the piezo- 
meter-reading m, at a given temperature under ordinary 
pressure, is chiefly dependent on the precision with which 
the temperature For the 
vations made use of here, the temperature at the bottom 
determined on with 3 different ther- 
mometers, and thus the probable error of the mean of 
has been determined.: obser- 
was each occasion 
their several indications admits of being computed. I 
call it dt. From the Table for the reading of the piezo- 
meter as function of the temperature, we find the value of 
l 1 ; ° 
i i Hence we get the following Table: — 
dt dm 
+ 02.070 + 0.0280 mm. 
8 56 
7 70 
18 180 
16 160 
19 190 
30 330 
58 638 
62 572 
+ 0.050 + 0.0500 
+ 09.033 + 0'0298 
= + 0.081 mm. 
Remark. — This value is less than the value found 
above, + 0.155, since in the latter the errors attaching to 
the soundings (the pressures) are also included. The error 
in the pressures corresponding to d (m’—m) averages 
dp = 2.239 d(m’—m), 
