Fejlene under Nejlads vestover bliver næsten den samme 
som Summen af Fejlene under Sejlads østerover — indenfor 
den enkelte Gruppe. 
Maximumsfejlen 1 Klokkeslettet rammer Observationen 
Kl. 12 Middag. Efter Observationstabellerne (S. 4996) 
udtog jeg, for hver benyttet hel Observationsdag, Længde- 
forandringen fra Middag til Middag (regnet positiv, naar vi 
rejste mod Øst, negativ mod Vest) og omgjorde denne til 
Tidsminutter. For hver af de hele Grupper toges Summer 
af de positive og Summer af de negative Verdier. Deres 
Forskjel, divideret med Antallet af Dage i Gruppen, giver 
den gjennemsnitlige Fejl 1 Klokkeslettet, som Middagsob- 
servationen har været udsat for. 
Denne overstiger. som 
man nedenfor vil se, i intet Tilfælde 5 Minutter, altsaa en 
temmelig fuldkommen Compensation. Desuden udregnede 
jeg den højeste positive og negative Verdi, som overhovedet 
forekom inden Gruppen. Disse dreje sig om 30 Minutter, 
og overstige ikke 36 Minutter. I 1878, da Expeditionen 
sejlede paa meget hoje Bredder mellem Spidsbergen og 
Grønland, stilledes undertiden Skibsuret efter Stedets sande 
Tid ogsaa ved Midnat, hvorved Fejlen i Klokkeslettet er 
yderligere formindsket. Hertil er dog ikke taget Hensyn 
ved Beregningen af Fejlens Maximumsstørrelse. 
Efter de corrigerede Middeltal for hvert af de 24 
Observationsklokkeslet i de valgte Grupper beregnedes For- 
melen for den daglige Periode efter den Besselske Methode. 
Antallet af Led, der ere beregnede, er noget forskjelligt 
etter de forskjellige Tilfælder. Da der paa Grund af Ob- 
servationernes Faatallighed alene kunde blive Spørgsmaal 
om at erholde en første Tilnærmelse til Kundskaben om 
de daglige Perioder, ere i Regelen, foruden Mediet, kun 2 
Led blevne beregnede. Perioderne for Lufttrykket, nogle 
Perioder for Luftens Temperatur, en enkelt (1876) for 
Vanddampenes Tryk, den relative Fugtighed og Vindens 
Hastighed ere beregnede med 4 periodiske Led. 
Efter Dr. H. Wilds Forslag" har jeg forsøgt. efter 
de paa Rudepapir afsatte observerede Værdier at trække 
paa Frihaand den sandsynligste Curve for de forskjellige 
Perioder. I de allerfleste Tilfælder faldt denne Curve 
meget nær sammen med den beregnede. Hvor der viste 
sig mere merkelige Afvigelser, stillede det sig for mig meget 
tvivlsomt, hvilken af begge jeg skulde foretrække. For at 
give Læseren fuld Anledning til at dømme og vælge, optører 
jeg 1 Tabellerne og i Figurerne PI. I til III saavel de ob- 
serverede som de efter de Besselske Formler beregnede 
Curver, og skal for hvert Tilfælde gjøre Rede for disses 
mindre Overensstemmelse med deraf flydende 
Bestemmelsen af Epocherne og Værdierne for 
Maxima og Minima samt Epocherne for Middelværdiens 
Indtreetten. 
større eller 
Afvigelser i 
' Die Temperaturverbåltnisse des Russisehen Reiches. op. cit 
l h 
oe) 
errors on the passage west being very nearly the same as 
the sum of the errors on ‘the passage east — within any 
individual group. 
The maximum error in the hour vitiates the observa- 
tion taken at noon. From the Observation Tables (pp. 49 
96), I took, for every whole “observation day,” 
the varia- 
tion in longitude from noon to noon (calculated as positive 
when steaming east and negative when steaming west), and 
converted it into minutes of time. For each of the whole 
groups was then taken the sum of the positive and the sum 
of the negative values. heir difference, divided by the 
number of days in the group, gives the average error in 
the hour to which the observation taken at noon has been 
exposed. This error, as shown below, will in no case exceed 
5 minutes, — therefore a tolerably perfect compensation. 
Furthermore, I computed the highest positive and negative 
values that occurred within the group. These values average 
about 30 minutes, and do not exceed 86 minutes. In 1878, 
when the Expedition was cruising in very high latitudes, 
between Spitzbergen and Greenland, the ship’s clock was 
sometimes also set to apparent time at midnight, thus still 
further reducing the error in the hour. In such cases, 
however, regard was not had to this circumstance when 
computing the maximum value of the error. 
From the corrected mean values for each of the twenty- 
four hours of observation in the groups selected, the formula 
for the diurnal period was computed according to Bessel's 
method. The number of terms computed is somewhat 
different in the several cases. Since, owing to the limited 
number of observations, it was out of the question to attempt 
more than a first approximation as regards our knowledge 
of the diurnal periods, as a rule 2 terms only, exclusive 
of the mean, have been computed. The periods for the 
pressure of the air, a few periods for the temperature of 
the air, one period (1876) for the force of vapour, the 
relative humidity and the velocity of the wind, are computed 
with 4 periodical terms. 
Acting on the suggestion of Dr. H. Wild.! I have 
tried, from the observed values laid down on ruled paper, 
to draw the most probable freehand curve for the different 
periods. In the great majority of cases this curve was 
found to coincide very nearly with that computed. Wherever 
any considerable deviations occurred, I was extremely doubt- 
ful which of the two to give the preference. With a view 
to afford the reader ample means of judging and selecting, 
I have set forth in the Tables and in Pl. I to III both 
the observed curves and those computed from Bessel’s for- 
mule, and shall in each case account for the 
or less agreement, and the variations resulting therefrom, in 
the determination of the epochs and the values for maxima 
and minima, as also of the epochs for the occurrence of the 
mean values. 
greater 
1 Die Temperaturverhåltnisse des russisehen Reiches, op. cit. 
13* 
