364 i I. P. Косн. 
4. Wherever the uncertainty exceeded a couple of hundred metres 
I have condemned the levellings, and have not indicated the 
altitude in the map. 
In many cases it does not occasion any inconvenience to form 
an estimate of the accuracy of the altitudes. This was, for instance, 
the case at the points levelled on both sides of the Annekssgen, 
where levellings were made at quite short distances and by means 
of a base measured on the lake. Here the errors in the altitudes of 
the points levelled could practically only proceed from the uncertainty 
of the altitude of the surface of the sea, which may amount to a 
few metres. In other places a considerable amount of work has been 
needed in order to form an estimate of the accuracy and at the 
same time to deduce the most probable values of mutually disagreeing 
single results. This was particularly difficult as far as Dronning 
Louises Land was concerned, because the distances here were, as a 
rule, so great, that for that reason alone one could not avoid a con- 
siderable uncertainty in the refraction. In order to be able to weigh 
the results of the levellings from the various stations against each: 
other, it has here been necessary to make a table of these results. 
An extract from such a table is given above. In this manner one 
gets the means of estimating whether certain stations, on an average, 
give too high or too low values, (the stations K. XII, B. XXIV and 
B. XXVIII thus seem to give too low values), and in the same manner 
one may also get the impression that certain stations give particularly 
reliable results (f. inst. B. XXVI). This in its turn becomes of im- 
portance when establishing the final value of the altitude, in parti- 
cular in those cases where there are. only two levellings or even 
only a single levelling at a certain point. 
That one ought not, in such circumstances as the ones shown 
in the table, to content oneself with the idea that the mean is the 
most probable value, is self-evident. After having compared the 
results from the single stations taken together, one must, as far as 
possible, argue the sources of error in the case of the single points. 
These arguments must be made with the working map and the ob- 
servation journals before one. 
Examples of arguments of this kind corresponding to points in 
the above table: 
Point 2100. The working map shows that the position of the 
point is determined with an uncertainty of a couple of kilometres. 
As the distances at which the point has been levelled vary from 
between 74.5 and 126 kilometres, the uncertainty of the position of 
the point, already in the correction for refraction and curvature of 
the earth, gives an uncertainty of 20—60 metres. At stations W. & L. VII 
