Survey of Northeast Greenland. 361 
stations there may be some difficulty. For these points the chief 
sources of errors in the altitude are ihe following: 
1. Errors in the determination of the position of the station. 
2. Errors in the altitude of the station. 
3. Ill-defined sighting points. 
4. Variable refraction. 
ad 1. Errors in the determination of the position of the station 
entail errors in the position of the points, the distance to these from 
the stations thus becoming wrong. In the case of favourable bearings 
this source of error does not play any great part, whereas at acute 
bearings it may exercise considerable influence. 
ad 2. When during a coast survey the height of a station is 
determined by barometric observations from the surface of the sea, 
one will in that manner generally have guarded oneself against 
serious errors. If, however, one leaves the coast for a longer period, 
such as for instance we did during the two sledge trips performed 
by the Danmark-Ekspedition across the inland ice to Dronning 
Louises Land, the barometric levelling becomes quite uncertain, 
though this does not appear visibly through the barometric observ- 
ations. In order to get the means of supporting the barometric ob- 
servations under conditions of this kind, one ought for periods of 
several days to level at the same sharply defined points. If one 
succeeds in doing this, one will be able to form an idea of the un- 
certainty of the heights of the stations. 
ad 3. Ill-defined sighting points are the most dangerous of the 
sources of error, because very often the result is a complete confusion 
of the points. But even if one only has two sights at such a point, 
and the levellings do not harmonize, it is, however, not always ne- 
cessary to condemn the observation. In case the one sight is short and 
very nearly horizontal, it may perhaps be used, if, for instance, in 
the sketches one has sufficient support for marking off the point in 
the map. On the other hand it would of course be quite wrong to 
use a mean of the two values of the altitude. 
ad 4. It has already been pointed out that the levelling becomes 
uncertain, when the distance exceeds 30 kilometres, but for that 
reason one ought of course not to omit levelling at greater distances; 
only one must fully realize that the uncertainty of the refraction 
nearly increases with the square of the distance, and so may be- 
come very large. The greatest distance at which levellings were 
made on the Danmark-Ekspedition was about 140 km. 
Sometimes it is possible to conclude from certain meteorological 
conditions that on the day in question there exist more or less ab- 
