I. P. Косн. Survey of Northeast Greenland. 163 
to the periods July 9th to August 12th, August 23rd to October 21st 
(autumn) and May 31st to June 22nd. In the same manner as in the 
table on pp. 156—157 we have made the terms d4z—= z- zm Where 
Zm stands for the mean value of all the zenith distances in the said 
direction; but whereas 4z in the earlier table gives the direct im- 
pression of the variation of the co-efficient of refraction, here where 
1--k is proportional to the quadrate of the angle of depression it 
only gives a direct impression of the variation of the zenith distance. 
As to direction II k is computed by every single observation, after 
which ДК is made from 4k = k- km (ДК is indicated in units of 
the third decimal place). 
When on account of mirages there were several horizons to 
choose between, it was as a rule impossible to decide which was 
the “real one”, and which one or ones must be characterized as 
mirages. This is owing to the complicated nature of the mirage, 
which often only appears in the shape of a confused column structure 
intersected by one or more horizontal lines, more or less effaced. 
The column structure and the horizontal lines will at times dis- 
appear altogether, in which case the existence of the mirage can only 
be proved by the fact that an observation gives z < 90°. The other- 
wise rather common, simple case of an upward mirage where the 
reflection of the sea horizon appears as a horizontal streak, which 
may elevate itself about 1/4 degree above the normal position of the 
horizon, has not occurred during the observations. On the other 
hand, we have in the course of the observations had several cases 
of a downward mirage, where the reflection of the sky appeared as 
a stratum of air under the real sea horizon, and so created a false 
and lower lying horizon. An example of this are the observations 
in the directions I and II on September 11th at 2P.20 (see Fig. 22, p. 195). 
As regards direction II the real sea horizon was just discernible 
with the naked eye, but it could not be observed in the telescope. 
In the table below the zenith distance given in direction II is to 
the false horizon; the computation of this zenith distance would 
have given k negative, but would have been illogical, as the line of 
vision does no longer, as presupposed in the formula, touch the 
surface of the sea, but crosses the latter. In all the cases now be- 
fore us there was in reality no other expedient open to us but to 
consider the false horizon as the real one. In the cases where the 
lower lying ones were sufficiently clearly defined to admit of an 
observation, I have in the table given the zenith distances in paren- 
thesis. 
In the case of direction III it happened on several occasions 
that the sea horizon lay deeper than the land which the sight tra- 
