RAPPORT 1905—06 ANNEXE A - — XXVIII — 
circle of longitude and determination of the points of intersection between the circle 
and the isohalines). In making the adjustments some help has been had from the 
mean deviation of the salinities from their mean value, which in connection with the 
number of observations that go to make up a mean number, indicates how far it 
is permissible, in drawing the curve that represents the adjusted mean salinity as 
a function of the longitude or latitude, to depart from the actual mean values when 
the object is to indicate the presumed mean salinity. Thus by means of the 
graphical adjustments two systems of points are determined; where the points do 
not give the isohalines simple forms the adjustments are improved with constant 
regard to the mean deviation of the salinity, and the isohalines are finally drawn 
with as simple a form as possible through the constructed points (Plate 1). 
Since the mean deviation of the salinity also has an interest of its own as 
a representation of the variation in salinity found in the same place, it is given 
on Plate II. First a calculation is made in the usual way of the mean deviation 
between all the observation-values in one field and their mean value; this mean 
deviation is then subjected to a reducing correction in order to eliminate from the 
mean deviation the part caused by the variation of the salinity within the field. 
The corrected mean deviation thus represents how great the salinity’s mean devia- 
tion from the mean value is at a certain point. But as the fields are chosen suffi- 
ciently small, the correction may as a rule be neglected. 
The map (Plate II) over the mean deviations is in other respects con- 
structed in the same way as the map (Plate I) over the mean salinity. For the 
construction of the two surface maps there have been about 7000 observations 
at disposal. 
The Salinity in the Depths (Plates III—XV) 
In order to get a representation of the mean salinity in the depths the mean 
value of the salinity has been calculated for every station at those depths where 
observations customarily are made. If the salinity should be lacking for a parti- 
cular depth, whenever possible a value has been inserted, determined by graphical 
interpolation, whereby all the mean values for a station are as a rule made to rest 
on the same number of observations — real or fictitious — taken at the same time. 
For those stations where the salinity varies from the surface to the bottom, 
curves are drawn which in rectangular coordinates represent the mean salinity as 
a function of the depth. These curves are to a certain extent adjusted; since the 
value of the salinity at the surface is taken from the surface map (Plate I). We 
have now the required means at hand for the construction of vertical sections to 
represent the mean salinity, for from the curves just spoken of can be determined 
for each station the points of intersection of the isohaline surfaces with the vertical, 
