( 462 ) 
be represented by the formula: 
y = a 4- bx , 
in which x represents the number of years elapsed since an assumed 
initial date, and b the yearly change. Making use of all these sea- 
levels, he calculates the value of b by the method of least squares, 
and finds for it + 0,13 cm. ; for the mean error of this result I find 
± 0,042 cm., whereas, if we assume the formula y = a-\-bx the 
mean error of a yearly mean is ± 2,52 cm. 
From the value of the mean error we should conclude that the 
calculated yearly change, a relative rise of the mean level of the 
sea of 0,13 cm., is real; from the graphical representation, however 
this does not appear so clearly, and in order to examine this matter 
further, I have repeated the calculations of b from part of the data. 
I have namely excluded the results of the four first and of the four 
last years, which greatly deviate from the others, as appears from 
the figure, and determined b from the series of the 27 other years; 
I find then b —— 0,05 cm., so a relative fall of the sea-level 
with a mean error of ± 0,046 cm. The great difference of the two 
results, —0,13 and -{-0,05, derived from series of 35 and 27 years, 
the latter of which is quite contained in the former, shows that only 
little value is to be attached to those results, and that the sinking 
of the ground, which is derived from the complete series of observa¬ 
tions, is practically only based on the results of 8 years, the first 
4 and the last 4, whereas the other 27 years would rather point 
Beside this first series of readings of the water gauges Mr. Ramaer 
has dealt with another, which readings were made during the years 
1862—1877 on 18 water gauges. (Table XXI of the paper). The 
16 mean yearly sea-levels derived from this series he has combin 
with those derived from the first series of 35 years and he deduce 
a new value of b from the whole number. It is clear that as m 
this combination the first series, is almost twice as long as t e 
second, the result for b will depend in a great measure on the 
found value and cannot be considered as a quite new resu t. 
avoid this difficulty I have calculated the 16 yearly means o 
second series separately, and I obtain then for b -{-0,30 cm., W1 
a mean error of ± 0,213 cm. For the mean error of a yearly mean 
I find zb 3,93 cm., from which it appears that Mr. Ramaer 
attaches more importance to the values from the first series vf 
are each based on the mean of 8 water gauges only, than t° ^ 
values from the second series, in spite of the latter being ellV 
from the mean of 18 water gauges. 
