1446 
errors in the sea levels, probably to a large extent caused by the 
circumstance, that the causes of deviations in the normal sea level 
are of lengthy duration, and thus can cause abnormally high or 
low sea levels during a long time. 
In order to investigate this, I have taken the means of a series 
of 42 months in a different way, by combining the heigbt in Jan. 
of the year a, with, that in Feb. of the year a+ 1, in March of 
the year a+ 2 ete. From this follows for the mean error of the 
unit of weight + 60.2 mm. which agrees much better with the 
value we found + 51.5. The real mean errors of p and gq therefore 
probably do not differ greatly from the values calculated. 
~ 
5. A second way of judging of the reliability of the results 
obtained is the caleulation of the same quantities from another 
combination of observations. For this purpose I chose the observations 
of 1855— 1892, which I had calculated in 1894, but had now redu- 
ced to the yearly means with better values for the deviations of the 
monthly means and further the observations of 1893 —1912. I found 
from both series of observations : 
h = 4.50 sin (p -+ 168°.59) . . . (1855 —1892) 
and 
B74 Rele B51 76° 19 'y hd e019). 
By the change in the reduction numbers and a more accurate 
calculation, the formula for the sea level during the period 1855— 
1892 differs somewhat from the formula found in 1894. The striking 
correspondence between the three formulas now found for the periods 
1855—1892, 1893—1912 and 18551912 is certainly largely due 
to accident, but it confirms the view that the variation in the sea 
level is real. 
6. In order to test the efficiency of the method that I had fol- 
lowed, | applied it to two cases in which one could not a priori 
expect a periodic variation, and to another case in which the existence 
of such a variation was certain. 
First 1 arranged the bi-monthly means in a period of 13 months 
or 395.75 days which is not a multiple of any period of a sun 
or moon tide, and in which therefore we could not expect any 
periodic variation of level. For this purpose I used the observations 
of 1855—1892, and got the following deviations of the sea level from 
- 
their general averages. 
