1443 
January, February, March, April, May, June, July, August, 
— 17.8 +285 +60.9 +1024 +929 +480 —16 — 384 
September, October, November, December. 
— 42.6 —85.2 —751 — 72.5. 
By the introduction of these corrections I have eliminated the 
influence of the yearly periodic variations in the water level. 
In order to inerease the accuracy of the values from which the 
results must be deduced and to remove entirely or partially the 
error that might arise, if the number of low waters in a month 
should be one less or more than the number of high waters, I-have 
always taken the averages of two consecutive months: Jan. and 
Febr., Febr. and March, ete. The further calculations are based 
upon these two-monthly means. 
Corrections for known tides are not introduced into these values. 
The influence of tides of short period is very slight upon the two- 
monthly means, and if, as is the case in my calculations, the average 
is taken of nearly 50 such means, it may be altogether neglected. 
Of the tides of longer period we must mention, besides the yearly 
and half-yearly sun tide, the influence of which has been taken 
into account, the tide Mm, with a period of over 27 days. It appears 
from the calculations that the influence of this tide upon the two- 
monthly means can rise to about + 6 mm. but as the amplitude 
and pbase constant of this tide are very little known, we cannot 
calculate the exact value of the correction. We may, however, assume 
that in an average of about 50 of these valnes, for dates that cor- 
respond to very various phases of this tide, its influence may be 
neglected. 
3. The length of the period of the latitude variation of about 
431 days (CHANDLER’s period) was deduced from long series of 
astronomical observations, by E. F. v. bp. SANDE BAKHUYZEN, Dr. 
Zwiers and me; the results obtained by us differ very little, but I 
take as the most accurate that deduced by Dr. Zwaers in a paper 
in These Proceedings of June 24, 1911, Vol. XIV, p. 111, that 
is 431,24 days. 
In order to determine whether a variation in the sea level takes 
place in that period, I have, starting from the first bi-monthly mean 
for 31 Jan. 1855, determined the dates of the days, which fall 
431,24 days later, or a multiple of that interval and then selected 
the bi-monthly means which are nearest to these dates, sometimes 
a little earlier and sometimes a little later, with a difference at most 
