TIDAL DATUM PLANES 63 
and Yakutat there are observations covering periods more than 5 years, and since 
along this coast sea level appears to be falling, a brief discussion of these will be of 
interest. In Figure 33 are plotted the yearly heights of sea level at the above-named 
four stations. 
From the plotting for Ketchikan it appears that sea level at that station has been 
rising slowly from 1919 to about 1940, less than a tenth of a foot during this period. 
Since that time the indications are for a falling sea level, though it will require further 
observations to bring this out clearly. 
For Sitka, Juneau and Yakutat the evidences for falling sea level are unmistakable. 
At Sitka there are available a year of tide observations from 1893 to 1894 and another 
year in 1925. In the latter year sea level was a quarter of a foot lower than in 1893-94. 
And in 1925 sea level was 0.08 foot higher than in 1939 at the beginning of the present 
continuous series. 
At Juneau a year of observations is available in 1912 ad this value of sea level is 
1.27 feet higher than in 1937 when the present series began. It is to be noted that in 
this latter series there is a break in the observations between the years 1939 and 1944. 
. A rising sea level along the Pacific coast of continental United States and a falling 
sea level along the Pacific coast of Alaska must be interpreted as due to relative land 
movements in the two regions. 
Primary Determination 
The variations in sea level discussed in the preceding sections may be summarized 
as follows: At any point on the coast, sea level varies from day to day, from month to 
month, and from year to year. From one day to the next, sea level may vary by a 
foot or more, and within the same year two values of daily sea level may differ by 5 feet 
or more. Monthly sea level is subject to variations of both periodic and nonperiodic 
character, so that within a year sea level for two different months may differ by as much 
asafoot. Yearly values may differ by as much as one or two-tenths of a foot from one 
year to the next and in addition may be subject to a slow progressive rise or fall. 
The determination of mean sea level therefore involves two problems. The first 
is, how long a series of tide observations is required to give an accurate determination 
of mean sea level? The second problem is, how can the sea level derived from a short 
series of observations be corrected to mean value? 
A period of 19 years is generally considered as constituting a full tidal cycle, for 
during this period of time the more important of the tidal variations will have gone 
through complete cycles. It is therefore customary to regard results derived from 19 
years of tide observations as constituting mean values. Hence sea level derived from 
19 years of observations may be taken to constitute a primary determination and as 
giving accurately the datum of mean sea level. 
If the mean level of the sea remained constant over long periods of time and if 
the coast were absolutely stable, we might expect sea level at any place determined from 
one 19-year series to be the same as that derived from another such series even if 
separated by a number of years. Apparently, however, this is not the case, and for 
precise purposes it is therefore necessary to specify the particular, epoch used in the 
determination of mean sea level. The six long series of observations available for the 
