850 Address of Sir Wm. Thomson at the Glasgow Meeting, 
yearly tide (according to the same law as the other long-period 
tides) is not easily explained without assuming or admitting a 
considerable degree of yielding. 
Closely connected with the question of the earth’s rigidity, 
and of as great scientific interest and even of greater practical 
moment, is the question—how nearly accurate is the earth asa 
time-keeper? and another of, at all events, equal svientific 
interest—how about the permanence of the earth’s axis of rota- 
ion? 
tion 
Peters and Maxwell, about thirty-five and twenty-five years 
ago, separately raised the question, how much does the earth's 
axis of rotation deviate from being a principal axis of inertia? 
and pointed out that an answer to this question is to be ob- 
tained by looking for a variation in latitude of any or every 
place on the earth’s surface in a period of 806 days. The 
model before you illustrates the travelling round of the instan- 
taneous axis relatively to the earth in an approximately circular 
cone whose axis is the principal axis of inertia, and relatively 
to space in a cone round a fixed axis. In the model, the 
’ . . 
amounting to about ;,”’ of the axis of rotation from the princr 
the subject, he kindly did so at once, and undertook to analyze 
a series of observations suitable for the purpose, which ha 
been made in the United States Naval Observatory, Washing: 
ton. <A few weeks later I received from him a letter a - 
me to a paper by Dr. Nysen, of Pulkova Observatory, 1 Wole 
a similar negative conclusion as to constancy of magnitude 0 
direction in.the deviation sought for is arrived at from nt Fl 
series of the Pulkova observations between the years 1842 > . 
1872, and containing the following statement of his conclu 
sions :-—_ ; fee 
“The investigation of the ten month period of latitude ie 
the Washington prime vertical observations from 1862 to 18 
completed, indicating a coefficient too small to ne 
with certainty. The declinations with this instrume 
ject to an annual period which made it necessary to discu : 
of each month separately. As the series extended eee ae 
full five years, each month thus fell on five nearly eq™ ” 
