606 PROFESSOR CHRYSTAL 
irregular character, due to the beats of seiches not very different in period. The ends 
of a lake are therefore the worst places for distinguishing seiches of higher nodality. 
If a lake had three periods 10’, 11’, 12’, the proper place to observe at, in order to 
establish clearly the seiche corresponding to the first or last of these periods, would be 
at a node of the seiche whose period is 11’. Thus Foret’s argument, that the 10’ seiche 
of Léman is a transversal and not a plurinodal longitudinal one, because it is not detected 
at Geneva, is by no means conclusive. I may also add, although I have little theoretical 
or experimental ground as yet for the opinion, that it seems to me unlikely that any 
transversal seiche would be so stable, especially in a lake of the shape of Léman, as the 
beautiful records of Foret’s limnograph seem to indicate. Nevertheless, great weight 
must be attached to the inclination of so sagacious an observer, who has all the data 
before him. 
The table also raises many interesting subjects of inquiry. Why, for example, are 
no seiches observed in Constance of the periods 22°4’ and 18°3’? Is this an accident, 
due to the position at which the limnograph was placed; or are these seiches unstable, 
owing to irregularities of the lake-bottom near one or more of the corresponding 
nodes ? 
§ 9. In a purely concave lake the ratio of the uninodal period to the binodal period 
is less than a half. In a purely convex lake, if such a thing could be found in nature, 
the corresponding ratio would be greater than a half. In lakes which are neither 
purely concave nor purely convex the value of T,/T, will be greater or less thana half 
according as the concavity or convexity predominates. 
§10. In the case of parabolic and quartic lakes the rule given by Du Boys for 
calculating the periods, viz., aT, = (2/r) | dl/,/(gh), where his the depth corresponding 
to the element di of the line of maximum depth, gives too high a value for purely con- 
cave and too low a value for purely convex lakes ; but it gives in many cases a good first 
approximation to the periods. This approximation is better for concavo-convex lakes 
than for purely concave or purely convex lakes ; and for purely concave or purely convex 
lakes, the approximation is better the higher the nodality of the seiche. For a purely 
concave symmetric parabolic lake Du Boys’ rule would be considerably out; in fact, 
for such a case gI',/T,=1'414. It may also err greatly in cases where there are great 
variations of the breadth of the lake, as the method of applying the formula takes no 
account of such peculiarities. * 
§11. In a lake of varying depth the uninode is not in general in the middle of the 
lake, and the uninode, middle trinode, middle quinquinode, etc. are not coincident. 
Also the ventral points are not midway between the nodes; and the wave length varies 
from node tonode. Thus, for example, in a symmetric parabolic lake the uninode is of 
course in the middle, but the binodes are displaced towards the shallow ends. It 
results from the calculations @ priori, made by myself and Mr WeppeErsury, that in 
* Dr Enprés has found a striking example in the uninodal seiche of the Waginger See, of which he was good 
enough to tell me by letter. 
