THE FORMATION OF ICE RAMPARTS 



159 



considerable depths. It is well known from studies of the "fatigue " 

 of materials under stress that they often yield to slowly acting 

 stresses that would be transmitted undiminished in intensity if 

 quickly applied. Snow blanketing of the ice, from the evidence 

 in Mr. Tyrrell's paper, would appear to be very general within 

 the districts which he studied. 



Further limitations upon the formation of ice ramparts are 

 imposed by the third condition — the incompetency of the ice 

 cover as a transmitter of stresses. With the ice serving as a strut, 



Fig. i. — Sketch map showing the position of ice ramparts and of buckled ice 

 ridge formed on Lake Mendota at Madison, Wisconsin (based on Buckley's Map). 



its push can be transmitted effectively only when the cover is main- 

 tained as a plane surface. Lack of homogeneity or of absolute 

 uniformity in strength, and variation in form of the surface at 

 which stress is applied, will with increasing length of beam intro- 

 duce an important stress component tending to buckle the beam 

 and dissipate the energy transmitted by it — the competency of a 

 strut to .transmit stresses is inversely as its length. Experience 

 shows that lakes or arms of lakes which are much over a mile and 

 a half across do not develop important ice ramparts. On Lake 

 Mendota at Madison, the best ramparts are found upon the shores 

 of University Bay, which is about three-fourths of a mile across. 

 Outside this bay the lake ice is raised each winter into a sharp 



