964 THE JOURNAL OF GEOLOGY. 
is precisely equal. In a viscous liquid receiving additions 
above, the horizontal is necessarily less than the vertical force 
(except locally). A horizontal force greatly in excess of the 
vertical does not therefore closely imitate the agencies of 
glacial motion, if it be true that a glacier acts as a viscous 
liquid. 
If a glacier be presumed to move by means of granular 
changes, the ratio of vertical and horizontal forces may still be 
essentially the same, for the granular action may be merely a 
mode of motion under gravity. If, however, the growth of the 
granules brings into play the forces of crystallization, it is con- 
ceivable that the ratio of vertical and horizontal forces may not 
remain the same. It may be conceived that the amount of 
crystalline force brought into action along any given line is 
somewhat correspondent to the number of granules lying in that 
line, or to the length of the line. If this be true, the axis of the 
glacier would represent the line of greatest force. Gravity 
would of course interpose its influence and the combined result 
would be merely a greater or less departure from the mode of 
action under gravity alone. Probably no one will doubt that 
gravity dominates the phenomenon. 
The experiments of Mr. Case more nearly reproduce the 
conditions of this phase of the granulation hypothesis than those 
of the viscosity hypothesis. 
It is obvious that the nature and mode of application of the 
forces employed are critically important in dealing with a sub- 
stance on the border line between the solid and liquid states. It 
is quite possible to manipulate a liquid so that it shall deport 
itself like a solid, and conversely, to handle a solid so that it 
shall deport itself like a liquid. A solid may be made to flow, 
or seem to flow like a liquid, and a liquid may be made to shear 
or to seem rigid like a solid. In these cases it is the manipula- 
tion rather than the nature of the substance that gives expres- 
sion to the result. A semi-solid is peculiarly sensitive to the 
mode of manipulation, and in experiments with bodies of this 
class the interpretation of results is quite as much to be guided 
