6 ON THE STRUCTURE AND MOTION OF GLACIERS 
, 
The application of the results here obtained to the “ viscous flow’ 
of glaciers, will perhaps be facilitated by the following additional 
experiments. 
A block of boxwood (A, fig. §), 4 inches long, 3 wide and 3 deep, 
had its upper surface slightly curved, and a longitudinal groove 
(shown in dots in the figure), an inch wide and an inch deep, worked 
into it. A slab of the wood was prepared, the 
Fig. 5. under surface of which was that of a convex 
a cylinder, curved to the same degree as the 
concave surface of the former piece. The 
Jagd sctbate betes arrangement is shown in section at B. A 
straight prism of clear ice, 4 inches in length, 
an inch wide, and a little more than an inch in 
B depth, was placed in the groove, and the upper 
\\ slab of boxwood was placed upon it. The 
< mould was submitted to hydraulic pressure, as 
N SS in the former cases ; the prism broke as a matter 
WN of course, but the quantity of ice being rather 
more than sufficient to fill the groove, and hence 
projecting above its edge, the pressure brought the fragments together 
and re-established the continuity of the ice. After a few seconds it 
was taken from the mould, bent as if it had been a plastic mass. 
Three other moulds similar to the last, but of augmenting curvature, 
were afterwards made use of, the same prism being passed through 
all of them in succession. A? the conclusion of the experiments the 
prism came out, bent to a transparent sentt-ring of solid tce. 
In this way, by the proper application of force, all the bendings. 
and contortions observed in glacier ice, and adduced in proof of its 
viscosity, can be accurately imitated. Any observer, seeing a straight 
bar of ice converted into a continuous semi-ring without being aware 
of the quality referred to, and having his attention fixed on the 
changes of external form alone, would be naturally led to the con- 
clusion that the substance is viscous. But it is plainly not viscosity, 
properly so called, which enables it to change its shape in this. 
way, but a property which has hitherto been entirely overlooked 
by writers upon glaciers. 
It has been established by observation, that a vertical layer of ice 
originally plane, and perpendicular to the axis of a glacier, becomes 
bent, because the motion of its ends is retarded in comparison with 
that of its centre. This is the fact upon which the viscous theory 
principally rests. 
In the experiments with the straight prism of ice, four successive 
