368 Fence Wall Geology. — Foerate. 
cause. For while the greater percentage of boulders travel 
along the path of the glacier (or with its gradient) a consider- 
able percentage deviate from this course ; many 5, some 10, 
and a few even as much as 15 degrees, thus affecting the per- 
centage of such boulders in the fence walls. Knowing the 
direction in which the glacial drift moved from the scratches 
it left on actual exposures of bed rocks, it is possible by means 
of an examination of the relative degree of angularity and size 
of rocks, the frequency of their occurrence in the fence walls, 
and a study of their distribution, to trace boulders back to 
their original source. 
It is evident that the study of very angular bowlders is alone 
of direct value in determining the original position of any 
class of rocks, since these alone lie near their original source. 
A record of the remaining boulders of the fence-walls is, how- 
ever, of value in determining their probable distance from the 
original source, and in guiding future search. When boulders 
are derived from rocks maintaining their lithological and 
palseontological characters over wide areas, the angular 
boulders derived from one locality within this area will be 
mingled with the more or less rounded boulders from some 
other locality in the same area, so that careful records axe 
always of value in reaching accurate conclusions. Note-tak- 
ing is chiefly confined to recording the varying percentage of 
the various rocks forming fence-walls and their degree of 
angularity. A record of their size, in addition to that usually 
already indirectly expressed in a record of their percentage, is 
usually of less importance. 
The distribution of the very angular rocks will determine 
the form of the original area, whether the exposure was local 
and limited in all directions, formed a long narrow band, or 
covered a wdde and extensive area. It will be of some assist- 
ance in this work to remember that the limits of any forma, 
tion in going against the glacial gradient are near the line of 
more or less abrupt disappearance of all boulders derived from 
that area, and that the limits of the same area on the side 
with the gradient are best determined by the similarly sudden 
appearance of boulders of a dWerent character. As the bound- 
aries of any area approach parallelism with the glacial 
gradient, boulders of nearly the same degree of angularity but 
derived from different sides of the boundary will become inter- 
