and still afford support to the face. The extensions also afforded
convenient working platforms. They were subject to severe bending
strains while the shield was being shoved, however, and the cast-iron
rams were frequently broken or jammed. The extensions did not last
beyond the edge of the ledge at Manhattan, nor more than about half
through the rock work at Long Island City. The fixed extensions
originally placed on Shields _A_ and _C_ at Manhattan were not
substantial enough, and lasted only a few days.

Wherever the rock face was sufficiently sound and high, a bottom heading
was driven some 20 or 30 ft. in advance of the shield. The heading was
driven and the cradle placed independently of the face of the soft
ground above, and in the manner described for all-rock sections. The
remainder of the rock face was removed by firing top and side rounds
into the bottom heading after the soft ground had been excavated. Great
care had to be taken in firing in order not to disturb the timber work
or break the rock away from under the breast boards. If either occurred,
a serious run was likely to follow. The bottom-heading method is shown
by Figs. 1, 2, and 3, Plate LXVIII, and the breasting and poling by Fig.
2, Plate LXX.

In the early part of the work, where a bottom heading was impracticable,
the soft ground was first excavated as described above, and the rock was
drilled by machines mounted on tripods, and fired as a bench. By this
plan no drilling could be done until the soft ground was removed. This
is called the rock-bench method.

Later the rock-cut method was devised. Drills were set up on columns in
the bottom compartments of the shield, and the face was drilled while
work was in progress on the soft ground above. The drilling was done