ed into the heading.

In one or two instances considerable ground was lost at the face. On the
evening of December 14th, 1906, as a heavy coal wagon was passing along
33d Street above the heading, the rear wheels dropped through the
asphalt pavement. An examination disclosed a cavity under the pavement
about 14 ft. long, 12 ft. wide and 14 ft. deep. Evidently, the fine sand
had gradually settled into the voids caused by the loss of material at
the face, and the settlement broke the brick sewer over the heading. The
sewer was temporarily repaired, and the hole in the street was filled
before morning. A tight bulkhead was built across the heading, and work
was abandoned at that point. The north drift was advanced to a point
108 ft. west of Fifth Avenue where sand was also encountered and a
considerable run occurred. After that time all work on the three-track
section was discontinued.

The Company then took up the consideration of changes in plan. To
determine the difficulties of driving a Twin Tunnel at a lower
elevation, an exploration drift, 8 ft. high and 12 ft. wide, was driven
on the center line of the street as a top heading on the proposed new
grade. Test holes were drilled above this heading and to the sides. The
results indicated that there was sufficient rock cover of fair quality
to enable the Twin Tunnel to be driven without great risk. The new plan
(continuing the Twin Tunnel westward at a lower grade) was adopted in
March, 1907, and work was immediately resumed at Fifth Avenue.

The relation between the cross-sections under the old and new plans at
that point is shown by Fig. 3. Before the new section was excavated it
was necessary to support the timber work in the old headings. The plan
adopted is also shown by Fig. 3. The rock was excavated under the center
heading, as shown in cross-section, for a length of about 3 ft. A girder
composed of two 18-in. I-beams was then put in position over each line
and supported on the sides by posts. The ends at the center lines
between the tunnels were supported on short posts bearing on the rock
bench. The support of the timbering in the headings was then transferred
to the girders by additional posts. Blocking was also inserted between
the tops of the beams and the rock walls between the headings. Fig. 2,
Plate LIX, gives a good idea of the timber work in the top headings
above the I-beams. When the roof had been made secure, the removal of
the bench was begun. As t