jecting equally beyond the supports. In order to prevent
crushing of the fibre at the points where the stress is applied
it is necessary to use bearing blocks of maple or other hard
wood with a convex surface in contact with the beam. Roller
bearings should be placed between the bearing blocks and the
knife edges of the crosshead to allow for the shortening due to
flexure. (See Fig. 29.) Third-point loading is used, that is,
the load is applied at two points one-third the span of the beam
apart. (See Fig. 30.) This affords a uniform bending moment
throughout the central third of the beam.

[Illustration: FIG. 30.--Two methods of loading a beam, namely,
third-point loading (upper), and centre loading (lower).]

_Measuring the deflection_: The method of measuring the
deflection should be such that any compression at the points of
support or at the application of the load will not affect the
reading. This may be accomplished by driving a small nail near
each end of the beam, the exact location being on the neutral
plane and vertically above each knife-edge support. Between
these nails a fine wire is stretched free of the beam and kept
taut by means of a rubber band or coiled spring on one end.
Behind the wire at a point on the beam midway between the
supports a steel scale graduated to hundredths of an inch is
fastened vertically by means of thumb-tacks or small screws
passing through holes in it. Attachment should be made on the
neutral plane.

The first reading is made when the scale beam is balanced at
zero load, and afterward at regular increments of the load which
is applied continuously and at a uniform speed. (See SPEED OF
TESTING MACHINE, above.) If desired, however, the load may be
read at regular increments of deflection. The deflection
readings should be to the nearest 0.01 inch. To avoid error due
to parallax, the readings may be taken by means of a reading
telescope about ten feet distant and approximately on a level
with the wire. A mirror fastened to the scale will increase the
accuracy of the readings if the telescope is not used. As in all
tests on timber, the strain must be continuous to rupture, not
intermittent, and readings must be taken "on the fly." The
weighing beam is kept balanced after the yield point is reached
and the maximum load, and at least one point beyond it, noted.

_Log of the test_: The proper log sheet for this test consists
of a piece of cross-section paper with space at the m