e
the centroid of compression there is nothing but a trifling amount of
embedment of the stirrup. If 1/2-in. stirrups were used in an 18-in.
beam, assuming that 30 diameters were enough for anchorage, the centroid
of compression would be, say, 3 in. below the top of the beam, the
middle point of the stirrup's anchorage would be about 8 in., and the
point of full anchorage would be about 16 in. The neutral axis would
come somewhere between. These are not unusual proportions. Analogy with
a riveted truss fails; even the anchorage above the neutral axis is far
from realization.

Major Sewell refers to shallow bridge stringers and the possibility of
failure at connections by continuity or deflection. Structural engineers
take care of this, not by reinforcement for continuity but by ample
provision for the full bending moment in the stringer and by ample
depth. Provision for both the full bending moment and the ample depth
reduces the possibilities of deflection at the floor-beams.

Major Sewell seems also to have assumed that the paper was a general
discussion on reinforced concrete design. The idea in pointing out that
a column having longitudinal rods in it may be weaker than a plain
concrete column was not to exalt the plain concrete column but to
degrade the other. A plain concrete column of any slenderness would
manifestly be a gross error. If it can be shown that one having only
longitudinal rods may be as bad, or worse, instead of being greatly
strengthened by these rods, a large amount of life and property may be
saved.

A partial reply to Mr. Thompson's discussion will be found in the
writer's response to Mr. Mensch. The fault with Mr. Thompson's
conclusions lies in the error of basing them on averages. Average
results of one class are of little meaning or value when there is a wide
variation between the extremes. In the tests of both the concrete-steel
and the plain concrete which Mr. Thompson averages there are wide
variations. In the tests made at the University of Illinois there is a
difference of almost 100% between the minimum and maximum results in
both concrete-steel and plain concrete columns.

Average results, for a comparison between two classes, can mean little
when there is a large overlap in the individual results, unless there is
a large number of tests. In the seventeen tests made at the University
of Illinois, which Mr. Thompson averages, the overlap is so great that
the maximum of the plain c