less for the design
of reinforced concrete arches, basing his objections on the shrinkage of
concrete in setting, the unreliability of deflection formulas for beams,
and the lack of rigidity of the abutments. The writer, noting that
concrete setting in air shrinks, whereas concrete setting in water
expands, believes that if the arch be properly wetted until the setting
up of the concrete has progressed sufficiently, the effect of shrinkage,
on drying out, may be minimized. If the settlement of the forms
themselves be guarded against during the construction of an arch, the
settlement of the arch ring, on removing the forms, far from being an
uncertain element, should be a check on the accuracy of the calculations
and the workmanship, since the weight of the arch ring should produce
theoretically a certain deflection. The unreliability of deflection
formulas for beams is due mainly to the fact that the neutral axis of
the beam does not lie in a horizontal plane throughout, and that the
shearing stresses are neglected therein. While there is necessarily
bending in an arch ring due to temperature, loads, etc., the extreme
flanges sometimes being in tension, even in a properly designed arch,
the compression exceeds the tension to such an extent that comparison to
a beam does not hold true. An arch should not be used where the
abutments are unstable, any more than a suspension bridge should be
built where a suitable anchorage cannot be obtained.

The proper design of concrete slabs supported on four sides is a complex
and interesting study. The writer has recently designed a floor
construction, slabs, and beams, supported on four corners, which is
simple and economical. In Fig. 7 is shown a portion of a proposed
twelve-story building, 90 by 100 ft., having floors with a live-load
capacity of 250 lb. per sq. ft. For the maximum positive bending in any
panel the full load on that panel was considered, there being no live
load on adjoining panels. For the maximum negative bending moment all
panels were considered as loaded, and in a single line. "Checker-board"
loading was considered too improbable for consideration. The flexure
curves for beams at right angles to each other were similar (except in
length), the tension rods in the longer beams being placed underneath
those in the shorter beams. Under full load, therefore, approximately
one-half of the load went to the long-span girder and the other half to
the short-span girder.