resist the sway of trains. Not only should the top flanges of adjacent girders be connected by rigid lateral bracing, but the bottom flanges should be stayed by occasional cross-bracing frames, one of the latter being invariably used at each expansion end of each track.

VII.   PIN-CONNECTED, PONY-TRUSS SPANS AND PLATE GIRDERS
WITH UNSTIFFENED TOP FLANGES.

These defective constructions are noticeable in some of the older lines, but, fortunately, not often in the newer.

What the ultimate resistance of the pony-truss structure is no man can tell without testing it to destruction; but, in the opinion of most engineers, it is much less than it is assumed to be by those designing pony-truss bridges.

VIII.  EXCESS OF EXPANSION JOINTS.

Too many expansion joints in an elevated railroad are nearly as bad as too few. In the former case the metal is overstrained by the vibration induced by the lack of rigidity, while in the latter case it is overstrained by extreme variations of temperature. There are elevated roads in existence with expansion joints at every other bent, and there is at least one with them at every bent. For long spans there should be expansion provided at every third bent, and for short spans at every fourth bent.

IX.   RESTING LONGITUDINAL GIRDERS ON TOP OF CROSS-
GIRDERS WITHOUT RIVETING THEM EFFECTIVELY THERETO.

This is by no means an uncommon detail, especially in the older structures. It is conducive to vibration, and its only advantages are ease of erection and a cheapening of the work by avoiding field-riveting.

X.   CROSS-GIRDERS SUBJECTED TO HORIZONTAL BENDING BY
THRUST OF TRAINS.

The resistance that can be offered by a cross-girder to horizontal bending is very small; nevertheless, cross-girders are