While it is undoubtedly difficult to procure a tight fit for stiffeners on rolled I-beams, it does not appear to the author safe to omit them at the ends of railway girders that are supported from beneath, or even from those used for carrying heavy highway loads to the masonry, because the unsupported webs are not of a shape satisfactorily to resist severe pounding.

Although it is true that turning the flanges of channels in makes the riveting somewhat more difficult, it need not prevent the use of power riveters, except in the case of a few rivets; while it facilitates greatly the detailing by bringing all the webs of main truss members in the same plane for the attachment of the gusset plates. Most of the author's riveted bridges are built in this way.

Again, it is important to have the batten plates inside of the gussets, and to carry them to near the ends of the members, both of which conditions the turned-in channels permit. Moreover, they generally involve an economy of weight of metal for lacing and battens. But one of the most important advantages of turned-in channels is that they permit the ends of the floor-beams to fit closely to the bottom chords without cutting either the chord or the beam, which is not practicable if the flanges of the bottom chords turn out.

In viaduct construction some manufacturers use their influence to have the girder-depth the same from end to end of structure, which is uneconomic of material, because the tower spans and the intermediate spans are nearly always of different lengths, which arrangement would call for different depths unless metal is to be wasted. The manufacturers' claim has apparently some justification if the tops of the columns are cut off so as to let the main girders be supported directly thereon; but, in the author's practice, the columns are carried up to the level of the deck, and both the longitudinal girders and the tower cross-girders abut into them; hence there is no valid objection to making the comparatively-short longitudinal-girders over the towers shallower than the long, intermediate longitudinal-girders. This layout certainly looks much better; and the corner brackets afford an excellent connection for the diagonals of the longitudinal bracing.

If the designing of details be left to the manufacturers of the metalwork, they often place the end or pedestal pin of a riveted-truss span below the bottom chord, forgetting that the thrust of a braked train, acting with a lever arm equal to the vertical distance between the center of the chord and the center of the pin, produces a large bending moment that has to be resisted by the stiffness of the bottom chord and that of the inclined end post.

When bridge superstructures are let to the manufacturers by the lump sum and they have the designing to do, they like to omit the end floor beams of through bridges, substitute instead inexpensive struts, and rest the stringers on the masonry; but the author believes that invariably end