A. An additional splice or two in the web, or else a slightly increased pound price for the large plates.

B. Larger outstanding legs for all stiffening angles.

C. Reduction in the number of cover plates.

D. Narrowing of flange angles and necessitating thereby either an additional bracing frame or an increase in sectional area of the compression flange, in order to compensate for the greater ratio of unsupported length to width.

E. Possible thickening of web because of its greater depth.

F. Possible encroachment on under-clearance in deck spnns, or raising of grade to avoid the same.

G. Possible difficulty in fabrication or shipment in case of long or heavy girders because of excessive depth.

Any one of these changes would be likely so to upset the economics of the case as to cause material decrease in the theoretical depth found by the preceding investigation. One will not often make an error in economy by following the old established rule in "De Pontibus" to the effect that the best practicable arrangement is generally to make the weight of the flanges equal to the weight of the web and its details; and there are occasionally cases where a saving of metal can be effected by making the web depth even smaller than that given by this old criterion, when by so doing a web splice may be avoided or smaller stiffening angles may be adopted. It should be borne in mind that there is quite a range in web depths over which the theoretic minimum weight is about constant, unless the thickness of the shallower web must be increased on account of the shear; hence one may often vary the dimensions of a plate-girder materially without affecting greatly the matter of economics. In Fig. 20a is given a diagram of economic depths of plate-girders with riveted end connections.

Concerning economic panel lengths for truss bridges, it is safe to make the following statement: Within the limit set by good judgment and one's inherent sense of fitness, the longer the panel the greater the economy of material in the superstructure. Of course, when one goes to such an extent as to use a thirty-foot panel in an ordinary single-track-railway bridge he exceeds the limits referred to, because the lateral diagonals become too long, and their inclination to the chords becomes too flat for rigidity. Again, an extremely long panel might sometimes cause the truss diagonals to have an unsightly appearance on account of their small inclination to the horizontal.

In plate-girder structures with floor-system of cross-girders and stringers, there is generally no economy in adopting long panels-in fact they are certain to involve an increase of total weight of metal; but, on the other hand, the cost of erection is probably lessened by reducing the number of field-driven rivets.

Warren trusses are cheaper than Pratt or Petit trusses for parallel chords, but not for those with a polygonal chord. The first-mentioned type