nickel steel, having an elastic limit of 55,000 lbs. per square inch and an ultimate strength of 90,000 lbs. per square inch.

The live loads employed were as follows:

For the floor-system,

Class 25 for the electric railway,

Class B for the roadways,

Class C for the sidewalks.

For the trusses a logical combination of these loads was adopted; and a ten-car subway train was assumed on each track when figuring the stiffening trusses of the wire-cable structure and the crescent trusses of the eye-bar-cable structure. When finding the moments and shears for the former, in order to give the wire-cable bridge the best possible chance to compete, the ends of its stiffening trusses were assumed to be anchored down, although, of course, not fixed, and the theory of stress determination adopted was the approximate method given in Johnson, Bryan, and Turneaure's "Modern Framed Structures," Part II, instead of the older method of Dr. Wm. H. Burr, which, for convenience and simplicity, was taken as standard by the author in writing Chapter XXVII of "Bridge Engineering." The results of the two theories do not differ greatly, especially for ends of trusses anchored, but the latter theory requires a little less metal.

In order to ascertain the weights of metal in the crescent trusses, there was assumed for each panel point a load of unity; and its effect was computed for every web member and every chord member thereof, thus rendering it easy to find all the live load stresses and dead load stresses by means of the slide-rule. These index stresses were checked by an independent computer.

After the first set of computations was completed, the question arose as to whether the assumption of a double-deck structure carrying a much greater proportion of electric-railway live load would have caused any material changes in the results; and it was decided to repeat the calculations for a set of five double-deck structures. The result indicated no serious disagreement, as is shown in Table 29a.

An analysis of this table shows how very little variation there is between single-deck and double-deck bridges in respect to the proportions of the various materials in wire-cable structures and the corresponding eye-bar-cable structures. For this reason in what follows the author, with a clear conscience, has drawn his general conclusions from computations on single-deck structures only, thus saving considerable time and expense. His so doing is a good illustration of the application of the "economics of bridgework" which he is endeavoring to expound.

The investigations made up to this point permitted the establishment of a number of formulae for quantities of materials involving variations in span length and loading, thus permitting of an extension of the study