anchor-arm of a three-span cantilever will generally be a little greater than twenty per cent of the total distance between centres of anchorages.

When, however, the problem is to determine the economic length of anchor-arm for a fixed distance between main piers, the result will be quite different; because, within reasonable limits, the shorter the anchor-arm the smaller will be its total weight of metal, and because trestle approach is much less expensive than anchor-arm. It would not, for evident reasons, be advisable to make the length of anchor-arm less than twenty per cent of that of the main opening, or say fifteen per cent of the total distance between centres of anchorages. With this length there would probably be no reversion of stress in the chords of the anchor-arm, even when impact is considered. Generally, though, the appearance of the structure will be improved by using longer anchor-arms than the inferior limit just suggested.

In respect to the best proportionate length of anchorspans, the latter weigh so much per lineal foot for all cases that the shorter they are made the greater the economy; but, as before stated, it is improbable that navigation interests would ever permit of their being made shorter than one half of the main openings.

In respect to his fifth and last problem, Mr. Hedrick obtained the following results:

The total weight of metal in a three-span cantilever railroad bridge, floor system included, is to the total weight of metal in a simple-span bridge of three equal openings, for which false work is to be used throughout, as unity is to 0.6. The corresponding ratio for the case of the centre span, erected without false work, is unity to 0.64.

For a very long bridge, composed of a succession of cantilevers and anchor-spans which are one half as long as the main openings, and which has a suspended span resting on each extreme pier, the ratio of weight of metal to that in a corresponding bridge of equal, simple spans and the same number of piers, the spans being erected on false work, is as unity to 0.75.  For  the  case  of  alternate  simple  spans erected  without  false