of the main piers, and saves placing the tops of the trusses at an abnormal height above the water.

G. If there be any choice between the riveted and the pin-connected types of construction for any cantilever bridge, it is generally better to adopt the latter; because, as cantilever bridges are usually employed for long spans only, pin-connected work is the more suitable. Again, it is a little lighter than riveted work; and, therefore, the dead load on the structure would be somewhat less. On the other hand, the riveted construction is so much more rigid than the pin-connected that it is preferable to adopt it whenever the conditions permit; besides, in the riveted work it is not necessary to stiffen any truss members for erection, although it might be obligatory to increase a few of their sectional areas.

H. Very large compression members should be made of box section so as to do away with latticing. This not only effects an improvement in the design, but also saves some metal, although the details required at the panel points to distribute the stresses from the cut cover-plates tend to offset the saving in weight of lattice bars and stay plates.

Professors Merriman and Jacoby present in their "Roofs and Bridges," Part IV, an excellent treatment of the subject of cantilever bridges, discussed mainly from the theoretical point of view. Their economic investigations, which are based upon chord weights only, show that for a three-span cantilever of Type A, each anchor arm should be about twenty-one and two-tenths (21.2) per cent of the total length of structure. This is quite a close agreement with the twenty (20) per cent minimum found by the more accurate and practical investigation that was made for "De Pontbus." The professors find, though, thirty-nine and four-tenths (39.4) per cent of the total length of structure for the economic length of the suspended span, corresponding to about sixty-eight per cent of that of the main opening, while the "De Pontibus" investigation made it only thirty-seven and a half (37.5) per cent thereof. Actual experience has repeatedly shown that the economic length of the suspended span is from three-eighths (3/8) to one-half (1/2) of the main opening, hence the professors' figures for this portion of their work have the appearance of being incorrect; but Prof. Merriman has explained to the author by letter that he assumed the truss depth to be the same throughout the entire structure. This assumption, combined with that of ignoring the effect of the weight of the web, will account for the large discrepancy; because the professors' mathematics have been checked and found to be faultless. As a matter of fact, though, no American engineer would think for an instant of making the truss depth constant throughout the structure, because for economic reasons it should generally be about twice as great over the main piers as in the suspended span. European engineers, however, often fail to make the truss depths, especially in the cantilever and anchor arms, great enough for economy.