occupied all of his spare time for a month and a half, representing altogether some 300 hours of steady figuring. As in the case of his paper on "The Possibilities in Bridge Construction by the Use of High-Alloy Steels," he did all of the computation work entirely unaided, checking the results himself, but relying for their correctness mainly upon the regularity of the platted curves.

As his data on weights of metal in cantilever bridges were primarily for double-track-railway structures, his first investigation was made for that class of bridges, using the live loads, impact, and specifications indicated in the two previously-mentioned papers. For convenience of comparison, he assumed Dr. Steinman's unit prices for metal in place, but for substructure estimating he adopted the method which he has employed for many years, viz., using a unit price for concrete above low water, another for the mass of the pneumatic caissons with their superimposed cribs below low water, another for the corresponding mass below the same in box cribs filled with concrete resting on piles, and a price per lineal foot for those portions of the said piles projecting below the bases of the cribs. These unit prices are as follows:

The unit prices for metal in place were as follows:

The costs of the railway tracks, the roadway pavements with their reinforced-concrete bases, and the reinforced-concrete sidewalks have been ignored when computing the total costs of structures, because they are common to the two classes of bridges compared.

In making the computations for this investigation, the author took the liberty of adopting several short cuts, such as assuming squared instead of rounded ends for all piers, using generally the method of "end areas" instead of that of the "prismoidal formula" when calculating volumes of masonry, carrying out quantities of materials and total costs to rather large limiting units, and estimating costs of certain parts by proportion from the previously-computed costs of similar parts of other structures. All these and many other short cuts for avoiding labor are perfectly legitimate when making comparative estimates, provided that they affect alike the compared types of construction, as they do in this case.

In the plotted curves of the accompanying diagrams no curve was