slide-rule from the formulae given in either Chapter XIV or Chapter XVI, as the case may be, or from the corresponding tables at the end of the book, and are written on a separate diagram.

Next are computed all the wind-stresses which could possibly affect the sizes of the sections of main-truss members, and these are recorded either on a separate diagram or on one of those already prepared, in the latter case care being taken to indicate that each such stress is marked as a wind-load stress.

Next the various combinations of all stresses are made and recorded on a new diagram, after which the required sections of all main members are figured according to the specifications, and are recorded on the same diagram; then the actual sections are proportioned and recorded there also.

The exact lengths of all members, including camber allowances, are next figured and recorded on the last-mentioned diagram.

Next the weight of metal in the trusses is estimated. For preliminary estimates, the weights of details are percentaged from recorded results of previous similar estimates; but if the structure be of an unusual type or size, the details are sketched and their weights are computed.

Next the total weight of metal in the structure is figured, and the dead load is checked. If it does not agree with that assumed within the limit of error set in the specifications, a new dead load is assumed, and the entire computations of total stresses, sections, and truss weights are made anew. It is very seldom, however, that it is necessary to make these calculations more than once, owing to the great mass of accumulated data concerning weights of metal in all kinds of bridges.

In making any set of calculations the computer should check back on his work at short intervals, so as to see that no error has been made, because the effects of such errors often extend over all succeeding computations.

In determining stresses graphically, the frame-diagram should be laid out on as large a scale as is convenient, and the load diagram should be made as small as practicable;  for the large