stresses were summed up for greatest tension and greatest compression on each piece, in order to determine by slide rule the live load stresses. The live load assumed for each track was the author's Class 60 loading.

In order to save time and labor, a constant percentage for impact was included in the live load itself instead of varying the percentage amounts to be added to the live-load stresses in the different web members. This approximation, of course, caused certain errors in web stresses; but their effects on the two contrasted types of structure were practically alike, and, therefore, did not affect the correctness of the comparison. The reactions for concentrated loads in the continuous-truss structure were obtained by the Theorem of Three Moments. No attempt was made to correct later the stresses thus found by the more exact method of least work; for the reactions obtained in that manner by the designers of the Sciotoville Bridge indicated that the difference in total weight of metal caused thereby was trifling.

The finding of the live-load stresses was a comparatively simple matter, but the determining of the dead-load stresses was much more arduous, because sometimes the correct distribution of the metal between the various panel-points was not ascertained until the third trial. No attention was paid to wind stresses; because, in double-track railway-bridges of long span and heavy live-loading, the excess intensities of working stresses allowed in modern bridge specifications for combinations of wind stresses and other stresses result in rendering wind stresses in the trusses entirely negligible.

After the live-load stresses and the dead-load stresses for both the continuous and the non-continuous spans had been computed, they were combined, and the maximum stress on each piece for both tension and compression was recorded. Then the sectional areas were determined by the specifications of Chapter LXXVIII of "Bridge Engineering," ignoring, however, all effects of reversion; after which the total weights of metal in main members were figured for both layouts, and to them were added the proper percentages to cover weights of details, thus giving the comparing weights of metal for the two types of structure under consideration. Much to the author's surprise, the weights thus found were so nearly alike that their difference amounted to a small portion of one per cent—so small, in fact, as to be negligible.

It had been intended to make an entirely new set of sectional areas and compute the resulting weights of metal for both types on the basis of caring for reversing stresses in accordance with the method provided in the before-mentioned specifications; but this was found to be unnecessary, because members in which reversion occurred were very few, and both the direct and the indirect effects thereof were readily determined. By "indirect" effect is meant in this case the increase in weight of metal due to augmentation of dead load caused by provision for reversal. Here again was a surprise, for the effects on the two types were exactly alike. These