Wooden rods were used at first, but proved entirely too light, as the mandrels used were a close fit, and it required considerable effort to push them through 400 ft. of conduit. Iron pipe with ordinary couplings was next tried, but the couplings broke quite often, as the threads became worn in uncoupling the sections to move the rods from one line to another, and the break was generally inside a duct line. The long couplings were finally adopted, and a set of rods was put in each line, that is, six sets in all, so that when coupled up they remained in the line until it was finished. The expense of the extra quantity of pipe thus required was more than offset by the decreased labor cost. It was thought necessary at first to run a cutter, Fig. 15, _E_, through the conduits ahead of the final rodding mandrel, but this was soon found to be unnecessary except in a very few instances, and, after a short experience, the cutter was only used at places where an obstruction was encountered by the mandrel. At such times as the pipe became uncoupled inside the duct line, the part remaining inside was recovered by the use of the tool shown at _D_, Fig. 15, called a “weasel.” In two instances, the mandrel became stuck in such a manner that the duct line had to be cut into in order to take it out. The best day’s work of the rodding gang (1 foreman and 4 men) was 20,400 duct ft. of the 4-way conduit in the telegraph and telephone line, and 19,200 duct ft. of single conduit on the low-tension line, an average day’s work under ordinary conditions being about 10,000 duct ft. The cost, including labor, material, and all tools, for rodding for the whole work was slightly less than 0.2 cent per duct ft. The average cost of the single conduit was about 0.25 cents per ft., and of the 4-way, 0.15 cents per ft. About 10% of the conduit lines were rodded twice, owing to partial sections having been rodded once before completion. The best continuous work on rodding was done between October 22d and 29th, 1908, when in 7 working days, 105,600 duct ft. were rodded, an average of a little more than 15,000 ft. per day. _Bench-walls._--The original design for the tunnels provided for the construction of a brick arch above a point 22° above the springing line, that is, the part above the side-walls (Fig. 10). It was thought desirable, therefore, in designing the bench-wall forms, to provide for placing the concrete in the side-walls and bench-walls at one operation. These forms, as first designed, are shown by Fig. 2, Plate XXV, and the details in Fig. 16, _A_ and _A’_; they were built of steel, the facing plates being 5/16 in. thick, in pieces 4 ft. 6 in. wide, and in length about 6 in. more than the height of the bench-wall.