The form and details of the expanding mandrel which was finally designed to meet the conditions, and proved so satisfactory in every way, are shown by Fig. 15, _C_. The mandrel consisted of two triangular pieces of hard pine, separated by wedges attached to one piece which fitted into slots in the other; these, when expanded, practically filled the whole of the inside of the ducts. One of these mandrels was placed in each line of single ducts and two in each 4-way duct, placed diagonally, as shown in Fig. 1, Plate XXV. This required 60 mandrels at each working point, or 240 for the whole work. The mandrels were 35 ft. long, so that they easily covered the whole of a 25-ft. section, projected sufficiently far back into the previously finished work to assure the continuity of the alignment, and allowed the ends to be racked out at the forward end to secure proper breaks between the joints. In laying the single conduits, as a rule, the (collapsed) mandrels were pulled ahead from the previous section as each line was laid, and the conduits were strung on it until the whole length was completed; the conduits were then pushed up tight together, so as to close the joints as tightly as possible, and then the mandrel was expanded. The conduits were thus held firmly in position, and the forward end of the line was lifted slightly so that the wraps could be placed around the joints. The 4-way conduits were generally laid in the ordinary way, except that no laying mandrel was necessary. One dowel was used between each of the pieces of conduit, at the center, and the joints were wrapped. When a line was finished, two mandrels were placed diagonally in each line and expanded simultaneously, so that any inequalities in the ducts themselves were divided as far as possible. In connection with the use of these mandrels, one of the points which was most carefully watched was that they projected back into the last completed section, thus insuring the continuity of the alignment. It was originally intended to wrap the joints of the 4-way ducts only, but it was found to be impractical to keep the grout from the wet concrete entirely out of the single ducts, and, after a short trial, it was decided to wrap these also. The expanding mandrel kept out a great deal of the cement, and, in the sections laid without wraps, the only difficulty from this cause seemed to be that a slight film of grout, from 1/16 to ⅛ in. thick, was deposited on the bottom of the inside of the ducts at some places, and although this was not considered a serious defect, it was thought that the slight extra cost of placing the wraps would undoubtedly be justified by the practically perfect results obtained by using them.