The explosive is placed in a sheet-iron tube about 1½ in. in diameter and 4 ft. long, and suspended by cords in a pit, 11 ft. deep and 16 ft. in diameter. This pit was once used as the well of a gas tank, Fig. 2, Plate VIII. In adapting the pit to its new use, the tank was cut in two; the top half, inverted, was placed in the pit on a bed of saw-dust, and the space between the tank and the masonry walls of the pit was filled with saw-dust. The cover of the pit consists of heavy timbers framed together and overlaid by a 12-in. layer of concrete reinforced by six I-beams. Four straps extend over the top and down to eight “deadmen” planted about 8 ft. below the surface of the ground. The recording device, known as the Mettegang recorder, Fig. 2, Plate VII, comprises two sparking induction coils and a rapidly revolving metallic drum driven by a small motor, the periphery of the drum having a thin coating of lampblack. A vibration tachometer which will indicate any speed between 50 and 150 rev. per sec., is directly connected to the drum, so that any chance of error by slipping is eliminated. The wires leading to the primary coils of the sparking coils pass through the explosive a meter or more apart. Wires lead from the secondary coils to two platinum points placed a fraction of a millimeter from the periphery of the drum. A separate circuit is provided for the firing lines. In making a test, the separate cartridges, with the paper trimmed from the ends, are placed, end to end, in the sheet-iron tube; the drum is given the desired peripheral speed, and the charge is exploded. The usual length between the points in the tube is 1 m., and the time required for the detonation of a charge of that length is shown by the distance between the beginning of two rows of dots on the drum made by the sparks from the secondary coil circuits, the dots starting the instant the primary circuits are broken by the detonation. At one end of the drum are gear teeth, 1 mm. apart on centers, which can be made to engage a worm revolving a pointer in front of a dial graduated to hundredths; by means of this and a filar eyepiece, the distance between the start of the two rows of spark dots on the drum can be measured accurately to 0.01 mm. As the drum is 500 mm. in circumference, and its normal speed is 86 rev. per sec., it is theoretically possible to measure time to one four-millionth of a second, though with a cartridge 1 m. long, such refinement has not been found necessary.