A somewhat similar case was that of a large horizontal boiler at Kidderminster, the tube of which collapsed in 1865, as shown in Fig. 76. It was heated by four furnaces, one of which worked into the tube, one under the bottom, and one on each side; and all the furnaces worked into the same end of the boiler. The tube was found to have partially collapsed at that end, and the top had dropped 11 inches. This was repaired in the first instance, but was afterwards again found injured by overheating, although not so seriously. It is very probable that the extremely rapid ebullition from the sides and bottom, from which the steam had to pass up the narrow space between the tube and the shell, produced such a foaming that very little solid water could reach the top of the tube where it was exposed to extreme heat. [Illustration: _Fig. 75._] [Illustration: _Fig. 76._] Many explosions have been attributed to deterioration of the iron through long use, as in an explosion at Durham in 1864, and another at Haswell, near Sunderland, 1865, where the boilers had worked constantly for 25 and 30 years respectively. When an explosion arises from the failure of a plate which has not been properly welded in rolling, there is no question that it was unsound when put in, and escaped notice; but when the plate that fails is found to be brittle and of bad iron, the fault is rather attributed to the effect of working than to original bad quality. Of course this is not always the case, as the injury done to plates by overheating has been already explained. Pieces of plate have in some cases been erroneously pronounced to be deteriorated by work, which have been taken from situations in the boilers where they were not exposed to any action of fire that could cause overheating; and therefore in reality the injury could only have taken place when the boiler was being made, by burning the iron in bending it to the required shape. A frequent cause of fatal injury to boilers is injudicious repair, whereby the crossing of the seams is destroyed, as in the explosion at Wolverhampton in 1865, previously referred to and shown in Fig. 36. Moreover the edges of the old plates, already tried by the first rivetting and the subsequent cutting out of the rivets, are frequently strained again by the use of the drift to draw them up to the strong new plates; and many a seam rip is thus started which ultimately causes explosion.