ANCIENT AND MODERN WORKS. 2$ 



the Elgin pipe a thickness of 6 inches or so near the bottom and 

 30 inches or more in the upper section within 30 to 35 feet range 

 of daily fluctuation of water-line. It is commonly the case that 

 the inlet of warm water melts away and prevents the formation 

 of ice shell some distance from the base of stand-pipes, the extern 

 of this action depending chiefly upon the temperature of the 

 water in the mains. Since this had been noted as about 32 

 previous to the accident, this process of melting the ice must 

 have been very slight in the Elgin pipe. Under these conditions 

 the circulation of water by convection must have been insignificant. 

 The increase in thickness of the ice walls from the base 

 upward was manifestly due to the increased exposure toward 

 the top. The ice was originally moulded close against the plates 

 and rivets. It was evident, however, that a film of water had 

 formed between the ice and metal shell, probably by action of 

 the sun and warm winds, for some days before the accident, 

 for the impressions of the rivet heads and joints, while perfectly 

 distinct, were not sharply defined. This initial thaw is further 

 evidenced by the thin layer of fresh ice, which clung with surpris- 

 ing adhesion to the inner surface of the plates of the top section 

 notwithstanding its tremendous impact with the frozen ground. 

 Although the J-in. film of new ice held thus tenaciously to the 

 metal sheets, no connection could be traced between the fresh 

 film and the fragments of the older ice. A careful examination 

 showed the top of the ice mass in the upper section just even 

 with the upper edge of the top ring of the plates and the imprints 

 of the rivet-heads in the ice near the top continuously matching 

 the rivets themselves, showing that the ice tube had not shifted 

 longitudinally in the metal shell. Since the ice-level was below 

 the point of buoyancy, it is certain that the ice mass was supported 

 top to top with the stand-pipe either by the continuity of the 

 ice shell to the bed-plate, or by a frozen connection between 

 the ice and the plates, or perhaps both. In any event, it seemed 

 absolutely certain that the main bulk of the ice moulded more 



