330 PROCEEDINGS OF THE AMERICAN ACADEMY. 



maximum pressure reached by three of the cylinders without breaking 

 was 24000 atmos., over 8 times the lowest breaking pressure of the 

 cylinders filled with mercury, and nearly 2^ times the maximum 

 breaking pressure. Greater pressure than 24000 atmos. was not ap- 

 plied because the limit of the press was reached. Of course the stress 

 produced in a cyhnder by internal fluid pressure depends only on the 

 pressure in the fluid and not on the nature of the fluid. It must be, 

 then, that the mercury takes some special part in producing rupture 

 quite apart from the stress produced by it. 



The magnitude of the fluid pressures mentioned here requires brief 

 comment, because without a word of explanation it may seem so large 

 as to cast discredit on the accuracy of all the data. In the first place, 

 a steel of the kind described above with the indicated heat treatment 

 is the only steel known to the author that will stand the pressures used. 

 The best grades of carbon tool steels will not stand much more than 

 18000 atmos. As to the measurement of the pressure, similar apparatus 

 has been used up to 13000 atmos. and the pressure measured directly with 

 an absolute gauge with an accuracy of yV per cent. Absolute measure- 

 ments have not been made above this, and it is possible that the fi-iction 

 of the packing may become unexpectedly large, although no evidence 

 of this has been found. However, it makes no difference what the fric- 

 tion above 13000 is ; the fact stands incontestable, for the breakage 

 produced by mercury is in a region open to easy direct measurement, 

 while the other cylinders have stood a pressure certainly several-fold 

 greater. 



All these results so far were obtained with hardened chrome nickel 

 steel. Search was now made for the same effect in other steels. A 

 cylinder of nickel steel of similar dimensions to the above, but left soft, 

 was tried. This was filled with mercury. Pressure was kept at 8000 

 for 3 hours, then pushed gradually to 15000, where the increasing non- 

 elastic stretch became so great as to let the mercury past the packing. 

 The behavior here was exactly like that of a similar cylinder filled 

 with water and stretched beyond the elastic limit : the elastic limit in 

 the two cases was the same, as also the manner of yield and the shape 

 into which the cylinder was deformed. Apparently, then, mercury exerts 

 no selective action on the soft nickel steel. A similar cylinder of bes- 

 semer steel filled with mercury was left exposed to 3r)00 atmos. for 

 14 hours, and subsequently the pressure was increased until the mer- 

 cury blew past the packing, exactly as for similar cylinders filled with 

 water. It should be noted that non -elastic yield occurs for bessemer 

 steel at pressures much lower than 3.")00, perhaps as low as 2000 atmos. 

 The explanation finally adopted attaches some significance to the value 



