328 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



The accompanying table (see Table I) shows the results found with 

 the six cylinders containing mercury. The pressure was increased 

 more slowly for the higher numbered cylinders ; with the four first the 



TABLE I. 



pressure was increased uniformly at the rate indicated, while with the 

 two others it was increased in discontinuous steps as shown. In mak- 

 ing the tests, pressure was first pushed to 1500 atmos. and kept there 

 for several minutes to make sure that there was no leak and every- 

 thing was in working order. The time given in the duration column 

 is exclusive of the time occupied by the preliminary application of 

 1500. It is evident that the data admit of no quantitative comparison 

 as the great discrepancy between tests 1 and 3 made under as nearly 

 as possible the same conditions shows. However, these two tests were 

 made with a high rate of increase of pressure. Those made with 

 a slower rate show much more consistent results. In general it 

 appears that the slower the rate of increase of pressure the lower the 

 bursting pressure, which apparently has as its lower limit about 3000 

 atmos. The data are not inconsistent with the view that there is a 

 critical pressure which will produce rupture if applied for an infinite 

 time ; pressures above this produce rupture in constantly less times. 



The location of the crack is an important consideration. The crack 

 in each cylinder is in an axial plane, extending the entire length of 

 the cylinder. It was located by measuring the angular distance from 

 a fiducial line marked the length of the bar from which the pieces were 



