414 PROCEEDINGS OF THE AMERICAN ACADEMY 



In the experiments on pillars with flat ends, much care was taken to 

 have the weight uniformly distributed over the whole area of the metal 

 at the ends ; this is not an easy matter to accomplish in so unyielding 

 a material as cast-iron ; it can be most readily done by interposing a 

 plate of more ductile metal, such as copper, between the two surfaces 

 of cast-iron. Although this was not done in testing the experimental 

 pillars, there can be no doubt but that they were much better fitted in 

 this respect than can be expected in ordinary practice. 



The experimental pillars were tested in a rigid iron frame, which 

 must have secured them from sensible vibration, except for the mo- 

 ments when the weights were applied. In many structures in which 

 cast-iron pillars are used, such as factories, they are pretty constantly 

 subject to vibration ; what effect this has in weakening the pillar, it is 

 not easy to estimate ; the additional strain tending to produce fracture 

 by lateral flexure must be in some proportion to the amplitude of the 

 vibrations ; in pillars of ordinary length and size, such as are com- 

 monly used in factories, the amplitudes of the vibrations are very 

 small; although they may be very sensible to the touch, they are 

 usually too small to be detected by the eye ; this being the case, it 

 would appear that, ordinarily, the additional strain due to the vibration 

 is very small. It has been a common opinion that long-continued vi- 

 bration, although very small in amplitude, causes a change in the 

 structure of iron, rendering it more liable to break ; of late, however, 

 this notion appears to have been abandoned by those best informed. 



In testing the experimental pillars, sufficient weight was applied to 

 break the pillar in a short time, and it is certain that a somewhat less 

 weight would have produced fracture in each case, if applied during a 

 sufficiently long period of time. Fairbairn made some experiments on 

 this point.* He loaded a pillar with 97 per cent of the weight which 

 had previously broken another pillar of the same dimensions ; it bore 

 this weight between five and six months, and then broke. In another 

 pillar, loaded with 75 per cent of the weight which had broken another 

 of the same dimensions, at the end of three years its lateral deflec- 

 tion was still increasing slightly ; this does not indicate, however, that 

 it would ultimately fail, as we find the deflection of bars of cast-iron 

 subject to a lateral strain increasing for a much longer time, although 

 loaded with from 70 to 96 per cent of the weight which broke another 



* Tiedgold OQ Cast-iron. ' Part II., 4th London edition. 



