OF ARTS AND SCIENCES : FEBRUARY 14, 1865. 419 



considering the axis to be the arc of a circle, the chord of which was 10 

 feet, the average versed sine was about 0.03 foot. In one pillar the 

 deviation was about 0.08 foot. These pillars were well-fitted with flat 

 ends, and loaded with less than one tenth of their computed breaking 

 weight. A small part of the observed curvatures must have been due 

 to the load. It may be necessary to admit a curvature when unloaded 

 of Tj^ij of the length as a maximum, which in ten feet would give a 

 versed sine of 0.033 foot. 



It is usual to make round pillars a little smaller at the upper end 

 than at the middle, frequently both ends are smaller ; this is advan- 

 tageous in strength with the same weight of iron, and also in appear- 

 ance ; the difference should not much exceed one tenth of the diameter. 

 If smaller at both ends than at the middle, the surface of the pillar 

 should be regularly curved from one end to the other ; when so made 

 a small deviation in the axis from a straight line is less observable. 



In estimating the amount of the greatest load to which a pillar may 

 be subjected, doubtful points should be taken on the safe side. In 

 warehouses and stores, it would not be safe to take the ordinary load 

 on the floors ; they are liable to be packed full of heavy goods. When 

 a warehouse or store is erected for a particular use, it will seldom be 

 safe to estimate the weight due to that particular use; it should be 

 made for the greatest weight due to any use to which such a building 

 may be applied in the ordinary course of business. In modern cotton- 

 mills with wooden floors I have found that the average weight, includ- 

 ing the floor itself, the machinery, shafting, stock in process of manu- 

 facture, &c., is about 50 pounds per square foot of floor, and a flat roof 

 when loaded with snow is liable to be as heavy. It is not safe, how- 

 ever, to take the average weight, as some pillars will have to support 

 more and some less than the average ; I think an addition of one fifth 

 should be made to the average weight for the possible excess. In 

 w^arehouses and stores, the weights to be provided for are usually much 

 greater than in factories. 



In pillars to support railway bridges, or other structures subject to 

 violent concussions or vibrations, a less proportion than one fifth of 

 the breaking weight should be allowed. In experiments on the deflec- 

 tions of railway bridges with a stationary load and with the same load 

 moving at a high velocity, a greater deflection has been observed in 

 the latter case, indicating a corresponding increase of the strain. The 

 increased strain will, however, be principally felt in the girders, and 



