DESIGNING. l8l 



form the floor of the balcony, which I consider a very important 

 feature of water-tower designs, as it enables convenient and 

 careful inspection to be made of the portion of the tank which 

 is most liable to failure, namely, the junction of the tank with 

 the supporting posts. In the case of the Fairhaven tank, if we 

 consider the horizontal radial components of the post thrusts 

 to be carried only by the lower flange of the circular girder, and 

 if this lower flange had been made continuous past the posts, 

 a bending moment of 331,000 inch-pounds and a thrust in the 

 line of the flange of 67,000 Ibs. would result immediately over 

 each post. Counting the lower flange as composed of two 3^X 

 3JXi-in. angles, and one 8Xi-in. cover-plate (see Engineering 

 News, Sept. 5, 1895), and also counting in 3^ ins. of the |-in. 

 web plate of the circular girder, the resulting stress on the outer 

 fibre would be 38,400 Ibs. compression at the post connection. 

 As built, the lower flange seems not to have been continuous, 

 and this bending moment would have to be carried by the rivets 

 of the connection between the segments of the circular girder 

 with each other and with the top of the posts. No data of the 

 connection have been published which would enable the resulting 

 stresses upon the rivets to be computed, but under any reasonable 

 assumption it seems probable that they must have all along 

 been stressed nearly to the breaking point whenever the tank 

 was entirely full of water. 



"As the formulas for the computation of the stresses given 

 above are not widely published, the writer here gives them, credit- 

 ing them to two students of the Civil Engineering Department 

 of Cornell University (Transactions of the Association of Civil 

 Engineers of Cornell University, 1896.) 



"At any point A in the circular hoop shown in Fig. 50 let the 

 bending moment from the pair of radial forces, P, be called M , 

 the thrust T, and the shear /. Then 



