OF ARTS AND SCIENCES : FEBRUARY 14, 1865. 417 



^ to be sufficient ; Mr. Rastrick, Mr, P. W. Barlow, Mr. R. Stephen- 

 son, and Mr. Joseph Cubitt adopt ^; Mr. Hawkshaw prefers y, ex- 

 cept in cases where great care is exercised in the selection of materials 

 and workmanship, when a smaller proportion would suffice ; and Mr. 

 Gljnn considers that, in structures exposed to concussion and vibra- 

 tion, the ultimate strength of a girder should be 10 times the greatest 

 load." 



Defects in material and Avorkmanship are as likely to occur in girders 

 as in pillars, and, with the exception of defective alignment, the defects 

 in girders must be at least as injurious, and require as large an allow- 

 ance of surplus strength to provide for them as is required to afford the 

 same degree of security in pillars. In determining the proper allow- 

 ance to be made for security, regard should be had to the consequences 

 of failure. In most of the large modern buildings erected for com- 

 mercial and industrial purposes, cast-iron pillars are extensively used 

 for supporting the floors, and in many cases the walls of the upper 

 stories are supported in the same manner. The failure of the pillars, 

 in such cases, involves such disastrous consequences, that no economical 

 considerations would warrant the use of pillars of insufficient or doubt- 

 ful strength. 



It is possible that a single pillai*, although ample to support the 

 superincumbent weight, may be broken by a lateral blow ; in such an 

 event, the neighboring pillars will usually have a large additional 

 weight thrown upon them, and if they have not considerable surplus 

 strength, the whole structure may be involved in ruin. 



These considerations apply to girders as well as pillars, and no doubt 

 they were taken into accpunt by the experienced men whose opinions 

 have been quoted. 



It is very obvious that a large allowance of surpliis strength must 

 be provided to insure the complete security of a building depending 

 for support on cast-iron pillars ; the various considerations which have 

 been offered do not, however, afford any definite data for determining 

 the amount of this allowance. The only safe course is to rely upon 

 the practice and opinions of experienced engineers and constructors. 

 Taking all things into account it appears to me that, in pillars made 

 with ordinary care to support walls and floors, to insure complete 

 security without using more material than is necessary, the greatest 

 load should be about one fifth of the computed breaking weight ; and 

 I have adopted this as the rule in computing the following tables. 



