640 PROCEEDINGS OF SECTION H. 



(4) The failure of column No. 1 would suggest the usefulness, 

 in certain cases, of bracing diagonally opposite longitudinal column 

 components together. 



(5) The columns tested were placed horizontally in the testing 

 machine. It might, therefore, be expected that their own weights 

 would cause slight initial downward bending, great enough to 

 influence the direction in which final collapse would occur. But, 

 •curiously enough, none of the columns bent downwards at collapse ; 

 most of them bent upwards, and the remainder bent sideways. 



(6) The following are a few quotations from the paper by 

 Talbot and Moore, referred to in the foregoing notes. (University 

 of Illinois Bulletin, No. 44 " An Investigation of Built-up 

 Columns under load." Also Proc. A.M. Soc, C.E., June, 1909.) 



P. 63. — " It seems futile to attempt to determine the stresses 

 which may be expected in column lacing for central loading by 

 analysis based on theoretical considerations or on data now 

 available." 



" No relation has been found between the stresses actually 

 observed and the stresses computed by column formulae. The 

 stresses do not increase towards the middle of the length of the 

 column, as may be expected from the Rankine form of analysis, 

 but are quite irregular in their location and distribution." 



P. 61. — " For the strength of the component angle, channel, 

 ■or other structural shape used in a built-up compression piece, 

 many engineers have been satisfied with the provision that the 

 slenderness ratio " {i.e., the ratio, length to least radius of gyration) 

 "of the component member shall be less for the length between 

 the points of attachment of lacing than the slenderness ratio for 

 the column as a whole, and have given little attention to the possible 

 non-integrity of the section or to the probable effect of imper- 

 fections of manufacture. Fortunately, the large influence of the 

 slenderness ratio in column formulae has given sections with which 

 failures have not occurred. Whether a column formula should 

 include a factor depending on the form of the section and the 

 relative thickness of the metal, or whether the allowable stresses 

 for any form of column should be based on experimental data for 

 the section used, will depend on future developments." 



P. 60. — " Within the critical length at which Euler's formula 

 governs, the general flexure of the column as a whole under load 

 has less influence upon the strength of the column than is ordi- 

 narily assigned to it, and therefore the influence of — (length to 



least radius of gyr.) is not as great as is represented in the usual 

 ■column formula." 



(7) A few small alterations and one or two additions have 

 been made to the text since the original paper was read. 



