638 PROCEEDINGS OF SECTION H. 



In the majority of cases the procedure followed in the prepara- 

 tion of a formula is to find a value for a bending moment and 

 calculate the corresponding shear, which is, of course, a maximum 

 at the ends and zero at the middle of the strut. Now, as Carl 

 Jensen points out, lacing designed on these principles might 

 prove dangerously weak, for the shear may be quite large at the 

 middle of a column's length ; for example, when the centres of 

 pressure on the two ends of a column are on opposite sides of 

 the column's geometrical axis, then the column will tend to fail 

 by bending in an S curve, causing heavy shear at mid-length. 



Burr's large model column failed in this manner. Instead, 

 then, of basing our calculations upon some assumed deflection 

 of a column as a whole — i.e., a bending such that the maximum 

 deflection is at the middle, we should remember that it may bend 

 in an S curve ; and, if for purposes of approximation, we employ 

 one of the many formulas, deduced as above-mentioned, we should 

 certainly not reduce the dimensions of the lacing near the middle 

 of the column, but maintain throughout the dimensions found 

 for the lacing at the ends. 



But, rather than trust to any formula or to any factor of 

 safety, which may be guessed to be sufficiently large to cover 

 unknown possibilities, the wTiter would recommend that those 

 concerned with the design of important compression members 

 should ascertain experimentally what would be safe dimensions 

 to employ for the lacing components of their columns. 



The results of a series of such tests might be used to base 

 designs upon in much the same way that the results of tests of 

 timber columns are employed. 



Addendum. 



Since writing the foregoing notes the writer has had the beneht 

 of the opinions of those who took part in the vigorous and able 

 discussion which followed the readmg of the paper. Prompted 

 by the remarks made by engineers havmg different points of view, 

 he thinks that the following facts and explanations may add to 

 the clearness of the original paper. 



(1) All the L bars employed for longitudinals, heads of 

 columns and lacing had sides of equal width. The columns were 

 thus not unlike those used in poppet-heads for important mines. 



(2) The ratio of length to least radius of gyration was, in the 



48 

 case of the complete column -— =22 to 1. The correspondmg 



ratio for the portion of each longitudinal member of a column, 

 contained between adjacent lacing points was, for columns Nos. 

 1 to 6, about 55 to 1. 



In columns 2A, 3A and 4A, this ratio was 28 to 1. 



