170 MEMOIR OF THE LATE PROF. E. HODGKINSON, F.R.S. 



^^ The details of these experiments are found in the fol- 

 lowing Table : — 



No. of 



Eatio of the 



Area of whole 



Strength per 



Experi- 



section of 



section in 



square inch 



ment. 



the flanges. 



square inches. 



of section. 



I 



I — I 



2-82 



2368 



2 



I 2 



2-87 



2567 



3 



1—4 



3 '02 



2737 



4 



1—4-5 



3'37 



3183 



5 



1— 5*5 



5'o 



3346 



6 



I— 6-1 



6-4 



4075 



'^ In the first five experiments_, each beam broke by tear- 

 ing asunder of the lower flange. The distribution by which 

 both were about to yield together, that is, the strongest 

 distribution, was not therefore up to that period reached. 

 At length, however, in the last experiment, the beam yielded 

 by the compression of the upper flange. In this experi- 

 ment, therefore, the upper flange was the weakest ; in the 

 one before it, the lower flange was the weakest. For a 

 form between the two, therefore, the flanges were of equal 

 strength to resist extension and compression respectively, 

 and this was the strongest form of section. In this 

 strongest form, the lower flange had six times the material 

 of the upper. It is represented in the accompanying figure. 

 In the best form of cast- 

 iron beam or girder used 

 before these experiments, 

 there was never attained 

 a strength of more than 

 2885 lbs. per square inch of 

 section. There was there- 

 fore by this form a gain 

 of 1 190 lbs. per square 

 inch of the section, or of 

 two-fifths the strength of the beam/^ (See Moseley^s 

 ^Engineering and Architecture,^ art. 411.) 



