338 Scientific Proceedings, Royal Dublin Society. 



irregular shape, of greater accuracy than can be obtained by calcu- 

 lation, unless great detail is gone into. Thus, in calculating I for 

 an zn. plate girder, the ordinary approximate method (in which a 

 sixth of the web is added to the flanges) gave for the moment of 



inertia 



I = 806-1. 



A very carefully detailed calculation made 



I = 852-0, 



showing that the approximate method erred by 5-4 per cent. My 

 experimental method gave with a wooden model 



I = 856-9, 



which is only 0-57 per cent. out. This is one of several examples 

 which show that the experimental method here recommended gives 

 in complicated cases much closer results than the methods of calcu- 

 lation which are practically used by engineers. 



It should be observed that the wedge need not be one which 

 comes to an edge, but may be of a truncated form. In all cases 

 the line m n, from which Y and y are to be measured, is to be the 

 intersection of the faces of the wedge, and in this case lies outside 

 the pillar, instead of being a tangent to it. However, the nearer 

 to the pillar that it can be conveniently placed the greater will be 

 the accuracy of the determination. Of course the surfaces of the 

 wedge must be flat, and the model of the pillar out of which it is 

 made of uniform section and density, m n may have two positions, 

 since two lines can be drawn, touching the pillar, and parallel to 

 the axis round which the moment of inertia is wanted. It is 

 advisable to place it at the side of the pillar which is the broadest, 

 if there is a difference in this respect. 



