35 



over all the fibers at this section, which may b(^ citlicr a riglit or an oblicine section. The section 

 of failure, however, is seldom al the very end. The slij;litcst siairce of weakness may determine 

 its position, as a very small knot for e\am[)!c, for knots are a source of wmkness in compression 

 as well as in tension. 



Some tests are made on colniiins t(l inches lon.^ by I inches s(|nareon ihe larfe beam machine 

 but these usually fail tiie same as the short bh)cks, and not liy bcndiiijif sidewise. 



Compression across the urn in. 



Si)eeimens 4 inches square and 6 inches long are tested in compression across the grain. An 

 arbitrary limit of distortion, namely .3 ])er cent of the height, has been chosen as a reasonable 

 maximum allowable distortion in practice. This limit is indicated in the test l)y the ringing of an 

 electric bell, and the load then on the specimen is called the com])ressive strengtii across the grain. 

 The same limit has also been found to be very near the maximum load in lateral compression, 

 which is also determined. 



The shearing tests. 



Since timber fails by shearing, or splitting, oft ener liian any other way this test becomes a 

 very important one. The six'cimeu is taken 2 inches s(iuare and S inches long, and rectangular 

 holes mortised 1 iucli from each end, and at right angles to each other, as shown on Plate V. 

 The specimen is then i)ulled, in the Universal Machine, by means of snita])le stiirups ami keys, 

 as shown in the plate. The ends are kept from spreading or splitting by iiutting on small 

 clamps with just enough initial stress in them to hold them in place. After one end shears 

 out two auxiliary hoops or stirrups are used to connect the key wiiich slieared out to a pin put 

 through the hole at the center of the specimen as shown. The other end is then slieared, and two 

 results are obtained on i^lanes at right angles to each other. In this way the shearing strength is 

 determined on two planes at right angles to each other. 



Tests of full-sized column!:. 



No set of experimental tests of timber would be complete without numerous tests on full- 

 sized columns. This requires a machine of not less than l,(MM),(l(M» pounds capacity, cajiable of 

 crushing to failure columns from 12 to 14 inches square and at least 30 feet long. Such a machine 

 has been built exjiressly for this work, and is shown on Plate VI. It is capable of exerting a com- 

 pressive force of l,(MMt,()00 pounds on a length of -'{(i feet or less. The sides or tension members of 

 this machine are made of four long-leaf yellow pine sticks {Pinus palustris), trom Georgia, each 8 by 

 12 inches and 45 feet long. The power is applied by the same hydraulic pum]> which operates 

 both the large beam machine and the l(Mt,0()0-pound universal machine. The loads are weiglu'd 

 on this latter machine the same as for the beam tests. The plunger in the column machine 

 has Just ten times the area of that in the weighing machine, and hence the loads in the column 

 tests are just ten times those indicated on the weighing beam, with a sliglit correction for the 

 friction differential, which has not yet been deterniined. The tail block is of cast iron, resting in 

 a spherical socket, which is carried on a car, and which can be held by struts resting in slots 

 in the timber. The outer ends of these struts !<re kept from sineading by means of tiebars, as 

 shown, and the whole cond)ination can be moved forward or back so as to make the distance between 

 face i)lates any even number of feet from two to thirty-six. The spherical socket in the tail 

 block will i)ioduce an accurate adjustment of the end bearings at the beginning of the test, but 

 after the load is on it is tliought that this joint will remain rigid, the same as a solid block, esiiec- 

 ially if precautions arc^ taken to increase the frictional resistanc<' between these bearing surfaces. 

 This spherical socket is provided to eliminate the effects of unequal shrinkage in the side tind)ers, 

 or any uneipialcom])ressioniu the bearing sockets, and not toserv(>as a nmnd-end bearing for the 

 column. When long columns are tested a part of their weight will be supported by means of lines 

 ^nd pulleys, so as to make the test correspond to a vertical load in actual i)ractice. 



