88 Kl'LLETIX 1216, U. S. DEPARTMENT OF AGRICULTURE. 
(b) The specimen shall he placed in the exact center of the 
heads and be secured in some positive manner so that slip and 
side bending stresses do not occur. 
(c) The exteusometer should be of a type to measure the 
elongation on two or more sides of the specimen. 
(d) It should read to 0.0001 inch or less. 
(e) It should be of such a design that no change of zero will 
occur upon release of the load in determining the real elastic 
limit. 
(f) The load shall be applied so slowly that simultaneous 
readings of elongation and load can be obtained with certainty. 
(g) The testing machine shall have previously been calibrated 
for accuracy and sensitiveness, and heads lined up and made 
parallel. 
II. METHODS FOR COMPRESSION TESTS OF METALS. 
Definition of Terms. 
1. Elastic limit is the greatest load per unit of original cross 
section which does not produce a permanent set. 
Note : — This determination is rarely made in the commercial test- 
ing of materials. 
Proportional limit is the load per unit of original cross section 
at which the deformations cease to be directly proportional to the 
loads. 
Note : — This determination is rarely made in the commercial test- 
ing of materials. 
Yield point is the load per unit of original cross section at 
which a marked increase in the deformation of the specimen oc- 
curs without increase of load. It is usually determined by the 
drop of the beam of the testing machine, or by the use of dividers. 
2. The test specimen shall be a cylinder having plane ends 
truly normal to its axis. 
Note — Only two replies from testing laboratories mention- cubes. 
A cylindrical specimen will usually be cheaper to prepare than a 
cube. The stresses are probably less uniformly distributed over a 
square than over a circular section, owing to the influence of the 
corners, this being especially the case with the internal shearing 
stresses which accompany the compression. 
3. The diameter of the specimen shall be not less than 1 inch 
nor greater than 1.18 inches. A specimen 1 inch in diameter is 
to be preferred. 
Note — The range of diameter mentioned in the replies from test- 
ing laboratories is from 1 inch to 1.129 inches. A diameter of 
1.1284 inches gives a sectional area of 1 square inch. 
4. The length of the specimen should be between 2.5 and 4 
diameters. 
Note. — Two testing laboratories use a length of 1 diameter, one a 
length of from 1.5 to 2 diameters, one a length of 2.6 diameters, 
and one a length of 10.5 diameters. It is believed that a length 
less than 2.5 diameters is not sufficient for the internal shear to be 
properly developed, and that such short lengths give a fictitious 
strength owing to the friction of the bearing plates of the machine, 
which causes the specimen to assume a barrel-like form. 
5. No bedding should be used for the ends of the specimen. 
Note. — Only one reply favors bedding. It is known by general 
experience that bedding modifies the breaking load and that dif- 
ferent kinds of bedding have different influences. 
6. The bearing blocks which transmit the pressure from the 
testing machine should be truly normal to the plane ends of the 
specimen. To secure this, one of the blocks should be provided 
with a hemispherical bearing which can turn freely. 
Note. — These requirements seem essential in order that the load 
may not be eccentrically applied to the specimen, and are generally 
recommended in the replies from testing laboratories. 
