86 BULLETIN 1216, IT. S. DEPARTMENT OE AGRICULTURE. 
12. The elastic limit should be determined with great care, 
but any inaccuracy will cause less proportionate error than in 
the case of the modulus. The yield point, being less well de- 
fined, can not be so closely determined, and it is believed that 
in most cases the use of dividers instead of an extensometer 
will give sufficiently accurate results. 
18. It is considered undesirable in accurate determinations of 
the modulus of elasticity to use a shorter gauge length than 8 
inches. It is evident that the greater the total elongation 
measured the less will be the error due to inaccuracy of the 
reading, and the accuracy thus appears to increase directly as 
the gauge length. 
14. That the difference between short and long gauge lengths 
has a greater influence in affecting results than other factors 
(personal error, inaccuracy of the testing machine, etc.) is 
shown by the closely agreeing readings obtained with the greater 
lengths. 
15. The effect of improper methods of holding specimens can 
not be established from the results of actual tests. The result 
of improper methods of gripping materials of low elongation, 
such as cast iron, is well known, and it is probable that in 
material of a more ductile nature the effect is largely local and 
does not extend to the portion of the specimen within the gauge 
marks. 
Conditions to Insure Correct Testing Machines. 
16. It is recommended that in machines on which specimen 
tests are made, whether the power be applied hydraulic-ally or 
by means of screws and gears, the load be measured by a sep- 
arate system of levers and knife-edges, or by a method similar 
to that employed in the Emery testing machines. 
17. All knife-edges shall be kept sharp, and free from oil and 
dirt, and the machine shall be sensitive to a variation in load 
of one two-hundred-and-fiftieth of the load carried. Design and 
workmanship on testing machines shall be good, and they shall 
be calibrated at least once every six months by the following 
method. 
Calibration of Testing Machines. 
18. Test for accuracy by loading the weighing table with 
standard weights and compare the actual weight at each addi- 
tion with the reading of the beam. If the table is uniformly 
loaded in this manner with the full amount of weights that it 
will accommodate, the proportionality of the levers and the 
weighing beam can be successfully established. This relation, 
in a properly designed machine, will remain constant for all 
loads, but as a further test for sensitiveness under greater loads 
than can be accommodated in this manner, the following pro- 
cedure is recommended : 
19. Place in the machine a tension bar of such cross section 
that the maximum capacity will not stress it to the elastic limit. 
Stress this bar to various extents through the full range of the 
machine and at each load balance the beam and place upon the 
weighing table standard weights of 100 pounds. A weight one 
two-hundred-and-fiftieth of the total load on the machine should 
produce a readable movement of the beam. 
20. Where evidence of the accuracy of the machine over its 
whole range is desired, a known load may be applied by means 
of an extensometer and calibrated bar whose modulus of elas- 
ticity has been determined with exactness. 
21. It is recommended that a device be adopted conforming 
to the following reqiurements. in which the extensometer and 
bar are permanently attached to each other: 
( a ) The bar should be of high-elastic-limit material, and of 
such cross section that this limit will be well above the total 
capacity of the machine on which it is to be used. 
