410 REPORTS ON THE STATE OF SCIENCE, ETC. 
Test ON 1-INCH DIAMETER PULLEY. 
Tension on wire, lb. . 10 15 30 35 40 
Number of bends to 
fracture j 2662 2630 2240 34 148 
142 62 
TEST ON 2-INCH DIAMETER PULLEY. 
Tension on wire,lb. . 10 20 30 40 50 
Number of bends to 
fracture ; AP EEE 8526 10,304 5733 61 
12,009 8269 
The next stage took the pulley diameter up to the theoretical minimum, and 
here the number of bends to fracture was not greatly reduced at pulls equal to or 
greater than the yield-point tension of the wire. For éxample: 
TrEstT ON 3-INCH DIAMETER PULLEY. 
Tension on wire, lb. . 10 20 30 40 50 
Number of bends to 
fracture . . 34,437 32,595 35,452 39,742 29,127 
Test ON 4-INcH DIAMETER PULLEY. 
Tension on wire, lb. . 15 30 40 50 60 
Number of bends to 
fracture ‘ . 67,348 63,737 105,172 78,050 53,627 
59,826 91,600 63,315 
When the pulley diameter slightly exceeded the critical value, 4-4 in., the un- 
expected result was obtained that the number of bends necessary to fracture the 
wire increased with the tension to which it was subjected. The following furnishes 
an example of this character : 
TrEst ON 6-INCH DIAMETER PULLEY. 
Tension on wire, lb. . 10 20 30 40 50 
Number of bends to | 
fracture : 107,359 163,232 214,747 268,693 468,351 
On still larger pulleys this wire was not in general broken after a million cycles ; 
in all cases the endurance was very considerable, and it appeared that the bending 
stresses had been so far reduced that they were negligible for single wires. 
The apparent inconsistencies in the results, illustrated where an experiment was 
repeated, were doubtless caused by the inequalities which were known to exist along 
a length of drawn wire, particularly in the finer gauges. 
No satisfactory explanation of this behaviour of wire under repeated bending 
is given here. The consideration of a single case on the usual lines is attempted, 
because it appears to show that these and the allied phenomena are not understood. 
The example chosen is the test of the 0-021-in. wire over a 6-in. pulley under a 
tension of 50 lb., in which it required nearly half a million bends to break the wire. In 
fig. 24 the strain in tension and compression is graphed against the diameter of the 
wire for the case when the wire is bent round the pulley under no tension. The next 
portion of fig. 24 shows the strain deduced as follows. As an approximation it is 
assumed that when the metal yields the stress remains constant. This is not quite 
correct, but it does not materially affect the conclusions. The wire was under 
the yield-point tension, consequently it appears that it was under this tension right 
across the section to allow the total tension in the metal to equal the pull on the wire. 
Therefore, it is concluded that the bending caused yield all over the section. The 
next stage was the straightening of the wire under tension, which, reasoning in a 
similar fashion, seems to have caused further tensile yield, equalising it across the 
wire. The wire had 1-4 per cent. elongation right across the section from one com- 
plete cycle of bending, yet it withstood half a million such bends. It is clear that 
