HEAT TREATMENT ON METALLIC SUSPENSIONS 33 
The suspended system used was a cylindrical brass weight with a mirror 
attached since this design would minimize the effects of convection currents. 
The weight of the suspended system was about half that required to break 
the wire. The upper end of the suspension was fastened to a brass rod which 
extended through the top of the furnace and was clamped outside so that a 
twisting or slipping of the furnace would not affect the position of the sus- 
pended system. 
In the case of tungsten wires carbon dioxide gas was kept flowing through 
the furnace during all annealings at temperatures higher than 200°C. In 
order to anneal at temperatures higher than 500°C the wire was suspended 
in a glass tube with a weight of 6 to 10 grams at the lower end dipping in a 
copper sulfate solution so that a current could be passed through the wire. 
The temperature coefficient, which is defined as the change in the equili- 
brium position per unit temperature change, was measured by heating the 
furnace about 20°C above room temperature and allowing it to cool slowly; 
the scale reading being noted at the beginning and at various temperatures 
during the cooling. Then a graph was plotted showing the scale reading as a 
function of the temperature. Since the relation between the scale reading and 
the temperature is linear, the temperature coefficient can be easily obtained 
from the graph. It was not possible to make readings during the heating be- 
cause convection currents disturbed the suspended system in an irregular 
way and the alternating magnetic field of the heating coil exerted a torque 
on it. The possibility of convection currents affecting the position of the sus- 
pended system even during the slow cooling was investigated by making two 
sets of readings on the same wire, one set using the furnace in the usual man- 
ner, the other set by changing the temperature of the whole room very 
slowly. The corresponding readings were the same, within experimental error, 
showing that the effect of convection currents was not appreciable. For a 
typical temperature coefficient determination see Fig. 1. The coefficient is 
reckoned positive if an increasing temperature causes a clockwise rotation of 
the suspended system, as viewed from above, and negative if the rotation is 
counter-clockwise. 
EXPERIMENTAL RESULTS 
The first observations were made at room temperature on changes in the 
equilibrium position with time. In one typical case the rate of drift immedi- 
ately after loading the wire was 4.4 mm per hour. After twenty-four hours 
the motion was still easily observable but much slower, about 0.7 mm per 
hour. In some cases a change in the direction of the drift would occur. It 
seems that this time drift is caused by a slow release of strains of some sort 
in the wire and that in a sufficiently long time a stable condition would be 
reached. Just how long a time would be required is not known but.Shaw and 
Lancaster-Jones! have given a case in which the drift was still going on after 
50 days. 
The next observations made at about 200°C showed that at this tempera- 
ture the untwisting process was greatly accelerated so that the stable posi- 
177 
