and another valve close to the regulator on the cylinder of 
compressed air. After 80°C, the pressure was not released 
and continued to rise with each successive reading until it 
was usually between 70 and 80 psig for the final reading at 
SOIC) 
Work by Korosi and Fabuss (20) and Bridgman (21) 
indicate that pressures of up to 100 psig would have a neg- 
ligible effect on the viscosity of water. This was checked 
by taking readings at the lower temperatures with varying 
pressures. After correction for the vapor column, there 
was no signigicant variation among the various viscosity 
values. 
Initially, the closed chamber was not only intended to 
provide buffer volume. It had been intended to accomodate 
a silver/palladium membrane in the form of a narrow tube 
wound in a spiral and with the end inside the chamber sealed. 
The other end was left open and was led out through the back 
plate to a vacuum line. With this, hydrogen could be sub- 
Stituted for lair jin the system and the liquid raised by 
diffusing hydrogen through the membrane. In this case, no 
water vapor would be lost. However, the membrane proved 
very easily poisoned by impurities. When it was first 
received from the suppliers, it could not pass any hydrogen 
at all and had to be returned for re-activation. The second 
time it was received it did work butby then it had been 
decided to continue with the alternative method. 
Constant Temperature Bath 
The 0il bath consisted of approximately 12 gallons of 
Dow Corning 510 Silicone 011 contained in a glass cylinder 
surrounded by a larger glass cylinder through which hot air 
could be circulated. The oil was circulated downward through 
ancentnall tube contained in the o11 bath: 
The bath temperature was monitored by two Hewlett 
Packard quartz thermometer probes and controlled by the 
resistance thermometer of a Hallikainen Thermotrol. The 
maximum temperature variation found was 0.08°C and the 
maximum temperature fluctuation was .02°C. The probes were 
Calibrated bat the dice) point and were accurate to .02°C 
absolute and sensitive to better than 0.001°C. Thus the 
maximum possible temperature uncertainty is 0.1°C. 
TS 
