MECHANICAL STRETCHING OF LIQUIDS. 
365 
It must not, however, be supposed that the condition of the liquid, when in a state 
of tension, is necessarily nearly unstable. Thus, before the method of detachment by 
heating the platinum wire was hit upon, it was often very difficult to effect a release. 
With the tonometer indicating a pull of from 8 to 12 atmospheres, the liquid could 
not be made to let go its hold by moving the apparatus about, nor by means of taps 
and jars as violent as it seemed safe to give, nor by strong local heating of the narrow 
tube. Meanwhile, specks of suspended impurity could be seen floating about in the 
interior, proving the liquidity of the substance ; and, when the release did take place, 
the immediate rise of the mercury in the tonometer to its normal position showed that 
no mistake had been made in the measurement of the tension to which the liquid was 
all the time subjected. 
The Observations. 
The accompanying diagram (Plate 10) is a graphic record of all the observations 
that I succeeded in obtaining. 
The first set were taken on May 17, 1890, and are distinguished thus 0. It was 
in making these that I became fully aware of the importance of aiming at steadiness 
of temperature before liberating the bubble, so as to permit of a little waiting till the 
liquid on the sides of the tube had drained down. Hence these measures are, with 
apparently one exception, probably all a little too high. The observations recorded + 
were made on May 19, with all possible care; two of them low down in the diagram 
were recorded as doubtful, and are marked (?). Those marked © correspond to cases 
in which the bubble formed at the top of the tube FE (fig. 1), from which position it 
could not be “ transferred” to the wire. This upper portion of the tube had not been 
calibrated with special attention, since its employment had not been anticipated, and 
the subsequent breakage of the apparatus prevented this being done afterwards. 
Nevertheless, I have thought it best to introduce the observations into the diagram. 
An eye estimate had to be made of the contraction of the tube due to the sealing at 
E, but in long bubbles an error on this score must have been comparatively small. It 
will be observed, however, that all the observations in question fall rather below the 
mean curve. The observations marked 0 were made on May 22, in order to fill in 
regions about which information was still needed. All of these were recorded as 
“good” observations. 
I interpret these observations as representing the straight line through the origin 
drawn in the diagram and making an angle with the axis of stress, whose tangent is 
0700. 
The vibrations set up, and the shock sustained by the apparatus on the sudden 
release from the highest tension reached (over 17 atmospheres) were so violent that it 
seemed unsafe to proceed further before making such observations as were necessary 
