Two different methods of subjecting a liquid to tension under dynamic conditions are described. In the first method a tension pulse is created ab initio in a column of water and the resulting pressure-tension cycles are monitored by a pressure transducer. These cyclic pressure variations have been observed by earlier workers but no adequate explanation of their occurrence has been previously given; such an explanation is proposed in the present paper. Secondly, a development of the bullet-piston method used by Couzens and Trevena (1974) is described in which a new way of obtaining and presenting the experimental results is reported.

The ab initio production of a tension pulse in a liquid has been described previously by Overton and Trevena (1981). The work described in this paper is a development of those earlier experiments. First a description is given of the way the amplitude of the tension pulse is influenced by stressing frequency, together with a determination of cavitation thresholds. Secondly, the influence of entrained gas on the amplitude of this threshold is discussed; and thirdly a study is made of the first positive pressure pulse resulting from bubble collapse.

Experiments are described with water in a new type of Berthelot tube, designed so as to enable the tube to be evacuated before the introduction of the test liquid. Secondly, some (pressure, temperature) curves for water under tension are reported. A new feature is that two of these experimental curves show a minimum at a temperature of about 5 degrees C. Thirdly, thermodynamic theory is applied to discuss the audible click of the tube when cavitation sets in.

Measurements have been made of breakdown potentials, primary ionization coefficients and formative time lags, under uniform field conditions in mercury vapour with a mercury pool as a cathode. Generalized secondary ionization coefficients (ω/α) were calculated as a function of E/p0 from the breakdown potentials and primary ionization coefficients. This curve is interpreted in the light of calculations made of the relative populations of atoms in the P states per ion pair and the application of Davidson's treatment of the temporal growth of ionization. It is concluded that the dominant secondary processes in mercury vapour for the pressure range 0·5-1·0 torr, when the cathode is a mercury pool, are collision-induced radiation from the destruction of metastables in the gas and positive-ion action at the cathode for values of E/p0 below 600 V cm−1 torr−1. At higher values of E/p0 the process is one of positive-ion action at the cathode alone, the ion acting by virtue of its potential energy.

The subjection of liquids to tension using the Berthelot tube method has been carried out by Temperley and Chambers (1946) and a number of subsequent workers. A common feature in all this work is the wide scatter obtained in the breaking tensions for a given specimen of the test liquid. A new type of investigation is described in which various thermal and mechanical variations were introduced into the method in order to study the effect on the measured breaking tensions, but the results still show a large scatter. Secondly, the role of surface nucleation sites in this matter is considered and a possible mechanism to explain this larger scatter is proposed.