D BY FIOT PLATINI'M IN DIFFERENT GASES. 29 



9. Special Properties of New Wires. 



The writer* has shown that a new platinum wire, even when well cleaned with 

 boiling nitric acid and distilled water, gives rise to an abnormally high positive 

 ionisation when heated in a vacuum. This initial ionisation gradually falls off to a 

 small value, but it is found that a wire which has lost most of this initial effect still 

 exhibits peculiar properties when the leak from it in an atmosphere of oxygen or air 

 is examined. These effects, which only seem to disappear after continued heating at 

 a low pressure, are characterised by great variation of current with electromotive force 

 together with time changes in the current under constant voltage. 



It has been shown that an old wire exhibits little variation of the current with 

 electromotive force with voltages above 40. Changing the voltage from +40 to +760 

 never more than doubled the current in the case of an old wire at atmospheric 

 pressure. In the case of a new wire, however, a change in the voltage generally 

 altered the current in the ratio of the applied voltage. The current did not, however, 

 remain steady at the new value. If the voltage had been increased it gradually fell 

 to a value much nearer that which it had at the lower voltage. The steady value 

 which the current seemed to be approaching was greater the higher the voltage, and 

 what may be called the steady increase with voltage was generally greater the newer 

 the wire. The converse increase in the current subsequent to lowering the voltage 

 was also sometimes observed, but was much smaller. 



The following numbers, which were obtained in air at atmospheric pressure at a 

 temperature of about 900 C., with a wire which was not very new and therefore did 

 not show the effect in its most exaggerated form, will illustrate the kind of thing that 

 occurred. The wire under +40 volts was giving a leak which oscillated from 33 to 

 44 divisions, the minimum value of 33 divisions being, however, fairly constant. The 

 voltage was changed to + 360, when readings every successive 3 minutes gave for the 

 values of the leak 117, 97, 79, 68, 70, 73. On changing the voltage back to +40, 

 the leaks at 3-minute intervals were 21, 25, 22, 23. 



The above experiments were carried out in the glass tube apparatus first described. 

 This apparatus relies for its insulation inside the tube solely on the surface of the 

 glass. It seemed possible that queer effects like the above might be obtained on 

 changing the voltage if the surface of the glass were getting charged up. This might 

 be especially likely to happen after a new wire had been sealed into the tube, owing 

 to the surface retaining a film of moisture, since it was necessary to introduce moisture 

 to clean out the tube. 



For these reasons it seemed advisable to test the question with a form of apparatus 

 which was not liable to these objections. The apparatus used was really designed for 

 some experiments on the leak from a platinum tube which will be described later (see 

 fig. 16, p. 54). The hot wire was insulated along the axis of three equal cylindrical 



* ' Phil. Mag.' [6], voL 6, p. 80. 



