with Mercury contained in Tabes. 511 



The change in the diagram due to alteration in length of 

 tube is shown in fig. 2, in which AB, AC, L>E refer to a tube 

 4 cm. long, and ab, ac to a tube of the same external and 

 internal diameter, but 20 cm. long. The line de was un- 

 represented in the latter case. The shorter the tube, the 

 less the resistance required to give a good necklace on any 

 given voltage. Further, the shorter the tube, the lower 

 the minimum possible voltage for the production of the 

 necklace. 



The diagrams (figs. 2, 6, & 7) all refer to tubes having the 

 same internal diameter. If the other circumstances are kept 

 constant, but the internal diameter is increased, the effect is 

 to make the lines AB and AC become less sloped to the axis 

 of volts, and to make the angle BAC less also. Increasing 

 the internal diameter makes the allowable range of resistance 

 less on any voltage, and also necessitates the actual resistance 

 being made less. 



If all else be kept the same, an increase in hydrostatic 

 pressure necessitates a decrease in externa! resistance on any 

 given voltage. This is shown in fig. 7, in which AB, AC are 

 the necklace limits for 81 cm. pressure, and a/>, ac are the 

 corresponding limits when the pressure is raised to 162 cm. 

 These results were obtained with the same filling of the 

 tube, this tube being the same as that used in fig. 6. 



Having referred to the general character of the changes in 

 the diagram produced by varying the length and bore of the 

 tubes and the mean hydrostatic pressure on their contents, 

 the changes in the frequency, current, fall of potential, and 

 length of arc must be dealt with. 



The frequencies have usually been determined by strobo- 

 scopic means. The speed of rotation of the disk was deter- 

 mined b} r a mechanical counter and a stop-watch, and was 

 regulated by a rheostat in series with the electric motor. An 

 electromagnetic eddy current break was subsequently fitted to 

 the instrument, to serve as a fine adjustment to the speed. The 

 axle which bore the stroboscopic disk also carried a four-sided 

 reflecting prism, and the tubes could be placed parallel to the 

 axis of rotation. By means of a photographic lens, a real 

 image of the tube was thrown, after reflexion, on to a ground- 

 glass screen. The apparently stationary image thus produced 

 could be studied in detail ; the method was found useful in 

 tracing the gradual changes in the phenomena as the circum- 

 stances were arbitrarily varied. In the necklace effect, the 

 image on the screen can be kept quite still, showing that 

 the phenomena are of very regular periodicity. 



In all the cases which have been investigated, the frequency 

 falls as the indicating point moves from A to B. Thus in 



