with Mercury contained in Tuhes. 509 



will be reabsorbed by the tides of mercury as these move 

 over the bedewed surfaces daring the latter part of the 

 period of approach. This shows why the curved bands meet 

 the sides of the beads at an angle. 



It should be noted that the part of the tube occupied by the 

 waist of the necklace is filled with hot vapour, which expands 

 outwards as the bead forms. In fact> there will be a hot wind 

 blowing each way from the centre of the arc. In some 

 photographs of necklaces, sharply defined narrow dark lines 

 are seen following very similar paths to the curved bands. 

 These are the tracks of globules of mercury, not large enough 

 to fill the bore of the tube, as they are blown outwards by the 

 hot mercury wind. 



The mean position of the necklace remains at rest near the 

 middle of the tube. The current may be reversed without 

 extinguishing the lamp, but the mean position will now be 

 nearer the new positive end of the tube. This is probably 

 connected with the fact that in all mercury arcs the positive 

 electrode tends to become hotter than the negative. 



The behaviour of such a lamp as regards changes in external 

 resistance, and in the electromotive force acting round the 

 circuit, can be shown on a diagram such as PI. VII. fig. 2, in 

 which the ordinates are proportional to the external resistance 

 and the abscissre to the electromotive force. On such a diagram 

 there is an area ABC indicating the ranges of volts and ohms 

 within which the necklace is produced. Between the lines AB 

 and DE the necklace is not perfect, but has a very thin waist, 

 which is occasionally broken. Above DE the lamps will not 

 keep alight. All along AB the necklace is characterized by 

 having a very thin waist ; while along AC the waist is very 

 wide. Just below AC the necklace is imperfect, the luminosity 

 usually ceasing while the arc is growing in length. 



The changes due to altering the external resistance are shou n 

 in the photographs, PI. YI. figs. 3, 4, & 5, taken with the same 

 tube as fig. 1, and with approximately the same voltage but 

 different resistances. With too little resistance (30 ohms), 

 the effect is as shown in fig. 3. The broken necklace pattern 

 is varied by the occasional presence of a fan. The invariable 

 characteristic of the arcs produced in regions in the diagram 

 indicated by points below line AC, is that the arc is never 

 maintained until the columns come into contact. It usually 

 ceases to be luminous while still growing in length. As the 

 resistance is increased, we get figures approaching more and 

 more nearly to the typical necklace pattern (fig. 1). The 

 necklace finally becomes perfect, and remains so until we 

 cross the line AB, when the waists, previously quite thin, 

 now occasionally break. This is shown in fig. 4, in which 



