THE CONSTITUTION OF THE ELECTRIC SPARK. 209 



An interesting question arises as to whether, in the case of an alloy, the different 

 components affect each other or not. The evidence, so far, goes to show that they 

 d<>: the zinc lines, for instance, are less curved when the zinc is amalgamated, and 

 similarly when electrodes of bismuth are moistened with calcium chloride the 

 displacement of the bismuth lines is reduced. 



7. Experiments without Prisnvttic Decomposition. 



Dr. FEDDERSEN took photographs of the entire spark by means of the revolving 

 mirror, and the appearances he obtained were very irregular, though his experiments 

 allowed him to draw some general important conclusions. We considered it to 

 be of interest to take some photographs after removal of the prism, retaining 

 the slit ; fig. 29 (Plate 12) shows the appearance with zinc poles under the normal 

 condition of our experiment, i.e., with the six jars and a pole distance of 1 centim. 

 The straight luminous initial discharge which passes the air gap is followed by a 

 number of curved lines, which represent the oscillatory discharges. Fig. 30, in which 

 the oscillations are spread out by the interposition of self-induction, gives a better 

 representation of the phenomenon. We notice, in the first instance, the alternation 

 at each pole between strong and feeble discharges : the strong discharge at one pole 

 being opposite a weak discharge at the other. This peculiarity was pointed out 

 already by Dr. FEDDERSEN. Our experiments do not allow us to decide at which 

 pole the discharge is most luminous, and it would l>e important to find this out, as it 

 would allow us to decide whether the discharge through the metal vapour resembles 

 more that of a vacuum tube or that of the voltaic arc. Fig. 30 shows the initial 

 discharge to lie followed by a second straight luminous band before the curved lines 

 liegin. The curvature is irregular, and the two poles do not l)ehave exactly alike. 



The distance between successive discharges is the same for all metals we have tried, 

 which shows that the resistance of the metal vapour cannot be a dominating portion 

 of the whole resistance, for doubtless different metal vapours will differ in resistance, 

 especially as they must be present in very varying quantities, owing to differences in 

 volatility. As we can count about ten discharges, we may take it that the damping 

 is small and therefore the time of an oscillation 27rx/LC, if L is the self-induction. 

 I-Yoni this we calculate that our self-induction was about 3000; for steady currents 

 a rough estimate of our circuit gave a self-induction of 1000, and we know that 

 for rapid oscillations it must be greater. If we adopt the higher number, we may 

 further conclude that the total resistance of our circuit must have been small 

 compared to 20 ohms, while, taking the lower estimate for the coefficient of self- 

 induction, the resistance was small compared to 12 ohms, a further proof, if one 

 were needed, of the fact that once the insulating property of air is broken down, 

 its conductivity may be large. We do not wish to enter further into some interesting 



VOL. CXCIII. A. 2 E 



