Cady and Arnold — Electric Arc. 395 



upon an oxide particle as before. This may be due simply to 

 the loose particles having a higher temperature than the elec- 

 trode itself, but it may indicate also that the oxide of iron, 

 like that of calcium, emits negative corpuscles more readily than 

 does the metal itself. "When the electrodes have become com- 

 pletely tipped with oxide, no glow discharge can be obtained.* 

 A similar spontaneous change back and forth between glow 

 and arc was also observed with terminals of copper, and of 

 platinum. 



§16. Fig. 3 shows the limiting values of current and volt- 

 age for the change from glow to arc, with platinum and iron 

 electrodes. The length of arc is about l mm . In the case of 

 platinum, AB is the line along which the change takes place 

 in going from glow to arc. The arc current can then be 

 decreased so that the change back to glow is along CD. This 

 is analogous to the "parallelograms" referred to in §4, and is 

 similar to an observation of Kaufmann's.f No such overlap- 

 ping was detected in the case of iron. The platinum arc is 

 very irregular, but as nearly as can be judged it is on the 

 second stage. 



When the glow between iron electrodes changes along A' B' 

 to an arc, the latter is on the first stage. Further decrease in 

 external resistance results farther on in the change to the sec- 

 ond stage, as shown in fig. 3. Points based on observations 

 made during this experiment are represented by circles. The 

 lines representing the first and second stages of the iron arc in 

 air are based on other data. 



The close agreement of the curve for the platinum arc with 

 the second stage of the iron arc is noticeable. It looks as if 

 the higher boiling point of platinum allowed the glow to per- 

 sist until the current was greater than is possible with iron 

 electrodes, so that when the change to arc occurred, the initial 

 arc current as determined by the slope of the line AB was 

 too great for the first stage to exist. In other words, cathode 

 and anode begin to volatilize at the same instant. If a higher 

 supply e.m.f. had been available, it is quite probable that the 

 first stage could have been maintained with the platinum arc. 

 With the arc between platinum terminals in nitrogen, no better 

 success in reaching the first stage was obtained, even at reduced 

 gas pressures. 



After the platinum electrodes had cooled, an examination 

 under the microscope revealed a system of grooves or striae on 

 their surfaces, somewhat like those mentioned by Arons.^: A 

 similar structure, apparently crystalline in nature, can be seen 



* Stark (Phys. Zeitschr., v, 81, 1904) found that the glow did not change to 

 an arc when the electrodes were clean. 



f Ann. Phys., ii, 158, 1900. J Ibid., i, 705, 1900. 



