102 
was then formed again by means of a third carbon, and the spectrum 
was photographed. No Ca lines appeared except those always present in 
the ordinary are. 
In the second part of the experiment a plain carbon was used as a 
negative pole, and the Ca was placed in the positive pole. The are was 
formed as before and allowed to continue for one minute. The positive 
pole containing the Ca was then replaced by a new plain carbon, the are 
was formed.and the spectrum photographed. The Ca lines came out 
very clear and strong, almost as strong as if the negative carbon had been 
filled with the salt. There can be but one conclusion. In the first case 
the Ca was in the negative pole, and there was no tendency for it to pass 
over to the positive pole. In the second case it was placed in the positive 
pole, and it freely passed over to the plain negative pole. The latter be- 
came so impregnated with it that it was capable of giving a strong spec- 
trum of Ca when it was afterwards used in an are with a plain carbon. 
Any of the elements, which show a marked tendency to cling to the 
negative pole, may be used instead of Ca in the above experiment. The 
result was doubtful in the case of Zn. 
The electrolytic nature of the are was further confirmed by a series 
of measurements of the voltage necessary to maintain an are of given 
length between unlike electrodes. It was found that a much higher volt- 
age was required when the metallic salt was placed in the negative pole 
than was necessary when the salt was in the positive pole. 
When Ca was placed in the upper pole only and it was made positive, the 
average of several readings gave 49 volts between the upper and lower 
carbons. With the same length of are and the current reversed in diree- 
tion the average reading was 65 volts. With a slightly shorter arc the 
readings were 35 and 48 volts, respectively. With the salt in the positive 
carbon of a horizontal arc the reading was 34 volts, and when in the neg- 
ative carbon, 45 volts. 
Similar results were obtained when Li and K were used instead of Ca. 
An are was formed between a carbon and a copper rod of the same 
size and shape. An average of many readings showed that about 10 
volts more were required to maintain an are with the current from the 
carbon to the copper rod than were required when the current was re- 
versed. 
When a third carbon was held with its point midway between the 
poles and in an are maintained at 55 volts, the potential difference between 
