174 TRANSACTIONS OF THE 
son. But the place occupied by the arc lamp in the past is 
not to be minimized. As is always the case in pioneering, 
things are usually done in a wasteful, more or less hap- 
hazard, and, judged from more highly developed standards, 
in a primitive style. So it was with the arc lamp, the 
pioneer in electric lighting. For the first twenty-five years, 
time, energy, and material were wasted so far as practical 
results were concerned. But it is a significant fact that 
almost immediately after the invention of the incandes- 
cent lamp the failures were put to good account and a 
really serviceable arc lamp arose out of the ruins of these 
failures. From this time on it was a race between the 
two types with the arc continuing to fall farther and farther 
behind, notwithstanding the fact that its efficiency al- 
ways was and is even yet considerably higher than any- 
thing yet attained by the incandescent lamp. Its failure is 
due among other things to the necessarily complicated 
mechanism of its regulating device, to the necessity of fre- 
quent attention in changing electrodes, cleaning, etc., and . 
the lowering of illumination by deposits from the electrodes 
on the surface of the globes. 
This brief survey of electric lighting is given in order 
to emphasize the lines along which development up to the 
present time has progressed, which are efficiency, illumina- 
tion, and durability. Tests are made by the manufacturers 
with these factors constantly in mind, but since the lamps 
are seldom used except for ordinary service they are not 
usually carried much above the normal load. 
It occurred to me that it might be interesting to carry 
the test from voltages considerably below normal up to the 
burnout point of the lamps in order to examine the illum- 
inating characteristics through as wide a range as possible. 
Therefore a large number of lamps both of the carbon and 
tungsten types were tested from 80 volts up to the burnout 
point (which was for carbon lamps from 200 to 220 volts 
and for tungsten from 270 to 290 volts). In nearly all 
cases the lamps tested were new. The primary object was 
to determine the candle power and efficiency characteris- 
tics through the entire range. The curves thus obtained 
