76 JOURNAL OF THE WASHINGTON ACADEMY OF SCIENCES VOL. 13, No. 5 
lamps. These lamps were originally calibrated by comparison with 
the pentane lamp mentioned, but published records indicate that 
only two series of such comparisons have been made, so that in 
effect the unit is maintained by the electric standards. In Germany 
the Hefner amyl-acetate lamp is presumably still the nominally 
recognized standard. 
In addition to the uncertainties inherent in all these flame standards 
all of them show systematic variations resulting from atmospheric 
conditions. Consequently in every case the actual light output of 
the flame standards under average conditions has been determined 
by means of incandescent electric lamps, which are free from these 
effects. The electric lamps have proved so much more reliable and 
convenient than the flames that there has been a marked tendency 
to use them as the real standards, comparisons with the flames 
being rarely made. In fact the international agreement mentioned 
has been obtained and is maintained by means of comparisons be- 
tween electric standards initiated by the Bureau of Standards after 
several attempts made in Europe to reach agreement by direct com- 
parison of flame standards had failed. 
This procedure is in part analogous to that followed in the case of 
the international ohm, which is maintained by means of wire resist- 
ance standards whose value is derived from mercury ohms set up at 
infrequent intervals, but there is the very important difference that 
there exists no reproducible primary standard of light analogous to 
the mercury ohm. In the adoption of the unit of candlepower in 
the United States this difficulty was faced, and it was decided not to 
adopt any of the existing nominal primary standards, but to base 
all values on a large group of electric incandescent standards pending 
agreement on a more satisfactory primary standard. 
Some sixteen years have passed since this decision was made. 
During that time extensive experimental! studies of the various flame 
standards have been made at the Bureau, and the net result of these 
investigations has been a confirmation of the wisdom of the course 
taken. Instead of the flame lamps being taken as a fundamental 
basis for the calibration of the electric standards, this relation is 
reversed, and the flames are used only as working standards for less 
precise measurements where facilities are not available for operating 
electric standards. 
For practical purposes the electric standards are highly satisfactory, 
and the probability of any important drift in the value of the unit 
maintained by them is almost negligible. Nevertheless it must be 
