8 ROBERT M. YERKES AND JOHN B. WATSON 



incandescent lamps, the metallized filament lamp is, for our 

 purposes, the most satisfactory. All carbon lamps yield a yel- 

 lowish light, which tends to become reddish with age, and they 

 exhibit a greater sensitiveness to fluctuations in current than 

 does the tungsten. In neither quality nor intensity are carbon 

 lamps as constant as tungsten and Nernst lamps. We recom- 

 mend them only in case the other lamps are not available, or, 

 for some reason, prove impracticable. 



e. Carbon arcs. — The carbon arc as a source of light is val- 

 uable for experiments on vision which require high intensities. 

 For all other purposes the Nernst and the tungsten are prefer- 

 able. The crater of the arc yields a light whose spectrum is 

 continuous and whose quality approaches sunlight more nearly 

 than that of any other artificial light. The light of the arc 

 itself, as contrasted with that of the crater, presents a banded 

 spectrum which is especially intense in the violet. Intensity 

 variations are almost always great, except in the crater of the 

 arc. and it is difficult to control them. These defects render 

 this form of lamp slightly, if any, more satisfactory than the 

 sun for the quantitative investigation of light vision. It should 

 be stated, however, that the arc lamp is eminently satisfactory 

 when the light from the crater alone is used. 



The faults mentioned above are at their minimum in lamps 

 of the " baby arc " family, to which belong the Siva arc, the 

 Lilliput, the Midget, and other lamps. They are small carbon 

 arcS; used chiefly in Europe, w^hich under high pressure furnish 

 an excellent quality of light in fairly uniform intensity. Unfor- 

 tunately, none of these arcs burns well out of the horizontal 

 position. 



/. Acetylene lamps. — The quality of acetylene light is fairh' 

 satisfactory, but its intensity is with difficulty kept constant. 

 In the absence of electricity, this form of lamp may be used to 

 advantage for many tests of vision. 



g. Gas lamps. — The light is likely to be over-strong in orange 

 and yellow. Under carefully chosen conditions, in the absence 

 of electric lights, incandescent gas lamps may be used for tests 

 of vision. 



h. Oil lamps and candles. — Like the light of the gas flame, 

 that of oil lamps and candles is likely to be over-strong in yel- 

 low. With respect to constancy, controllabiUty, and measur- 



