Radiation from a quartz Hg lamp was passed through a large monochromator, and through an exit slit so adjusted as to allow isolation of a narrow band of wave lengths. Erythemic response of human skin to various wave-lengths in the XJ.-V. spectrum was measured by adjusting the slit so as to allow the radiation from a particular region of the U.-V. spectrum to fall on the untanned skin of the arm, just long enough to produce a perceptible and brief (24 hrs.) erythema. Readings of the minimum time of exposure required to produce such an erythema, at different wavelengths, give a curve which shows that the erythemic response is nil at below 313 m[mu]; increases rapidly with decrease in wave-length, reacting a maximum (taken as 100) at around 297 m[mu], and falling sharply to a minimum at around 280 m[mu]; to this point, the present data agree admirably with those of Hausser and Vahle and of Luckiesh, Holladay, and Taylor; at wave-lengths shorter than 280 m[mu], the erythema response increases rapidly with decrease in wave length, but the present data show a much more gradual increase of sensitivity, and a lower value at 250 m[mu]. than the data of the other investigators. Reducing the intensity (by a glass filter) to 1/4 when exposing 4 times as long gave, at 279 m[mu], substantially the same effect as the standard intensity and duration. To produce minimum perceptible erythema on 1 sq. cm. of untanned skin, using monochromatic radiation of 297 m[mu], required about 500,000 ergs. The authors comment on the need and difficulty of devising a standard unit of radiation, and an inanimate dosage-meter.