The lumen is based on the CIE photopic luminous efficiency function. It was derived at high light levels and for on-axis (foveal) vision. This definition is adequate for specifying the luminous stimulus for foveal tasks, but it does not strictly apply to certain nighttime applications where low-light level, off-axis vision is most critical (e.g., night driving). This paper compares visual performance, using reaction time (RT), for two light sources, metal halide (MH), and high pressure sodium (HPS). A small, 2-degree target was flashed either on-axis or off-axis at background luminances ranging from 0.01 to 10 cd/m(2). The results of the experiment showed that RT increases exponentially with decreasing background luminance. There was no difference between light sources when the target was flashed on-axis, or when the target was seen at high background luminances (above 1 cd/m(2)). When the target was flashed off-axis, reaction times th HPS became systematically longer than MH as background luminance decreased below 1 cd/m(2). Our results support the conclusion that rods play an increasing role in peripheral vision as background luminance drops below photopic levels, but play little or no role in the fovea at any background luminance. These data led to the development of a preliminary mesopic luminous efficacy function for off-axis detection. Under a representative asphalt luminance of 0.1 cd/m(2), MH was calculated to be approximately 60 percent more efficacious than HPS.
