The daylight which enters buildings is often considerably altered in color by reflection from other buildings and from vegetation, and after it enters a room it is sometimes modified by reflection from colored surroundings. It may be commonly noted that the light reflected from green grass through a window to the upper part of a room is very much tinted with green and the light reflected from a yellow brick building is tinted yellow. Besides these alterations, sunlight varies in color from the yellow or red of dawn through white at noon to orange or red at sunset. Throughout the day the amount of light from the sky does not change nearly as much as the amount of sunlight, so there is a continual variation in the proportion of direct sunlight and skylight reaching the earth. This is further varied by the changing position of the sun. For example, at a north window in which the direct sunlight may not enter throughout the day, the amount of sunlight which enters by reflection from adjacent buildings and other objects may vary greatly. Thus it is seen that daylight not only varies in quantity but also in quality, and an artificial daylight, which is based upon an extensive analysis, has the advantage of being constant in quantity and quality as well as correct in quality. Modern artificial-daylight units which have been scientifically developed not only make mankind independent of daylight in the discrimination of colors but they are superior to daylight. Although there are many expert colorists who require an accurate artificial daylight, there are vast fields of lighting where a less accurate daylight quality is necessary. The average eyes are not sufficiently skilled for the finest discrimination of colors and therefore the Mazda "daylight" lamp supplies the less exacting requirements of color matching. It is a compromise between quality and efficiency of light and serves the purpose so well that millions of these lamps have found applications in stores, offices, and industries. In order to make an accurate artificial north skylight for color-work by means of colored glass, from 75 to 85 per cent. of the light from a tungsten lamp must be filtered out. This absorption in a broad sense increases the efficiency of the light, for the fraction that remains is now satisfactory, whereas the original light is virtually useless for accurate color-discrimination. About one third of the original light is absorbed by the bulb of the tungsten "daylight" lamp, with a resultant light which is an approximation to average daylight.