Several writers [Dubois (1914 book)], Fischer (1888), Molisch (1904 book) have noticed that the light of luminous bacteria changes in color if grown on different culture media. Light which is "silver white" on dead fish becomes "greenish" on salt-peptone-gelatin media and more yellow on salt-poor media. Peron (1804) and Panceri (1872) describe the light of _Pyrosoma_ as yellow to greenish after death of the animal and reddish on stimulation; then fading out through orange, yellow, greenish and azure blue. Polimanti (1911) describes the normal light of _Pyrosoma_ as greenish, and states that as the animals die, or if they are kept at temperatures above the optimum, the light becomes more red. McDermott (1911, _b_) noticed that the light of fireflies placed in liquid air became decidedly reddish just before going out and on rewarming the first light to appear was reddish followed by the proper shade at higher temperatures. I have frequently observed a more reddish color from luminous tissues of the firefly upon the addition of coagulants such as alcohol, and have noted that the light of _Cypridina_ becomes weaker and more yellow at both low (0°) and high (50°) temperatures. The meaning of these color changes will be discussed in Chapter VII. The efficiency of any light may be defined in several different ways: (1) By the percentage of visible wave-lengths in the total amount of radiation emitted, _i.e._, visible radiation divided by total (heat, visible, actinic) radiation; (2) by considering, in addition to visible radiation ÷ total radiation, the sensibility of the eye to different wave-lengths, visible radiation × visual sensibility ÷ total radiation. Visible radiation × visual sensibility is spoken of as luminosity; (3) by the amount of light (expressed in candles) produced in relation to a given expenditure of energy or in relation to the cost of the energy expended. Thus, of the radiation emitted from an incandescent electric lamp only a small per cent. is light, the rest being heat and actinic rays. It is therefore very far from being 100 per cent. efficient. If there were no infra-red or ultra-violet in the radiation from an incandescent lamp its efficiency would be 100 per cent. if we disregarded visual sensibility. But if we take into account the fact that the eye is most sensitive to yellow green, a source of light, even though emitting only visible radiation, would not be 100 per cent. efficient unless its maximum of emission corresponded also with the maximum of visual sensibility. We shall return to this question in a later paragraph. Looking at the question from the standpoint of energy consumption, the carbon incandescent lamp gives one mean spherical candle for 4.83 watts (watt = 10^7 ergs per sec.), while the tungsten lamp gives one mean spherical candle for 1.6 watts, about one-third the energy, and the latter is consequently more efficient.