Up to 1893 the carbons of an arc lamp operated in the open air, so that they were rapidly consumed, lasting from eight to sixteen hours depending on their length and thickness. Louis B. Marks, an American, found that by placing a tight fitting globe about the arc, the life of the carbons was increased ten to twelve times. This was due to the restricted amount of oxygen of the air in the presence of the hot carbon tips and thus retarded their consumption. The amount of light was somewhat decreased, but this was more than offset by the lesser expense of trimming which also justified the use of more expensive better quality carbons. Satisfactory operation required that the arc voltage be increased to about 80 volts. [Illustration: ENCLOSED ARC LAMP, 1893. Enclosing the arc lengthened the life of the carbons, thereby greatly reducing the cost of maintenance.] This lamp rapidly displaced the series open arc. An enclosed arc lamp for use on 110-volt constant potential circuits was also developed. A resistance was put in series with the arc for use on 110-volt direct current circuits, to act as a ballast in order to prevent the arc from taking too much current and also to use up the difference between the arc voltage (80) and the line voltage (110). On alternating current, a reactance was used in place of the resistance. The efficiencies in lumens per watt of these arcs (with clear glassware), all of which have now disappeared from the market, were about as follows: 6.6 ampere 510 watt direct current (D.C.) series arc, 8¼ l-p-w. 5.0 ampere 550 watt direct current multiple (110-volt) arc, 4½ l-p-w. 7.5 ampere 540 watt alternating current (A.C.) multiple (110-volt) arc, 4¼ l-p-w. [Illustration: OPEN FLAME ARC LAMP, 1898. Certain salts impregnated in the carbons produced a brilliantly luminous flame in the arc stream which enormously increased the efficiency of the lamp.] [Illustration: ENCLOSED FLAME ARC LAMP, 1908. By condensing the smoke from the arc in a cooling chamber it was practical to enclose the flame arc, thereby increasing the life of the carbons.] The reason for the big difference in efficiency between the series and multiple direct-current arc is that in the latter a large amount of electrical energy (watts) is lost in the ballast resistance. While there is a considerable difference between the inherent efficiencies of the D. C. and A. C. arcs themselves, this difference is reduced in the multiple D. C. and A. C. arc lamps as more watts are lost in the resistance ballast of the multiple D. C. lamp than are lost in the reactance ballast of the multiple A. C. lamp.