The 110-volt current can push an ampere of electricity through eight inches of carbon thread seven thousandths of an inch in diameter, and when this happens the filament gets hot enough to give out as much light as sixteen standard candles. In the place of the 16-candle-power lamp, we put a 32-candle-power 110-volt lamp. The ammeter indicated one ampere. The carbon filament was larger (No. 30, diameter = .01 inch), so as to allow more current to pass. An 8-candle-power 110-volt lamp was substituted; one quarter of an ampere passed. A 4-candle-power 110-volt lamp was used; one eighth of an ampere passed. A 100-candle-power 110-volt lamp was substituted; three amperes of current passed through it. In all these cases the lamps which passed the larger current had the larger filaments. A little practice would enable one to distinguish between these lamps without labels by examining their filaments. Among these 110-volt lamps, it is to be noted that the amount of light which they give is proportional to the amount of current which they pass. And it is convenient to remember that one ampere of electricity for one hour costs about one cent. We introduced into the socket a "Hylo" lamp (Fig. 96). The filament, _A_, took half an ampere of electricity, gave 16-candle-power of light, and cost half a cent an hour. When the lamp was turned in its socket the current was switched off of the filament _A_, and on to the filament _a_. This took .03 of an ampere, gave one candle-power of light, and cost .03 of a cent an hour, or at the rate of about $3.00 a year, burning continuously day and night. [Illustration: Fig. 96] The uses of such a lamp are apparent in rooms which have no daylight. However, a wall switch at the entrance of such a room, making it easy to throw on and off the light entirely, seems to be a more satisfactory arrangement. One of the boys connected a wattmeter in the circuit with a hylo lamp and found that the small filament did not pass current enough to move the armature of the wattmeter. Hence that may be burned alone without affecting the consumer's bills. We took a 16-candle-power 220-volt lamp, and lighted it by a 220-volt current. The meter showed that it allowed only one quarter of an ampere to pass. The filament was very much smaller than that in the 110-volt, 16-candle-power lamp. The pressure was twice as great as before, but the resistance was four times as great, and hence only half as much current passed. We find that it costs just as much to generate one quarter of an ampere at 220-volt pressure as it does to generate half an ampere at 110-volt pressure.