Fig. 13.-Life Tests on Edison-Swan 8 C.P. 100-volt Lamps. Curves showing the relation between Cost of Light and Time. Price of a Board of Trade unit 9d. Cost of a new lamp ls. 9d. 3.8 0 100 200 300 400 500 600 700 800 900 Time in Hours. 1000 1100 1200 1300 1400 The sudden breaks in the curves connected by perpendicular lines show the sudden rise in price due to a lamp breaking and a new one being substituted. The price of a kilowatt-hour we took as 44d in fig. 12, and as 9d. in fig. 13, the cost of a new lamp being taken as 1s. 9d. in both sets of curves. Of course, it will be seen from the formula quoted above, that whenever a lamp breaks and a new one is put in its place, there must be a sudden, though temporary, rise in the cost per candle-hour due to the cost of the new lamp; but this rise becomes less marked as the number of lamps included in the calculations increases, and would have become inappreciable if the tests could have been carried on for a sufficient length of time. The first three lamps to break in both the 102 and the 104 volt groups were replaced by the lamps which we had previously run at 102 and 104 volts respectively for 110 hours; and in calculating the cost-curves we have added in the watthours and the candle-hours due to these previous runs; we have also added to the abscissæ of the curves the 110 hours. On examining the total-cost curves on figs. 12 and 13, it is at once apparent that they nowhere show any such rise in the cost of producing light, as was seen in the similar curves for American lamps (fig. 5). The curves in fig. 12, for which the price of a Board-of-Trade unit was taken as 44d., do not turn up at any point, not even for the three lamps run at 101 volts, which lasted unbroken for over 1300 hours. In fig. 13, however, for drawing the curves on which the price of a Board-of-Trade unit has been taken at 9d., there are slight indications of the curves turning up: for example, the curve for the 100-volt group rises a little at 850 hours, at which time the lives of the three lamps being tested were 850, 850, and 278 hours; and, again, at about 1600 hours, when the lives of the three living lamps were 1028, 643, and 433 hours, the cost per candle-hour rises a little. In the group run at 101 volts the cost of lighting began to rise a little after the lamps had been lit for 1100 hours; and in the 102-volt group the curve (fig. 13) turns up a little at 1700 hours, at which time the ages of the lamps alight were 928, 678, and 552 hours. But at no one of these points is there a sufficient rise to indicate that a time has been reached when it would be good economy to discard lamps. The remaining total cost curve on fig. 13, viz. that for the group of lamps run at 104 volts, when the price of the Board of Trade unit is taken at 9d., does not show any indication of a minimum point. Hence, generally, we may conclude that when the modern Edison-Swan 8 C.P. 100-volt lamps such as we have been testing, costing 1s. 9d. per lamp, are supplied with energy costing 9d. per Board of Trade unit, no economy will be gained by discarding the lamps before the filaments break. This important result is of course due to the fact, already pointed out, that there is no marked decrease in the candlepower of the lamps during their lives. Owing to the extent to which the individual lamps varied among themselves, it is difficult to say which of the four pressures at which we tested lamps would be the most economical to use. The three lamps tested at 101 volts seem to have been quite abnormal, their efficiency being very low, and, as a consequence, their lives very long. With a Board of Trade unit costing 44d., the lamps tested at 100 volts gave the cheapest light, and as they do not seem to have been above the average in quality this result may probably be safely taken as correct. With energy costing 9d. per Board of Trade unit (fig. 13) there does not seem much to choose between the total cost of producing light with the three pressures of 100, 102, and 104 volts, excepting that the use of 104 volts would introduce an extra expense arising from the labour of frequently replacing broken lamps. The two conclusions, then, to which we are led by a study of these total-cost curves on figs. 12 and 13 are :— 1. The modern Edison-Swan 8 C.P. 100-volt lamps such as we tested should not be discarded until the filaments break. 2. No marked economy can be gained by over-running this type of lamp. Addition made January 25, 1895. The question arises-what is the cause of the marked rise in candle-power which occurred with all the lamps tested by us during 1893 and 1894? In the early summer of 1894, when our tests were approaching completion, Mr. Howell, in contributing to a discussion at the American Institute of Electrical Engineers, stated that such a rise in candle-power was due to the vacuum of the lamp improving during the early part of its life. Therefore, after we had stopped the main investigation, we made some experiments to test the validity of Mr. Howell's explanation. We applied the induction-coil vacuum-test to some dozen Edison-Swan lamps when quite new, and again after they had been run for about 80 hours at their normal pressure of 100 volts. In every case we found that the vacuum had apparently improved; for using a spark of about 8 millimetres in length, the lamps when new were entirely filled with a bright blue glow accompanied with some bright white patches on the glass; whereas, after a run of about 80 hours at a pressure of 100 volts, this phosphorescent effect with the 8 millimetre spark was either entirely absent, or consisted of only a faint nebulous glow which did not fill the lamp-bulb, and there was very little light on the surface of the glass. Thus it appeared as if Mr. Howell's explanation of the great increase in the light given out by a glow-lamp during the early portion of its life was correct. But this conclusion which we arrived at last year has been modified by tests that we have been carrying out on EdisonSwan glow-lamps purchased since our paper was sent in to the Physical Society in November 1894. For we find that, while the induction-coil vacuum-test applied to these lamps when they are new, and after a run of 80 hours at the normal pressure of 100 volts, shows that the vacuum apparently improves, this improvement in the vacuum is not accompanied with the great rise in candle-power which we found with all the Edison-Swan lamps purchased in 1892 and 1893, the tests of which form the subject of this paper. And yet the lamps are all marked in the same way, "100 E.F. 8," and they all look alike. In the case of a set of these Edison-Swan lamps, purchased on January 21, 1895, not merely were the current and the light periodically measured, but the induction-coil vacuumtest was applied at intervals during the 80 hours' run; and it was found that the change in the phosphorescent appearance of the lamp, and the consequent apparent increase in the vacuum, was produced slowly and not suddenly on the first application of the pressure of 100 volts. The following are the results of the tests of this batch of six lamps. The light was measured with a Harcourt-Pentane Standard, the potential difference with the Ayrton and Mather reflecting electrostatic voltmeter, and the current with the strip of manganin shunted with the Ayrton and Mather moving coil-galvanometer; this voltmeter and the ammeter combination being the same as those employed with all the lamps tested previously. But both instruments were carefully calibrated on January of this year, to make quite certain that the different behaviour of these newer lamps was not merely an apparent one produeed by some change having occurred in the sensibilities of the measuring-instruments. The insulation of the lamp-holders was carefully cleaned and the contact-surfaces scraped before these tests were made. |