turing department. As soon as Doctor Bolton's experiments showed that the originally brittle tantalum could be made ductile enough to draw into wire by the usual methods, and that this wire could be bent and coiled like a thin steel wire, it became possible to test it thoroughly as to its usefulness for incandescent lamps. The first trials with wires of about 0.3 millimeter diameter gave most promising results. They confirmed the fact that tantalum has a very high melting point and that it is but slightly subject to disintegration in a vacuum, even when subjected to a heavy current.

The first tantalum lamp that proved moderately satisfactory in that it admitted of an exact measurement of the electric photometric conditions and stood a burning test for some time, was completed just over two years ago, viz, on December 28, 1902. This lamp had a loop-shaped filament made of the first tantalum wire ever drawn. The diameter of the wire was 0.28 millimeter, its effective lighting length 54 millimeters, and its electrical resistance when cold 0.29 ohm. This corresponds to a specific resistance (1 meter length, 1 square millimeter section) of 0331. The photometric measurements made at efficiencies of 2, 14, and 1 watt per Hefner candlepower .showed potential differences of 4·9, 4·95, and 59 volts, currents of 5, 546, and 62 amperes, and illuminating values of 11, 18, and 37 Hefner candlepower, respectively. On being burnt at 1 watt per candlepower the lamp had a life of twenty hours, during which it blackened considerably.

As the chemical and mechanical manufacturing processes developed and the material became purer and the wires more uniform, the results obtained also improved. The lamps lasted longer and blackened less; at the same time the specific resistance decreased until it had dropped to the present figure of 0·165 for the pure metal. It is clear that the material used for the first lamps still contained a considerable quantity of impurities, probably niobium and carbides, which caused the great disintegration and the nearly double specific resistance. During these first trials we looked very carefully into the question as to what dimensions the filament of a tantalum lamp ought to have for ordinary voltages and illuminating values. From the dimensions of the filament used in the first lamp we calculated that, with this rather impure material, we should require a filament about 520 millimeters long and 0.06 millimeter diameter for a lamp for 110 volts, 32 Hefner candlepower, and 15 watts per candlepower. These unusual figures increased when the specific resistance of the material had diminished to the present value of 0·165, at which, for a 32 Hefner candlepower lamp, a filament of about 700 millimeters in length by 0.055 millimeter in diameter was required; for a 25 Hefner candlepower lamp a filament of about 650 millimeters by 0.05 millimeter diameter was required. Thus, in order to con

struct a practical and useful lamp for standard voltages and illuminating values, we had to solve the problem of drawing the tantalum wire in sufficient length down to a diameter of 0.05 millimeter to 0.06 millimeter; this we succeeded in doing after long and laborious trials.

In July, 1903, we possessed the first tantalum lamp with a filament of about 0.05 millimeter diameter. It had a loop-shaped filament 54 millimeters long, and it took 0.58 amperes at 9 volts and gave 3-5 Hefner candlepower at 15 watts per candlepower. On the basis of these figures a lamp having the same quality and diameter of wire and working at the same efficiency on a 110-volt circuit would have a filament 650 millimeters long and would give 43 Hefner candlepower. The experiments thus far had proved that the task of producing lamps for 110 volts and a maximum of 25 to 32 Hefner candlepower was not an easy one in several respects. We had to solve the problem of suitably and reliably fixing a filament rather more than 2 feet long within a glass globe which should not exceed to any great extent the dimensions of the usual incandescent lamps. The first and most obvious attempt was made, of course, by adhering to the loop shape and accommodating the required length of wire by connecting several such bows in series within the lamp. However, lamps made according to this plan with two to four tantalum loops gave results which were anything but satisfactory.

It appeared that, like all other metallic filaments which have hitherto been used for incandescent lamps, tantalum wire softens sensibly at the temperature attained when worked at 15 watts per candlepower. To use loop-shaped or spiral filaments similar to the carbon filaments of the common incandescent lamps was, therefore, out of the question. There was no difficulty in suspending the loops, but in that case the lamps would have to be used exclusively in a vertical position, a limitation which we wished to avoid in all circumstances. Besides, such a construction would necessitate staying the loops firmly to prevent them from becoming entangled with each other during transport of the lamps. Nor did lamps made with loops of corrugated wire (fig. 1) or of plain or corrugated metal ribbon give satisfaction; for although the loops were certainly shortened in this way, there were other drawbacks which caused us to abandon this construction. It soon became apparent that the one road to success lay in the direction of dividing the filament into a number of short straight lengths supported at their ends by insulated holders. In this manner we succeeded at last, in September, 1903, in producing the first really serviceable lamps for about 110 volts. This lamp is illustrated in figure 2, and it will be seen that it contains two glass disks cast to a central wire holder; each disk carries laterally twelve arms having small hooks at their ends and insulated from each other. Through these

twenty-four hooks the thin tantalum wire is drawn up and down between the two disks. This is believed to be the first metallic incandescent lamp for nearly 110 volts which, like the common carbon glow lamp, can burn in any position whatsoever. This lamp supplied about 30 Hefner candlepower on a 94-volt circuit at 15 watts per candlepower. It lasted for 260 hours, and lost during that time 9.5 per cent of its illuminating power.

After this first practical success we redoubled our efforts to improve the lamp further. As far back as about the middle of October, 1903,

we succeeded in making the first 200-volt tantalum lamp, which was of a design similar to the lamp just described, but with eighteen arms on each disk and with a greater distance between the two disks. I may add at once that it is of interest only as a curiosity, for it has served no practical purpose. The length of its filament was 1,350 inillimeters and the illuminating value about 60 Hefner candlepower. In the course of further development the form of the frame. of wire filament for the 110-volt lamp went through different stages, the principle of subdivision being always followed. Among other constructions, we tried some in which, instead of one long filament,

a number of short pieces of wire were fixed on a supporting frame; these pieces, connected in series, made up the total length required. Figure 3 represents a lamp thus constructed, the wire being fixed obliquely in sixteen straight pieces between two insulated supporting stars. Such lamps offer the advantage that short pieces of filament can be used in the manufacture, but they are only reliable if the wires used in the same lamp

are absolutely uniform in diameter and quality. In the end we arrived at the shape represented in figure 4, which is for 110 volts, 25 candlepower, and 1.5 watts per Hefner candlepower. In this form, differing from most of the previous constructions, the central support consists of a short glass rod carrying two disks, into which the arms, bent upward and downward in the shape of an umbrella, are cast. The upper star has eleven, the lower twelve arms, each upper arm being in a vertical plane midway between the vertical planes in which two adjacent lower arms lie. Between these eleven and twelve arms, which are bent into hooks at their ends, the entire length of the filament is drawn in a zigzag fashion. Its extremities, held by two of the lower arms, are connected with the foot of the lamp by means of platinum strips.

The standard type for 110 volts 25 Hefner candlepower and 15 watts per candlepower has a filament 650 millimeters long and 0.05 millimeters in diameter. The weight of this filament is 0.022 gram, so that about 45,000 lamps contain together 1 kilogram of tantalum. The shape of the glass globe is adapted to the frame described above. Care has been taken to make it of a size not exceeding the usual maximum dimensions of common incandescent lamps of the same candlepower

(25 Hefner candlepower 110 volts). This shape offers a number of noticeable advantages. In the first instance it is very stable and will stand strong shocks without damage to the lamp. A considerable number of such lamps sent across the sea to test their ability to withstand the hardships of transport came back unhurt, although they had been packed just like common glow lamps, and no special care in any respect had been taken in their handling. The lamp burns, of course, in any position, and can therefore be held in any kind of fitting. The light is rather white and agreeable, and its effect is particularly uniform if the lamp is provided with a groundglass globe.

We shall now proceed to describe the electric and photometric properties of the lamp and its behavior in actual use. Numerous trials for lengthy periods of time at 1 to 3 watts per candlepower have proved the vast superiority of the tantalum lamp over the carbon filament lamp under equal electric and photometric conditions. Expressing this fact in figures, we can state that the tantalum lamp consumes about 50 per cent less current at the same voltage, with the same intensity of light and the same useful life; or that, at the same economy, its life is several times that of the carbon type. Moreover, at an initial efficiency of 15 volts per Hefner candlepower the tantalum lamp has an average life quite sufficient for all practical requirements, so that this rating has been standardized for the 110-volt lamp. Trials have also proved that the lamps have a life of several hundred hours at 1 watt per Hefner candlepower, but in that case they were very sensitive to variations of pressure and often showed an early decrease of illuminating power. The useful life of the tantalum lamp-i. e., the time within which it loses 20 per cent of its initial illuminating power-averages between 400 and 600 hours at 15 watts per Hefner candlepower. Some specimens have proved to have a useful life of as much as 1,200 hours. The absolute life, in general, amounts to 800-1,000 hours under normal working conditions. Further, we have to remark that the tantalum lamp blackens but little unless it has been strongly overheated during work in consequence of partial short-circuiting of the filament.

It is very interesting to observe the behavior of the tantalum lamp during the whole course of its life. The first fact worthy of note is that, like some carbon lamps, the illuminating value increases at the beginning, generally after a few hours, by 15 to 20 per cent. In the same way the consumption of current rises by about 3 to 6 per cent, while the consumption of energy drops to 13 to 14 watts per candlepower. After that the illuminating value gradually decreases, while a corresponding increase of the consumption of energy occurs. The average behavior of the 25-candlepower lamp