amperes, the are being about 5" long, while the carbons burn nearly flat instead of taking the shape as in the case of the open arc. This makes a change in the distribution of the light, com FIG. 1.-Carbon Feed Lamp. Lamp is about 30" long, whereas lamps with Carbon Rods come as long as 46". pared with the open arc, which will be shown by curves later on. Fig. 2 shows the difference in appearance of the carbon in the two styles of arc. The enclosed are burns flat on the top or positive. carbon, the lower or negative one becoming slightly convex. The arc itself does not remain in any one spot, but wanders all around the flat ends. In the case of large carbons, this is objectionable, as it causes rather heavy shadows to be cast, which to a large extent, can be prevented by the use of proper globes. The experiment was tried of rounding the ends slightly. This arrangement did away with the shadows, but the carbons burned flat again in a very short time. + FIG. 2. The current passes through the extra resistance, the solenoid and the carbons, in series. This series arrangement makes a very simple lamp, with very few parts, and working much steadier than the open arc lamp. When no current is passing, the carbons touch; the moment the current is thrown on, the core is drawn up, carrying the positive carbon by a friction clutch. The negative carbon is fixed in a holder or socket, at the bottom of the lamp. FIG 3. and having at each end an arm so pivoted that these two arms cross each other like a letter X (as shown in Fig. 3). At each lower end of these arms is pivoted a friction piece, with either a rounded or a V-shaped surface, that grasps the rod holding the upper carbon. Lifting the upper ends causes the lower ends to come nearer together, thus gripping the rod. The carbon and its rod are dropped when the device falls far enough to touch a stop that prevents the lower part of the clutch from coming down any further. A novel departure from this form of clutch consists of a cylindrical piece of brass, fitting around the rod, and having its upper end grooved to hold a row of steel balls. These are held in place by a conically-shaped cup. When the inner cylinder rests on the stop, the cup, which is attached to the plunger, descends far enough, by reason of its sloping sides, to release the friction on the balls and the carbon rod falls. When the core is drawn up, the cup follows, the balls being caught in the apex of the cup, and the rod is drawn up. Fuses should be put in circuit with each lamp, and as the first rush of current is from 40 to 240 per cent. of the steady value, according to the make of the lamp, the fuse must be heavy enough not to blow when the lamp is first thrown in. In the case of lamps having a large flush of current, the lamp is not so well protected against an accidental heavy current, as the fuse must be of greater carrying capacity than in the case of lamps with a small flush of current, and may not blow until the lamp has been seriously overheated. One effect of high temperature around the working parts, is to make the brass rod that carries the upper carbon, stick, the surface losing its smoothness, so that in some cases it is necessary to take the rod out and polish it. The current is usually carried to the upper carbon by means of two brushes pressing on this rod. As the carbons wear away, the armature of the solenoid gradually descends, the current remaining practically constant whatever the position of the plunger, until it reaches the stop that releases the carbon from the friction clutch; the carbon then falls and strikes the lower one, but is immediately picked up again,—the light sometimes going out and sometimes remaining. This process, however, occurs at comparatively long intervals as the core has a play of 3" to "before the stop is reached. The positive carbon wears away at the rate of .05 of an inch, approximately, per hour, and the negative at half this rate; there will consequently be from 5 to 8 hours between the times of the resetting of the core to its top position. The open arc lamp, allowing a consumption of one inch per hour, feeds about 15 to 20 times as much. A good lamp should have the following points, the determination of which was one of the objects of our experimental work: Long life for one set of carbons; Simple and light, but strong mechanism; Lamp must be short, for use in low-ceiling rooms; Must cast no shadows from the carbon points, and it must not be necessary to use very dense globes to obtain this result; Smallest possible amount of deposit on the inner globe ; Smallest possible flush of current at start; Lamp should pick up immediately when the carbons have fallen together; Carbons should drop immediately at feed and if the arc should break; Minimum hysteresis in the core and no friction of the moving parts sufficient to cause sticking; Lamp should be so insulated that when the current is on, there is no uninsulated portion exposed; The shell of the lamp should be readily removable so as to expose the working parts; The outer globe should be easily removable and so fastened that, when hanging down it cannot be dropped or broken; Old carbons and inner globe must be easily removable; Both globes must admit of easy cleaning; It should be easy to replace and centre the carbons; The lower carbon holder and inner globe holder must be held firmly in place, and the arrangement for making the small bulb air-tight at the bottom should have no tendency to crack the glass or get it out of centre; Regulating mechanism should give the smallest possible flickering of the light; The dash-pot should be firm enough to resist sudden changes, but must not be so much so that it is slow in getting to its normal position. LENGTH OF ARC. In order to measure the length of the arc for different values of the voltage across it, the image was projected on a screen, on which the diameter of the carbon and the length of the arc were measured. The former being known, the true value of the length of the arc was determined by simple proportion. The results are given in Table I, and shown graphically in Fig. 4. The simple regulating mechanism of the enclosed lamp has a very great advantage over that of the ordinary open lamp. The series mechanism is not applicable to short are lamps, as a given variation makes a large proportional change, whereas this is obviated by using a long arc. The current and voltage variation in |