Nearly all the car companies build their own cars, and all have repair shops at their stables, and as soon as a car is damaged by a collision it is sent at once to the shop and repaired. Men are detailed to keep a strict watch over all the working parts of cars. No metal or plate has yet been found of which to make a hand railing that will keep bright and untarnished. Many experiments have been tried, but the hardest plate that can be obtained will not stand the friction of the hands longer than two months, before the plated metal will show through. Cars are painted and varnished at least once a year. The various parts of the car last different periods. The wheels average about eighteen months on long routes; on short routes, about two years. Steps and platforms last about five years. There is no particular limit for the floors and framework, as they are but little worn. Cars are frequently built up from an old floor or framework, but at the end of about fifteen years there is but little left of the original car. RINGS OF SMOKE. [Illustration: FIG. 1--APPARATUS FOR PRODUCING RINGS OF SMOKE.] When, by means of a tube of from 2 to 5 millimeters in diameter, we gently blow tobacco smoke against a wet pane of glass, we produce very fugitive rings. If we operate with a closed vessel the rings are fixed, the current being itself uniform. But the experiment that shows the phenomenon perfectly is the one that consists in rendering the current automatic by means of an aspirator--an arrangement analogous to that devised by Mr. Nickles for analyzing the flame of a candle. A tapering glass tube or, better, a metallic blow pipe traverses a cork which hermetically closes a large bottle having a cock beneath and filled with water (Fig. 1). The nozzle of the blow pipe entering the center of the flame, and the cock being open, the liquid flows, and a column of white smoke descends vertically to the surface of the water, where it forms several concentric rings whose relief soon increases with the thickness of the heavy smoke, which finds no exit. These rings have a diameter so much the greater in proportion as the current is stronger (Fig. 2). Unfortunately, the number of the rings soon diminishes in measure as the stratum of smoke that remains upon the surface of the water becomes thicker. Finally, there remains but a single ring, which has a thickness in the center of more than 0.015 m. (Fig. 3). [Illustration: FIGS. 2 TO 5.--DIFFERENT ASPECTS OF RINGS OF SMOKE.] Instead of the smoke of a candle, we may employ that of a cigar or of a tobacco pipe. We thus avoid a deposit of fatty matter, which, in the first case, soon clogs up the tube, if it is too fine a one, and thus puts a stop to the experiment.