(Fig. 193) is open to neither of these objections. Instead of a rectangular tank up aloft, we now have a large copper cylinder situated in the kitchen near the range. The flow and return pipes are continuous, and the cold supply enters the bottom of the cylinder through a pipe with a siphon bend in it. As before, water is drawn off from the flow pipe, and a radiator may be put in the circuit. Since there is no draw-off point below the top of the cylinder, even if the cold supply fails the cylinder will remain full, and the failure will be discovered long before there is any danger of the water in it boiling away. [Illustration: FIG. 193.--The "cylinder" system of hot-water supply.] Boiler explosions are due to obstructions in the pipes. If the expansion pipe and the cold-water supply pipe freeze, there is danger of a slight accumulation of steam; and if one of the circulation pipes is also blocked, steam must generate until "something has to go,"[38] which is naturally the boiler. Assuming that the pipes are quite full to the points of obstruction, the fracture would result from the expansion of the water. Steam cannot generate unless there be a space above the water. But the expanding water has stored up the heat which would have raised steam, and the moment expansion begins after fracture this energy is suddenly let loose. Steam forms instantaneously, augmenting the effects of the explosion. From this it will be gathered that all pipes should be properly protected against frost; especially near the roof. Another cause of disaster is the _furring up_ of the pipes with the lime deposited by hard water when heated. When hard water is used, the pipes will sooner or later be blocked near the boiler; and as the deposit is too hard to be scraped away, periodical renewals are unavoidable. HOW A LAMP WORKS. From heating we turn to lighting, and first to the ordinary paraffin lamp. The two chief things to notice about this are the wick and the chimney. The wick, being made of closely-woven cotton, draws up the oil by what is known as _capillary attraction_. If you dip the ends of two glass tubes, one half an inch, the other one-eighth of an inch in diameter, into a vessel of water, you will notice that the water rises higher in the smaller tube. Or get two clean glass plates and lay them face to face, touching at one end, but kept slightly apart at the other by some small object. If they are partly submerged perpendicularly, the water will rise between the plates--furthest on the side at which the two plates touch, and less and less as the other edge is approached. The tendency of liquids to rise through porous bodies is a phenomenon for which we cannot account.