Two difficulties are, however, encountered in this method. In the first place it is impossible to feed a fire continuously with a shovel. There will be intervals between the charges which are thrown in, so that the supply is not uniform, even if the charges do not consist of more than a portion of a shovel-full at a time; and if the fire was fed in this way as uniformly as possible it would then be necessary to open the furnace door every time fresh coal was put on the fire, and so much cold air would thus be admitted that more would be lost by lowering the temperature of the boiler than would be gained by the improved combustion. Another difficulty also is encountered in this method of burning coal in locomotives. In order to admit enough air through the fire it is necessary to keep the latter so thin on the grates that the violent draft produced by the blast lifts the coal from the grate-bars and carries the lighter particles through the flues unconsumed. It is thus extremely difficult to keep the grate uniformly covered with coal, and if it is not, the air will enter in irregular and rapid streams or masses through the uncovered parts, and at the very time when it should be _there_ most restricted. Such a state of things at once bids defiance to all regulation or control, so that it is found almost uniformly that firemen of locomotives keep enough coal on the grates to avoid the danger of “losing their fire,” as they express it; that is, having all the burning coal drawn through the tubes by the blast. _Now, on the control of the supply of air depends all that human skill can do in effecting perfect combustion and economy_; and unless the supply of fuel and the quantity on the bars can be regulated, it will be impossible to control the admission of the air.[95] [95] The Combustion of Coal, by C. Wye Williams. Another method of feeding locomotive boilers is to pile up the coal in the back part in a thick layer and slope it downward towards the front, so that there is a comparatively thin fire in front. The mass piled up at the door becomes converted into coke, and the production of gas from the coal is more gradual and uniform than it is when only a small quantity is thrown in at a time, and therefore a more uniform supply of air is needed for its combustion. But it is apparent that very little air can pass through the thick heap of coal at the back part of the fire-box, and that therefore all, or nearly all the air which enters it must come in through a comparatively small portion of the grate. It will of course be difficult to admit the requisite quantity, for the reasons already stated.