“Thus far the dependence for draft has been on the burner jet and the suction of the smoke-stack in virtue of the inertia of the moving gases. But even with this ventilated boiler, this method is limited to certain dimensions of the boiler. Thus a boiler 80 cm. long yielded about the same quantity of steam as a boiler half as long and otherwise similar. Only the initial parts of the boiler are, therefore, relatively efficient, and the reason of this seems to be that, apart from shape, etc., the flame as a heat-producing agent is practically defunct, when a certain amount of heat has been taken out of it: in other words, even with fair ventilation the flame is eventually chilled off by the voluminous products of combustion continually accumulating in the boiler. The same choking action accompanies the presence of unburnt gases. If, for instance, the flame be burnt in the air, it is slender and much smaller in volume than in the boiler. The flame is also of small volume and burns completely in a wide boiler, but the steam is always deficient, because of the distance between flame and coils (see above). With the above apparatus about 1/2 lb. of dry steam per minute per square foot of heating surface was attained. [Illustration: FIG. 15.] “This introduces the final condition for rapid steam generation. There must be artificial suction at the smoke-stack. By passing the exhaust steam in the form of a central jet through the smoke-stack the yield of steam was increased 20 to 30 per cent. In fact as the supply of gas from the burner is given, the artificial suction in question means more air in the boiler for the same amount of gas and it means also a more rapid removal of the exhaust gases. The experiments with steam suction are yet to be completed, and with them the boiler question is to be finally laid at rest. The chief points at issue are these: “1. Seeing that the jet suction increases with the length of the smoke-stack, up to a certain length at least, how long and how wide must the efficient smoke-stack be made? Thus a smoke-stack 10 cm. long is all but useless. Good results are obtained when the stack measures 30 cm. in length beyond the end of the steam jet. [p075] “2. What is the relative efficiency of the initial and final halves of the length of the boiler? This will show in how far it is useful to increase the length of the boiler for a given burner and steam jet. It will also show what advantage is to be gained from triplicate boilers with three burners, as compared with duplicate boilers with two burners, or single boilers with one burner, ‹when the same weight› of tubing is used throughout.