Following the development of the Fuel Research Station “smoke eliminator” fire doors for hand-fired, natural-draught Lancashire boilers, a series of trials was carried out to obtain figures for their performance under a variety of conditions. Most of the trials were made with one or other of two sizes (”singles” and “smalls”) of a Northumberland coal, which was chosen because of its tendency to make heavy smoke; similar results were obtained with both sizes. The figures obtained during these trials, together with those of a few supplementary trials, are used to illustrate the effects of certain variables upon the performance of a boiler of that type. The information presented includes (1) the correlation of smoke intensity with the composition of the flue gases and with thermal efficiency, (2) the use of secondary air and suitable methods of firing to reduce the heat losses caused by incomplete combustion, (3) the effect of too little and of too much excess air upon efficiency, and (4) the effect of load upon efficiency. With the special fire doors, the admission of secondary air over the fire could easily be controlled and a simple method of firing could be employed. In consequence, the efficiency of combustion was maintained at a consistently high level without calling for exceptional skill or effort on the part of the fireman. Maximum efficiency was obtained when the quantity of secondary air was just sufficient to eliminate smoke: decreasing the percentage of excess air by a small amount below the optimum had as bad an effect upon efficiency as a considerable increase. Varying the load also affected efficiency, the maximum being reached well below rated load, but the variation in efficiency between 40 and 100 per cent of full load was comparatively small.