(.821 .026 .016) W = 34.56 (---- + (.0425 - ----) + ----) = 10.88 pounds. ( 3 8 8 ) If the amount of carbon which is burned and passes away as flue gas is 80 per cent, which would allow for 2.1 per cent of unburned carbon in terms of the total weight of dry fuel burned, the weight of dry gas per pound of carbon burned will be from formula (16): 11 × 10.7 + 8 × 9.0 + 7(0 + 80.3) W = --------------------------------- = 23.42 pounds 3(10.7 + 0) and the weight of flue gas per pound of coal burned will be .80 × 23.42 = 18.74 pounds. The heat lost in the flue gases per pound of coal burned will be from formula (15) and the value 18.74 just determined. Loss = .24 × 18.74 × (500 - 60) = 1979 B. t. u. The percentage of heat lost in the flue gases will be 1979 ÷ 14500 = 13.6 per cent. The ratio of air supplied per pound of coal to that theoretically required will be 18.74 ÷ 10.88 = 1.72 per cent. The ratio of air supplied per pound of combustible to that required will be from formula (14): .803 ------------------------- = 1.73 .803 - 3.782(.09 - ½ × 0) The ratio based on combustible will be greater than the ratio based on fuel if there is unconsumed carbon in the ash. Unreliability of CO_{2} Readings Taken Alone--It is generally assumed that high CO_{2} readings are indicative of good combustion and hence of high efficiency. This is true only in the sense that such high readings do indicate the small amount of excess air that usually accompanies good combustion, and for this reason high CO_{2} readings alone are not considered entirely reliable. Wherever an automatic CO_{2} recorder is used, it should be checked from time to time and the analysis carried further with a view to ascertaining whether there is CO present. As the percentage of CO_{2} in these gases increases, there is a tendency toward the presence of CO, which, of course, cannot be shown by a CO_{2} recorder, and which is often difficult to detect with an Orsat apparatus. The greatest care should be taken in preparing the cuprous chloride solution in making analyses and it must be known to be fresh and capable of absorbing CO. In one instance that came to our attention, in using an Orsat apparatus where the cuprous chloride solution was believed to be fresh, no CO was indicated in the flue gases but on passing the same sample into a Hempel apparatus, a considerable percentage was found. It is not safe, therefore, to assume without question from a high CO_{2} reading that the combustion is correspondingly good, and the question of excess air alone should be distinguished from that of good combustion. The effect of a small quantity of CO, say one per cent, present in the flue gases will have a negligible influence on the quantity of excess air, but the presence of such an amount would mean a loss due to the incomplete combustion of the