results will be worked out if the coke exceeds the above amount although I have met with coals giving 87 per cent. of coke which were perfectly manageable, though in other cases the coal did not burn completely. It will be noted that the non-volatile residue of anthracite is never as low as 86 per cent., and this, together with the very dry steam coals and bastard anthracite (found over a not inextensive tract of the South Wales Coal field), form a series of coals, alike difficult to burn in Thompson's calorimeter. Considerable experience has shown that in no single instance was the true comparative heating power of anthracite or bastard anthracite indicated. With a view to accelerate the perfect combustion of these coals, sugar, starch, bitumen, and bituminous coals--substances rich in hydrogen--were employed, mixed in varying proportions with the anthracitic coal, but without the anticipated effect. Coke was also treated in a like manner. Without enlarging further upon these futile trials--all carefully and repeatedly verified--the results of my experiments and experience show that for coals of an anthracitic character, yielding more than 87 per cent. of coke, or for coke itself, Thompson's calorimeter is not suited as an indicator of their comparative calorific power, for the simple reason that some of the carbon is so graphitic in its nature that it will not burn perfectly when mixed with nitrate and chlorate of potash. A sample of very pure anthracite used in the experiments referred to, gave 90.4 per cent. of non-volatile residue, and only 0.84 per cent. of ash. This coal was not difficult to experiment with, as combustion started with comparative ease and proceeded quite rapidly enough, but in every instance a portion of the carbon was unconsumed, and consequently instead of about 13° of rise in temperature only 10° were recorded.