2. When the orifice of the blowpipe is somewhat large, or when (the orifice being capillary) the current of air is very strong, or the beak is somewhat removed from the flame, the jet of fire, instead of being prolonged into a pointed tongue, is blown into a brush. It makes then a roaring noise, and spreads into an irregular figure, wherein the different parts of the flame are confounded beyond the possibility of discrimination. This flame is very proper for the working of glass, and particularly of glass tubes; it ought to be clear and very brilliant, and above all should not deposit soot upon cold bodies suddenly plunged into it. The _maximum_ of temperature in this flame is not well marked; we can say, however, that in general it will be found at about two-thirds of the whole length of the jet. As this roaring flame contains a great quantity of carburetted hydrogen, and even of vapour of oil, escaped from combustion, it possesses a disoxidizing or reducing property in a very high degree. PLACES FIT TO WORK IN. Every place is adapted for a workshop, provided it is not too light and the air is tranquil. The light of the lamp enables one to work with more safety than day-light, which does not permit the dull-red colour of hot glass to be seen. Currents of cold air are to be avoided, because they occasion the fracture of glass exposed to them on coming out of the flame. MEANS OF OBTAINING A GOOD FIRE. The lamp should be firmly seated upon a steady and perfectly horizontal table, and should be kept continually full of oil. The oil which escapes during the operation, from the lamp into the tin-stand placed below it, should be taken up with a glass tube having a large bulb, and returned to the lamp. When you set to work, the first thing you have to do is to examine the orifice of the beak. If it is closed, or altered in form, by adhering soot, you must carefully clean it, and open the canal by means of a needle or fine wire. In the next place, you freshen the wick by cutting it squarely, and carrying off with the scissars the parts which are carbonised. You then divide it into two principal bundles, such as C, K (pl. 1, fig. 21), which you separate sufficiently to permit a current of air, directed between the two, to touch their surfaces lightly, without being interrupted in its progress. By pushing the bundles more or less close to one another, and by snuffing them, you arrive at length at obtaining a convenient jet. It is a good plan to allow, between the two principal bundles and at their inferior part, a little portion of the wick to remain: you bend this down in the direction of the jet, and make it lie immediately beneath the current of air.