The most convenient way of reducing the point to a suitable length (say 1.5 inch) is to fuse it off in the flame, but this must be done neatly; if a tail is left it may cause inconvenience by catching, or even piercing the finger and breaking off. The blow-pipe flame being turned down to a suitable size, and the shoulder of the "draw" having been kept warm meanwhile, let the tip of the flame impinge on a point where the diameter is about half that of the undrawn tube, and let the temperature be very high (Fig. 11). The tube is to be inclined to the flame so that the latter strikes the shoulder normally, or nearly so. Then, according to circumstances, little or much of the glass can be removed at will by drawing off the tail (Fig. 12), till, finally, a small drop of melted glass only, adheres to the end of the now closed tube (Fig. 13). Fig. 11. Fig. 12. Fig. 13. Fig. 14. When this is satisfactorily accomplished, heat the extreme end of the tube most carefully and equally, holding it in such a position that the glass will tend to flow from the bead back on to the tube, i.e. hold the closed end up to the flame, the tube being, say, at 45 degrees to the horizontal. Then when the temperature is such as to indicate complete softness lift the tube to the mouth, still holding the tube pointing with its closed end a little above the horizontal, and blow gently. A beginner almost always blows too hard. What is wanted, of course, is a continued pressure, to give the viscous glass time to yield gradually, if it is uniform; or else intermittent puffs to enable the thinner parts, if there are any, to cool more, and hence become more resisting than the thicker ones. In any case a little practice will enable the operator to blow out a round and even end--neither thicker nor thinner than the rest of the tube. Sec. 21. To make a Weld. To begin with, try on two bits of glass of the same size, i.e. cut a seven-inch length of glass in half by scratching it with the knife, and pulling the ends apart with a slight inclination away from the scratch. In other words, combine a small bending moment with a considerable tensional stress. It is important to learn to do this properly. If the proportions are not well observed, the tube will break with difficulty, and the section will not be perpendicular to the main length. If the knife is in good order it will make a fine deep scratch--the feel of the glass under the knife will enable the operator to decide when the scratch is made. The operation of cutting large tubes will be treated further on. The two halves of the tube being held one in each hand, and one tube closed at one end, the extremities to be united will be warmed, and then put in the flame as before. Fig. 16.