If in spite of these precautions such a lump does form, proceed as follows. Take a rod of glass, at least one-eighth of an inch thick, and warm it in the flame at one end. Heat the imperfect joint till it softens all round, and then bring the flame right up to the thick part, and heat that as rapidly and locally as possible. The oxygas flame does this magnificently. Press the heated end of the glass rod against the thick part, and pull off as much of the lump as it is desired to remove, afterwards blowing the dint out by a judicious puff. Finish off as before. Sec. 27. Occasionally, when it is seen that in order to produce a joint closed all round, one side of the tube would be too much thickened, it is better to patch the open side. For this purpose take a glass rod about one-sixteenth inch in diameter, and turn the flame to give its greatest effect, still keeping rather an excess of air or oxygen. See that the side of the joint already made is kept fairly hot--it need not be soft; interrupt any other work often enough to ensure this. Then, directing the flame chiefly on the thin rod, begin to melt and pull the glass over the edges of the gap. When the gap is closed get the lump very hot, so that all the glass is well melted together, and then, if necessary, pull the excess of glass off, as before described. It must be remembered that this and the method of the previous section are emergency methods, and never give such nice joints as a manipulation which avoids them, i.e. when the ends of the tubes are perfectly straight and true to begin with. Also note that, as the tubes cannot be kept in rotation while being patched, it is as well to work at as low a temperature as possible, consistently with the other conditions, or the glass will tend to run down and form a drop, leaving a correspondingly thin place behind. Fig. 20. Sec. 28. A very common fault in cutting a tube of about an inch in diameter is to leave it with a projecting point, as shown. This can be slowly chipped off by the pliers, using the jaws to crush and grind away the edge of the projection; it is fatal to attempt to break off large pieces of glass all at once. Sec. 29. It will be convenient here to mention some methods of cutting large tubes. With tubes up to an inch and a half in diameter, and even over this--provided that the glass is not very thick--we may proceed as follows: Make a good scratch about half an inch long, and pretty deep, i.e. pass the knife backwards and forwards two or three times. Press a point of melted glass exactly on one end of the scratch; the glass point even when pressed out of shape should not be as large as a button one-twelfth of an inch in diameter. If this fails at first, repeat the operation two or three times. Fig. 21.