The inner support of B should be removed as soon as practicable after the joint is made, or, at all events, should not be perfectly rigid; a tightly-fitting cork, for instance, is too rigid. The reason is, of course, that in cooling there may be a tendency to set B a little to one side or the other, and if it is not free to take such a set, the joint most probably will give way. Good annealing both with flame and asbestos is a sine qua non in all inserted work. Fig. 34. Sec. 35. Bending Tubes. I have hitherto said nothing about bending tubes, for to bend a tube of a quarter of an inch in diameter, and of ordinary thickness, is about the first thing one learns in any laboratory, while to bend large tubes nicely is as difficult an operation as the practice of glass-blowing affords. However, even in bending a narrow tube it is possible to proceed in the wrong way. The wrong way is to heat a short length of the tube and then bend it rapidly, holding the plane of the bend horizontal. The right way, per contra, is to use a batswing burner to heat, say, two inches of the tube with constant turning till it is very soft, and then, holding the glass so that the bend will be in a vertical plane passing through one eye (the other being shut), to make the bend rather slowly. If an exact angle is required, it is as well to have it drawn out on a sheet of asbestos board. In this case bend the glass as described till it is approximately right, and finish by laying it on the asbestos board and bringing it up to the marks. A suitable bit of wood may be substituted for the asbestos on occasion. N.B: The laboratory table is not a suitable piece of wood. A right-angled bend is often wanted. In this case the corner of a table will serve as a good guide to the eye, the glass being finished by being held just above it. If great accuracy is wanted, make a wooden template and suspend it by a screw from the side of the table, so that the vertex of the gauge for the interior angle projects downwards, then finish by bending the tube round it. The wood may be about half an inch thick. If a sharp bend is required, heat the tube in the blow-pipe, and bend it rapidly, blowing out the glass meanwhile. The reason why a long bend should be held in a vertical plane is that the hot part tends to droop out of the plane of the bend if the latter be made in a horizontal position. To bend a tube above half an inch in diameter is a more or less difficult operation, and one which increases in difficulty as the diameter of the tube increases.