The reason of this is chiefly that the under side cools most rapidly (according to Faraday, Chemical Manipulation, Sec. 1194), and consequently can expand no further; but also it is not unlikely that the glass tends to flow somewhat from the upper side, which remains hot, and consequently the bulb, when the next puff reaches it, will tend to yield at this point. By heating several zones the tube will become gradually expanded. Fig. 24. Fig. 25. Fig. 26. When the length of the thickened part of the tube only slightly exceeds its diameter (Fig. 25), let the whole be brought to a temperature which, with flint glass, should be just short of that of perfect fluidity; and then, holding the tube horizontally and constantly turning it, let the bulb be blown out to its full size, noting the appearances and correcting too great protuberance on any side by the means above mentioned. If the bulb appears pear-shaped turn the tube so that the melted mass is directed upwards; if the bulb have the contrary fault, correct in the corresponding manner. The bulb when finished may be lightly tapped on the table, when, if there is any weak place owing to inequality of thickness, the bulb will break, and the operation may be started afresh. "A good bulb is round, set truly on the tube, and free from lumps of thick glass or places of excessive thinness." When the amateur has succeeded in blowing a bulb two inches in diameter on the end of a strong bit of thermometer tube--say for an air thermometer--he may well seek the congratulations of his friends. In case the bulb is not satisfactory on a first attempt, it may be melted down again, if the following precautions are taken. Directly creases begin to appear in the bulb let it be withdrawn from the flame, and gently blown till the creases come out. By alternate heating and blowing the glass can be got back to its original form, or nearly so, but unless the operator shows great skill and judgment, the probability is that the glass will be uneven. By heating and keeping the thicker parts in the higher position, and blowing a little now and again, the glass may be got even, and a new attempt may be made. It must not be supposed that this process can be carried on indefinitely, for the glass tends to lose its viscous properties after a time, or, at all events, it "perishes" in some way, especially if it has been allowed to get very thin; consequently too frequent attempts on the same glass are unprofitable. Two or three trials are as many as it generally pays to make. As a rule the largest possible flame may be used with advantage in this operation. Sec. 32. To blow a bulb in the middle of a tube, the procedure is much like that already treated, but the manipulation is, if anything, more difficult, for the further end of the tube must be carried and turned as well as the end which is held to the lips. Sec. 33. To make a side Weld.