The tip of the inner cone of the blow-pipe is brought to bear directly upon one of the fragments, and if the operation is performed boldly it will be found that the surface of the fragment can be fused, and the fragment thus caused to hold together before the lower side gets hot enough to suffer any contamination from the tile or brick. A second fragment may be treated in the same way, and then a third, and so on. Finally, the fragments may be fused together slightly at the corners, and a stick may thus be formed. Of course a good deal of cracking and splitting of the fragments takes place in the process; the best pieces to operate upon are those which are well cracked to begin with, and that in such a way that the little fragments are interlocked. An alternative method which has some advantages is to arm a pair of forceps with two stout platinum jaws, say an inch and a half long, and flattened a little at the ends. The fragments are held in these platinum forceps and the blow-pipe applied as before. This method works very well in adding to a rod which has already been partly formed, but the jaws require constant renewals. The first fragment which is fused sufficiently to cohere may also be fused to a bit of tobacco pipe, or hard glass tube or rod, and the quartz stick gradually built up by fusing fresh pieces on to the one already in position. Fig. 64. Since the glass or pipeclay will contaminate the quartz which has been fused on to it, it is necessary to discard the end pieces at the close of the operation. A string of fragments having been collected and stuck together, the next step is to fuse them down into a uniform rod. This is easily done by holding the string in the blow-pipe flame and allowing it to fuse down. Twisting the fused part has a good effect in assisting the operation. It is desirable to use a large jet and as powerful a flame as can be obtained during this part of the operation. The final result should be a rod, say two or three inches long and one-eighth of an inch thick, which will in most cases contain a large number of air bubbles. Since the presence of drawn-out bubbles cannot be advantageous, it is often desirable to get rid of them, and this can conveniently be done at the present stage. The process at best is rather tedious; it consists in drawing the quartz down very fine before an intense flame, in order to allow the bubbles to get close enough to the surface to burst. A considerable loss of material invariably occurs during the process; for whenever the thin rod separates into two bits the process of flame-drawing of threads goes on, and entails a certain waste; moreover, the quartz in fine filaments is probably partially volatilised.