The operation is now simple. Fuse the junction and press the tubes lightly together, being careful not to collect more glass than can be helped; finally, blow out the joint and reduce the thickness by mild drawing (Fig. 33). In order to make a really good joint, two points must be particularly attended to--the rim must be thin and its plane perfectly perpendicular to the axis of tube B; the end of tube A must be cut off quite clean and perpendicular to its axis before B is inserted. So important are these conditions--especially the latter that the writer has even occasionally used the grindstone to get the end of A into a proper condition, an admission which will probably earn the contempt of the expert glass-worker. Fig. 33. Now for the second method, which is often practised in Germany, where soda glass is chiefly used. With this glass the chief point is to get a very even and not too thick ring at the junction, and consequently the extra thickening produced by making a rim on B is rather a drawback. The method consists in cutting off from B the length which it is desired to insert, slipping this into A (which may be an otherwise closed bulb, for instance), and then gradually melting up the open end of A till the piece of B inside will no longer fall out. By holding the joint downwards so that the inserted portion of B rests on the edges of the opening, a joint may be made with the minimum thickening. The external part of B, previously heated, is then applied, and the joint subjected to a "general" heat and blown out. Very nice joints may be made by this method, and it is perhaps the better one where the external part of B is to be less in diameter than the inserted part. It was in this manner that the writer was taught to make glass velocity pumps, one of which, of a good design, is figured as an example. In all cases good annealing should follow this operation. If the inserted part of the inner tube (B) is anything like an inch in diameter, and especially if it is of any length, as in some forms of ozone apparatus, or in a large Bunsen's ice calorimeter, the arrangements for supporting the inner part must be very good. A convenient way of proceeding when the inner tube is well supported is to make the mouth of A only very little larger than the diameter of B, so that B will only just slip in. Then the mouth of A and the zone of B may be heated together, and B blown out upon A. This, of course, must be arranged for, if necessary, by temporarily stopping the inner end of B.