To facilitate choice, there is appended a diagram of sizes from the catalogue of a reliable German firm, Messrs. Desaga of Heidelberg, and the experimenter will be able to see at a glance what sizes of glass to order. It is a good plan to stock the largest and smallest size of each material as well as the most useful working sizes. Fig. 1. Sec. 11. Testing Glass. "Reject glass which has lumps or knots, is obviously conical, or has long drawn-out bubbles running through the substance." If a scratch be made on the surface of a glass tube, and one end of the scratch be touched by a very fine point of fused glass, say not more than one-sixteenth inch in diameter, the tube, however large it is (within reason), ought to crack in the direction of the scratch. If a big crack forms and does not run straight, but tends to turn longitudinally, it is a sign that the glass is ill annealed, and nothing can be done with it. If such glass be hit upon in the course of blow-pipe work, it is inadvisable to waste time upon it; the best plan is to reject it at once, and save it for some experiment where it will not have to be heated. The shortest way of selecting glass is to go to a good firm, and let it be understood that if the glass proves to be badly annealed it will be returned. Though it was stated above that the glass should not be distinctly conical, of course allowance must be made for the length of the pieces, and, on the other hand, a few highly conical tubes will be of immense service in special cases, and a small supply of such should be included. The glass, as it is obtained, should be placed in a rack, and covered by a cloth to reduce the quantity of dust finding its way into the tubes. It has been stated by Professor Ostwald that tubes when reared up on end tend to bend permanently. I have not noticed this with lead glass well supported. Each different supply should be kept by itself and carefully described on a label pasted on to the rack, and tubes from different lots should not be used for critical welds. This remark is more important in the case of soda than of lead glass. In the case of very fine thermometer tubes it will be advisable to cover the ends with a little melted shellac, or, in special cases, to obtain the tubes sealed from the works. Soda glass can generally be got in rather longer lengths than lead glass; the longer the lengths are the better, for the waste is less.