For many purposes the small dimensions of the flame render it very convenient, and the high temperature which can be attained at exact spots enables glass to be fused together after a certain amount of mixing, which is an enormous advantage in fusing lead glass on to hard glass. The lead glass should not be heated hot enough to burn, but, short of this, the more fluid it is the better for joints between dissimilar samples. It will be noticed that the blow-pipe can be rotated about a vertical axis so as to throw the flame in various directions. This is often indispensable. Sec. 15. In general the oxygen flame does not require to be delivered under so high a pressure as for the production of a lime light. In England, I presume, most experimenters will obtain their oxygen ready prepared in bottles, and will not have to undergo the annoyance of filling a bag. If, however, a bag is used, and it has some advantages (the valves of bottles being generally stiff), I find that a pressure produced by placing about two hundredweight (conveniently divided into four fifty-six pound weights) on bags measuring 3' x 2'6" x 2' (at the thicker end) does very well. To fill such a bag with oxygen, about 700 grms of potassium chlorate is required. If the experimenter desires to keep his bag in good order, he must purify his oxygen by washing it with a solution of caustic soda, and then passing it through a "tower" of potash or soda in sticks, and, finally, through a calcium chloride tower. This purifying apparatus should be permanently set up on a board, so that it may be carried about by the attendant to wherever it is required. Oxygen thus purified does not seem to injure a good bag--at least during the first six or seven years: In order to reduce the annoyance of preparing oxygen, the use of the usual thin copper conical bottle should be avoided. The makers of steel gas bottles provide retorts of wrought iron or steel for oxygen-making, and these do very well. They have the incidental advantage of being strong enough to resist the attacks of a servant when a spent charge is being removed. The form of retort referred to is merely a large tube, closed at one end, and with a screw coupling at the other; the dimensions may be conveniently about 5 inches by 10. The screw threads should be filled with fireclay (as recommended by Faraday) before the joint is screwed up. Before purchasing a bottle the experimenter will do well to remember that unless it is of sufficiently small diameter to go into his largest vice, he will be inconvenienced in screwing the top on and off. Why these affairs are not made with union joints, as they should be, is a question which will perhaps be answered when we learn why cork borers are still generally made of brass, though steel tube has long been available. Fig. 4.