THE experimental production of oxygen upon a small scale with a test tube or glass retort Is very useful in demonstrating a chemica! point, where only a small quantity of oxygen is needed. When a larger quantity of this gas is required to fill a dozen large vessels or more, for ledure purposes, or for use with Ihe oxyhydrQgen blowpipe, for the fusing of metals and other refractory bodies, then resort must be bad to a more substantial type of generator. The type of retort and the wash bottles here described have been used by the writer for several years successfully, particularly for the production of oxygen for limelight, for stereopticon and dissolving view apparatus, and for lecture room demonstration. There are many handy men who will be able to make both the retort and the wash bottles, these latter being made of zinc. Fig. 1 shows the body of the retort, which is made of moderately stout sheet iron, with a flange bent outward at the bottom. B represents the bottom, made of sheet copper. The flange of the iron cone being filed clean, the copper is beaten over this flange, then brazed all around in a blacksmith's fire with spelter, using borax lor the flux. The screw socket at C consists simply of an ordinary iron pipe socket, such as is used for connecting wrought-iron gas pipe. This is also spelter brawd into the top of the cOne. D is a piece of iron gas pipe of one-inch bore, with a screw cap at the top, an'l E is a piece of three-eighths inch bore iron pipe, with a serew at one end, so as to ft into a hole that has been tapped to receive it at the top of D. This outlet p L pe. also answers the purpose of a handle for removing D, so as to place the charge in the retort, as well as acting as a lever to screw the joint quite tight to prevent the escape of oxygen. Fig. 2 shows the arrangement of retort and wash bottles when in action generating oxygen. The advantage of this lcind of retort is that it will resist almost any amount of wear and rough usage. At F in Fig. 2 is one of the wash bottles in section, showing the oxygen passing down the inlet pipe at G and through· the outlet pipe at H, then down the inlet pipe at J of the second wash bottle, thence out at K, to the receptacle, which may be either a metal gas holder or an India rubber gas bag. These inlet and outlet pipes are simply made of the ordinary composition metal gas tubing, frmly soldered into the zinc bottles. The connecting tube is a good quality of vulcanized India rubber. Each of the bottles should hold about three pints. To prepare the oxygen, mix one pound and a half Making oxygen in moderate quantities. of potassium chlorate in crystals, and half a pound of peroxide of manganese (the ordinary black oxid8 manganese). The use Of the manganese prevents the chlorate from fusing into a mass, and aids the production of oxygen. The wash bottles must be charged with a pint and a half of cold water. The object 0[ this is to absorb and retain any trace of chlorine that might pas'S over in case the retort became over-heated, and to cool the gas. The quantity of potassium chlorate here given will produce well over six cubic feet of pure oxygen-in fact, enough to allow of some waste, which is sure to take place in preparing this ,as unless exceptional care is taken. As sCOn as the gas ceases to be evolved, disconnect the tubing, remove the retort from the fire 01' gas stove, place it upon the ground or stone floor, press one foot upOn the, sIde of the cone, give the tube E a sharp rap with a hamme'!', this will start the screwed joint; remove D entirely and pour a pint of water (no matter whether it is hot or cold) into the hot retort. This will 100sen the contents quickly. The retort should be well washed out and dried, so as to be ready for future use. The water in the wash bottles must be poured off, and the bottles allowed to drain. For those who wish to prepare their own oxygen, the foHowing facts may prove useful: Eight ounces of potassium chlorate will make, under ordinary circumstances, rather more than two cubic feet of oxygen. The chlorate may vaTY a little in quality, but for all practical purposes these figures will supply the necessary information. Those who wish the information a little more accurate for making oxygen on a small scale will find that one hun.ded grains of pure potassium chlorate will yield one hundred and eight cubic inches of pure oxygen. There are many other cher.i(:als that will yield oxygen UPOll heating, but there is nothing that will surpass the potassium chlorate and leroxide of manganes e mixture.