If the quantity of the air, the goodness of which I want to ascertain, be exceedingly small, so as to be contained in a part of a glass tube, out of which water will not run spontaneously, as _a_ fig. 15; I first measure with a pair of compasses the length of the column of air in the tube, the remaining part being filled with water, and lay it down upon a scale; and then, thrusting a wire of a proper thickness, _b_, into the tube, I contrive, by means of a thin plate of iron, bent to a sharp angle _c_, to draw it out again, when the whole of this little apparatus has been introduced through the water into a jar of nitrous air; and the wire being drawn out, the air from the jar must supply its place. I then measure the length of this column of nitrous air which I have got into the tube, and lay it also down upon the scale, so as to know the exact length of both the columns. After this, holding the tube under water, with a small wire I force the two separate columns of air into contact, and when they have been a sufficient time together, I measure the length of the whole, and compare it with the length of both the columns taken before. A little experience will teach the operator how far to thrust the wire into the tube, in order to admit as much air as he wants and no more. In order to take the electric spark in a quantity of any kind of air, which must be very small, to produce a sensible effect upon it, in a short time, by means of a common machine, I put a piece of wire into the end of a small tube, and fasten it with hot cement, as in fig. 16; and having got the air I want into the tube by means of the apparatus fig. 15, I place it inverted in a bason containing either quicksilver, or any other fluid substance by which I chuse to have the air confined. I then, by the help of the air pump, drive out as much of the air as I think convenient, admitting the quicksilver, &c. to it, as at _a_, and putting a brass ball on the end of the wire, I take the sparks or shocks upon it, and thereby transmit them through the air to the liquor in the tube. To take the electric sparks in any kind of fluid, as oil, &c. I use the same apparatus described above, and having poured into the tube as much of the fluid as I conjecture I can make the electric spark pass through, I fill the rest with quicksilver; and placing it inverted in a bason of quicksilver, I take the sparks as before. If air be generated very fast by this process, I use a tube that is narrow at the top, and grows wider below, as fig. 17, that the quicksilver may not recede too soon beyond the striking distance.