The mechanism for removal of air from the sampling cans must provide not only for absolute uniformity and regulation of removal, i. e., the removal of a volume of air equivalent to the amount discharged into the sampling cans through the 10-mm. openings, but it must likewise force the air thus removed through proper purifiers for the quantitative absorption of water vapor and carbon dioxide, respectively. This may be accomplished by means of either a reciprocating pump or, as with the universal respiration apparatus, by a rotary positive blower. This blower, shown as F1 in figure 8, is belted with the motor, G, each sampling can having its corresponding blower. (See also fig. 11.) The blowers are actuated by the same electric motor, from the same shaft and hence at the same speed, for since the openings into the bottoms of the cans, C1 and C2, are exactly the same diameter and a like amount of air is discharged through both, the volume of air to be taken care of is exactly the same with both blowers. While other and less costly types of blowers have suggested themselves to us and have been subjected to preliminary tests, as yet nothing has been found that compares with the regularity of performance of the somewhat expensive Crowell blower.

The blower is adjusted to remove somewhat more than 45 liters of air per minute, the rate being determined entirely by the size of the pulleys and the speed of the motor. As at present employed, the diameter of the pulley on the blower is 20 cm. Any considerable excess in the amount of air removed in this manner may be controlled by the wheel by-pass shown as h1 and h2 in figure 11. This by-pass consists of a connection between inlet and outlet side of the blower, made up of standard brass pipe and fittings provided with a wheel valve. By opening the valve the effectiveness of the blower can be altered at will. Indeed, if it is opened wide no air will pass through the purifying train, since it simply will run back through the by-pass. Thus the amount of air withdrawn by the blowers from the sampling cans may be grossly regulated.

From the construction of the sampling cans and their rubber diaphragms (see fig. 13) it can be seen that, if the air is withdrawn too slowly, the rubber tops will gradually rise and become distended. Conversely, if the air is withdrawn too quickly, they will become more or less collapsed. From the fact that the actual weight of the diaphragm is in large part counterpoised by a spiral spring, considerable fluctuations in the general shape and size of the rubber top of the can may actually take place without a material alteration in the internal pressure on the diaphragm. A slight rise and fall, such as 1 or 2 mm., when, so delicately counterpoised, is absolutely without influence upon the tension inside the diaphragm. This fact has been taken advantage of to actuate a second by-pass on the blower, which is controlled by an electrical magnet.

This by-pass is constructed in the following manner: The pipes on each side of the by-pass, h1 and h2 (fig. 11), are extended and reduced in size to permit the insertion of a short length of thin-walled soft rubber tubing, m1 and m2. By placing this tubing back of the armature of a small magnet (an inexpensive telegraph sounder has been found suitable for the purpose), air may be allowed to pass at will through the rubber tube by opening or closing the magnet. A vertical movement of the rubber diaphragm of 1 mm. opens or closes a simple electrical contact which is actuated by two small disks (d1 and d2, figure 13) on the suspension rod of the diaphragm on each sampling can. This in turn opens or closes the magnet release and hence leaves the passage free or obstructed through the rubber tube of the by-pass m1 or m2. The details are shown in figures 11 and 13.

With this arrangement, if the blower draws somewhat too much air out of the sampling can, i. e., more than is delivered to it, the diaphragm tends to fall; this closes the electric contact which, in turn, draws down the telegraph sounder. The normal spring tension against the rubber tube m1 or m2 is released by this movement, thus allowing air free passage through the tube. Under these conditions the blower draws too little air from the can and the discharge from the wind chest is somewhat greater than its removal from the sampling can. This results in a slight raising of the rubber diaphragm, the breaking of the electric contact, and the closing of the by-pass m1 and m2. This alternate opening and closing is practically continuous, occurring several times each second. Under these conditions the position of the rubber diaphragm remains nearly constant and the pressure of the air inside the sampling chamber is atmospheric.

This constancy in atmospheric pressure may be tested by an extremely delicate Sonden petroleum manometer P (fig. 8), which is so mounted that, by turning a 3-way stopcock, it can be connected with either of the sampling cans at will. If the pressure inside the sampling can is atmospheric, the petroleum manometer indicates 0. The adjustment at the start is easily made by increasing or decreasing the tension on the spiral springs above the rubber diaphragm by means of the small regulating nuts, n1 and n2.

If the manometer shows that the pressure is too great inside the can, the weight of the diaphragm has not been suitably counterpoised; hence it is necessary to turn the nut so as to produce greater tension on the spring. When the pressure inside is below atmospheric, the operation is reversed. The small disks which actuate the electric magnet are attached to the vertical rod by tiny set screws and can be adjusted for any elevation. When once adjusted, however, very considerable alterations in the tension of the spiral spring may be made without the necessity of altering the position of the contacts materially.

Occasionally, through error or otherwise, the blower delivering air to the wind chest is connected before the rubber stoppers1 in b1 and b2 are removed. There is then a great distention of the diaphragm with a tendency to lift the rods so that the contact slips out from between the two small disks (d1 and d2, fig. 13). To prevent this a third disk, d3, is attached to the rod below the cross arm support in such a manner that the rubber diaphragm can not be raised sufficiently to do any harm to the contacts. It has been found advantageous to insulate all the parts by means of fiber washers. Hard-rubber bushings are placed in the top cross-arm, which insulate the entire electrical system from the table.

It is thus seen that we have provided a means for removing air from the sampling cans as fast as it is delivered, to maintain absolute and uniform atmospheric pressure inside them, to insure uniform discharge of air from the wind chest through the two 10-mm. openings and, furthermore, to drive the respective volumes of air withdrawn from the sampling cans through suitable purifying vessels to remove the water vapor and carbon dioxide. The absorption of the carbon dioxide and the weighing of the purifying vessels still remain to be described.