is a very large proportion of the total pressure and reduces the efficiency. Besides these few suggestions, only the requirements of good engineering need be demanded. In Figs. 42, 43, 44, 45, 46, and 47 we show a fan, two blowers, and an air-compressor suited to the requirements of pneumatic-tube service that can be found in the market, and that are built by responsible concerns. We believe they are all good of their kind, but do not recommend any particular make. =The Tube, Line Construction, etc.=—Up to the present time we have found no material better suited for the straight parts of pneumatic tubes than cast iron, machined upon the interior. It gives a smooth and accurate tube. It can be made in most convenient lengths. It is strong and not easily deformed. The bell-joint, calked with lead and oakum, having the tubes fitted together male and female at the bottom of the bell, is the best joint yet devised for pneumatic tubes. It is slightly yielding, accommodating itself to slight changes of length of tube due to changes of temperature, and it allows slight bends to be made at each joint. The joints are very accurate, presenting no shoulders to obstruct the passage of carriers. The joints can be made by men accustomed to laying water- and gas-pipe. The cast iron is so stiff that it is not distorted in calking, as may be done with wrought-iron tube. The principal objections to its use are the expense of boring and the readiness with which it corrodes upon the interior. [Illustration: FIG. 47. RAND COMPOUND COMPRESSOR OF MODERATE SIZE.] We are always hoping that wrought-iron or steel tubes will be so much improved in uniformity of dimensions and smoothness of interior that we can use them, but our experiments thus far have been discouraging. It may be that some of the new processes of making tubes will give us what we want, but we have not yet found it. Small tubes and the short bends of large tubes are made of brass, it being the most suitable material. It would be very difficult to bend iron tubes without involving great expense. The thickness of the bent portion of an eight-inch tube is usually three-sixteenths of an inch and never less than one-eighth of an inch.