It is impossible to lay down a rigid system equally well adapted to all places and purposes. We must accommodate ourselves somewhat to circumstances. For example, the post-office department may require one size of tube, arranged to operate in a particular way, while the requirements of a parcel delivery business would be utterly different. The geographical location of the stations will have much to do with the general arrangement; also the condition of the streets. Some of the streets of our large cities are so filled with water- and gas-pipes, electrical conduits, sewers, steam-pipes, etc., etc., that it is almost impossible to find space for pneumatic tubes, especially of large diameter. Railway or water facilities have much to do with the location of a central pumping station, on account of the coal supply. All of these and many other things have to be taken into consideration in planning a system for any locality. We have an example of a peculiar location and conditions in a proposed line of tubes over the New York and Brooklyn bridge connecting the main post-offices of those cities. This would be in many respects a unique plant. Two air-compressors would be used, one at each office. In order to give a general idea how a large number of stations can be connected into one system, the diagram Fig. 21 has been drawn. We have already referred to the attempts of Clay and Lieb to devise means whereby several stations could be located along a main line and carriers be sent from any station to any other through the main line. Their method was to use branch tubes leading off from the main line with switches at the junctions. They deflected the air-current into the branch by placing an automatic closed valve in the main line just beyond the junction, returning the air from the branch to the main line just beyond the valve. The carriers were to open and close this valve automatically as they passed.