The branch and switch system has many attractions for the inventor, and upon first thought it would seem the most feasible solution of the problem. It has been the dream of more than one inventor, as the records of the patent-office show, but no one has succeeded in working it out. The current of air cannot be divided; carriers passing from the branch into the main line must not collide with other carriers running in the main line; a certain minimum distance must always be maintained between the carriers in the same tube; when a carrier is despatched it must go directly to the station for which it is intended without further attention from the sender and it must not interfere with other carriers; expense of manufacture prohibits the use of any but round smooth tubes up to eight inches in diameter, hence projections cannot be placed upon the carrier to give it an individuality and cause it to operate a switch at any particular point along the line; the carrier is free to rotate in a round tube about its longitudinal axis, therefore, its individuality must be indicated by some symmetrical marking about this axis, if it is to be automatic in its operation; the speed of the carrier is so high that electrical contacts placed in distinctive positions on the carriers cannot be used while it is in motion, for mechanism having inertia could not be moved during the short time that the electric circuit would be closed; only the simplest attachments can be made to the carrier, for constructional reasons and because of the rough usage that they receive. These and numerous other reasons make the problem most difficult. We have not attempted to solve it by the use of branch tubes and electrically operated switches, but have adopted the simpler and equally effective method of carrying the main line through each of the stations that it unites. In our system each carrier has an individuality determining the station at which it will be discharged from the tube. By a simple attachment to, the front end of the carrier, consisting of a circular metal disk, the sender so marks the carrier that it will pass all stations until it arrives at the station for which it was destined and will there pass out of the tube. In addition to this a method has been devised whereby carriers can be inserted into the tube without the possibility of collision with carriers already running in the tube. [Illustration: FIG. 21. A DIAGRAM SHOWING VARIOUS METHODS OF CONNECTING THE STATIONS OF A LARGE SYSTEM WITH PNEUMATIC TUBES.]