I take it that there is only one truly tenable theory of castering; this when applied will obviate “sensitiveness” completely and under all circumstances; it is as follows: The pivotal connection must be such that the line _a b_ strikes in front of the point of support, as before spoken of, and it must also be so constructed and placed in such a position that no motion of the handle-bar will cause the machine to lower its centre of gravity. If by turning the handles any weight is lowered, you can depend upon it that the force of gravity, always tending to lower this weight, will inversely cause the handle-bar to turn. It will be noticed that when the machine stands upright the steering apparatus is not in a state of stable equilibrium; that is to say, the weight of the machine tends to shift the wheel, and it can hardly keep straight by means of such castering element as results simply from the line of the pivotal connection striking in front of the point of contact. The necessary conditions are as follows (see Fig. 2): The pivotal line _a b_ must strike at _c_ in front of _d_, and the line _a b c_ must be vertical in order that no motion on its axis can lower any weight when the machine stands upright. Now, it follows from these conditions that the head must be vertical and no part of the pivotal line in the rear of a vertical through the centre of the wheel.[6] The automatic steering devices do not work as successfully on a bicycle as on the leading wheel of a tricycle. There are two principal plans which have been in use; in one of which a spring forces the steering-bar into a position for running straight ahead; the other plan for the same purpose consists in a V slot and pin. In the latter the weight of the rider keeps the wheel straight by forcing the pin into the bottom of the V slot, and it will rest there until forced out by the action of the handle-bar. Either of the above devices is objectionable in a bicycle, because the constant working of the steering-bar for the purpose of balancing is so continuous, as compared with that of steering pure and simple, that any force tending to hold it in any one position will soon tire the arms and make riding more laborious. A new form of the rear-driving Safety was shown in the season of 1887, invented by a German. I give herewith a cut of the same, citing what he claims. [Illustration: The “Rothigiesser” principle.] “Can be ridden any distance and on any road without using the handle-bar. The new principle—pedals on the rear wheel and saddle on the front wheel—is just the reverse of the construction of the ordinary bicycle, and is the only true principle for a Safety; the fault of the common rear-driving Safeties being that both saddle and pedals are fixed on the rear wheel, so that the front wheel must be controlled by the arms of the rider.”