however, the guides compelled the block to travel round a curve, then
the power, instead of being divided between the two sides in such
proportion as is necessary to relieve the guides of all strain, is
suddenly applied only to the inside, and the effect is that of a single
driver only. This is the clutch. Lastly, if the last-mentioned beam,
instead of being pushed along parallel to itself, were pivoted in the
middle, and that pivot only pushed, the same power would be applied to
each side of the block, and no strain would be thrown on the guides,
whether straight or curved, so long as the resistance opposed to the
block on the two sides were equal; if, however, one side met with more
resistance than the other, then the guides would have to keep the block
straight. This is the differential gear.

I have assumed that in the last case the force was applied to the middle
of the beam; this corresponds to every evenly-balanced gear. In the gear
employed by Singer, which is not evenly balanced, but which derives its
good qualities from its simplicity, the same effect is produced as
if the beam were pivoted on one side of the center instead of on the
center. Thus, though both sides are driven, one is driven more than the
other. On the whole, there is no doubt that the balanced gear gives a
superior action to the clutch, for except when the two sides of the
machine meet with very different resistance, and then only when running
straight, the clutch will not compare with the other. The clutch also
gives rise to what is considered by most riders a grave defect, the
inability to back treadle, while the free pedal, which is an immediate
consequence, is considered by others a luxury.

On the other hand, this same free pedal can be obtained on
differentially driven machines to which speed and power gear have been
applied.