works. Around the inside of each shaft extended a
spiral track upon which ran rollers attached to revolving frames
underneath the cars. When the frames were made to revolve, the rollers,
running around the track, would raise or lower one car, the other
traveling in the opposite direction (fig. 23).

[Illustration: Figure 23.--Backmann's proposed helicoidal elevator for the
upper section of the Eiffel Tower. The cars were to be self-powered by
electric motors. Note similarity to the Miller system (fig. 20). (Adapted
from _The Engineer_ (London), Aug. 3, 1888, vol. 66, p. 101.)]

In the plan as first presented, a ground-based steam engine drove the
frames and rollers through an endless fly rope--traveling at high speed
presumably to permit it to be of small diameter and still transmit a
reasonable amount of power--which engaged pulleys on the cars. The design
was remarkably similar to that of the Miller Patent Screw Hoisting
Machine, which had had a brief life in the United States around 1865. The
Miller system (see p. 19) used a flat belt rather than a rope (fig. 20).
This plan was quickly rejected, probably because of anticipated
difficulties with the rope transmission.[9]

Backmann's second proposal, actually approved by the Commission,
incorporated the only--although highly significant--innovation evident in
his designs. For the rope transmission, electric motors were substituted,
one in each car to drive the roller frame directly. With this
modification, the plan does not seem quite as unreasonable, and would
probably have worked. However, it would certainly have lacked the
necessary durability and would have been extremely expensive. The
Commission discarded the whole scheme about the middle of 1888, giving two
reasons for its action: (1) the novelty of the system and the attendant
possibility of stoppages which might seriously interrupt the "exploitation
of the Tower," and (2) fear that the rollers running around the tracks
would cause excessive noise and vibration. Both reasons seem quite
incredible when the Backmann system is compared to one of those actually
used--the Roux, described below--which obviously must have been subject to
identical failings, and on a far greater scale. More likely there existed
an unspoken distrust of electric propulsion.

That the Backmann system should have been given serious consideration at
all reflects the uncertainty surrounding the entire matter of providing
elevator serv