This is David Bell's original message from the life list (actually it's from October 2006). I hope this is OK:
4757 From: David Bell
Date: Sun Oct 8, 2006 6:57 am
Subject: Idea for various speed Caterpillars based on standard spaceships
It might be possible to make Caterpillar spaceships out of just standard
spaceships with no exotic crawling reaction required. The idea is very
simple, so please let me know whether it could really work! If it does
work, it should be possible to create Caterpillars with lots of different
To illustrate the idea:
Using an appropriate helix, create two downward moving streams of
standard spaceship convoys one of which can perform a glider duplication
reaction and the other a glider reflection reaction. (These convoys take
the place of the blinkers in the existing Caterpillar. They are much
more complicated to build from a helix than just blinkers, but it should
still be possible.)
For the "exotic" crawling reaction, just use a glider! It just bounces
back and forth between the glider duplication and reflection convoys,
slowly moving upwards, liberating an output glider at each encounter with
a glider duplication reaction.
Multiple gliders can be running through the glider duplication convoys at
the same time to create lots of output gliders, just like multiple PIs
run through the existing Caterpillar.
These output gliders are then used to construct the helix which then
completes the loop.
I don't think that there is a need to multiply the base period of the
reaction like the 6x in the existing Caterpillar. The helix only has to
generate a single set of convoys since the spacing of the glider reactions
can be adjusted to match. Similarly, there is no need for a filter stream
for the output gliders.
The speed of the Caterpillar is determined by the upwards speed of the
gliders between the times that they are duplicated and turned, so it
would be near c/4, but slightly different due to the reactions involved.
Since the spacing of the two convoys is adjustable, the actual speed can
also be adjusted. The further apart the convoys are, the closer the speed
would be to c/4. I can think of ways to adjust the speed further, such
as having one of the glider paths go backwards to make a VERY slow
spaceship (but I don't know if this can work), or using a "catch and throw"
reaction to delay when the glider is sent back on its way.
Finally, in practice, more convoys would actually be needed. For example,
for the left and right side, for forwards and backward glider duplications,
for phase and color capabilities, and for "catch and throw" reactions.
This are details which I haven't thought about and which someone else
[I just realized that even glider bouncing isn't really necessary. A still
life can be activated by a passing convoy to create an output glider,
and then recreated a bit forwards in the spaceship so that the next convoy
can reuse it (to create a VERY slow spaceship again.) So this is something
else which can be investigated.]
Here is a sample spine:
x = 351, y = 771, rule = B3/S23