[huge snippage for brevity, apologies for rubbish formatting]
I'm not 100% sure I'm right in what follows (it's been a long time) but
improvements are welcome. It builds on much of what has already been said.
Termination has been covered by various contributors - termination reduces (but may not
completely eliminate) reflections.
Reflections in a single segment setup (two boxes, one cable) are relatively simple to cope
with in a setup from the Qbus era where things aren't particularly fast.
In a three box (two cable) setup my recollection is that the configuration rules require
the cables to be of significantly different lengths, and the reason for this is to ensure
that the two sets of reflections are timed signifcantly differently and canot make Bad
Things happen by arriving at the same time as each other.
Consider the middle box (of 3) is driving a bus transition. Signals will propagate from
the middle box to each of the ends. When the moving rope er sorry voltage transition
reaches the end of the cable, it will be reflected to some extent. If the cable segments
are both the same length(ish), the reflections will come back to the middle box at round
about the same time, superimpose on each other, and potentially cause confusion. (Does
that sound plausible?)
If the cable segments are of significantly different lengths then the reflections will
arrive back at the middle at significantly different times and the reflections will be
more manageable - less risk of Bad Things happening when they superpose.
Or something along those lines.
Anyway, hopefully the "different cable lengths so the reflection timings are
different" will ring a few bells even if it's not actually right.
Have a lot of fun
John Wallace