Many years ago (20?), I was musing on circuits to implement Manchester
coding.

  As I understood it, there was a "Type I" and a "type II" Manchester
coding, type I being regular clock transitions with data transitions
(present or not present) between.  At 1200 bps, this was the Kansas City
Standard for tape storage and other things.  The alternate description
of it was that one state was coded as a half cycle of 600Hz square wave
while the other state was a full cycle of 1200Hz.

  If it could fit a tape recorder, it should have passed down a
telephone line too!  The type II encoding I understood to be a
transition at clock time for one state, no transition for the other
state, alternatively called NRZ or NRZI depending on which state is
regarded as which.  This requires bit-stuffing and flags or an
asynchronous protocol (with guaranteed periodic stop bits corresponding
to transitions) to maintain clock synchronisation and is used in HDLC
protocol.

  With filtering, this is used as the basis for the 4800 baud HAPN
protocol on Amateur Packet Radio, but it very touchy to tune.  It
appears that we will be reverting to 1200 baud (Bell 212) FSK on our
packet network in this area for the near future!

  Cheers,
        Paul B.