The Manchester encoding I know about is slightly different than the one Rancid_Pickle wrote about. First of all, there are two 'polarities' associated with Manchester encoding: one can choose low-to-high to represent either a digital '1' or a digital '0'. I normally associate low-to-high with digital '0', but that's a choice that varies from individual to individual (and 'standard' to 'standard').

More significantly, however, the Manchester encoding I know always has the transition exactly in the middle of the bit. Therefore, a digital '0' is always represented by half a bit period of low signal followed by half a bit period of high signal, and a digital '1' is always represented by half a bit period of high signal followed by half a bit period of low signal. The test data in the above writeup then look like this:

```Data   <-0-> <-1-> <-0-> <-1-> <-1-> <-1-> <-0-> <-0->
-----       -----    --    --       --    --
|     |     |     |  |  |  |  |     |  |  |
|     |     |     |  |  |  |  |     |  |  |
--       -----       --    --    -----    --
Time  0     1     2     3     4     5     6     7     8
```

A lot more regular-looking a waveform than in the above writeup. It has a number of features worth noting. Every pulse is exactly half a bit wide or exactly one bit wide. When the data are unchanging, the signal is a regular square wave with a frequency equal to the bit rate. When the data are alternating, the signal is a regular square wave with a frequency equal to half the bit rate. A sequence of '1's always begins with a high pulse a full bit period wide, a sequence of '0's always begins with a low pulse a full bit period wide (except for the very first bit transmitted).

A more technical name for Manchester encoding is fully-coherent bi-phase encoding. Differentially-coherent bi-phase encoding is achieved by performing a gray code to binary transformation on the data, and transmitting the result with Manchester encoding.