Curiously, I didn't say it wasn't evolution, just that it was not an example of a mutation arising that shows a benefit -- the mutation was already extant in the population, and the melanic variety was known about well before hand.
There are actually a variety of genes that lead to the melanic form; and these will arise by spontaneous mutation from time to time - a fact which is likely to be partiarly responsible for the maintainance of the melanic form in normal populations - so it is also likely that some of the melanic forms selected during the industrial period arise by new mutation during that period so both you and Greyseal are partially correct.
The short answer is not exactly, but most current genetic research points to retroviruses being derived from transposable elements native to organismal genomes called Long Terminal Repeat (LTR) retrotransposons.
Really? Interesting. I was under the impression the generally held view was that retroviruses pre-dated the domain split, since certain retroviral proteins from different domains of life are most closely related to each other than they are to any host protein. I was also under the impression that transposons were thought most likely to be degenerate viruses whose evolutionary pathway had taken them away from having a separate infective stage.