​Right -- what gets a little messy is that you need to understand the *roff, the C compiler and the base OS were all being changed and released in parallel of each other.   How different features made it into the wild was not linear.

I mentioned the roff to nroff to troff to ditroff stream earlier today.  What I do not remember is the independent troff releases.   I think I remember that there was an early troff that went out independent of one of the kernels.  The problem is that like Noel I know we had it at CMU and UCB, but I do not remember its provenance.   As Noel points out, few of us outside of Bell Labs had C/A/T typesetters.   But there was  C/A/T typesetter emulator for the Versatec and Variam plotters called 'vcat' which converted the typesetter codes into pixels for the plotter and used the 'Hershey Fonts' which were the first generally available bit maps around at I believe 200 dpi [this was originally distributed in an early USENIX tape IICR and later came out as part of 2 or 3 BSD -- I believe Tom Ferrin at UCSF had his hand in its creation].

As Noel and I have talk about in the past, V6 beget PWB 1.0 - the former from MH and the later out of Whippany IIRC - and there are some small difference in the kernels.  Ken and Dennis​ continued there kernel work at MH and a number of kernel changes were done by them that were substantial from V6 -- this kernel is generally called the UNIX/TS kernel.

There was never an official UNIX/TS release, but it `leaked' from MH primarily by the AT&T employees on their OYOC year [One Year On Campus - or some times One Year in California ;-)    For instance Ted brought it to us at CMU and Noel and team had that kernel at MIT.      UNIX/TS would become the basis of the kernel that went out in V7; with a number of other additions and changes.

Dennis and Steve of course were working on changes to the language.   These compilers would come out inside of AT&T as needed but we less available directly to the rest of us.   Again how they made it outside was sometimes a tad random.   Sometime, I want to say 1976-77 time frame Dennis and Brian were also writing the 'White Book' - aka K&R [again I saw a xerographic draft that Ted had at the time and wishes it had been around a few years earlier].   Please remember that until the publishing of K&R there was not stdio in the language (v6 had the portable C library a precursor and there were some others).

As I said, the typesetter (troff) support was available independent of the OS at one point for V6 (and inside the Bell System PWB 1.0).    K&R had not yet been published but that version of troff needed stdio, IIRC.  So as part of the package, the version of the Ritchie Compiler that could compile it was included in the distribution by AT&T.   This version was >>close<< to the C that would be come out in V7 but not 100% the same compiler.   Hence it is referred to as 'typesetter C.'

A versions of ditroff would be released that did not include the C compiler, because by that point the language was wide spread.

What I do not remember is which version of troff was included in the original 'typesetter' distribution.   I want to say it was troff not ditroff, but I have forgotten.

Clem