Sure the sources and a man page or two exists. Ken Thompson wrote bas(1) and bs(1) was Dick Haight - which IIRC correctly, was ~78/79. Haight's program which would go out as part of System V --- was first distributed outside of the labs in System III. I would have expected man pages in either source distribution. I don't have those online to check and I don't tend to bother with System V as its historically less interesting.
That said, one thing I do remember is that the primary version of bs from the PWB family (certainly the one we had at UCB) was a little different from the version I saw in a couple of the BTL sites later. My >>memory<< here (which could be flawed) was the CB Unix folks added a few features -- like a set of plotting function that Whippany picked up etc. I only remember this because we were working with a number of different places (from different AT&T labs to IBM, Intel, NS, HP, Tek, etc.) when we developed the UCB CAD tools. I remember and an incident where one of the Bell folks we had at UCB had brought some thing with him that relied on the strange bs dialec -- i.e. did not run on the UCB CAD systems - which had the version of PWB. It was close, but made calls to some extra stuff (plot and the like IIRC).
My suggestion, as always start in the TUHS archives and look there -- that will be the most complete of what was released. That said, Disk4 of Kirk's set will have all the stuff we had in /usr/local (like cpio etc..) on the UCB systems. FWIW: I'm fairly sure the sources and the man pages will be with Kirk's trove. But I don't believe BSD ever released it, as it was not part of V7, so you will not find it in a formal BSD distribution tape.