If the first couple of years of work on the first web server are missing then that seems to me to be a minor tragedy, given how absolutely vital this technology has since become. It would be instructive to be able to track the development of the server and understand the learning process by comparing versions.
I wonder if its worth asking Tim Berners-Lee about this, since he wrote it? http://www.w3.org/People/Berners-Lee/#Before
Maybe Henrik Frystyk or some of the other early contributors have a copy laying about.
EDIT: Just sent a mail to Frystyk. Someone with a direct or indirect connection to TimBL should also try and contact him.
EDIT1: Apparently not. email@example.com doesn't work anymore :( Trying a different route.
Now that these versions have been found, what should be done with them so we don't have to do this again in twenty years time?
I think the found versions should be available on the W3 ftp server along with the later versions. The CERN HTTPD page (http://www.w3.org/Daemon/) has "Dan Connolly firstname.lastname@example.org" at the bottom. He might be the person to approach to arrange this.
I also wonder if all the versions should be of the server and browser should be archived on archive.org. They have the infrastructure and organisation to ensure that important cultural artifacts are not lost. There may be copyright issues though.
Given my negligible role in this compared to the people who did the real work, I don't want to jump in and contact these organisations unless nobody else wants to.
What do you think?
As we speak, there are torrents, sourceforge projs, and a github repo with the files. But those should be considered backups; The best place for the tarballs (if we find them all) would be at ftp://info.cern.ch/pub/ (the original source), and the src should be on w3.org as you suggested.
So, please do jump in :) I'm already behind on several "real" projects because of this :)