Actually, that is not a screenshot of build 43d, saying so is misleading. It is a screenshot of Word for Windows 2.0 which just so happens to include files from the said build of Windows 3.1 that make it display that about dialog when installed on Windows 3.0. This is also a case with a few other Microsoft programs from 1991 and early 1992. Please correct this misconception. Additionally, there is no indication in the emails that this build was ever shipped to testers as you claim, the bug report for debug compile of this build comes from a Microsoft employee as seen by the name stated - "mikegi" is an abbreviated name and surname, this was used in all internal MS emails back then. The fact that 43e exists and is the official Beta 2 released outside MS strongly opposes what you've written. --DeFacto
My response to your message:
1. A screenshot of the About Windows dialog box from Windows 3.1 is still Windows 3.1 regardless of how you look at it. Since the About Windows dialog is from this same version of Windows 3.1, it still counts as a screenshot since it not only demonstrates what the particular version looked like but also shows the progression of development seen in Windows 3.1 at that point.
Regarding the SHELL.DLL file that you mentioned, I believe that it might have had something to do with TrueType support being required for that particular version of Word for Windows, but I would have to look back at it to be sure.
2. You realize of course that there were far more pre-release versions shipped to testers than just the Beta 1 and Beta 2 releases, right? ISVs (independent software vendors), for example, would need to test their applications with the upcoming Microsoft Windows 3.1 product, so why should the individual pre-release builds be any exception? Especially when you consider that one seemingly small change can have a major effect on an application's performance.
3. Your logic seems to be "3.1.043e" exists and was the official Beta 2 release shipped to users, therefore anything in between Beta 1 and Beta 2 was only kept to a few people. This is a false argument since it basically ignores the fact that even testing the various Beta 2 "candidates" was in itself a major part of testing applications for compatibility with Windows 3.1. For all you know, even a very small revision could have just as easily fixed problems with data corruption in utilities, for example, or possibly, users inside of other companies could have also reported other bugs found in between the different revisions.