I don't know so much about screws and washers but certainly there are any number of components that we don't have the tools to make the tools any longer--and we probably wouldn't want to use those components even if we could. Would be want to use the Apollo guidance computer? Raytheon had better gear back up to make core memory in that case.
The whole "Saturn V plans are lost" meme seems to be an urban legend but it's a red herring in any case. We couldn't easily build a 100% authentic 1965 Corvette either but that doesn't mean that we've somehow lost the technology to do so.
"doesn't mean that we've somehow lost the technology to do so"
I can think of two examples.
The LM had some truly weird chemical milling technology invented mostly because they didn't have the CNC mills and CNC EDM gear we have today. Design and production floor feed back on each other and some design decisions were optimized to production realities such that you could at enormous expense either reinvent the chemical milling processes or simulate them with EDM and/or CNC mills but it would be enormously cheaper to scrap the design for a chemically milled door or whatever and replace it with a door designed specifically to match the modern technologies of EDM and milling machines.
The other example I can think of is I don't think we have the tech anymore to make aerospace grade core memories. Too much info was in the brains of people who died decades ago. To a first approximation the cost would be something like the entire fixed capital expenses of the whole computation industry from 1940-1960 plus many man years of R+D and more importantly reverse engineering and reinventing the QA/QC that man rated aerospace grade core memory would require. It would be a heck of a lot simpler and cheaper to use modern tech.
There is an obvious computing analogy. For a good time check out the "soylent news" project as in soylentnews.org who spent weeks re-implementing modperl 1.0 and apache 1 and all that so as to re-implement /. using the last public release of slash code. Its non-trivial to bring up old software while simultaneously applying a decade or two of security patches and best practices.
I meant that statement in the sense of it's not really as if we've lost the knowledge from some golden age of engineering. I'm guessing it would be enormously difficult and expensive to build an IBM 360 as well. (Or, if that's not a good example for some reason, certainly any number of other obsolete computer systems.) But people don't normally lament the fact that we can no longer manufacture a 1960s-era computer.
But, that said, fair comment. It's the reason that there are always various military projects ongoing to basically replace old technology (such as guidance systems) with modern tech. Not necessarily to upgrade capabilities, but simply because we can't build the old stuff any longer.
ArsTechnica has an interesting article [1] on efforts to recreate the F1 using intact engines and the original design documentation. Reading it, I was surprised how difficult this proved to be in practice, given the extensive information available.
For any engineering effort, there is the documentation and there is the never well documented wisdom of actually building the thing. The plans may say a certain alloy has to be used or part P and the Helium cooling system should not run at more than X psi, but the reasons why it is so could very well be lost in a multitude of memos on failed tests. Sometimes, the reasons of those failures themselves are never known because someone figured out a way that works. Not knowing the complete history of the development of a given machine and why it is the way it is may very well make it almost impossible to continue evolving the design and kill the technology.
The whole "Saturn V plans are lost" meme seems to be an urban legend but it's a red herring in any case. We couldn't easily build a 100% authentic 1965 Corvette either but that doesn't mean that we've somehow lost the technology to do so.