>>8
I'm not so sure anything should be undervalued here to be called an accident, it's not that unstructured and unexplained. Even lower down from what you're referencing this manual describes part of the pure theory of LISP.
I want to say it helps explain how an meta-expression capable LISt Processor 1.5 would go symbolic and be self hosting using evalquote. Further reading slightly answers by saying m-expressions are used in this manual to describe the actual workings of the interpreter and that pure LISP has a universal function written here as an m-expression for interpreting an application written as s-expressions, evalquote, is that function, I'm guessing. Different conclusions can be brought together from the previous papers but I think that's further from what you're saying here.