"Recursive Functions of Symbolic Expressions and Their Computation by Machine, Part I" https://www-formal.stanford.edu/jmc/recursive.pdf
This is a fascinating paper; many "new innovations" from the next 70 years of computer science are already laid out here, such as functional programming and garbage collection. It blew my mind the first time I realised that the short LISP m-expression on page 17 was the entire LISP interpreter itself.
Why write this?
In my opinion, there is quite a jump from the original paper to all later practical LISP implementations. I couldn't find any minimal code examples of theoretical LISP from the original paper.
I wrote this as a way for programmers to get their head around the mathematics of the paper, from a programmer's perspective, and to better appreciate how theoretical LISP became a practical programming language.
I tried to write as much as I could without reference to any future work, only using what is found in the paper.
The Python files
I built multiple implementations of increasing complexity, since the paper goes through multiple steps to arrive at a "final" language.
The original paper does not give a definitive syntax for a usable language; it is a theoretical work. It was later adapted into something practical. Therefore, these LISP variants are not to be particularly useful, correct, or elegant; they are possible interpretations of S-expression LISP from the original paper, and should be read in conjunction with that.
Disclaimer
Please do not take this code as a good example of a LISP implementation. I am not a mathematician, or particularly steeped in compiler and grammar theory beyond the undergraduate level. I went for simplicity and faithfulness to the paper over good design.