IBM Selectrics had a symbol ball [𝛼] you could swap for the standard one. It only takes a few seconds to change balls, though when I've seen people doing it they would normally type all the prose on a page, then go back and type all the math. Super/subscripts were done by rolling the paper up and down half a line. The big symbols like the square root were done later with pen and a ruler. Working from a handwritten manuscript, of course.

[𝛼] https://www.duxburysystems.org/downloads/library/texas/apple...

We had an amazing technical secretary in our Stanford research group (spouse of a Nobel Prize winner). She quickly visualize the order of symbols on the IBM Selectric math ball and typed the equations flawlessly. Then laser printers became affordable in 1980 with various UNIX equation hacks. Until Knuths magnus opus TeX.

The Selectric came out in 1961, after this paper.

There were some interesting Selectric balls, e.g. one specifically for writing APL! One of my CS professors wrote his dissertation on something to do with APL and had a copy of the manuscript and an APL type ball that he liked to show off.

I was wondering the same, until I realised that typewriters were presumably also sold in Greece.

In addition to Selectric symbol balls, you could use a "normal" typewriter with a set of extra symbols on tiny plastic sticks. You would place the symbol head on the stick in the way of the typewriter's strike.

I found a Math Overflow thread full of stories about this: https://mathoverflow.net/questions/19930/writing-papers-in-p...

And see page 7 of this for pictures: https://etconline.org/backissues/ETC099.pdf

Indeed, pre-TeX, Phyllis Winkler prepared all of Knuth's papers using TYPIT sticks for all the math symbols. Unfortunately, I didn't nab her typewriter and TYPIT box after her retirement, but here's a good set of photos and explanation: https://twitter.com/mwichary/status/1098850604640755712?lang...

Also, a TYPIT advertisement on the bottom left of the third page of https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=6500767

I don't know how NASA did it, but in the 60s and 70s the IBM Selectric typewriters had those replaceable balls and you could swap one in with special math symbols on it.

https://www.duxburysystems.org/downloads/library/texas/apple...

https://mathoverflow.net/questions/19930/writing-papers-in-p...

Heh, that's a really interesting question. It's easy to forget that not everybody lived in the world of PC's running TeX / LaTeX, and Postscript enabled laser printers, etc. I have no idea how math got typeset back in those days... I'm too young to have any appreciation for that era in that regard.

TeX was specifically written because Knuth was unhappy when his books went from being delicately hand-set by someone to being produced by a computer.

It was the transition of commercial printing technology from "hot lead" typesetting machines (big brothers of the Linotype beasts I helped maintain while in college) to photo-offset, during which the fonts Knuth loved got "left behind" (as so often happens as proprietary tech evolves), that led him to write his own typesetting system. In the early 90s I used both vde with troff and TeX through grad school on a Kaypro 1 running CP/M. TeX was definitely the superior system for the kind of research papers I was writing (Chicago Style cites, mixed Latin and Greek text, annotated bibliographies).

There were typewriters with math symbols:

https://www.mrmrsvintagetypewriters.com/products/hermes-3000...

Notably, Dijkstra used a modified Hermes with a mixture of different type bars to get the output he wanted in his earlier EWDs.