> if it actually wrapped around on itself with a half-twist, wouldn’t the letters wind up backwards too? Then 6-Down wouldn’t be MESSENGER but rather MESSENGER, and so on.

This premise is assuming the Mobius strip is transparent, which the header image clearly isn't. No letter-reversals or any symmetry is needed. On normal, opaque paper, you can just make a crossword as a normal long loop. (proof: Write GGGGGGG on one side of a paper, and write it upside-down on the other side. With a twist, you'll get one long GGGGGGGGGGGGGG... looping as if it were one boring single-sided loop).

Interesting article. Note that many of the clues in the article are mis numbered.

Not sure why. Refer to the original puzzles linked for the correct clues.

A long long time ago I had a fun time making Mobius strip structure out of link list(s).

Each edge of the ribbon was a linked list, connected to each other by rungs. It looped back on itself but connected crosswise .

Then if you nuked all the rungs it was clear you still got a connected piece that you could traverse over.

qxw [https://www.quinapalus.com/qxw.html] is a wonderful open source app that lets you make crosswords with interesting topologies, complete with grid filler support, if anyone wants to experiment with them.

My partner publishes crosswords and it's always fun to see how much artistry goes into authoring a puzzle. You can play with the structure/grid, the theme of course, every minute detail of the fill, etc.

Never would have thought of a Mobius-strip though, insane.

For anyone trying to solve the examples, the toroidal crosswords example image has a copy-paste error from the post author (not present in the original): the numbers for Down are 1, 3, 5 and 6 (and not 1, 2, 4, 7).