Dan Luu

There's obviously some limit -- a 40 foot long line surely reduces reading speed, but it seems like no one's tried to find the limit? When I talked to designers/typographers, they pull rank and tell me that "experts" know that shorter=better, but they also can't refer to sources

Did you actually read these? The study in the first link is a literature review that finds highly mixed results, definitely not that long lines are obviously worse .

The 2nd is some person's opinion that cites a book from 1967 that cites a study from 1929 that find that 3", 10pt. font is optimal. The author uses this as evidence that their website (20 pt. font) is good.

I can't seem to find the full text of that 1929 study, but the abstract (https://psycnet.apa.org/record/1929-03857-001 …) seems to indicate you're misreading it. It doesn't compare different font sizes and say 10pt is optimal, it only compares different lengths (and uses a 10pt font size).

Ah, I've got my hands on the contents now, and it does say that in a _previous_ experiment they demonstrated that 10pt was better than alternatives - at a fixed line-length of 80mm. Then they fix the size to 10pt and find that 80mm is optimal. But they didn't vary both at once.

Since they didn't vary both at once, one hypothesis consistent with their evidence is that line-length should vary in proportional to font size (unsurprising if true), and that there's nothing magical about 10pt font or 80mm line-length but only the ratio.

This is a problem in basically every study I've found on this. Almost all quantities that you'd care about are fixed (and different from what you'd use/see today). People then cite these studies to justify whatever they prefer, but it's not known if the results are generalizable.