[0] is a good article about this; not least that this has been happening since at least 1889 (to the point where I'd say we could now probably consider it a valid alternate spelling.)

[0] https://itre.cis.upenn.edu/~myl/languagelog/archives/001166.... (no, I've no idea why they're behind the wrong subdomain certificate)

Because they don't really live on that server any more. The correct URL for the languagelog archives is https://languagelog.ldc.upenn.edu/~myl/languagelog/archives/..., but none of the links will keep you on that domain.