Thanks. Is that a better link to use for future downloads?
What I see in the revised “Part Four” file is that the five rogue files have not actually been recreated afresh. Last week’s version has simply replaced the dodgy versions created earlier this week, so their creation dates are 3rd or 4th August.
It is an interesting “feature” of this system that usually every file is re-created each time (*). As the file creation date is included within the file as metadata and the service start date is typically incremented by a week, this means that software designed to identify unchanged/duplicate files cannot help quickly identify files with substantive changes. But there you go - open data piggyback on what exists for TfL’s internal purposes, not the other way round!
(*)The same applies to the WTTs, though in that case new and old pdf files should be identical.