Made some headway on the FreeBSD PR I mentioned the other day having to do with mktime() giving strange results for the US Central timezone - discovered while fooling with Zope3 and Python.
I found the FreeBSD timezone description files that contain historical information about DST transitions, and saw that Chicago, representing US Central, had a strange entry for the year 1936 that the Eastern, Mountain and Pacific zones didn't have - which would explain the unique behavior.
By fiddling with that single line in the timezone description file, I found you can affect mktime()'s result even though the year you're passing isn't 1936.
Kind of wonder now if NetBSD and OpenBSD have the problem too. Setting the FreeBSD box to Canadian Central time seems to be a workaround.