In [c.mb.wcs] paragraph 8 (describing the return value from calls to mbrtoc8), bullets 8.2 and 8.3 describe storing of a "value" (for a multibyte character) and a "next character". In both of these cases, what is being stored is a code unit. For 8.2, it is the first (or only) code unit associated with the UTF-8 representation of the multibyte character just consumed. For 8.3, it is a code unit that continues the sequence for a multibyte character consumed by a prior call.