https://bugs.python.org/review/25916/diff/16210/Doc/howto/unicode.rst File Doc/howto/unicode.rst (right): https://bugs.python.org/review/25916/diff/16210/Doc/howto/unicode.rst#newcod... Doc/howto/unicode.rst:195: 2. A Unicode string is turned into a :class:`bytes` object containing no embedded zero I think this is talking about the UTF-8 encoding in general; linking to the Python bytes() object is probably not right. IMO “string of bytes” was a reasonable description, e.g. U+00A9 is encoded as a string of two bytes 0xC2 0xA9. https://bugs.python.org/review/25916/diff/16210/Doc/library/logging.config.r... File Doc/library/logging.config.rst (right): https://bugs.python.org/review/25916/diff/16210/Doc/library/logging.config.r... Doc/library/logging.config.rst:148: send it to the socket as a :class:`bytes` object preceded by a four-byte length Again, I am not familiar with the API, but I think this is talking about a socket-level protocol, and linking the bytes() class seems out of place. I thought “string of bytes” was fine. https://bugs.python.org/review/25916/