A few of my colleagues using Mail, and myself, are unwillingly and without any knowledge of this misbehaviour, sending messages encoded with a charset of windows-1252:

Content-Type: text/plain; charset=windows-1252
Content-Transfer-Encoding: quoted-printable
Mime-Version: 1.0 (Mac OS X Mail 7.2 \(1874\))
X-Mailer: Apple Mail (2.1874)

I see the same problem coming from many different versions of Mail. This is a serious problem since for some anti-spam and anti-virus filters, messages encoded with this historical coding are identified as coming from SPAM sources or viruses.

Of course, I found an easy way to choose the format used to encode one message. But this function doesn't scale for people dealing with tenths or hundreds of messages.

How is it possible to avoid sending messages with this historical and malware symptomatic format? How is it possible to definetly and easily choose the format used for all sent messages?

  • → Tom: what is the ratio of spam and viruses you receive relative to legitimate E-mail when only looking at messages encoded with Windows-1252? – dan Apr 15 '14 at 20:01
  • Almost all my spam is Arabic and other non-latin junk in UTF-8. But my point is this: Anybody who runs a spam filter that rejects win-1252 messages will wind up rejecting a huge number of legitimate messages generated by Windows default settings, so I can't imagine a filter would be run that way. Instead it would use other criteria and ignore the encoding. – Tom Gewecke Apr 15 '14 at 23:04
  • As a typical example, on a set of > 20 000 E-mail, just 4 % have an header with charset=windows-1252, and among these 74 % are total SPAM or malware, and 26 % are coming from Mail on ≠ versions of MacOS X. – dan Apr 20 '14 at 11:57
  • Thanks, daniel. It appears I am very much behind the times regarding how Windows behaves. Found a reference indicating that even 5 years ago email default was ios-8859-1 instead of win-1252. I'd be interested in the technique you used to analyse the charset headers, would like to try it on my mail. – Tom Gewecke Apr 20 '14 at 16:25
  • → Tom: with Mail, Preferences > Rules > Add Rule, create a new header to survey with Edit Header List… and enter Content-Type. Then define your rule as Content-Type Contains windows-1252. To easily count them, choose an action of Move Message in a temporary mailbox. – dan Apr 21 '14 at 13:15

As far as I know, the only way to guarantee the encoding is to make sure every message includes an Unicode dingbat (e.g. in the sig). This will ensure that the message goes out as UTF-8. Dingbats are in the Unicode 2700 range, e.g. ✈ . You need to add this from the Character Viewer, not by just changing to a "dingbat" font, which may not be Unicode.


Turning off smart quotes and dashes seems to fix this problem. When composing a message, Edit -> Substitutions -> Smart Quotes (and Smart Dashes).

  • 1
    Turning off these 2 functions will remove a lot of uncontrolled coding with windows-1252, but not all. For example message containing only: (=⌥⇧]). – dan May 2 '14 at 13:10

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .