The quote character (") in particular takes up six bytes. Enter takes up six. The ampersand (&) takes up five. Less than (<) and greater than (>) four each. There are a few other standard encoded characters, plus any non-ISO-8859-1 character is also so encoded.
So character count as the software sees it to stick in the database won't be the same as the character count as you might see in whatever editor you're using. The actual limit is 65,535 bytes, not characters, and escape sequences can be up to nine or so bytes long for e.g. some rare Chinese characters.
Sticking with ISO-8859-1 over UTF-8 there is a security and performance consideration, but even switching to UTF-8 isn't going to get rid of the three most common offenders there.
The new software stores all strings in compressed form, so the above will become much less of an issue - most English text tends to average to about ~two bits per character or so when compressed.