To expand a bit on NetWallah's remarks, if you are using 5.8.0, and have a locale-aware environment that happens to be set currently to "utf-8", then perl will try to interpret non-ASCII data as if it were utf-8 data by default, and you'd need to expressly tell it not to do that, via the "use bytes" and/or "no utf8" pragmas (or by changing your locale setting). See the "perluniintro" and "perlunicode" man pages if this applies to you.

You say you are not really doing any utf8 stuff, but it would appear that your text file contains non-ASCII data (e.g. latin1 "accented" characters). If the file uses the "upper table" of iso-8859-1, then the "0xb2" byte is a superscript "2". If it's some other character set, then it's probably some other "special" character. Or maybe it's just noisy data...

In any case, upgrading to 5.8.2 (the current version) will help. 5.8.0's "interpretive preference" based on the locale setting was ultimately viewed as a bad idea, and was changed in 5.8.1.