I think they would have to read out as unicode. If we read some non-ascii text through a FileReader in 2.7.2 we get the file contents through the default encoding. On my machine (WIndows) it is not ...