Java parses character escape codes in source code, not just strings.
This allows you to use Unicode identifiers without a Unicode encoding.
Therefore, the \u000d in the comment is parsed as a newline, ending the comment and beginning an instance initializer.