First two bytes of a Unicode UTF-16 (UCS-2) file is a BOM (Byte Order Mark) to distinguish between Litlle-Endian (Windows) and Big-Endian (Unix/Java):

• Little-Endian file - BOM:     FF FE
• Big-Endian file - BOM:     FE FF

Unicode format note:

• UTF-16 (UCS-2) is a hardware-dependent implementation of word value:
  1. Windows machines - Little-Endian byte order
  2. Unix machines; Java - Big-Endian byte order

  Note. UTF-16 is the same as UCS-2 (which is 16-bit - fixed-length - entity)
  unless a UTF-16 value utilizes the surrogates to extend the repertoire - for some (rarely used) languages and codes recently added to Unicode.
  

• UTF-8 is a variable-length (multi-byte) encoding

Paul Gorodyansky