What I find incredible about this is that it decides that the file ending ".CSV" must not be a CSV file but SYLK. Then loading it as SYLK fails, and it doesn't then try and load it as a CSV file instead.