Remarks on a Recent Paper on Round-Off Errors in Regression Analysis