lammert - 4:46 am on May 22, 2010 (gmt 0)
The options provided to the tar command in this script are valid. Your problem is probably the directory where the backup is stored, not the backup options. The tar command makes a backup of $HOME/html/* which is the same directory tree where you also store the backup file. What happens is that the backup tries to store the resulting backup file in itself. That is where the backup probably fails and the tar command aborts, leaving an invalid gzip file.
Some other remarks about this script: The TERM=linux, export TERM and clear commands have only a valid meaning when the command is executed from a command line. In your situation where a cron file is used they have no function, and worst case they may interfere with the automated way the script executes.
Another advice is to add the --ignore-failed-read option to the tar command line. This causes the tar command to proceed execution when a read fault occurs (for example due to open files, files deleted by other processes while the backup is running etc.).