We've noticed a problem with different users using 3dRefit on datasets. It seems that when 3dRefit is run, the byte order information is rewritten in the default order for that particular user. If another user with the opposite default byte order had originally created the dataset, then the result will be a mismatch between the actual data byte order, and what's written in the header.
Shouldn't 3dRefit only rewrite the header values that the user wants to change, and simply copy the other header values?
Thanks,
Tom