>>Actually, counting with no index involved would scan through the whole table and bump with an error when a damaged portion of a table is encountered.
>
>There are at least two separate cases here: damage to a record in a table and damage to an index file. The method I described was designed to uncover damage to an index file. In that case counting with no index would not see any problem at all.

Yes, and this is what you get: en error if the table is damaged, and discrepancy in count (or a different error code) if the index is damaged. Actually "killing two flies with one hit" (I guess the corresponding English phrase mentions "two birds with one stone").