Thanks. I wasn't aware of the existence of MERGE tables yet.
If, in the situation sketched above, I created MERGE tables, they would be ReadOnly, as I only want to provide a global Search. Indexing would only be done in the subtables.
Correct me where I'm wrong. A Search is roughly done as follows: the query words entered by the user are looked up in the keywords table, giving key_ids that the engine table can translate in spider_ids, which correspond to URLs of pages where the query words occur.
For this to work correctly with MERGE tables, there should not be overlaps between the values used for key_id in the merged subtables, similarly for spider_id values etc.
These are AUTO_INCREMENT fields and there is a user comment in the MySQL manual stating how to set an initial value for an AUTO_INCREMENT field:
http://www.mysql.com/doc/en/example-AUTO_INCREMENT.html
So it would seem to me that if we define the subtables with widely separated start values for these id fields, a global Search on MERGE tables might function!
What do you think?