You might want want to give your db some good thought first. Have you had a look at the links I posted? If you solve this problem without taking into consideration the whole db, then you may end up doing yourself more bad than good. In any case, as you wish.
The Companies table will just be a list of companies, PK could be a unique 4 letter of the company (you would need to device what pattern) or it could be an autonumber PK. The advantage of having a 4 or 5 letter PK is that if you have a pattern, the chances of entering the same company twice would be slim, however, it would require more work from you when creating new records in your main table. So if you're not too bothered about that then just use an autonumber as the PK. I don't know what the date field is for but if you want one then include it.
This Companies table would now be referenced in your main table via a CompanyID field, so all the IBMs will be 1 in the main table if the first company you entered in the Companies table was IBM. Don't forget to ENFORCE REFERENTIAL INTEGRITY with Cascade Update.
So:
1. Create the new Companies table
2. Add all the companies into it
3. Duplicate say your IBM table, get rid of all the records and call this table something meaning because it will be the master
4. Add a new field, say CompanyID, Datatype - Number
5. In your original two tables, IBM and the other (can't remember the name), add the same new CompanyID field and in the IBM table write the number that IBM represents from the Companies table across all records. Do the same for the other table
6. Copy all the records from both tables into the new master table.
7. Create a query based on this new main table and call it qrySearch (you obviously would have deleted the original qrySearch query before doing this)