Best Practices for One to Many to Many relationship

I have a situation where I'm writing an application that has a 1 to many to many relationship.

1 Audit -> n number of BatchIDs -> m number of possible errors.
so per audit that has average 40 batchids and average 8 errors

Version 1
Create 1 audit in table 1, create n number of batchid data in table 2 repeating m times for same batch id but different error
average 1 record table 1 and average 320 records in table 2

version 2
create 1 audit in table 1, create 1 batchid in table 2, create many errors in table 3
average 1 record in table 1 and average 40 records in table 2 and 8 records in table 3

Version 3
create 1 audit in table 1, create 1 batchid in table 2 and also populate m number of errors as seperate columns
average 1 record in table 1 and average 40 records in table 2

version 4
create 1 audit in table 1, create 1 batchid in table 2 and also populate an int using bit level math on an INT column
average 1 record in table 1 and average 40 records in table 2


The error list is populated via another table that only shows me the possible errors for this type of audit.

Which is the best method? I prefer method 4 because I can support up to 64 different errors per audit type and it's database driven...the lookup and data entry just does a conversion to an int.

Version 3 may be more readable, but I would have to do an alter table for each possible error
and since 1 error for 1 audit type isn't the same for another this chould get out of hand

Version 1 I don't like because of disk space

version 2 is proper? as in normalized?