You may create dummy IDs and descriptions for the skipped levels, such as "N/A" or "None".  In order to keep the relationships intact for all levels, you cannot omit any levels.  But you can show the users the skipped levels as "None" or simply blank, so they will understand it.

Yeah, that's exactly what I would do, and that would be the easiest way to do it: to have, like Kevin suggested, a dummy series level, for example.

The only other way you can do it would be if you came up with a generic table structure where any item can be a collection, or a file, or a sub-series, and you would just define that collection in the item itself. But this is adding a whole other level of complexity. It's like having a personnel database where a person can be an employee, a customer, management, and so on. You can keep the record level generic.

But again, I would go with Kevin's idea first, especially if you're a beginner.

Thank you both for your quick reply. That's what I will do.