>I agree with what you are saying but, I have seen a number of textbooks that emphasise the former. Of course many textbooks ignore surogate keys altogether.

Well diferences of opinion is what makes life interesting. My decision to use surrogates is based o the writings of Codd, but my commitment to using them is based on empirical experience or how well they work and how few problems they present.

>Raima (recently bought by Centura) claims to have SQL for Network Databases, maybe they have forward and backward pointers? They also claim the ability to mix Relation and Network Models in the same schema, do you have any experience with Velocis or Raima IV database systems?

Nope, I have no experience with the products you refer to.