Cliff,
the external balance is absolutely necessary to partly counter the reciprocating mass effects. You can achieve the same by internal balancing, but find it difficult to get enough counterbalance in there as per Bill and Howard's posts.
I think of the big picture in the following way, forget the reciprocating masses for the moment, and just consider the rotating masses without any external balance.
If we took the crank and balanced it at the counterweights, then added the flywheel and balanced there, then the clutch, then the pulley, balancing at each stage, we would have a perfectly balanced rotating assembly, and could achieve tight limits with no need for external balance. However, the engine would vibrate badly because the reciprocating masses have not been accounted for.
So where the need for external balance comes in, is for the reciprocating masses. In the 90deg V8 configuration, these result in a residual rocking couple (as per my previous diagram) that is countered by the 50 oz.in (or 28) external balancing.
However, there is another really important point that needs to be appreciated to get a more compete picture. Because the reciprocating forces are in only in line with the cylinder, and any balancing we do has an effect at right angles too, the best we can do is to balance 50% of the reciprocating masses. If we balance 100%, we only swap the balanced forces from in line to at right angles. By balancing 50%, we reduce unbalance by 1/2 and that is the best we can do, if we go further, all that happens is forces at right angles increase and the balance gets worse. How the forces ultimately balance out is down to engine configuration, in our case we are left with the rocking couple, so if we do a 50% balance and end up with a 50 oz.in counterweight, then there is still 50 oz.in left in there that can't be balanced. This is true whether you go internal or external.
I can reinforce this point by looking at the balancing process, which is to add counterweights to each throw. These counterweights are typically half the reciprocating mass, and yes, with the counterweights on, we can balance it down to 1gramme or whatever, but there is effectively 50oz.in still left in there, so what we are actually doing is balancing to a residual of 50oz.in, with an accuracy of 1 gramme (0.2 oz.in on a 5in radius).
I should explain that I have many years experience resolving balancing issues in our factories and global service centres, from small 15000 rpm down to 10ft diameter 120rpm rotating machines, so I have to understand the fundamentals very well.
The attached diagrams visualise the reciprocating unbalance, and the effect of the optimum 50% balance etc
Sorry to bang on about this like some boring lecture, but these points are important to understand the true nature of this balancing process.
Dave