The starting point is that you need certainty of objects to have a valid trust.
The leading authority is the House of Lords in McPhail v Doulton. McPhail says that this means that you must be able to tell whether a given person is or is not an object of the trust or power. An "is or is not" test.
That is the starting point. There is a little difficulty surrounding how that test is to be applied - hence the different approaches in Re Baden 2.
Stamp LJ says that the McPhail "is or is not" test means you must be able to tell with certainty whether a given person is or is not an object of the trust. If there is anybody about whom it is unclear whether he is or is not an object, the trust fails for uncertainty of objects. In other words, you MUST be able to put EVERYBODY into a "yes" or "no" box. If there is anybody who goes into the "don't know" box, the trust fails. This appears quite harsh. However, Stamp LJ wasn't so harsh on the facts - he interpreted "relative" to mean "next-of-kin" - which is a little bit fishy.
Sachs LJ takes a more lenient approach. Sachs LJ thinks that the McPhail test only applies to conceptual uncertainty, not evidential uncertainty. In other words, if the court knew all the facts in the universe like a God, it must always be able to say whether a given person is an object of the trust. If the problem is simply that the person has difficulty proving that there are an object, there is no problem. If a purported trust uses a word like "friends", that would be conceptually uncertain because the word "friend" is inherently ambiguous. If a trust used the words "anybody who gave me money on the 1st October 1951", that would be evidentially uncertain but not conceptually uncertain. It will be extremely hard, if not utterly impossible, for anyone to prove that they gave money to the settlor 50 years ago - but the trust is at least theoretically certain.
This boils down to using burdens of proof to get around the problem. As long as it is theroetically possible for people to prove that they are objects, that is fine. The burden of proof is on people who think they are objects to show this: if somebody's name goes into the "don't know box", they are assumed not to be objects.
Megaw LJ thinks that the McPhail test is satisfied if a substantial number of people are definitely objects of the trust. This is very difficult to reconcile with House of Lords in McPhail.
Out of the three, I back Sachs LJ. His approach is pragmatic and sensible. Megaw LJ's interpretation is clearly not what the House of Lords said in McPhail. Stamp LJ's interpretation is very harsh and relies on playing semantic games (i.e. treating "relatives", an ambiguous term, as meaning "next-of-kin") to get around its harshness.
Alastair Hudson has excellent podcasts on this and other trusts issues. Google Alastair Hudson and listen to them.