Not sure, Malcolm, what Armitage v Nurse  would bring to HMRC’s supposed argument, which the way it is depicted borders on the fabulous assertion of a sham, to the extent that the beneficiaries were unaware of the trust. I may be missing your point.
The beneficiary of settled agricultural land in that case knew she was a beneficiary, she appears to have had no access to the trustees’ documentation. It was a question of whether and on what basis the pleadings were amendable at that stage to plead “fraud”:
“and that, accordingly, since without amendment the pleadings could not support a plea of fraud, clause 15 of the settlement operated to absolve the trustees from liability for the alleged breaches so long as they had not acted dishonestly; but that the plaintiff would be allowed to examine the trust documents and investigate the trustees’ management in order to re-amend her statement of claim.”
There was no question of the settled property remaining in the settlor’s estate by definition.
If the trust is constituted by delivery of the property to the trustees, it is constituted, even if it might be what is inaccurately termed a sham. The sham or fictional trust argument does little more than revise an apparent trust into a bare trust for the purposes of an insolvency or ancillary relief proceedings in any event and involves more than the beneficiaries simply being unaware.
The existence of an “object”, the beneficiaries, is all that is needed, they do not need to be aware of the trust’s existence or agree to it for it to exist. The term “subject” would have been used in that case. There is a risk of the current, fashionable Americanism of attempting to reinvent a trust in a banking trust company context as a three way contract creeping in which should be left in the fog on the Grand Banks on the other side of the pond.
That has been insidiously making its way into offshore trust administration “agreements” between the settlor and the trustee, but has yet to involve beneficiaries.