THE THREE CERTAINTIES REQUIRED TO DECLARE A TRUST – OR IS IT FOUR? “DISTRIBUTIONAL CERTAINTY”

Abstract This article argues certainty in trusts is better understood by recognising a fourth certainty: “distributional certainty”. Distributional certainty is required in private trusts that involve dividing the property between beneficiaries: their shares must be clear. Distributional uncertainty is not, as usually understood, merely an instance of uncertainty of property: it has differing consequences, special resolution techniques, and may explain “administrative unworkability” in discretionary trusts. Distributional certainty is not required in charitable trusts. But this is not, as usually understood, merely an instance of the rule that charitable trusts do not need certainty of objects: it is an independent proposition.

what" certainty. Since this rule only applies to trusts containing a purported division of the propertyprobably the vast majoritythe statement that only "three certainties" are required to declare a private trust remains true: the fourth only arises for a subset of trusts, albeit a major one. 3 We shall see below that currently, in so far as distributional certainty is recognised at allvery indistinctlyin the context of private trusts it is invariably seen as merely part of certainty of subject matter; while in the context of public, charitable, trusts it is typically seen as simply part of certainty of objects. This contrastthe same issue classified in two different wayshelps to expose the confusion at play. Neither viewpoint is correct: it will be suggested that distributional certainty is best understood as a separate matter. The focus will be on private trustswhere distributional certainty is requiredand on separating it out from the requirement of certainty of property with which it is currently intermixed. Public, charitable, trusts will then be consideredwhere distributional certainty is not requiredseparating that proposition out from the rule that "Charitable trusts do not require certainty of objects".

A. Certainty of Property
But first, the certainty of property requirement must be defined: the overall property going into a trust must be clear. The leading case, widely cited, for the rule that trust property must be certain, or ascertainable, is Palmer v Simmonds. 4 Kindersley V.-C. said "my residuary estate at death" was ascertainableit is calculable by deducting expenses from assets. But he held "the bulk of my residuary estate" (emphasis added) was not certain or ascertainable property.
This general rule has been qualified since, almost unnoticed by text writers; and in a way that may mean the specific point regarding "the bulk of my residuary estate" would be decided differently today. 5 In Choithram International S.A. v Pagarani, 6 the Privy Council held that declaring a trust of uncertain property, but mentioning a specific item as included, creates a valid trust of the item. A settlor expecting to die shortly was alleged to have declared a trust of "all my wealth", and to have expressly mentioned some properties as included. Lord Browne-Wilkinson, delivering the judgment, said: It was submitted that a gift of "all my wealth" was void for uncertainty. Their Lordships express no view on that point since there can be no question but that 3 Or it could be said there is always a requirement of distributional certainty; but it is automatically satisfied in a trust for a sole beneficiary.

350
[2020] The Cambridge Law Journal the [expressly mentioned properties that were the subject of the dispute] were identified by [the settlor] as being included in the gift and the gift of them is pro tanto valid. 7 Following the logic of this "pro tanto" reasoning, arguably a trust of "the bulk of my residuary estate" must include at least half of the residue, plus one penny. The trust should therefore be valid to that amount. The only element of uncertainty is over how much more might have been intended; and the declared trust should only fail as to the remainder of the residue. After all, it is a very short step from (1) certain property explicitly mentioned within a stated uncertain whole, to (2) a certain quantity of property logically implicit within a stated uncertain whole. And the courts say they only insist on the minimum certainty needed to enable them to enforce a trust. 8

B. Consequences If There Is No Certainty of Property
If the overall property designated for a trust is uncertain there is no trust: the settlor remains owner. If the settlor is alive, this gives the opportunity to re-declare with certainty. If dead, the property is part of their estate. If property was given away, with an invalid trust added over an uncertain part, just the initial gift takes effect. 9

II. A FOURTH CERTAINTY: DISTRIBUTIONAL CERTAINTY
Even if the intention to create a trust is manifest, the overall trust property is clear, and all trust beneficiaries are identified, there can still be uncertainty affecting a private trust. That is, where the terms of a trust provide for division of the property between beneficiaries, the law also requires distributional certainty: sufficient certainty as to the beneficiaries' shares; or "who gets what". In his famous judgment in Knight v Knight listing the three certainties, Lord Langdale M.R. does seem to have briefly adverted to the need for distributional certainty, speaking about the need for certainty regarding "the interests to be enjoyed by the objects". 10 The leading case demonstrating the requirement of distributional certainty is Boyce v Boyce. 11 A trust of houses gave X whichever one she chose, the rest to Y; but X died without selecting, leaving the division unclear: so Shadwell V. for the settlor's estate. It is therefore clear that if a settlor allocates property shares to beneficiaries, it must be certain, or ascertainable, who is entitled to what: otherwise, at least that part of the trust fails. A modern example is Pensions Regulator v A Admin Ltd., 12 where purported trusts of an occupational pension scheme were held void because its benefits were uncertain.
Textbooks usually mention Boyce v Boyce under the heading certainty of property. 13 On the facts of the case the need for distributional certainty does appear to be an issue of certainty of property. The overall property going into the trust was clear (the houses); the identities of the beneficiaries were known (the testator's daughters X and Y); what was uncertain was the shares into which the property was to be divideda property-related uncertainty. However, the need for distributional certainty can just as easily appear to be an issue of certainty of beneficiaries. Suppose, again, the overall property and beneficiaries are known, but this time the trust says "my home to go to the daughter who took greatest care of me in my old age, the remaining houses to my other daughter", and it is unclear which daughter provided the greatest care. Now the division of the property is clearinto the home as one share and the remaining houses as the other shareand the uncertainty is over which beneficiary is entitled to each share. Now the issue looks more like one of certainty of beneficiariesa beneficiary-identification uncertainty.
In other words, the issue of distributional certainty can arise either looking like a problem of certainty of property (although it is really a problem of identifying property shares, the overall trust property being known) or looking like a problem of certainty of beneficiaries (although it is really a problem of identifying beneficial interests, the trust beneficiaries being known).

A. Why Separate Out Distributional Certainty?
Distributional certainty is not, therefore, simply part of certainty of property, as it is currently regarded in the context of private trusts: it appears to merit individual attention. Several reasons can be given why it is helpful to separate out distributional certainty. First, it helps to expose the potentially differing consequences of distributional uncertainty from those of uncertainty of property, which it is usually seen as merely part of. Second, it helps to highlight that special judicial techniques exist for resolving distributional uncertainty. 14 Third, isolating the idea of distributional uncertainty may also help us to understand the puzzling concept of "administrative unworkability" in discretionary trusts. , who perhaps does most to separate out distributional certainty, is typical in treating it as part of certainty of property; admitting to the potential for confusion this causes (p. 51, note omitted): "This requirement of certainty of subject is somewhat ambiguous, because the phrase may mean that the property subject to the trust must be certain, or that the beneficial interests of the cestuis que trust must be certain." 14 Although their use is not necessarily restricted to resolving distributional uncertainty in private trusts.

B. Consequences If There Is No Distributional Certainty
A distributional uncertainty issue usually presents itself looking like an uncertainty of property issue. But it is important to separate out the two in terms of possible consequences. Uncertainty of property first: if the property designated for a trust is uncertain, there is no trustwhat cannot be identified cannot be held on trust. There is no question of a partially valid trust. It is true that a single trust declaration might purport to include several assets, some identified with certainty and others not, so that the declaration succeeds only with respect to the property identified with certainty. In that sense, the declaration will be "partially valid". But the point is that, with respect to each asset, there is either a trust or there is not. However, distributional certainty is different. If the overall property going into a trust is certain, and there is merely distributional uncertainty over "who gets what" within that overall property, the trust only fails to the extent of the uncertainty. This may be total failure, so that there is no trust; but it will often be only partial failure, in a situation where a trust of sorts has been created. For example, if an asset is transferred to be held on trust, with a clear life interest for X, but a remainder interest for Y and Z affected by distributional uncertainty, a valid trust of the asset has been created; although it includes a resulting trust for the settlor as to the remainder interest.
In the rarer situation, where a distributional uncertainty issue presents itself a looking like an uncertainty of beneficiaries issue, there will be no difference in consequences from those attending uncertainty of beneficiaries.
The consequences if there is distributional uncertainty can be stated as followsand by way of contrast with those for uncertainty of property. To the extent that the beneficiaries' shares are uncertain there is no trust: the settlor remains owner. Or, if property was transferred to a trustee, they hold on resulting trust for the settlor, to that extent; or, if dead, the settlor's estate. If the settlor is alive, this gives the opportunity to re-declare with certainty. Or if property was given away, to the extent that an invalid trust was added, just the initial gift takes effect. 15

C. Judicial Techniques for Resolving Distributional Uncertainty
The courts use several mechanisms to resolve distributional uncertainty.

Court's duty to infer intention where possible
The first techniqueseeking to discern the settlor's underlying intentionis common to all the certainties. 16 But the courts perhaps give themselves particular leeway over inferring intentions when it comes to resolving 15 Under the so-called rule in Lassence (1849)  C.L.J. 353 The Three Certainties Required to Declare a Trust distributional uncertainty. Gold v Hill 17 says that if declared beneficiary shares are unclear, the court has a duty to infer the settlor's probable intention, to make the trust certain, if possible. A trust was declared for a mother and children, without specified shares: the court inferred an intention to give the mother discretion how the trustee was to spend money on the family. Paul v Constance 18 seems to be another example. A man repeatedly told the woman living with him that money in a bank account in his name was "as much yours as mine". This was held to create a trust of a half share for her. The court appears to have inferred a tenancy in common of the equitable interest was declared, so that at the man's death, the woman got half of the account; there was not a declaration of a joint tenancy, giving her all of it by right of survivorship. 19 Is there evidence the courts allow themselves particularly wide scope for inferring intentions to resolve distributional uncertainty? An extreme example, deciding a "reasonable income" could be identified as a sufficiently certain share of trust property, is Re Golay's Will Trusts. 20 Ungoed-Thomas J. said: It is common ground that in this case the trustees are not given a discretion so that if "reasonable income" does not fail for uncertainty then it would be open to a beneficiary to go to court to ascertain whether any amount quantified by the trustees was a "reasonable" amount in accordance with the provisions of the will . . .. The court is constantly involved in making such objective assessments of what is reasonable and it is not to be deterred from doing so because subjective influences can never be wholly excluded. In my view the testator intended by "reasonable income" the yardstick which the court could and would apply in quantifying the amount so that the direction in the will is not in my view defeated by uncertainty. 21 An interesting question is whether a "reasonable" sum of capital, rather than of income, would be sufficiently certain. 22 Is it fanciful to speculate that a purported declaration of trust over "a reasonable part of my residuary estate" would fail as too uncertain; but, if overall trust property is stated with certainty, a "reasonable part of it" might be upheld as a beneficial share?that is, there might be a difference in approach to certainty of property as contrasted with distributional certainty? It would be very difficult to 17 Gold v Hill

354
[2020] The Cambridge Law Journal justify such a distinction as a matter of principle; however, it is not inconceivable as a practical outcome.
But sometimes inference cannot provide a solution to distributional uncertainty, as in Boyce v Boyce itself; the ruling there would apply and the trust fail, to the extent of the uncertainty.

"Equality is equity"
Rowe v Prance 23 shows that if intended shares cannot be found, the maxim "Equity is equality" can be applied to give equal shares. Re Steel 24 says that "Equity is equality" usually leads to equal division, as in the case itself; but it might lead to division in proportion to some scale: for example, the differing size of other gifts previously given by the settlor to the same people. This proportionate approach seems to start with equality, but blends it with inferred intention.
But the maxim cannot be invoked to resolve distributional uncertainty where inequality was clearly intended by the settlor, as in Boyce v Boyce itself; that ruling applies and the trust fails, to the extent of the uncertainty.

Beneficiary's implied right of selection
Re Knapton 25 shows that one unidentified item among several (for example, "one of my houses") can be made a certain share: through the courts' readiness to find that the beneficiary was given an implied right to select. With a series of gifts (for example, "one to A, one to B") beneficiaries select in the order they were named by the settlor. If they were described as a class (for example, "one to each of my children") and they cannot agree, lots are drawn for the order in which they select.
But sometimes an implied right of selection cannot cure distributional uncertainty, as in Boyce v Boyce itself; leaving the ruling in that case to apply and the trust to fail, to the extent of the uncertainty. This, of course, would be by analogy with the power of beneficiaries under the rule in Saunders v Vautier. 27 That case held that a beneficiary who is sui jurisadult and of sound mindand is entitled to the whole beneficial interest, can terminate a trust and take the property out, even though this violates the terms of the trust. (A trust said the beneficiary should receive property at 25; he was held able to take it out as soon as he was adult.) It follows that several beneficiaries can do this: if they are all sui juris, between them entitled to the whole beneficial interest, and unanimously agreed. And it follows that beneficiaries can use this power to simply vary the terms of a trust, rather than terminating it. But this Saunders v Vautier power only arises if there is a valid trust in the first place; whereas what is in issue here is whether or not there is a valid trust at all for these parties. So, this must be a questionable proposition. 28 And it would require a very bold understanding of the court's inherent jurisdiction, or interpretation of the Variation of Trusts Act 1958, for the court to have power to consent on behalf of a party who was not sui juris or ascertained.

D. Discretionary Trusts and Distributional Certainty
On first impression, distributional uncertainty seems impossible in a discretionary trust. One would expect that the beneficiaries' shares should always be ascertainable: by the trustees carrying out the duty to exercise their discretion. But distributional uncertainty may possibly explain the enigmatic rule that a discretionary trust will fail if "administratively unworkable".
In McPhail v Doulton, 29 Lord Wilberforce said in the leading judgment, obiter, there may be cases where the description of the class of potential beneficiaries in a discretionary trust is certain, but the trust still fails, because it is "so hopelessly wide as not to form 'anything like a class' so that the trust is administratively unworkable". He gave as an example "all the residents of Greater London". The Divisional Court later applied this dictum, holding a trust for a county's inhabitants administratively unworkable. 30 So far, this rule has only been applied to large areas of population.
This rule is usually included in chapters of textbooks dealing with the three certainties; but often with little coherent explanation why it is there what the rule has to do with certainty. However, Emery has argued the reason these trusts fail is that there are no apparent criteria for selecting 27 Saunders v Vautier (1841) 49 E.R. 282. 28 D. Hayton, P. Matthews and C. Mitchell, Underhill and Hayton Law Relating to Trusts and Trustees, 19th ed. (London 2016), para. 8.22, would go even further and allow agreement to resolve uncertainty of propertythat is, uncertainty as to the overall property designated for a trustnot merely distributional uncertainty as to beneficiary shares.