Bailment on a Ledger

Possession, delivery, and digital tokens after the Property (Digital Assets etc) Act 2025

The contemporary literature on ledger-recorded assets has fallen into a persistent and consequential category error. It treats the private key, or analogous informational artefact through which control is exercised, as though it were the asset itself. From that misidentification it imports the familiar proposition that “information is not property” and then proceeds, almost mechanically, to the conclusion that bailment cannot arise in relation to token transfers. The resulting analysis denies the applicability of orthodox private-law doctrine not because the doctrine is inadequate, but because the object of analysis has been misconceived from the outset.

The error is twofold. First, it confuses an instrument of control with the subject matter controlled. A key, password, or credential enables interaction with an asset; it does not constitute the asset. Secondly, it misapplies authorities concerned with the status of information as such to assets whose legally relevant characteristics are exclusivity, rivalrous control, and transferability, notwithstanding that those characteristics are operationalised and evidenced through an informational system. OBG Ltd v Allan is routinely cited for the uncontroversial proposition that information, however valuable or confidential, is not property. What that authority does not support is the much stronger claim that rights constituted, transferred, and extinguished through an informational system are therefore incapable of being possessed or held in custody. The House of Lords was concerned with information itself, not with assets whose existence is mediated by information but whose control is exclusive and rivalrous.

A similar misreading attends Your Response Ltd v Datateam Business Media Ltd. That case marks a boundary for “pure information” by denying a possessory lien over a database. The Court of Appeal’s reasoning turned on the non-rivalrous nature of data: control over one copy does not exclude control by others, and the law therefore refuses to treat data as susceptible to possession. The decision does not abolish possession where the law recognises an asset and where the facts supply exclusive dominion. To treat Your Response as standing for a general incapacity of digitally mediated assets to be possessed is to erase the very boundary the court was at pains to draw.

These errors are compounded by a tendency to frame the problem as metaphysical rather than doctrinal. Much of the literature asks whether a digital asset is “really” a thing, or whether it can be “physically” possessed. Those questions are misdirected. Bailment has never depended upon physicality in any crude sense. It is concerned with delivery, control, and the allocation of responsibility that follows from the assumption of custody. Classically, bailment is the delivery of possession of a thing upon terms that it be redelivered or otherwise dealt with in accordance with the bailor’s directions, as the common law has recognised since Coggs v Bernard. The core inquiry is therefore functional. It asks whether a particular mechanism effects a transfer of exclusive control sufficient to ground the law’s possessory consequences. That inquiry governs outcomes in cases of mistaken transfer, custodial loss, unauthorised dealing, conversion, and restitution.

The statutory landscape now removes any residual incentive to evade that inquiry. The Property (Digital Assets etc) Act 2025 confirms that certain digital assets are capable of being objects of personal property rights notwithstanding that they are neither things in possession nor things in action. Parliament deliberately declined to codify the incidents of this “third category” of property, leaving their development to the courts. The classificatory question has therefore been resolved at the threshold. The remaining task for private law is to determine which factual states attract the incidents traditionally associated with possession, custody, and control. For third-category digital assets, exclusive practical control performs the same juridical work that possession performs for chattels. The label matters less than the function. What matters is whether the facts disclose a state of exclusive dominion that justifies the imposition of duties, the allocation of risk, and the availability of remedies.

The argument that follows is simple, orthodox, and immediately usable by a court. The relevant “thing” for private-law analysis is the tokenised right constituted and evidenced by the ledger state, not the private key or other informational means by which interaction with that right is enabled. The key is an instrument; the asset is the right whose state the ledger records. Delivery occurs at the moment of system recordation. Recordation disables competing claims to the same asset and confers on the recipient exclusive practical dominion to dispose of it. At that point, and not before, the transferor relinquishes control and the recipient acquires the factual state that attracts the law’s possessory consequences. Copying or sharing a key does not deliver the asset at all. It neither transfers control nor extinguishes the transferor’s power of disposition. It merely allocates a power over another’s property, engaging fiduciary, trust, or agency analysis depending on the arrangement, but not bailment because there has been no delivery of the thing itself. Once exclusive control is acquired through delivery, orthodox consequences follow: rights and liabilities attach by reference to control rather than ownership, consistent with The Winkfield, and a person who has assumed custody may have enforceable rights even absent title, as illustrated by The Aliakmon.

To see why this is not innovation, it is enough to recall what bailment does, and what it does not do. Bailment exists to regulate circumstances in which control and responsibility are dissociated from ownership. Title may remain with the bailor; possession passes to the bailee; duties arise because the bailee has assumed custody. That separation is not incidental; it is constitutive. The doctrine developed precisely to allocate risk and responsibility where property is held by one person for another, whether for storage, carriage, repair, cleaning, or safekeeping. In Morris v C W Martin & Sons Ltd, the Court of Appeal reaffirmed this structure by fixing responsibility on a professional bailee for loss caused by an employee: custody had been assumed in the course of business, and liability followed the assumption of control, not any proprietary interest in the goods.

Delivery is the doctrinal hinge. Until delivery, possession remains with the transferor. Once delivery is effected, possession passes and the incidents of bailment attach. English law has never confined delivery to physical handover. It asks whether custody and control have moved so that the recipient acquires dominion sufficient to exclude others, including the transferor, subject always to the terms of the arrangement. That is why bailment may arise through objective systems of deposit that operate without contemporaneous interaction. Where a mechanism disables the depositor’s unilateral control and brings the thing within another’s custody, delivery is complete because control has shifted, not because hands have met.

Ledger recordation fits squarely within that orthodox framework. A token transfer is not the copying of an object. It is a state change. Whether the system is output-based or account-based is irrelevant for present purposes. What matters is that the system enforces a single valid state for the asset at any given time, and that a recorded transfer extinguishes the transferor’s dispositive power while conferring it on the recipient. Prior to recordation, the transferor retains the practical ability to redirect disposition by authorising an alternative valid state change. After recordation, that ability is extinguished. The recipient alone can authorise any subsequent disposition. The point is not philosophical. It is operational. The system enforces rivalrous exclusivity: only one person can validly control and transfer the asset at a time.

Once that is stated, the legal inference is immediate. Recordation is delivery. It is the objective event by which the transferor’s control ends and the recipient’s exclusive power begins. The fact that the asset’s state is evidenced through information does not make the asset “information” for the purposes of private law, any more than the fact that a share register records ownership makes shares “mere information”. The boundary drawn in Your Response remains vital: data as such, capable of replication without exclusion, is not susceptible to possession. But an asset whose defining feature is exclusive, rivalrous control is not “pure information”; it is an object of rights whose state is recorded informationally.

Possession in English law is not mystical. It is an inference drawn from control and context. Parker v British Airways Board illustrates the point in the context of found objects: possession may be constituted through an organised framework of control over premises and processes, notwithstanding the absence of immediate knowledge of the particular item. Costello v Chief Constable of Derbyshire Constabulary reinforces that possession is sufficient to ground enforceable rights against all but those with a superior claim, even where the possessor lacks lawful title. Together these cases show the shape of the inquiry. The law asks whether, viewed as a whole, the circumstances disclose dominion sufficient to justify the legal consequences that attach to possession. Knowledge may matter to the inference in particular contexts, but it does not convert possession into a purely subjective state. Control does the work.

That is why the “keys” rhetoric is so destructive. A key is not the asset. Key-copying is not delivery. Copying a credential does not extinguish the transferor’s control and does not confer exclusive dominion on the recipient. It creates concurrent capability. That is the opposite of delivery, because delivery is the mechanism by which the transferor is excluded and the recipient becomes exclusively capable of disposal. When commentary proceeds from “information is not property” to “keys are information” to “therefore the asset cannot be possessed”, it has simply smuggled the conclusion into the premise by treating the instrument as the thing. OBG is then made to do work it does not do. Your Response is then treated as a general bar rather than a boundary case. The analysis collapses because it never identified the right object.

Once the object is correctly identified as the tokenised right (the ledger-constituted state), the consequences of delivery fall into place. First, bailment can arise even where title does not pass. A mistaken transfer is the cleanest illustration. Where recordation occurs, the recipient acquires possession but not title. That is not a paradox; it is the normal structure of bailment. Possession is the trigger for custodial duties, including duties of care, restraint on dealing, and redelivery where appropriate. The recipient’s subsequent unauthorised dealing is then analysed through orthodox wrongs. Conversion is the obvious route where dominion is assumed inconsistently with the rights of the person entitled to immediate possession, as classically stated in Fouldes v Willoughby. Receipt alone need not be wrongful; it is the act of dealing as though entitled that engages strict liability. The law does not need to invent “digital” conversion, because conversion is already a wrong of unauthorised dominion, not a wrong that depends upon physical handling.

Secondly, possession-based remedies function in the digital context in the same way they function everywhere else. The Winkfield confirms that a bailee in possession has a sufficient interest to sue a wrongdoer for interference and recover the full value, holding the proceeds on account for those ultimately entitled. The case is a reminder that private law often protects possession pragmatically without pausing to adjudicate ultimate title at the threshold. The Aliakmon illustrates the complementary point: possessory rights and responsibilities can exist independently of ownership, and standing may turn on who has assumed custody and immediate possessory entitlement. The doctrinal architecture is already there. It simply requires that the locus of control be identified correctly.

Thirdly, the same analysis distinguishes custodial from non-custodial arrangements without slogans. A provider does not become a bailee by virtue of “touching the keys” in some abstract sense. The question is whether the provider has assumed exclusive practical dominion over disposition, excluding the client from unilateral control except through the provider’s processes. Where that is true, custody has been assumed and the provider is, in functional terms, the possessor of the asset. Where it is not true—where the client retains unilateral ability to dispose—there is no delivery of custody to the provider and bailment does not arise. In mixed architectures, the inquiry remains the same: who can actually cause a valid disposition, and who can actually prevent it? Private law has always tied responsibility to real control, and it can do so here without doctrinal contortion.

It follows that blocked or non-recorded transfers do not crystallise bailment. Until recordation, the transferor retains the power to redirect disposition; the intended recipient has no exclusive dominion. There has been no delivery, no transfer of possession, and therefore no bailment. This negative proposition matters because it prevents the attribution of custodial duties based on intention or expectation rather than fact. The ledger supplies a clean evidentiary boundary between attempted transfer and completed delivery, much as receipt records and deposit systems do in conventional bailment contexts. The question “was it recorded?” often answers the delivery question in a way that is externally verifiable and operationally grounded.

None of this requires metaphysics. It requires discipline. The central discipline is to keep distinct: the asset (the tokenised right), the instrument of control (the key), and the mechanism of delivery (recordation). Confusing these categories causes predictable doctrinal damage. It causes OBG to be misused. It causes Your Response to be read as abolition rather than boundary. It causes possession to be treated as if it must be tactile. It causes bailment to be treated as if it must be contractual. And it causes modern custody disputes to be analysed as if the common law is helpless until it invents a new doctrine. The opposite is true. The common law is not helpless. It is being misdirected.

The Property (Digital Assets etc) Act 2025 makes that misdirection harder to sustain. The statute confirms that certain digital assets are objects of personal property rights even though they do not fit comfortably within the older taxonomy of things in possession and things in action. Parliament then declines to specify the incidents of that property, leaving the courts to develop them. That legislative choice does not invite doctrinal invention for its own sake. It invites the orthodox task of identifying which factual states attract which legal incidents. Control is central because control is what makes the asset rivalrous, exclusive, and transferable. That is precisely why control is capable of doing the work of possession for the purposes of bailment: it is the factual state that justifies duties, allocates risk, and grounds remedies.

The essay’s claim can therefore be stated in one sentence without loss of precision. For ledger-recorded assets recognised as property, exclusive practical control is the functional analogue of possession, and ledger recordation is the event that effects delivery by transferring that exclusive control. Everything else follows from that proposition by orthodox reasoning. If recordation disables the transferor and vests exclusive dominion in the recipient, possession passes and bailment is possible, with duties and remedies attaching in the usual way. If recordation has not occurred, delivery has not occurred, and bailment does not arise. If a key is merely copied or shared, the asset has not been delivered at all; authority may be delegated, but custody has not been transferred. If an asset is dealt with inconsistently with the rights of the person entitled to immediate possession, conversion supplies the wrong. If a possessor suffers interference, The Winkfield explains why possessory standing matters. If custody is assumed within a commercial structure, Morris v C W Martin & Sons Ltd explains why responsibility follows control, not labels.

The payoff of this approach is continuity rather than innovation. The law already has the tools required to govern custody, misdelivery, mistaken transfer, and unauthorised dealing once the preliminary category error is corrected. Courts are not asked to recognise novel forms of possession or to stretch doctrine to accommodate technology. They are asked only to apply existing tests to an objective mechanism of delivery. That approach preserves doctrinal integrity while delivering practical certainty. It ensures that responsibility tracks control, that custody attracts duties, and that interference attracts remedies, as it has done for centuries.

The debate about ledger-recorded assets has been made unnecessarily obscure by treating the wrong object as the asset and then declaring that the law cannot possess it. The better approach is blunt. Identify the thing. Identify the event that transfers exclusive dominion. Apply the law of delivery and possession to that event. Bailment does not fail because the asset is “digital”. Bailment fails only where delivery fails; and delivery fails only where exclusive control has not moved. Once the analysis is anchored to those facts, the common law’s response is neither fragile nor novel. It is simply itself.