One plausible first reaction to the 1926 Slavery Convention Article 1(1) definition of slavery, to wit:
Slavery is the status or condition of a person over whom any or all of the powers attaching to the right of ownership are exercised
is that it is a noteworthy example of law’s framing of definitions in terms more obscure than the definiendum.2 It has been argued, for instance, that the common law has no concept of ‘ownership’,3 and to the extent it does, its function is to distinguish one kind of property interest from others, for example an ownership right in land from an easement. The civil law notion of ownership as ‘dominium’ may not really provide us with firmer ground, since the most significant aspect of this concept of ownership is the way in which it prohibits the existence of the sorts of lesser proprietary rights which abound in the common law. There are also problems with framing the status or condition of slavery in terms of the exercise of powers of ownership, which I shall discuss in some detail below.
In view of these considerations, it is a genuine question whether such a definition was required at all. After all, one of the standard ways in which the common law at least proceeds in such cases is to begin with a dictionary definition, and then sort out what legal implications the concept, there defined, might indicate. Moreover, it is not as if there were no legal precedents for the institution of slavery; there is, shamefully, an embarrassment of riches here, as the contributions to this volume by Honoré (slavery in ancient Rome), Freamon (slavery in Islamic Law), Finkelman (slavery in the antebellum United States) and Patterson (slavery everywhere else)4 attest. In trying to determine what the Slavery Convention was trying to ban jurists could easily have proceeded by analogy from the various slave institutions on the record.5 Nevertheless, we do have a definition.6
What I hope to suggest here is, first, that however unpromising the task might appear at first glance, we can elaborate the definition in a defensible way, and second, that in so doing we may actually capture the evil of slavery in a way which rings true to the nature of the phenomenon.
Why is the definition in terms of the powers attaching to the right of ownership problematic? As Hickey’s chapter in this volume shows,7 normally lawyers think in terms of the ‘incidents’ of ownership, or incidents of title, not only the powers that go with having title. In law the difference is important. While it is one of the incidents of my title to the laptop I am writing this with that I have the right to immediate and exclusive possession of it, when I take advantage of that right to write this essay I am not exercising a power of ownership of any kind. ‘Power’ has a specific meaning in law; it refers to an ability in law to alter the legal norms (rights, duties, liabilities, other powers, and so on) of oneself or others. I do not do this when I simply take advantage of my rights, by typing on my laptop, for that is merely doing something I am entitled to do. By contrast, if I let you use my laptop, I am exercising a power in law. By authorising your use of it, I am granting you a licence to use it. I am permitting you to do what would otherwise be a trespass on your part, and so by giving my permission I am exercising a power in law in that I have altered your rights by doing so. Prior to my grant of permission, you had the obligation not to do anything which conflicted with my right exclusively to possess my laptop, but afterward you did not. The sorts of powers that go with ownership of a tangible, like land or chattels, are typically (in the common law) to permit access, to transfer possession of it to another (a ‘bailment’), to transfer title to it by way of gift or contract, and to declare a trust over it.
With this understanding under our belt, we can now identify the problem with the definition. We would presumably want to recognise slavery in cases where the slaveowner exercises none of these powers. As Patterson has shown,8 every possible variation in the rules governing the slave relationship can be found in one place or time or another, and the fact that a slaveowner could not sell the slave, or grant access to or possession of the slave to another, does not entail that the relationship is not one of slavery. So it would appear that the definition is simply inapt for prosecuting many cases of slavery. However, I disagree. I hope to show that if we attend to the way that the relation of the slaveowner to the slave is, as Patterson emphasises, one of corporeal possession, we can interpret the definition as it is written fairly straightforwardly to capture the cases we would want it to. However, to show that will require that we look more closely at what we think slavery is.
We can begin with the definitions of slavery proposed by Honoré and Patterson. Though Honoré’s definition is the more legal of the two, perhaps for that reason it is difficult to fit into a definition of slavery which must apply as much to de facto as it does to de jure slavery. Patterson’sdefinition is less amenable to formulation in terms of law, but I think it ultimately works better as a justificatory foundation for the proper interpretation of the Convention, framed though it is in terms of powers of ownership. This is so despite the fact that Patterson is well known for objecting to identifying slaves as the objects of property rights.
According to Honoré, what identifies a person as a slave is his or her unlimited subordination to another or others, coupled with his or her legal and institutional disability. As to the first element, Honoré quite rightly points to the fact that whilst my obligations to others, including the state, under the general law or under the contracts I make, may be more or less unspecified, they are not unlimited. This distinguishes me from a person of whom anything, in principle, can be demanded at the will of another. And this gets at the Convention definition insofar as an owner of a tangible typically has unlimited scope as to how he will exploit it. However, assuming that Tang9 was rightly decided, ie that the prostitutes in that case really were de facto slaves, it seems clear that not all de facto slaves are subject to unlimited subordination, if by that we mean being subject to an unlimited scope for exploitation. The point here is not that there can be de jure restrictions or limitations of various kinds upon slaveowners, regulating what they could demand of their slaves without diminishing the unlimited nature of their powers over their slaves. De jure slave institutions show us as much. In the same way that the state can place limits upon what can be done with chattels, such limitations do not in principle diminish the unlimited scope of what an owner can do with his own.10 And so long as the de jure norms in principle allow the unlimited exploitation of the slave, it doesn’t matter whether de facto a slave is treated well, even very well, and there are recognised de facto norms that limit the sorts of services that can be demanded of the slave. The point is rather that there is a problem in translating the notion of such a de jure norm to the de facto condition of slavery. This takes a little explaining.
‘Subordination’ is a normative term; I am subordinated to your will whenever I must comply with your wishes. I am in such a position, normatively speaking, however little you actually make use of the power to exploit me. ‘Exploitation’, however, is factual; I can be subordinate to your will, yet not be exploited. But de facto slaves are not de facto slaves in the absence of actual, severe exploitation sufficient to reflect a servile condition.
In certain cases of modern-day slavery, for example the cases of persisting classical slavery,11 one finds a de facto norm, custom or practice, of this kind, ie a de facto norm, custom or practice in which, in principle, the subordination of the slave to exploitation which is unlimited in scope is the common understanding of the situation as between these slaveowners and their slaves, and perhaps more importantly, amongst the slaveowners themselves, indeed amongst all who are free, whether slaveowners or not. But such a norm, custom or practice would appear to be neither necessary nor sufficient for de facto slavery. It was clearly not a defining aspect of the condition of the women in Tang. Nor would it be sufficient, since such a norm, custom or practice would appear to characterise the situation at least in some cases of those subject to forced labour and debt bondage. None of this is to say, of course, that the absence of such a de facto norm or custom is somehow protective of the welfare of the slave. The most extensive exploitation is compatible with the absence of any de facto custom or norm which authorises exploitation which is in principle unlimited in its scope. The difficulty, then, is translating ‘unlimited subordination’ as a criterion for slavery from the case where there is a norm, whether de jure or de facto, which frames the slave relationship in this way, to the case where there is no such norm in custom or practice, but the person in question is indeed a de facto slave owing to the nature and extent of their exploitation. This is a point we will return to when, in Part C, I ask whether all slavery reveals a quasinormative psychological basis in view of the notions of honour and dishonour that Patterson identifies as criterial for a slave relationship.
More straightforward is the second element of Honoré’sdefinition, the slave’s legal and institutional disability. From the perspective of the law, such a person is essentially a non-person. The difficulty here is in explaining quite what this disability amounts to in the de facto condition of slavery. As Siliadin v France12 makes apparent, making sense of the contours of de facto legal and institutional disability is problematic where there is no de jure legal and institutional disability.
Let us consider now the socio-historical definition of slavery advanced by Orlando Patterson. Patterson has refined his definition over the course of his writing, and it is useful to look at two of his formulations of it. According to an earlier one: