On the Role of Signs in Epicurus’ Legal Theory

Epicurus holds, in Key Doctrine 31, that what is just according to nature is a súmbolon or sign of the interest there is in neither harming one another nor being harmed. Certain readings of this maxim equivocate this legal sign with other signs found in nature, thereby failing to give sufficient weight to the role of reciprocity in its production. Other readings simply import a legal sense from outside of Epicurean doctrine, thereby failing to explain what makes Epicurean súmbola legal. A final set of readings attempt to find a legal rule as a kind of innate concept or Kantian ‘scheme’. This article identifies new sources for understanding súmbolon, drawn principally but not exclusively from Aristotle’s Eudemian Ethics. This article offers an original argument that Epicurus adopts Aristotle’s’ image of the tally stick (symbolon) as a meeting of often divergent interests which constitutes something new and particularly valuable to the Epicurean: friendship. A theoretical argument is also advanced to support this reading which claims that one person’s reflection on a ‘divine image’ of the end (telos) is insufficient to constitute a súmbolon; rather real (as opposed to abstract or ideal) individual interests are filtered via the mechanism of the tally, that is via the meeting of two given individuals who together generate a sign (symbolon) of reciprocal interest in neither harming nor being harmed.


Introduction
Epicurus (341−270 BCE) developed forty short but authoritative Key Doctrines ('KD'), of which eight-Key Doctrines 31-38-deal with central problematics of Greek legal theory, such as the nature of justice and punishment, and why laws differ between nations. At the root of these eight statements is the first (KD31) which operates as a definition of what is 'just according to nature' and as a bridge between the signs of nature operating in earlier Key Doctrines and signs specific to human 'political' friendship. The just according to nature, Epicurus tells us, is a sign (symbolon) of reciprocal interest, for not harming one another. This elicits the question: just what role do these signs (symbola) play in Epicurean legal theory?
There is no clear and distinct view on this question in the modern legal-theoretic literature. De Witt (pp. 142-150 [10]) goes too far in equivocating legal signs with signs from nature, rendering them arbitrary expressions disclosing no distinct content by which they might be identified as specifically normative. On this view, legal signs signify transitory subjective fears-at best an individual's intimate conviction (p. 28, [14]). A consequence, already identified by Cicero, is that what is purportedly legally significant is nothing more than what a given individual regards as useful i.e. what is just (dikaion) collapses into subjective desire. At the other extreme, Goldschmidt (pp. 40-41, [14]) argues that signs of the just are expressions if not of innate ideas, then of a Kantian rule-a move which demotes signs in favour of rational regulae (to which Epicurus makes no reference). A third textualist strand (LS22A 1 [20]) admits that these signs are specifically legal but overcompensates in the law's favour by occluding reference to any 'sign' with positive legal terminology drawn from outside of Epicurus' work.
In the semiotic literature (see e.g. [36]), attention has understandably focused on the detailed Epicurean semiotic treatise De signis by Philodemus and likely written c.40 BCE, some two generations after the KD. Yet as Asmis [35] argues, this text is a response to sustained Stoic criticisms of the theory of Epicurean signs, evident in its submission to the demand for a technical logical apparatus of para-conditionals not find in Epicurus' own works. The modern focus on Philodemus leaves the import of Epicurus' own use of symbolon in his 'legal' doctrines under-researched from the perspective of semiotics.
This article will argue that while these readings offer careful textual evaluations of the wider Epicurean doctrine, they discount the specificity of the notion of legal sign (symbola) and its centrality to Epicurus' definition of 'what is just according to nature'. In particular, it will argue that the distinct feature of Epicurean legal signs is that they (a) reflect the interest of a given sentient being, and (b) only find their validation in reciprocal recognition of these interests. Condition (a) distinguishes legal signs from signs of external nature and brings them closer to what have been termed in the literature 'divine images'; condition (b) distinguishes symbola as properly legal from the wider set of images of the divine.

3
On the Role of Signs in Epicurus' Legal Theory This article proceeds as follows: Sect. 2 sets out and situates the relevant text of Epicurus' doctrine before evaluating the readings given by modern scholars. Section 3 builds on Goldschmidt's close reading of the text by explaining the role the Epicurean theory of knowledge of signs plays in the legal theory. It critiques Goldschmidt's contention that legal signs are rules in a sense that evokes Kantian practical philosophy, and argues for a minimalist reading of what the human finds significant in signs such as symbola of the 'just according to nature'. In effect, this article claims that symbola share with other 'divine images' a reflection of the Epicurean telos or blessed end of neither troubling nor being troubled. From this minimalist reading it argues, in Sect. 4, that Epicurus has deliberately used a new term-symbolon-to name legal signs because he wishes to distinguish these signs from the wider set of 'divine images'. It should therefore be made clear that 'legal' is being used in this article in just this sense: it names the subset signs defined as symbola by Epicurus. This usage of 'legal' is intended to pick up both the explicit link to 'the just according to nature' in the definition Epicurus gives of symbolon, but also the place of this definition as a foundation for Epicurus' immediately subsequent account of justice (dikaiosyne), authority and punishment, and the origins of the great variety of laws (nomoi) in direct response to Hellenistic debates on the status of natural and positive norms.
Following a discussion of existing sources for understanding symbolon, this article advances a reading based on sources newly identified in this article, drawn principally but not exclusively from Aristotle's Eudemian Ethics. It is argued that Epicurus adopts Aristotle's' image of the tally stick (symbolon) as a meeting of often divergent interests which constitutes something new and particularly valuable to the Epicurean: friendship. A theoretical argument is also advanced to support this reading which claims that one person's reflection on a 'divine image' is insufficient to constitute a symbolon; rather real (as opposed to abstract or ideal) individual interests are filtered via the mechanism of the tally, that is via the meeting of two given individuals who together generate a sign (symbolon) of reciprocal interest in neither harming nor being harmed. On this realist view, symbola are highly specific and contextual arrangements, and as such Epicureans must study them in their context and so accumulate experiences from which to form preconceptions that may ground a legal theory. Hence the Epicurean study of law proceeds via an account of these contexts of socio-technical change (genealogian makran).

The Text and its Readers
Our focus in on a single maxim of Epicurean philosophy: the definition of what is 'just according to nature' which forms one of Epicurus' so-called Ratae Sententiae or Key Doctrines (hereafter 'KD') recorded by Diogenes Laertius (10.139-54 [11]). Let us set out the relevant text and my suggested translation.
My Preliminary Translation: The just according to nature is a sign of reciprocal interest, for neither harming one another nor being harmed. This is an extremely dense definition, and it is probable that no translation will convey the full sense of what is going on. Our focus is on symbolon which has been tentatively rendered as 'sign'. Translating symbolon however, is difficult, as Long & Sedley readily admit (Vol.2, p. 129 [20]). It seems that this definition may receive and indeed has received at least three interpretations which turn on whether the definition's inputs already involve some legal or normative constituent(s). The question is: in determining what is just, is Epicurus telling us that we need to refer to some other normative source, namely a symbolon, or does the desired normativity arise with the constitution of the sign? We look at the proponents of each view in turn.
On one contemporary reading, symbola are regarded as signs like any other bearing no essential normative force. Rather, they are just those signs which pretend to normativity i.e. they are arbitrary utterances which claim to be normative, and which people may well believe to be normative, but have nothing in themselves which makes them normative and derive their status from arbitrary and transitory belief (which may coalesce as laws at times by convention (nomô)). On an alternative view, a symbolon is already legal in itself, and where it is present, we know we are dealing with matters of the just. This position divides in two ways: (i) a textualist position which overreads positive law into symbolon by arguing that Epicurus is deploying a term exogenous to the theory in its fullest legal sense, as we might use 'contract' or 'guarantee'; (ii) a purportedly Kantian view that symbolon expresses a rule of practical reason. The alternative view, advanced from Sect. 4, is that symbola are generated as normative signs, and KD31 is indicating just how by using symbolon in a very specific way. Let us first review the existing candidate readings of the text.

Symbola as Arbitrary Normative Utterances
On this view the legality of the symbolon is essentially circular: if a requisite subset of persons are prepared to accept that a symbolon is normative, then it is normative until such time as they cease to accept it. There is nothing per se normative beyond a mere convention that a certain state of affairs should subsist now or in the future. It is customary. This view, already attributed by Plato to the Sophists, becomes a principled legal doctrine during the Hellenistic period with Pyrrho's disciple, Timon of Phlius: …nothing exists which is good or bad by nature, 'but these things are decided on the part of humans by convention (nomô) (11. 140 [27]) Following this line of thought, we should read KD31 as defining the just as simply utterances of interest in mutual non-harm, with the implication that these utterances will be understood by convention to be normative until such time as they are not. The argument for the arbitrary nature of constituting norms by convention (nomô) dogged Hellene legal philosophy, and it is unsurprising that those hostile to Epicurus' School should seek to conflate his doctrine with that of Sophists, Pyrrhonists and hedonists such as the Cyrenians. Thus, it appears Cicero is addressing the Epicureans when he writes: But if Justice is conformity to written laws and national customs, and if, as the same persons claim, everything is to be tested by the standard of utility, then anyone who thinks it will be profitable to him will, if he is able, disregard and violate the laws. 2 (I.xv. 42, [6]) Dyck writes (p. 186 [12]) that Cicero offers us a crude caricature of the Epicurean position. For a start, utility is not Epicurus' measure, but a much narrower 'interest' with respect to 'mutual non-harm'-this is not about gain but about avoidance of a worse position. Secondly, in KD33 Epicurus is at pains to speak of Justice (dikaiosyne) as not being absolute but arising in a specific place and time as a kind of contract (synthêkê tis) that comes about in course of dealing over mutual non-harm. This repetition of the notion of mutual non-harm must link back to the definition in KD31 which builds on the interest we have in mutual non-harm, an interest which in its expression through a symbolon defines 'the just according to nature'. The two horns of Cicero's attack miss their mark: Justice is not conformity to just any written law or convention, but nor is it grounded in any arbitrary and egotistical whim. Epicurus does wish to speak of 'the just according to nature' and he seeks to formulate this using an identified subset of symbolized interests which are subject to the qualification of mutual non-harm.

Symbolon as a Rule of Practical Reason
Victor Goldschmidt (pp. 40-41 [14]) concludes his detailed analysis of KD31 by drawing on Horace and Kant: …Horace, in the…text where he speaks of the just and unjust, and where he unjustly reproaches the Stoics for confounding every fault, big and small, recommends that they have recourse to a rule in order to ensure an exact equilibrium between whichever fault and whichever punishment: Adsit regula… 3 ['Let us have a rule to assign just penalties to offences, lest you flay with the terrible scourge what calls for the strap.']. Yet a rule is precisely the thing by which Kant will define the scheme. We thus authorise ourselves, by the text of Horace, to translate [KD31] so: 'Right is according to nature the rule of the interest that there is in mutual non-harm.' One implication of this view is that symbolon has normative content, but this does not derive from engaging some pre-existing (positive) legal token such as a guarantee, but a practical rule which regulates ethical behaviour by balancing respective interests in non-harm. Now Goldschmidt's structuralist analysis of Epicurus' legal doctrine is generally excellent, but one cannot but remark that following a word-by-word analysis of KD31, the above conclusion on the meaning of symbolon appears almost 'out of the blue' and is simply stated without further justification. Admittedly, Goldschmidt opens the conclusion by reiterating the function prolepsis plays in Epicurean theory of knowledge, and so presumably the rule in question is an Epicurean kanon or measure. Yet Goldschmidt pushes this very far by referencing Kant and speaking of the 'equilibration of interests' as a 'concept' (p. 40 [14]).
This article accepts that prolepsis is critical to our understanding of symbolon but rejects an overly Kantian view that it must be a rule grounded in a concept. First, Epicurus, author of Concerning Judgement and Canonic (Περὶ κριτηρίου ἢ Κάνων), deliberately chooses symbolon and not kanon or any cognate term to construct KD31. Second, in speaking of social matters, Epicurus is concerned to explain how the man in the street recognises the just, not the Sage. Not only would the theory lose explanatory range by denying normative force to anything not recognised by Epicurean philosophers, but it would also defeat any agreement a philosopher may make with the world at large for the man in the street would not be able to recognise its normative force. Third, Goldschmidt, in a footnote to this conclusion, perhaps reads into Epicurus what Cicero chooses to find there. He notes that Cicero translates the title of Περὶ κριτηρίου ἢ Κάνων as De regula et iudicio -Of Rules and Judgement, all as part of a discussion of Epicurean prolepsis. This may be so, but this only tells us about Cicero's theoretical preferences (for Stoicism with a Middle Platonist accent), and in any event is a translation of a term Epicurus does not use in KD31. Fourth, the relevant discussion by Cicero in De Natura Deorum, in the mouth of his Epicurean spokesman Gaius Velleius, proceeds: …the gods exist, because nature herself has imprinted a conception of them on the minds of all mankind. For what nation or what tribe of men is there but possesses untaught some 'preconception' of the gods? Such notions Epicurus designates by the word prolepsis, that is, a sort of preconceived mental picture of a thing… (I.xvi.43 [7]) As we say, prolepsis is central to our understanding of symbolon, but Cicero's account here is misleading: whatever the Epicurean doctrine of the gods was (see Sect. 3.3), this account offers only confusion. Preconceptions are not hypostatic Platonic ideas 'Epicureanized' by having nature implanting them ab initio in our minds. Rather, they are built up from atomic combinations through storing natural sensations, initially from the primary things, and become 'universalized' only through a process of concrescence of impressions (10.33 [11]; LS17C-D). New impressions ought to be measured by reference to those already safely constructed. Whatever issues one might have theoretically with this proto-empiricist account of knowledge, it is assuredly dangerous to equate symbolon with anything akin to pre-existing and eternally universal rules, pseudo-Platonist, Kantian or otherwise, which regulate the measurement of practical matters. To read KD31 in this way tempts one to state precisely what is always the measure of the just, and so the rule; a temptation to which Goldschmidt succumbs by interpolating a notion of equilibrium where there is none.

Textual Renderings of Symbolon
We term a third approach to KD31 the 'textualist strand' if only because it covers those whose position is disclosed in their direct translations of the text. Indeed, rendering symbolon as some variation of 'legal agreement' appears to be the default position of Epicurus scholars for whom KD31-38 merit only passing comment (see recently e.g. O'Keefe, 139−41, [37]). Long & Sedley (LS22A) render KD31 as: 'Nature's justice is a guarantee of utility with a view to not harming one another and not being harmed' (my emphasis). Now, the Greek for guarantor would be enguêtês (ἐγγυητής), and using this to translate symbolon (σύµβολον) seems quite interventionist, importing as it does a strong legal meaning to the text. A guarantee is primarily a legally binding surety, and if used in a wider context that usage surely derives from its primary technical meaning in a way which we would not accept, as with a term such as 'obligation'. Long & Sedley (p. 129, Vol. 2 [20]) admit translating the term is difficult, and that the expressionistic meaning of sign is possible, as Goldschmidt also notes (pp. 27-28 [14]). Yet the choice of 'guarantee' is not wholly unjustified, for Liddell & Scott [18] indicate the breadth of senses of symbolon: i. a mark or sign; ii. a beacon-fire; iii. a token (such as a tally stick) used by parties to indicate indebtedness, or that a fee was due; or. iv. a commercial convention or treaty between states.
The fourth sense-of commercial treaty-should be discounted because from KD32 onwards Epicurus will use synthêkê tis for positively made pacts including between peoples (ethnoi), whereas in KD31 he speaks of a state of affairs between individuals deriving from their interest in mutual non-harm. Epicurus is defining what is just according to nature, and would fall into Sophist arbitrariness if he is read as defining what is just by what is said to be just in a treaty (because what makes a treaty's statement just according to nature?). This leaves us with one sense of a token of indebtedness (not necessarily a guarantee, for the term might cover moral indebtedness), and more general senses of positively performed sign or mark. Does 'guarantee' go too far in supressing these other senses? Wider context suggests as much. We need only refer to Aristotle's account in the Politics ([1280b11] [2]) of Lycophron the Sophist's views on law to see a difficulty. Lycophron is reported as saying: The law (nomos) is merely a convention or agreement of convenience (synthêkê) a guarantee (enguêtês) to one another of what is right and just … which law has no real power to make the citizens themselves good and just. 4 Lycophron can be seen as advocating for a negative state, its function here (p. 109 [4]) being only to guarantee adherence to a set of agreed norms through its apparatus, not determine what those norms are. It seems improbable that Epicurus, who delights in taking existing epigrams and varying them creatively through modifying one or two words, should have meant 'guarantee' just when he avoided using the term (enguêtês) which Lycophron deployed to define law. Moreover, when in KD34-35 Epicurus does make use of 'those who have authority to punish [infringement] of the terms of mutual contract (synthêkê)' this enforcement apparatus is a new addition to his theory of law and was no part of the definition of the 'just according to nature' in KD31. These thoughts suggest that rendering 'symbolon' as 'guarantee' is problematic: it imputes a third-party apparatus of recognition and enforcement of what is just into Epicurus' definition, it occludes the semiotic import of symbolon by reducing it to a positive legal instrument, and ignores that Epicurus could have drawn on Lycophron's use of enguêtês but deliberately avoids it in favour of an alternative. This all suggests that the signifying content of symbolon should be taken much more seriously even at the level of textual analysis and translation.

Evaluation, and Related Views from the Literature
Interpretations of symbolon have either reduced it to a mere sign, having no normative content save by changing custom, or have regarded it as a normative term of art, either because a 'guarantee' or 'treaty' is generally accepted and even positively enforced as a legal sign and Epicurus is drawing on that, or because it expresses some universal rule about what is just (e.g. equilibrium). None of these interpretations is satisfactory. In what follows this article advances a reading in which reciprocity and reflection are expressed in the symbolon i.e. it is a symbolon of these moments.
It must be stated, however, that the function of reciprocity in KD31 has been considered in the literature. Goldschmidt already leans in favour of reciprocity (p. 32, 40 [14]), but as noted this view gives way to the regula interpretation. Anton-Hermann Chroust [p. 222, [5]) also notes reciprocity as a feature of Epicurean legal signs. Yet his interpretation is that reciprocity is used polemically to critique Platonic Justice (dikaiosune) in KD33, and that it is merely part of the general relativism of the wider Epicurean doctrine: laws differ as between these parties, places, times, and related particularities. On this view, reciprocity offers nothing positive in KD31; its definitional import is diminished in favour of its polemical use in KD33 and Chroust passes over it. This article takes the view that reciprocity plays a positive filtering role in KD31's definition that it does not elsewhere, and this reciprocity (of what is reflected) is the signified of the symbolon that defines 'the just according to nature'.

Defining What is Just According to Nature
This article follows Goldschmidt (p. 25 [14]) in reading KD31 as effectively a definition-of the just according to nature-on which is built the remaining legal theory set out in KD32-38. This is the most literal meaning of the words, and moreover the Key Doctrines deploy this form of words at several earlier points. It is 'effectively' a definition because to tês physeôs dikaion is not merely an empty definiendum to be filled by the definiens 'a sign of reciprocal interest, for not mutually harming one another and not being harmed'. Rather to tês physeôs díkaion is itself doing work by creating an apt parallel between the content of the sign (mutual non-harm) and the reasons for the selection of this content: its accordance with the wider Epicurean doctrine. In short: the 'just' in which Epicurus is interested is only that 'of nature'.
As Goldschmidt has shown, we find a form of words similar to to tês physeôs dikaion already in KD7 (to tês physeôs agathon) and the context is clear that the good people here obtain is security, and that this is 'nature's good' or 'the good according to nature'. Likewise, we have KD15 (ho tês physeôs ploutos) which can be rendered 'nature's wealth' or 'wealth according to nature'. One might contrast KD30 which has 'para tên eautôn physin' which can be rendered 'against (or not in) its own Nature' in the context where someone's false opinion misleads them away from natural goods. These KDs strongly suggest the just and nature are distinct. Nature is not in itself just. There is a relationship between the just and nature, indicated by the genitive. But nor should we say that the just is a possession of nature, as if nature acts justly. This is contradicted by the limitation of the just to a subset of sentient beings. There is a definite sense that the just is consistent with nature in some way, but not part of it.
Looking again at KD7 (LS 22 C(1)): Consequently, if such people's life was secure, they did obtain nature's good [i.e. security]; but if it is not secure, they are not in possession of the principle [arkhê] they originally sought. This suggests that it is possible for people to make designs for themselves which do, or do not, end up providing nature's good. Their human design is or is not in accordance with nature. Hence our translation as 'just in accordance with nature' or 'the naturally just'. The thinking here is to use naturally in the same way that 'natural' is used in 'natural healing method' (pharmakon). The method of healing is a human technology, but this technology is geared towards what achieves healing by working with nature, and by being led by nature's indications. So it seems with the just: a subset of sentient beings should be able to craft a technique of what is just which works with the nature of that subset of sentient beings.
It would be misleading to translate, as Robert Philippson [21] does, to tês physeôs dikaion as Natural Law or even Natural Justice. This would imply that Justice or Law somehow reside in Nature and that it is just a matter of the wise discovering it. Law and nature are distinct; there is nothing legal pre-existing in nature and the law is entirely a positive production of certain animals. But nor is it a case of a simple opposition between nomos and physis such as we find in Plato's Gorgias [22]. The indications of previous KDs are that nature is used as a kanon (criterion) by which to measure positive law (i.e. law created by sentient beings) and so differentiate between right according to and against nature. This interaction between nature and positive law is confirmed in subsequent KDs.
Yet while Goldschmidt places great weight on the textual parallels in the KDs, we would add a theoretical parallel between KD31 and KD1, which specifies that state to which all beings should aim and defines it principally by just what needs to be avoided to achieve it: That which is blessed and immortal has no troubles itself, nor does it cause trouble for others, so that it is not affected by anger or gratitude (for all such things come about through weakness). 5 On this reading 'being blessed and immortal' is defined just as 'not being troubled and not troubling others' with the remaining content offering a gloss on the main sense of 'trouble'. The parallel with KD31 is manifest: one might regard the latter as a narrowing of the general doctrine to a subset of behaviours by sentient animals (primarily humans). All beings have an interest in mutual untroubling; therefore, some beings have an interest in preventing those forms of trouble peculiar to their constitutions. Insofar as humans are similar, one might observe this trouble can be grouped under the term 'harm'. Like KD1, KD31 defines a blessed state of a kind but narrowed to the largely human social context.
One might conclude that absence of mutual harm defines something akin to a juridical beatitude: 'the just in accordance with nature'. Yet if this is so, then frankly there would be nothing distinctly legal about Epicurean legal theory. On this view we might say that when humans experience trouble we are apt to call this harm, and its absence is what people signify when the speak of 'natural justice/law' and their cognates. KD31 amounts to no more than anthropological description, adding nothing philosophical to KD1. Such a reading would be to stretch the parallel with KD1 too far, for KD31 interposes symbolon toû sympherontos-the just according to nature is not absence of mutual harm, but 'a sign' of the 'reciprocal interest' there is in the absence of mutual non-harm. Absence of sentient animal-specific trouble (harm) is only part of the picture; the sign of the interest is doing significant additional work.

3
On the Role of Signs in Epicurus' Legal Theory

The Preconception of the Interest in Mutual Non-harm
This article agrees with Goldschmidt's claim that if we are to understand KD31 we must account for symbolon by means of the Epicurean scientific method of preconception or prolepsis (πρόλεψις). Again, we distinguish our position from Goldschmidt, here on two key points: (a) as already mentioned, Goldschmidt adopts too Ciceronian a reading of prolepsis, one which openly flirts with Kantian schematism; and (b) we argue that the prolepsis at work in KD31 is of a particularly reflexive nature, closer to the Epicurean preconception of divine images -a refinement of the general doctrine of knowledge about nature which perhaps explains Epicurus' adoption of a new word (to symbolon) to name his object of analysis. Having briefly outlined the nature of Epicurean prolepsis in this subsection, we argue for each of points (a) and (b) in turn.

Prolepsis as Method
It is necessary to make a slight detour to understand prolepsis as the principally Epicurean account of how we understand signs in nature, and how we can develop semiology in a philosophically rigorous manner. Diogenes Laertius reports that: 'Prolepsis, [the Epicureans] say, is as it were an apprehension, or concordant prehension (κατάληψιν), or notion (δόξαν), or universal 'stored reason' (καθολικὴν νόεσιν) (i.e. memory), of that which has frequently become evident externally: e.g. 'Such and such kind of thing is a man'. (10.33, [11]) The root -lepsis implies a seizing or grasping, but also a receiving, so perhaps it is closest to 'a catching hold of'. As a preliminary, one might usefully imagine that someone throws you an unknown object-it could be a ball or a snake or anything. You have only moments to react and adopt an appropriate pose to catch whatever is coming; your grasping of it is almost entire responsive to the new information. But now imagine we are playing basketball and I throw you an object. It is almost certain the object will be a basketball, so you automatically place your hands a suitable distance apart to catch a basketball. You pre-empt what is coming and so recover, from memory, the way in which you will receive the object. And indeed, your catching reaction to any object thrown is likewise a learned response to data. This is the initial sense of preconception-it is a disposition of the memory of a body such that it is prepared to receive the collision of a body similar to one encountered many times before (see further Epicurus' Letter to Herodotus (46-53 [28]). The Epicureans are keen to stress that all humans deploy prolepsis; the role of philosophy is to refine it.
For that refinement, Philodemus' On Signs provides the most detailed methodological account (LS18F) in the context of debates with the other Schools. The Epicurean method could readily be termed inductive in the scientific sense that a general rule is held to be true until such time as a conflicting instance is encountered in the world. By contrast, the Stoa advanced a logico-deductive method of sign inference, a key tool of which was the elimination method. For example: (1) If Plato is a man, then Socrates is a man; (2) but [assume] Socrates is not a man; (3) Therefore Plato is not a man. The point here is that the Plato is co-eliminated from being a man as 1 3 a logical consequence of eliminating Socrates. The Epicureans take issue with this method's extension to nature: Socrates not being a man does not physically cause Plato not to be a man. This criticism is particularly incisive if one holds, as the Epicureans do, that preconceptions such as 'being a man' are not purely logical but can only derive from physical impressions. The Epicureans first accept that whatever the individual Plato is, it must derive from Plato's existence and the sensations his existence produce in the observer. Second, they see in 'manhood' simply a similarity of certain attributes of Plato and Socrates and any other empirically encountered 'man'. Critically, the Epicurean may only infer that Plato is not a man if Socrates is not a man if first she has remarked that Plato and Socrates possess some similar attributes given the name (or sign) 'manhood'. Hence Philodemus: 'If Socrates is not a man, Plato is not a man either' comes out true…not because by the [Stoic] elimination of Socrates, Plato is co-eliminated, but because it is impossible to conceive of Socrates not being a man but Plato being a man. And that belongs to the Similarity Method. (LS18F (5)) The Similarity Method is reinforced by the 'insofar as' (τὸ καθὸ) premise: a sign is said to be 'insofar as' precisely because its inductive validity extends only as far as the aggregate of previous instances. But the 'insofar as' also pertains to the relationship between signs because we compare relevant similarities, implying also differences. No two concrete things are identical because of the inherent individuality of atoms in the void, so any comparison can only equate a subset of secondary attributes (colour, speed, pain etc.).
All this points to a prerequisite of the Epicurean method of signs: sign inference is made by a person that has formed a prolepsis based on experienced similarities insofar as they subsist. But how can two things be similar if one does not already have a reference measure of similarity? Are we drawn back into an infinite chain of similarities in which one knows this is a horse only because one previously saw a horse, because one previously saw a horse and so on? Diogenes Laertius appears to confirm Epicurus' view that a preconception is anchored in something primary and self-evident: we at some point saw our first actual horse (not knowing what it was) and from this started building preconceptions based on apparent similarities.
A problem for the scholar is that Diogenes Laertius also reports these as 'universal stored notions' (10.33 [11]) echoing Epicurus' 'universal sensations for universal matters' (Letter to Herodotus, line 82 [28]). This appears very close to suggesting that prolepses are inbuilt concepts. That if not horse-ness, then there are some basic preconceptions in us, not born of experience, which allow us to intuit similarities in the world. Is this the case? The question is particularly important because while the self-evidence of an individual and present horse may be taken as an empirical anchor for subsequent inferences, the signs we are interested in are legal-the symbola of interests in mutual non-harm. People's interests are neither self-evident nor ostensibly universal, yet Epicurus wishes to define 'the just according to nature' by means of our inferences about a generic property of a sentient animal. It seems we all, and not just philosophers, have preconceptions about these interests; that these interests have some core similarity 'insofar as', permitting Epicurus to ground a universal legal theory. Whether or not preconceptions of natural world phenomena are pre-existing and innate concepts-a matter for epistemologists-there is a strong concern that Epicurus does implicitly regard prolepses of the reciprocal interest for mutual non-harm as universal to all humans, whatever experiences they actually have. We find this reading in Cicero and in the Kantian reading by Goldschmidt.

Against Reading Symbolon as a Schema
As already noted, Goldschmidt believes that we must see in symbolon a recourse to prolepsis and through prolepsis to a scheme or rule. To emphasise this point he claims that a 'rule is precisely the thing by which Kant will define the scheme' and so translates KD31 as: 'Right is according to nature the rule of the interest that there is in mutual non-harm.' One wonders if this insight says more about the context of French history of philosophy 6 than about Epicurus: Pierre Aubenque's Kant et l'epicurisme [3] is referenced as a source of inspiration for Goldschmidt's insight, and the view is echoed by Goldschmidt's fellow structuralist Jules Vuillemin (p. 132 [31]). Nevertheless, Goldschmidt adopts the schematic interpretation without qualification.
Goldschmidt's principal support for his view rests on reading symbolon as a synonym for typos, and by interpreting typos as a schema by which the mind regularly understands phenomena, perhaps even regulates its understanding of phenomena. Given the recourse to Kant with which his analysis ends, there is a sense that a typos is a kind of acquired category of any possible perception. The difficulty with Goldschmidt's use of typos is it is severely weakened by three manifest shortcomings: i. Goldschmidt admits that Epicurus does not use typos; the term is repeatedly applied by Diogenes Laertius, surely demonstrating the latter's interpretation of KD31. ii. Goldschmidt evidently wishes for typos to be read as 'type', as in 'typology' or 'model', so emphasising its schematic and regulatory nature. This stretches the sense of the word quite far. Typos means tap or hit, or impression, whence also mark where that mark has been impressed in, say, wax. One can start to say the genetic linkage to type in Plato's discussion of education in Republic, where acceptable patterns of thought are said to be impressed into pupils and which the metaphor of impressed wax is explicit (377b, 403e [22]). This is then used to speak of 'types' of citizens based on their training. Yet such a typos is received as a passion to control thought and is distinct from the innate Platonic genera of thought and Ideas. Even if we assume that symbolon is a synonym for typos, had Epicurus wished to emphasis a rule by which to measure impressions, surely, he would have deployed his own term of art: kanon. iii. Even if we accept that Epicurus is using symbolon for typos, reading it as impression rather than formal scheme or model accords more closely with Epicurean thinking than an attempt to find innate Kantian categories. On this view, we accrue preconceptions through repeated impressions of certain acts, and so are symbola are no different from many other notions an Epicurean believes we might form about the world through experience alone.
Goldschmidt appears to have been aware of these difficulties because his position varies just before publication of his [14] in 1977. 7 In a September 1976 conference paper [15], Goldschmidt seeks to distinguish the Stoic position on prolepsis he ascribes to Cicero, from the original Epicurean source of this term. In this paper the purportedly Epicurean typos is merely an image used to identify nominal, otherwise unrelated species, whereas Stoic prolepses break free of their instances to form conceptual parts of the whole cosmic order. They become, in effect, signs of the natural law of the cosmos. Goldschmidt's refinement is subtle, but important. There is a retreat from innate schemata and a return to empiricism: symbola are signs or images gathered from similar experiences insofar as similar. Goldschmidt makes no mention, however, of his schematic reading of symbolon as a prolepsis of the just.
Consequently, our preference would be to regard prolepses as signs generated from repeated apprehensions. There are two aspects to this: (i) if we wish to know what the naturally just is, then we need to attend to the preconception of this right; but (ii) to develop this preconception, we must have repeatedly apprehended certain impressions from which to construct a sign, according to the principles of similarity and elimination and the insofar as premise. Yet it is one thing to derive the preconception of 'mortal' or 'swan' from the impressions of countless physical experiences, quite another to formulate a notion of something so abstract and intangible as legal behaviour. Framed either as an interest in what should not happen in the future, or as an understanding about an incorporeal state of affairs between bodies (a Stoic lekton), it is unclear what empirical content forms the true canon of truth by which we should recognise our symbola. 8 Goldschmidt evidently senses the need for this missing thought component, but errs in interposing too Kantian a schema, bridging back to the Hellenistic philosophers via Cicero's reading of prolepsis.

Symbols of the Divine, and Reflection
There is something beyond the sensation of the statement 'I will not harm you if you do not harm me' which allows us to grasp our interest in the state of affairs constituted by my reciprocal assent. It allows us to measure our interest, and to see the justness of that state of affairs coming to subsist (and subsequently the (in)justice of (not) maintaining that state of affairs). This missing component is not a scheme, rule or category that would found a Kantian imperative; rather, we claim, Epicurus holds that the symbolon is a particular subset of expressions of the rational animal's capacity to reflect on its own impressions and so uncover what is divine in them.

On the Role of Signs in Epicurus' Legal Theory
We proceed by augmenting our reading of the Epicurean doctrine of prolepsis with a notion of epilogismos, here explained by Philodemus: As to what concerns things which are just, good and beautiful, those who philosophise according to our School affirm that they are indeed what the man in the street thinks they are: with this difference however that philosophers conceive things, not only in a passive manner (pathetikôs), but reflectively (epilogistokôs), and do not let them fall to be forgotten. (Philodemus, On Rhetoric, quoted from Sudhaus, I, 254 [1] As Phillip De Lacy [8] has shown, Epicurus uses epilogismos several times, 9 having borrowed the term from his teacher, Nausiphanes, and we find it deployed in Hermarchus' genealogical account of laws reported by Porphyry (LS22M(3)). 10 For Allen (p. 234ff [1]) epilogismos is the requisite apprehension drawing together the similar, and in this he draws on Schofield's conclusion that the term be rendered 'assessment' (p. 237 [25]). De Lacy (p. 179-180 [8]) notes a variety of uses, emphasising that epilogismos as not merely empirical but is deployed where there is a movement to seek after similarity and even form a preconception which is an epilogismos. De Lacy further suggests this purposiveness 'perhaps explains' why epilogismos also appears in moral treatises (p. 181 [8]).
Our second step is to pursue this purposive meaning, seeking to establish that it is purposiveness which superadds an epilogistic evaluation to sensations. In Epicurus' Letter to Menoecceus we find the following familiar phrase: For indeed who, think you, is a better man than he who holds reverent opinions concerning the gods, and is at all times free from fear of death, and has reflected upon the end according to nature (kai to tês physeôs epilelogismenou telos)? (LS20A(1), my translation).
What is this end? As the passage already suggests, these considerations pertain to the gods and death, and lead us back to Cicero's account of prolepsis and Epicurean views on the gods in De Natura Deorum 43-44 (cited above). From the mouth of Velleius we learn that we have a 'prolepsis, or prenotion, of the gods' and that they are blessed and immortal is 'engraved on our minds'. It is at this point that Velleius cites KD1: blessedness and immortality consist in not troubling and not being troubled. Yet we also know that Cicero's understanding of prolepses risks hypostatizing them, at least as universal notions expressing harmony with the whole natural order, or perhaps worse as ideas in their own right. So what value rests in Cicero's account of the end?
This passage from Cicero has generated vigorous scholarly debate, focused on whether it discloses an Epicurean doctrine that the gods exist or are mere prenotions formed by humans. David Sedley argues that Epicureans think that 'gods are our own graphic idealization of the life to which we aspire,' (p. 52 [26]). For his part, David Konstan holds that 'the Epicurean gods are real, in the sense that they exist as 1 3 atomic compounds and possess the properties that pertain to the concept, or prolepsis, that people have of them.' (p. 71 [16]). Both views are not necessarily contradictory: if the Epicurean gods are absolutely unperturbed and unperturbing, then, as Lucretius counsels, we scarcely have any impression of them (V. 146 [19]). It does not preclude divine atoms from being. This negative theological position becomes positive doctrine: if we consider divinity, as in KD1 (blessedness and immortality), as the minimal limit case 11 of our experiences (dare we say: trauma) of troubling and being troubled, then our reflection upon (epilelogismenou) the greatest good, and so the end, informs our evaluation of every experience in our lives. This perhaps explains an anonymous Epicurean scholiast's otherwise heterodox gloss on KD1 that the 'gods are seen by reason (logô)' yet also are 'in human form (anthropoeideîs)' (LS23G).
On this account the philosopher unfolds from the primary experiences of all things first the similar need or usefulness we find in avoiding harm, second the blessedness of achieving the limit of absolute unperturbability, third that the experiences of gods and sentient beings are not distinct but relative 'insofar as' perturbed, and fourth that each sentient being, in reflecting on avoiding or causing harm during each moment, reflects this similarity with the divine onto that experience. In other words, each sentient being, however unwittingly, projects this need for avoiding perturbation onto experience and thereby marks sense data as being of (dis)value constituting it as a kind of sign or symbol. It is this projection of a purposive aspect of ourselves which, we claim, differentiates the sense of epilogismos in which we are interested in the ethical and legal texts from more basic forms of knowledge built purely from accumulation of impressions.
We wish to highlight two consequences of this view: first, the divinity we find reflected in things as our end is minimal and simple: a lack of perturbation or ataraxy. We must resist temptations to discover in Epicurean thought a grand categorical scheme or even a system of duty, for everything that is built upon this most simple preconception of blessedness is particular and generated. That the philosopher should see in the reciprocal interest for mutual non-harm a reflection of the end of neither troubling nor being troubled is to see the generic link between the former and the latter, but also to see how the latter blessedness, applicable to all atoms, has been differentiated and specified for the particular context of the social being and its law. Friendship is immortal only insofar as it expresses blessedness; the rational content-the concern of wisdom-is merely mortal (Epicurus, Vatican Sayings 78 (LS22F(7))). 12 Second, similarity now works at two levels: two sets of sense data may be similar because of their secondary attributes (colour, magnitude, etc.), but they may have different attributes yet may nevertheless be reflectively similar in virtue of our finding them similarly reflective of the end. Hence it is possible for us to bring together a great variety of empirically dissimilar experiences by reflecting on them and unfurling their usefulness to us in achieving the end. And because the Epicurean claims to 1 3 On the Role of Signs in Epicurus' Legal Theory be able to find reflective, or 'epilogistic', similarity between otherwise very different sensory data, we can see how it would be possible to extend their scientific method beyond what we would regard as the physical realm to apparently intangible and radically divergent matters of human (dis)value, such as the social field of law.

Reflection and Reciprocity: An Aristotelean Reading?
We have argued that epilogismos, at least in its ethical and legal application, involves a reflection upon sense data that marks the latter with a sign of the end. The role of the philosopher is to unpack this reflective process, to proceed to the limit of what the end is, and to pursue that end in a prudent and reflective manner. One might therefore venture that the symbolon in KD31 is the product of just such reflection: what makes the content of the symbolon especially 'symbolic' is that it reflects KD1-it reflects the end of not troubling others nor being troubled. If there is an Epicurean jurisprudence, it would then be founded in the reflection on these symbola. Yet there remains one piece of the puzzle to place: why does Epicurus use a distinct word-to symbolon-for the sign or token expressive of the reciprocal interest in mutual non-harm? We claim that this is because legal signs possess an additional degree of reciprocity which raises them beyond those signs which any given individual may alone find reflected to them in the cosmos. Arguments that reciprocity plays a supplementary role in distinguishing symbola are well-known, and this literature is discussed in Sect. 4.1. It is in Sect. 4.2 that we advance original support for a reading from certain other of Aristotle's works which ties together the existing reliance on the word allêlous (to each other) in KD31 to Epicurus' deliberate use of symbolon.

Reciprocal States of Affairs as Symbola
Chroust appears to have been first to highlight the function of reciprocity in KD31 (pp. 223ff, [5]), but places the entire emphasis on KD32's critique of Justice (dikaiosyne) in apparent contradiction to the text: This µὴ βλάπτειν µηδὲ βλάπτεσθαι makes it quite obvious that law and right in their application are based upon the principle of reciprocity; and that they can exist and function only within an established society. According to its true nature, justice (dikaiosyne), as contrasted with law and right (dikaion), is, therefore, an essentially relational concept.
The careful reader will notice Chroust has even suppressed the ἀλλήλους in his quotation from the Greek of KD31. A footnote suggests the path of Chroust's thought: he references Aristotle's definition of Justice from the Nicomachean Ethics: 'This form of justice…is…virtue, not absolutely, but in relation to our fellow men' [1130a4]. The effect, unmerited given the text of KD31, is to resubordinate the Epicurean just according to nature back under hierarchical Platonic Justice as a concept that governs all. So doing reverses the genetic trajectory of Epicurean thought, which claims that Justice (dikaiosyne) is nothing more than the positive agreements we find here and there.
It is Goldschmidt who correctly identifies that reciprocity is already a key feature of KD31. His argument will mirror our own in that it has a textual and a theoretical component: Goldschmidt (a) notices reciprocity features in a possible source for symbolon in Aristotle; and (b) argues that his typos or schema plays an important role for any interest or arrangement which aims at mutually preventing harm.
As noted in Sect. 1, one sense of symbola, almost always in the plural, is as international treaty. Often such treaties dealt with the status of one's city's citizens before the law of another city. Goldschmidt draws our attention to Aristotle's Politics III, ix [2], where the Stagirite discusses commercial treaties (symbola) between nation states, and considers the responsibilities of the states-parties: Nor does one state take care that that the citizens of the other are such as they ought to be, nor see that those who come under the terms of the symbolon do no wrong or wickedness at all, but only that they do no injustice to one another (ὅπως µηδὲν ἀδικήσουσιν ἀλλήλους) [1280b3 -5].
'Injustice' is used here negatively. Aristotle is trying to define a state as being concerned with the good of its citizens; this he opposes to what is merely positively legal. Aristotle is concerned that what may formally be called a 'state' and its 'laws' may well enjoy these names but they are not truly deserving of them. Acts may be considered 'just' if they perform a treaty, but only in a limited sense.
Goldschmidt contends that Epicurus' otherwise unexplained use of symbolon could be derived from his reading of the Politics, and weight is given to that by the similarity of the phrase importing mutuality or reciprocity that follows. Most interesting though is that if Epicurus has indeed lifted the idea of symbolon from this particular text of Aristotle, he makes a small but critical variation: '…of the interest there is in not harming each other and not being harmed'. 13 Epicurus, according to Goldschmidt, has replaced Aristotle's 'injustice' with 'harm' (blabê) and used the latter to define 'the just according to nature'.
Aristotle continues by confirming that positive laws are just so many words: …law is only a convention, 'a guarantee to one another of justice', as the Sophist Lycophron says, and has no real power to make the citizens good and just.
[1280b10-14] 14 Whereas Aristotle the natural lawyer says that positive laws are merely called just and have no power to ground a true, natural justice, Epicurus is prepared to state the precise opposite: the synthêkê-a term he will use from KD32 onwards to make his symbola more precise-are signs of the naturally just, and indeed derive their 'power' from nature viz. our being troubled by harm.
This derivation from Politics III finds further support in Epicurus' use of sympheron (interest). As Delba Winthrop has established (p. 414, [32]), Politics III presents a political dispute involving 'some' who do not keep their contracts because they were undertaken not with a city but a tyrant. The some are more concerned with doing what is fitting than with obedience to positive law, and they do this by distinguishing regimes which exist for 'the common interest' (το κοινε συµφέρον). Aristotle's common interest appears to range widely, but again Epicurus has narrowed interest to a conception of mutual non-harm, even if this does appear to be a philosophical backward step from 'injustice' or even legal 'damage'. By replacing (a) dikaion with blabê, Epicurus refuses to hang his theory of the naturally just either on a generally inaccessible Platonic Idea of Justice, or on some (powerful) subgroup's claims that their superiority makes things right. Rather blabê is a sensation to which every sentient being has immediate access-Epicurus has democratised justice.
Does this not though engender a kind of anarchy? If everyone's one-sided fear of particular harm is a valid ground for the just according to nature, how would it even be possible to construct a preconception from such diversity? This concern engages the final qualifier of KD31: reciprocity. Combined with symbolon's role as expressing mutual obligation, often under a commercial treaty between states, there can be little doubt that the only relevant harms are these determined reciprocally. Following Aristotle, the search appears to be for a mutual interest. The notion of 'common interest' (of all citizens) appears to be put on hold in favour of a bilateral mutual interest determined by reference to harm. One can compare KD36, where it is only then that 'common interest' is reintroduced. Goldschmidt's writes: The concept of this interest [in mutual non-harm] can only be determined by that of reciprocity (allêlous). Yet the equilibration of diverse and often divergent interests requires a measure which, insofar as typos, can be sketched out by an image, like that, for example, which represents justice by a set of scales. (p. 40 [14]) Setting aside the distinctly unhelpful imagery of scales, the reference to typos indicates that Goldschmidt regards the preconception of the just according to nature as performing a truly regulatory function. From the great range of interests, it is only through this innate typos that we can identify our reciprocal interest in mutual non-harm. On such a view, it seems, each sentient being is capable of recognising in mutual non-harm something of value. They can do this because each sentient being to some extent reflects on their experiences also the end: the blessedness of neither troubling nor being troubled. From the diverse set of possible interests, any set of reciprocal interests which advances towards this end stands out, is remarkable, is a symbolon of reciprocal interest in mutual non-harm.
We have already considered (in Sect. 3.2) the difficulties with this Kantian reading of Epicurus. One might add the following: Having highlighted reciprocity as important to KD31 because of the text of Politics III, it is not now clear what Aristotle's positivist account of symbola adds to a theory grounding the just in nature. If according to the Peripatetics what is unjust in international relations is simply what peoples agree it to be in symbola, why does Epicurus remove injustice from the content of the treaty and replace it with reciprocal non-harm? Has this not so hollowed out the notion discussed in Politics III that many actual treaties would not fall under KD31? Moreover, why of all the terms in the passage does Epicurus choose to draw on symbola-treaties-as an image (rather than Lycophron's 'guarantee'), before immediately abandoning the term for synthêkê having adduced no theory relevant to public international legal affairs? There is a troubling lack of concordance between KD31 and the purported source text.
The overarching difficulty is that reading symbola as 'treaty' leaves us none the wiser as to what that particular word adds to KD31 other than a legalistic accent. If 'treaty' is meant in its narrow legal sense, why in the following KDs does Epicurus exclusively deploy synthêkai which arguably would cover a wider set of legal agreements than just treaties. If symbolon is not being used as 'treaty', then the case for drawing on the other features of the surrounding text of Politics III is severely undermined-Epicurus must be thinking of something else. Faced with an explanatory void, Goldschmidt fills it with a reading, inspired via Cicero, that privileges a Kantian practical schema. In what follows this article attempts to revisit both strands of Goldschmidt's arguments (the textual and theoretical) to address these difficulties. If we can establish an alternative source for symbolon, one which better harmonises with Epicurus' doctrine, then the 'treaty' reading could be deemed of much less relevance for our understanding.

Symbola as Reciprocal States of Affairs
This article's argument now proceeds in two stages. First, it suggests a new source for Epicurus' symbolon in two texts from Aristotle. As Bignone [33] and latterly Merlan (ch.1, [34]) have shown, Epicurus' thinking on hedonism and (political) friendship bears a clear stamp of Peripatetic views on these subjects. The Aristotelean texts on philia suggest alternative legal imagery which, it is claimed, approximates far better with Epicurus' doctrine. Building on this insight, it then argues that the Epicurean symbolon is only found in the co-creation or performance of reciprocal non-harm, and that it is this mutual filtering of interests which distinguishes legal signs from, say, divine signs.
In Sect. 1 we remarked that Long & Sedley had noted the difficulties of translating symbolon but had decided to stress its legal sense by using 'guarantee'. From a lawyer's perspective that felt perhaps too positively legal, but as against 'commercial treaty' it does have the merit of indicating the relatively personal nature of the legal frame. To guarantee is to vouchsafe and credit-to indicate one's belief in another by staking one's own social creditworthiness in favour of a third party. Yet the technicalities of a guarantee were perhaps too legal for Epicurus, and moreover-unlike commercial treaty-it does not appear to feature as an image in the legal-philosophical texts. On the other hand, there is another sense of symbolon which, while legal, is readily comprehensible to contemporary Greeks and which does feature in the texts: the tally stick.
A tally stick is an ingenious credit device, historically used wherever wooden sticks can be sourced. Two persons may agree e.g. that creditor will provide crop seed to debtor, and later, when debtor harvests and sells the crop, he will repay creditor either in kind or in coin. To signify this debt-relation some social groups, such as the Ancient Greeks, would take a stick and mark at either end the debt owed (such as one scratch per bushel due). They would then break the stick in two. Now each stick breaks in a unique way, like a fingerprint, and only those two stick pieces can be re-joined seamlessly. The debtor took one piece (the 'stub') and the creditor the other (the 'stock'-hence 'stockholder'). The stockholder then held a publicly understood sign of the debtor's obligation to him. Only on repayment of the debt would the creditor give up the stock to the debtor, who would then destroy both ends in the fire. The contemporary Greek word for tally stick was symbolon.
Aristotle deploys the tally stick in two theoretically significant images in his work. The first can be found in the Eudemian Ethics [2]: But in a way love of the opposite is also love of the good. For opposites strive to reach one another (allêlôn) through the middle point (dia to meson), for they strive after each other (allêlôn) as tally sticks (hôs symbola) because in that way one middle thing is generated (ginesthai) by the two.
What we then have are two theses on desires and natural desires (KD29, 30; pp. 77-78 [28]) which lay the ground for the discussion of the just according to nature.
It is the heart of the image, as sketched by Aristotle, to which particular attention is drawn, for it indicates an appropriate reading for symbolon in KD31. Aristotle says: the opposites are as tally sticks because 'in that way one thing is generated by two' different interests. If one of our guiding questions has been what is it that marks the symbolon as not just any sign but as that of the just according to nature, seeing symbolon through Aristotle's metaphor suggests that it is the meeting of different particular interests in non-harm which is generative of a new kind of sign.
It is worth emphasising the role different interests are playing here, because it allows Epicurus' doctrine to do as it claims: to account for normative behaviour by all, even some animals, and not just by philosophers. It also recommends to us a realist account of symbola. It is not necessary that individuals A and B have reflected on blessedness and immortality; it suffices (i) some A fear that some B could do y; and (ii) B fear that A could do x. Neither fear of x nor y may be philosophically motivated, provided that they are (however unwittingly) motivated by natural desires (KD30). It is enough that A apprehend in B a willingness and capacity not to do y to A, and for B to apprehend in A a willingness and capacity not to x. The only measure of whether A's (or B's) interest in avoiding y is just according to nature is whether really B (or A) recognises in this the tallying or exact reciprocity of their particular interests in non-harm. It is this tallying of interests which is the sign or token of which Epicurus speaks in KD31. It is this tallying which raises the symbolon beyond the very many signs we encounter in wider nature and makes it sociolegal. Yet nor should we ignore the genealogy that led us here: if each individual is fundamentally motivated by an originary trauma of being troubled and troubling in the great clash of atoms in the void, then for us some measure of therapy for that trauma is afforded by the mutual recognition of fear of harm and harming in the symbolon.
On this realist account we no longer have individuals being struck as if by lightning by the natural justness of reciprocal non-harm arrangements. Rather, we begin, as the Epicurean genealogies tell us (Lucretius, De Re. Nat. V. 953−61, V. 1011-27 [19]; Porphyry reporting Hermarchus, De abst. 1.7ff [23]), with a pre-legal motivation to avoid harm and so natural affinity (oikeiôsis). 17 It is only through accumulated experiences of modes of cohabitation and control that humans come to 'understand the usefulness of the law' (Hermarchus, ibid.). These genealogies also tell us that the wise were able to discern what was or was not useful to the general structure of life, but even they did so 'after studying men's social life and their dealings with one another' (Hermarchus, ibid.). All this suggests that rather than the symbolon being a single typos or category of practical thought permitting the philosopher at least immediately to grasp the natural justness of some arrangement, like all Epicurean science we learn through the accumulated experience of multiple human dealings. We reflect on their similarities, insofar as similar, and identify the outlines of common features in this time and this place, between these people. It is the various parties themselves, through iterated dealings, who come to delineate the sign of what is naturally just, grounded always in whatever particular notions they may have about what is actually harmful here and now.
If we wish to find the naturally just (and not according to some ideal or sectional interest) experience teaches us that we need to look for a special class of signs. Being born of experience these signs are not all the same: they represent in each particular case the specific interest of two or more sentient beings. That interest is reciprocal and draws on the apprehension of perceived harm by an individual. Reciprocity does not mean that the two particular individuals have an apprehension of the same harm, or the same kind of harm; the harms apprehended by each may be radically different. The reciprocity arises just when the two individuals apprehend that their respective needs for non-harm tally. Hence the symbolon is not merely a reflection of harm or even fear thereof, but is generated via reciprocity, and therefore is properly a sign of that reciprocity, of that tallying of individual perceptions of harm in each case.

Conclusion
This article has argued that while various readings of KD31 offer careful textual evaluations of the wider Epicurean doctrine, they often discount the specificity of the notion of legal sign (symbola) and its centrality to Epicurus' definition of 'what is just according to nature'. Some of these readings find an innate preconception of the naturally just in Epicurus' thinking, but in so doing impose a Ciceronian and even Kantian reading of the doctrine. Others identify the legal character of symbola but simply import legal terminology as if Epicurus sought an exogenous but unexplained basis for his legal doctrine. We have argued the distinct feature of Epicurean legal signs is that they (a) reflect the interest of a given sentient being, and (b) only find their validation in the real reciprocal tallying of these interests. We advanced new textual support for reading symbolon as a tally sick, noting that it is just this image which Aristotle uses to define friendship and protect the common interest where there are divergent desires-themes which are openly discussed in the Key Doctrines immediately surrounding KD31. We have proposed condition (a) so as to distinguish legal signs from signs of external nature; but it is condition (b) which distinguishes these signs from images of the divine and raises symbola beyond individualistic considerations to properly social signs. Symbola are generated in each case by the meeting of otherwise divergent interests: what is expressed is not simply an abstract exercise in philosophical reflection, but the real generation of a sign through the coming together of two individuals. In view of the above, we suggest the following reading of KD31: The just according to nature is a tally (symbolon) of reciprocal interest, for neither harming one another nor being harmed.
Being capable of generation even when the parties are not philosophers allows the Epicurean doctrine's explanatory power to extend across the entire social field.