PHIALAI INSCRIPTIONS: LISTS, NOT INVENTORIES

The Athenian phialai inscriptions defy easy classification. The surviving fragments constitute at most 19 different stelae, their content divided into columns and short entries. The roughly 421 entries preserved in whole or in part must be only a fraction of the original figure.Footnote ² On most (but not all) of the stelae, each entry records the dedication of a 100–drachma libation bowl (φιάλη; Fig. 1).Footnote ³ On all the stelae, each entry lists two or more names, framed as prosecutors and defendants in a trial, often accompanied by the word ἀποφυγών (‘acquitted’) indicating the verdict. Some fragments lack any indication of verdicts, although the courtroom setting remains apparent (see SEG XXV 180, esp. ll. 12–19). Entries of this format omit the word ἀποφυγών and reverse the order of names, placing the prosecutor's name first (in the nominative) and the defendant's name second (in the accusative). (Interestingly enough, the same format can be found among the manumission inscriptions of Oiniadai, with the manumission verb omitted, the manumitter's name in the nominative, and the slave's name in the accusative [Powell Reference Powell1904, 181–6].) Despite occasional debate over the precise significance of this ‘reversed formula’, scholars up to and including Meyer have always understood the basic truth of the matter – those entries which omit ἀποφυγών list the prosecutor first, and the defendant second (Lewis Reference Lewis1959, 235, 237; Reference Lewis1968, 372; Meyer Reference Meyer2010, 12–13, 136). Whatever the entry format, the type of case reflected in the phialai inscriptions is traditionally identified as dikē apostasiou on the basis of the broken word [-]στασίου preserved in a fragmentary heading.Footnote ⁴

Fig. 1.

Greek. Libation Bowl (Phiale), 5th century bc. Silver. Overall: 7/8 × 6 3/16 in. (2.2 × 15.7 cm). Evan F. Lilly Memorial. Eskenazi Museum of Art, Indiana University 69.102.2. Photograph by: Kevin Montague.

In fourth-century Athens, slave-owners had the right to bring apostasiou prosecutions against their freed slaves. According to Valerius Harpocration, the second-century ad Alexandrian scholar who produced a glossary to the Attic orators (s.v. ἀποστασίου):

Vulnerability to apostasiou prosecution, then, entailed a freed slave failing to honour certain legal requirements and service obligations towards the former master. Those convicted on such charges, Harpocration explains, were re-enslaved, while the acquitted achieved ‘absolute freedom’ from the various requirements and obligations, removing any risk of future apostasiou prosecution (τελέως ἤδη ἐλευθέρους).

In the phialai inscriptions, each entry includes a non-citizen defendant, identified by oikōn en nomenclature and often by occupation title, and at least one prosecutor, who is usually a citizen, but sometimes a metic, an isotelēs, a proxenos, a foreign refugee, a minor acting through a kyrios, or even a public slave. In an apostasiou context, these two roles will correspond to the freed slave and the slave-owner, respectively. Most of the stelae expressly identify acquittal as the outcome of the case; conviction is never mentioned. The inscriptions seem to date roughly to the Lycurgan era of Athens (338–322 bc), and another fragmentary heading containing [-]ούργου suggests that a policy of Lycurgus himself may have generated these documents.Footnote ⁵

With all this in mind, we can take up questions of epigraphic genre. How should we classify the texts, and what function do they serve? In recent years scholars have tended to describe the phialai inscriptions as inventories of dedications (D. Harris Reference Harris1992, 637–8; Meyer Reference Meyer2010, 59–69; E. Harris Reference Harris and Faraguna2013, 172). This notion carries a certain intuitive appeal, given the prominence of dedicatory phialai throughout the documents. Yet it is, I believe, a misclassification. That some of the fragments categorically omit any mention of dedications from their entries precludes a simple identification of the inscriptions as inventories. Surely an ‘inventory’ could not omit the very objects it seeks to record. Among the stelae that omit phialai from their entries (‘Typ U’ in Kränzlein Reference Kränzlein and Wolff1975) is IG II² 1576 – which in fact accounts for two, if Meyer (Reference Meyer2010, 129) is correct that its two fragments come from separate stelae. IG II² 1578 is probably another; as Lewis (Reference Lewis1959, 236–7) notes, ‘although the phialai have been restored, there is no reason to suppose their presence.’Footnote ⁶ Another possibility is IG II² 1577, which ‘may not have phialai, but is so fragmentary that nothing is certain’ (Meyer Reference Meyer2010, 13 n. 9). Meyer (Reference Meyer2010, 136) proposes that the lost headings of these stelae would have originally stated that each entry corresponds to a dedication. Lewis's (Reference Lewis1959, 237) solution – and I agree – is that this format predates the imposition of mandatory dedications upon courtroom manumissions. Either way, the phialē-free format would seem to prioritise the trial verdict content (and the individuals named therein) over any function as an inventory.

Moreover, the inscriptions do not conform to any standard class of ancient Greek inventory: they are not magistrates’ annual paradosis accounts, nor ad hoc exetasmos inventories, nor kathairesis accounts of melting down old dedications (on these three types see Aleshire Reference Aleshire1989, 103–10). Of these options only the last appears promising, since phialai exeleutherikai (commonly associated with the dedications mentioned in the phialai inscriptions) were in fact melted down, starting in 321/320 bc (IG II² 1469, ll. 3–12). But this appears to be the very first melting of phialai exeleutherikai, and the year is too late to reasonably assign a kathairesis function to inscriptions dating back to the 330s.Footnote ⁷ The 321/320 date suggests policy changes implemented by the post-Lamian War oligarchy rather than the continuation of a Lycurgan melting programme. While it may be inadvisable to insist that all inventory documents must rigorously conform to the three standard types, the challenge of classifying the phialai inscriptions remains.

But the problem is not intractable. Another class of phialai inscription provides a close parallel – so close, in fact, that one stele includes both types of phialai record on opposite faces (IG II² 1575). I refer to the phialai leitourgikai inscriptions, which record the 50-drachma phialai dedications of liturgists (Lewis Reference Lewis1968, 374–80; Lambert Reference Lambert2012, 224–34). These documents show great interest in the dedicatory vessels, recording their weight in each entry, meticulously noting deviations in weight as small as one drachma, and even listing a sum total of phialai (IG II³ 1 550, ll. 27, 31; SEG XXV 177, l. 51). Yet they are not inventories, at least not primarily, and this is clear enough from the documents’ own words. The heading of IG II³ 1 550 provides a crucial insight into their function: ‘the following were liturgists’ (οἵδε ἐληιρτού[ργ]ησαν, l. 5), it says, before launching into the body of the text. As important as the dedications may be, the emphasis is above all on the people involved. The same must be true of the phialai inscriptions that record trial verdicts, which can omit dedications but never the names of prosecutors and defendants. Both types of phialai record, then, are not inventories, but rather lists of individuals and publicly consequential actions, which happen to record (on most of the stelae, at least) related dedications as well.Footnote ⁸

Fiscal matters are thus of secondary importance in these inscriptions, though hardly insignificant. Recording the dedications and their precise weights on the stelae would have reflected positively on the achievements of Lycurgus, giving a sense of the volume of income paid to Athena thanks to his new financial measures. But the primary function, it seems, was to publicise information in the interest of the individuals named on the stelae. The phialai leitourgikai records proclaim the deeds of the liturgical class, providing an outlet to make their philotimia manifest to the Athenian people. A stele attesting not only to the performance of a liturgy, but also to the dedication of a metal vessel, would stand as an enduring testimony to one's civic-minded achievements. The phialai inscriptions concerning trial verdicts will have played a comparable role: propagating certain information favourable to the individuals named on the stelae. If, as I argue below, the idea of manumission through apostasiou is correct, that role comes sharply into focus. The inscriptions publicised information of vital importance to freed slaves, with lists of apostasiou verdicts and corresponding phialai dedications certifying their new freedom.

EVIDENCE FOR A MANUMISSION CONTEXT

Numerous pieces of evidence suggest a link between the phialai inscriptions and manumission.

4) Families and children

Seventeen children and around nineteen family groupings occur among the defendants in the phialai inscriptions (see Appendix; Table 1). Though the interpretive value of this evidence has rarely been recognised, the presence of families and children is one of the inscriptions’ strongest indications of a manumission context – so strong, in fact, as to verge on a diagnostic feature.

Table 1

Families in the phialai inscriptions.

First of all, the presence of so many children is an obstacle to any argument that the phialai verdicts are unrelated to manumission. Minors could, it seems, be held legally responsible in certain circumstances, such as involuntary homicide (Golden Reference Golden2015, 35; see also [Plutarch] Lives of the Ten Orators 842e). Yet the phialai inscriptions concern not only children, but young children – in 16 out of 17 cases, the child is identified as a paidion. That the remaining child is called a pais (on the same stele as several paidia, no less) shows that the documents do intend an age distinction between the two terms, with paidion conveying its literal meaning of ‘young child’ (cf. Rosivach Reference Rosivach1989, 368–9; see also Golden Reference Golden2015, 10–12). A paidion could be as young as a baby and was, according to a fragment of Hippocrates, no older than seven years (Philo, De Opificio Mundi 36.105).Footnote ¹⁴

If the idea of genuine prosecutions seems to crumble under such circumstances, the presence of families and children is entirely consistent with manumission. It is completely intuitive – and amply attested in slave systems throughout history (e.g., Mouritsen Reference Mouritsen2011, 285–6; Hünefeldt Reference Hünefeldt1994, 79–85; Whitman Reference Whitman1997, 119–39; Groth Reference Groth2017, 59) – that manumission would accompany a concerted effort to secure freedom for one's enslaved family members (even if those family ties, formed within the bonds of slavery, lacked legal recognition). Significantly, the phialai inscriptions do not only furnish examples of families residing together in the same deme; two or three families include a child living apart, readily explainable as families broken up by slavery, with the parents managing to secure their child's freedom following their own manumission (see Appendix, nos 3, 6, 7). If one rejects a manumission context for the inscriptions, these cases seem to elude explanation.

Furthermore, the preponderance of paidia over older paides (Table 2) very plausibly reflects the economics of manumission: it would be cheaper to purchase the freedom of a paidion than of a pais. This price disparity shows up in the ‘Attic stelae’, when a paidion sells for 72 drachmas, while a pais sells for 174 drachmas, ‘the average fee for an adult’ (Pritchett Reference Pritchett1956, 276). It is also significant that a majority of the child defendants are female: 10 girls, 6 boys, and 1 unknown. Though hard figures are lacking to prove the point, Athenian slave-owners may have assigned less value to enslaved girls than to boys, and set a lower manumission fee accordingly. Such is apparent in the Delphic manumissions from the second and first centuries bc, in which the average price of freedom is higher for boys than for girls at a roughly 55:50 ratio (Hopkins Reference Hopkins1978, 159). A modern example will nicely illustrate the kind of economic calculations I see at work in the phialai inscriptions. Consider the case of early national Maryland, ably treated by T. Stephen Whitman, where manumission documents exhibit the same trend with respect to both age and gender:

Free black parents whose children had been born as slaves would have known that buying out their sons and daughters would become more costly as they grew older and that girls, even infant girls, could be freed more cheaply than boys … Nearly twice as many children were freed before their fifth birthday as between ages five and nine, and more than twice as many were freed before age five as were liberated between the ages of ten and fourteen. (Whitman Reference Whitman1997, 133)

The factors that Whitman adduces to explain the Maryland evidence could apply just as well to Classical Athens: risk of child mortality; lack of economic productivity on the part of young children, despite the necessity of expending resources to raise them; expectations that, on average, boys would yield greater returns than girls over the course of a lifetime (Whitman Reference Whitman1997, 132–3). To this we can add the risk that girls would grow up to die in childbirth. There is, in short, ample reason why young children and girls would have been freed at a higher rate than older children and boys. Thus, manumission offers a good explanation for the appearance of so many young children in the phialai inscriptions, as well as the slight preponderance of girls over boys. There is no obvious reason why lawsuits unrelated to slavery should disproportionately target very young children – in fact, their regular presence as defendants seems most unlikely – and the inclination towards girls over boys suggests a price of freedom that varied according to gender.

Table 2

Children in the phialai inscriptions.

Finally, it is necessary to state that the occasional manumission of families is entirely consistent with the available Athenian evidence. In ancient Greece more broadly, it was not unusual to find enslaved family members set free together: the Delphic manumissions include numerous instances of mothers manumitted together with their children, and other locations throughout the Greek world provide further examples of manumission respecting family ties (Zelnick-Abramovitz Reference Zelnick-Abramovitz2005, 163–7; Reilly Reference Reilly1978, 111 n. 5). Classical Athens was no exception. The will of Aristotle promises freedom to a father and his child following the marriage of Aristotle's daughter (Diogenes Laertius 5.15). The will of Theophrastus contains provisions for Pompylos and Threpte, perhaps manumitted simultaneously as a romantic couple, since they are described collectively as ‘long since emancipated’ (Πομπύλῳ δὲ καὶ Θρέπτῃ πάλαι ἐλευθέροις οὖσι; Diogenes Laertius 5.54). In Hyperides’ Against Athenogenes, the plaintiff describes purchasing an enslaved father from a perfume shop, along with his two sons, in order to free them; a family of perfume dealers in the phialai inscriptions is remarkably similar in composition (Appendix, no. 19). In the recently published fragment from Hyperides’ speech Against Timandros, the speaker asserts that even slave-traders (ἀνδραποδοκάπηλοι) avoid splitting up families, to the point of willingly suffering economic loss (Jones Reference Jones2008). Whatever the extent of rhetorical exaggeration here, the claim's persuasive force depends on an audience of fourth-century Athenians assigning some value to maintaining the integrity of slave families, and slave-owners may have been in a position to feel these social pressures rather more acutely than slave-traders. But one need not attribute the existence of slave families purely to the benevolence of masters; allowing slaves to form family structures could just as well function as a form of social control (Xenophon, Oeconomicus 9.5). And masters eager for profit might have turned familial affection to their advantage, motivating a slave seeking freedom to procure a self-purchase fee not only for himself, but for an entire family. Thus, the presence of families in the phialai inscriptions points to the social and economic logic at work behind the scenes of manumission.

KITTOS AT ATHENS AND DODONA

By now, I hope to have shown that the phialai inscriptions have something to do with manumission, even if it remains difficult to see exactly what was going on behind the scenes (I will defer that discussion until the penultimate section of this paper, ‘Apostasiou as Manumission’). We can now ask what evidence Kittos adds to the conversation. Here is his entry in the phialai inscriptions (Meyer Reference Meyer2010, nos 2–9, face A, ll. 100–3 [= IG II² 1554, ll. 10–13]):

These lines inform us that Kittos, a non-citizen metalworker (χαλκεύς) residing in Piraeus, has been acquitted in a prosecution by Dionysios, himself a non-citizen enjoying the honorary isotelēs status in Athens. A phialē weighing one hundred drachmas has been dedicated. Immediately following these lines, we find an entry for a woman named Mnesithea, a textile worker (talasiourgos) living in Piraeus, acquitted from the same Dionysios. Mnesithea was probably the wife or cohabitant of Kittos.Footnote ²⁰ According to the traditional interpretation of the phialai inscriptions, these two entries inform us that Dionysios has manumitted both Kittos and Mnesithea.

Compare the phialē entry of Kittos to the following oracular tablet from Dodona, dated by its editors to the mid-fourth century (Dakaris, Vokotopoulou and Christidis Reference Dakaris, Vokotopoulou and Christidis2013, no. 1411; printed provisionally by Eidinow Reference Eidinow2007, 104 [= SEG LVII 536]):

Here we again find a Kittos and Dionysios, and the question at hand is manumission. Kittos, presumably the questioner (referring to himself in the third person), inquires whether Dionysios will set him free as he promised.Footnote ²¹

Although the name Dionysios is very common, there is reason to believe that these two inscriptions refer to the same individuals. Supporting the identification of Kittos and Dionysios is, first of all, the date of each document. The Dodona tablet dates to roughly the middle of the fourth century bc, and the phialai inscriptions date to the 330s–320s.Footnote ²² No archon year survives on the stele listing Kittos's phialē, but one of the other manumission acts recorded on the same face may date to the 320s.Footnote ²³ While precise dating may be impossible for both documents, the dates seem to support a chronology in which Kittos first inquired about – and was later granted – his freedom.

Second, it seems clear that the Kittos in the Dodona tablet was a slave from Athens. Not only is the tablet written in the Attic dialect, but the name itself betrays an Athenian origin – Κίττος, not Κίσσος, with the Attic double tau rather than a double sigma. The name Kittos is otherwise unattested outside of Attica during the Classical period, and thereafter it is limited to only four non-Athenian attestations.Footnote ²⁴ True, the double tau is not exclusive to Attic, appearing also in the Boeotian, Cretan, and Euboean dialects (Buck Reference Buck1955, 69–70), but none of these can be said to apply to Kittos's oracular tablet, which is fully consistent with Attic.Footnote ²⁵ Thus, Kittos – seemingly a slave from Athens, and owned by a certain Dionysios – inquired at Dodona about his potential manumission around the time when homonymous individuals appear in the Athenian phialai inscriptions. Arguably, they are identical.

The situation is somewhat more complicated, however, because this pairing of names is no accident. The word κισσός (κιττός in Attic) means ‘ivy’. As a pair, the names Kittos and Dionysios betray a meaningful onomastic relationship, making reference to the sacred ivy associated with the god Dionysus. (In Attica this aspect of the god found expression in the cult of Dionysus Kittos at Acharnai.Footnote ²⁶) It is not uncommon for the name Kittos to coincide with a variety of Dionysian names. Excluding the phialē entry and Dodona tablet, I am aware of five such examples.

1. (1) A second-century bc Attic gravestone commemorates Kittos, son of Dionysios Sinopeus (IG II² 10331 [= IG II² 10350]).

2. (2) A Kittos and Dionysiodoros occur together in a list of actors from the late fourth or early third century bc (IG II² 2418; Ghiron-Bistagne Reference Ghiron-Bistagne1976, 72). These are perhaps stage names, perhaps relatives in a family of actors – but in either case the names signal their theatrical trade, an art associated with the god Dionysus.

3. (3) Members of a family of potters in fourth-century Athens used the names Bakchios and Kittos, ultimately emigrating towards the end of the century (Robertson Reference Robertson1992, 292–5).

4. (4) The preceding family must be separate from the Bakchios and Kittos targeted in an Attic curse tablet from the third century bc (Audollent Reference Audollent1904, no. 72). The curse does not specify their relationship to each other, but the Dionysiac onomastic connection may extend to Kados, listed a few lines above in the same curse, whose name denotes ‘a vase used to store and transport wine’ (Amyx Reference Amyx1958, 186–90).

5. (5) Kittos son of Kados on a second-century bc gravestone would then exemplify this same phenomenon (IG II² 11845a [= Agora XVII 879]).

What does all this mean for the phialai inscriptions? Chiefly, it shows that the pairing of the names Dionysios and Kittos as prosecutor and defendant is probably not coincidence. Finding both names together suggests a personal relationship. A master naming a slave, a father naming a son, or a more distant family connection could all account for such onomastic similarities. But the pairing in the phialē entry cannot, I believe, be explained by appeal to family ties. It is clear enough from context that we are not looking at two family members. This would require positing that Kittos, seemingly a humble artisan, belonged to the same family as Dionysios, a recipient of an honorary status typically reserved for the wealthy (Niku Reference Niku2007, 31). But it is not at all implausible to imagine Dionysios amassing wealth, and ultimately procuring a grant of isoteleia, through ownership of a metalworking business based on slave craftsmen. For comparison one naturally thinks of the shield-making workshop owned by the family of Lysias, or the knife-making workshop inherited by Demosthenes, both of which utilised dozens of slaves in their day-to-day operations (Lysias 12.8, 19; Demosthenes 27.9). A donation of shields was one of the benefactions that led Athens to honour Pasion with citizenship (Demosthenes 45.85; [Demosthenes] 59.2). Thus, the Dionysios and Kittos of the phialai inscriptions suggest neither family ties nor an onomastically random encounter in the law-courts, but a master–slave relationship.

At this point one might acknowledge the onomastic connection between Kittos and Dionysios, but dispute the likelihood of identifying the slave at Dodona with the freedman at Athens. After all, Dionysios was one of the most common names in the Greek world, and even though Kittos was not (the names occur 1139 and 25 times respectively in LGPN IIA), the name-pair clearly is not random. Perhaps Dionysioi at Athens often named their slaves Kittos, and the Dodona tablet is simply another example of this trend. But at present this appears unlikely, as one crucial point reveals: the available evidence suggests that slaves in Classical Athens were sometimes named after their masters, but only rarely (Vlassopoulos Reference Vlassopoulos, Taylor and Vlassopoulos2015, 111–14; above, note 15). The unique naming convention apparent in the case of Kittos, then, somewhat mitigates the risk of coincidental homonyms. Not only are the phialē entry and Dodona tablet linked by chronology, by emancipatory context and by homonymy of master and slave, they are also linked by the rarity of the naming convention at work. This strengthens the argument that the slave who inquired about his freedom at Dodona is identical to the Kittos recorded in the phialai inscriptions.Footnote ²⁷

Moreover, one can point to a specific historical event around the time of Kittos's manumission that may well have drawn the enslaved metalworker to Zeus's oracular precinct. At some point in the period 331/330–325/324, Athens dispatched a special religious procession to Dodona (Hyperides, Defence of Euxenippus 24–6; Whitehead Reference Whitehead2000, 155–7, 223–9). Such processions may have been a recurring event, but on this occasion it took on an added significance: Zeus himself had ordered the Athenians to refurbish a cult statue of Dione at Dodona.

The Athenians prepared three types of new adornment for the statue: (1) a ‘face’ (prosōpon), (2) related objects accompanying the face and (3) kosmos. If not the others, the last of these would have depended on the efforts of metalworkers, since a statue's kosmos regularly included metal objects: rings, bracelets, gold crowns, tiaras and the like (Mansfield Reference Mansfield1985, 506–7). The passage implies that the objects were crafted in Athens before the procession set out. If the city engaged a workshop of Dionysios to forge the kosmos, Dionysios may well have chosen or needed to see the task of kosmēsis to completion, accompanying the procession to Dodona and bringing Kittos along.

Of course, it may instead have been somewhat more mundane circumstances that drew Kittos to Dodona. Such a prominent temple facility would surely require metalworkers on a regular basis. (Dodona was particularly renowned for its enigmatic Δωδωναῖον χαλκεῖον.Footnote ²⁸) Though it might seem strange for the sanctuary to recruit metalworking talent from so far afield, Greek temple authorities often enlisted the aid of foreign craftsmen (Feyel Reference Feyel2006, 341–68). In 334/333 bc, several metalworkers voyaged from Athens to Delphi to work on a krater and perirrhanterion, working for twelve days at a daily wage of three obols (CID 2.79A, ll. 23–32). And what if the motivation to visit Dodona was not economic, but purely religious? The great diversity of dialects preserved in the oracular tablets confirms that the site attracted believers from across the Greek world; variants of Doric Greek prevail, but Attic, Ionic, and Aeolic also occur (Lhôte Reference Lhôte2006, 363–80). Esther Eidinow feels it is plausible that slaves who lived apart from their masters had sufficient autonomy to make the voyage to the oracle, and slaves’ questions about their potential freedom are a recurring theme among the inquiries (Eidinow Reference Eidinow2007, 102–5; Parker Reference Parker2016, 83–4).

Still, if any episode of the Lycurgan era offers a likely context for Kittos's trip to Dodona, it is the embellishment of the statue of Dione. This event may even offer a clue about the origin of Dionysios's grant of isoteleia. The preparations for adorning the statue would have sparked public interest domestically, arising as they did from an oracle to the Athenian state. The procession that followed was of such a scale as to stir up commotion at Dodona – Olympias later fired off angry missives urging the Athenians to stop interfering in Molossia – and Athens must have incurred great expense preparing the adornments for the statue (Hyperides, Defence of Euxenippus 24–6). If Dionysios had, for instance, let his workshop forge the kosmos at cost, such a service would be grounds for an honour like isoteleia. Suddenly thrust into the public eye, Dionysios's reputation as a business-owner would have been on the line. A promise to manumit Kittos for a job well done may have served both their interests.

As it happens, Dionysios arguably turns up again on a fourth-century Attic gravestone (IG II² 7868 [= CAT 2.885]):

Three names are inscribed on the stone, corresponding to the three figures depicted in relief: Eutychis, Philippos and (inscribed later in a different hand) their son Dionysios isotelēs. The name Dionysios is so common that any attempt to associate this isotelēs with the owner of Kittos must remain tentative. It is possible that two contemporaries of the same name enjoyed isotelēs status in Athens. With that disclaimer in mind, several details make the identification worthy of consideration. First, it appears that no other known Dionysios isotelēs apart from the manumitter of Kittos can make a claim to this tombstone.Footnote ²⁹ Though Dionysios was a very common name, the number of mid-fourth century isoteleis named Dionysios was certainly a much smaller figure. In the judgment of Michael Osborne (1981–3, vol. 2, 33), ‘Isoteleia was not very frequently granted in Athens in practice.’ Next, the tombstone itself yields several details about the life of Dionysios which seem compatible with the master of Kittos. Clairmont observes,

The inscription further reveals that Dionysios earned his isoteleia himself rather than inheriting the status from his father, since Dionysios alone is described as isotelēs, and his name was added to the stele later. This all raises chronological questions: for the two Dionysioi to be identical, their biographical details must be compatible with the date of the tombstone. Kirchner assigns a date of ‘med. s. IV a.’, while Clairmont dates it on iconographic grounds to 375–350 bc. This dating window would accommodate a timeline in which Dionysios, a child when his mother died, grows up, accumulates wealth (including a metalworking business), earns a grant of isoteleia and ultimately frees Kittos and Mnesithea in the 320s before dying himself, at which point his name was added to his parents’ gravestone.

Kittos makes one final appearance. A joint account of the treasurers of Athena and a board overseeing the golden Nikai, dating to roughly the 330s–320s, records the work of several goldsmiths who melted down old dedications. Among them was a man named Kittos, who took over 2950 drachmas in golden objects, produced 2531 drachmas of pure gold upon melting them down, lost 419 drachmas in the process, but was able to recover 116 drachmas ‘from the ashes’, ultimately giving back a total of 2647 drachmas of pure gold to the treasurers (IG II² 1495, ll. 3–10).Footnote ³⁰

As Diane Harris (Reference Harris1988, 330–1 n. 3) has observed, this Kittos χρυσώτης may be identified as Kittos χαλκεύς of the phialai inscriptions. Indeed, the different occupation title (χαλκεύς vs. χρυσωτής) is no cause for alarm, since χαλκεύς was a generic word for ‘metalworker’.Footnote ³¹ In fact, Kittos's apparent proficiency in working with gold is reminiscent of the work he (possibly) performed crafting kosmos for the statue of Dione. There is some reason to believe that Kittos took on this melting project before his manumission, although certainty is impossible (compare the dates above in nn. 23 and 30). Either way, this document highlights two points. First, we see how the policies of Lycurgus permeated the age in which Kittos lived. Finding employment with the same board of treasurers who oversaw the dedicatory phialē commemorating his freedom, Kittos came into contact with a celebrated Lycurgan initiative: the restoration of golden Nikai that were melted down to mint coinage during the Peloponnesian War ([Plutarch] Lives of the Ten Orators 841d, 852b). Second, an economic observation: between taking on a contract for the treasurers of Athena and (perhaps) working temporarily at Dodona, one gets a sense of the various short-term metalworking projects that helped Kittos earn a living. Given his residence in Piraeus, one even wonders if the production of bronze naval rams for the fleet was among Kittos's diverse sources of income.

APOSTASIOU AS MANUMISSION

The available evidence points in favour of associating the phialai inscriptions with manumission, and Kittos presents an opportunity to examine the life of an individual slave in some detail. Still, we have hardly begun to address what has traditionally been the principal controversy surrounding these documents: do the phialai dedications correspond to genuine dikē apostasiou trials, with slave-owners prosecuting their former slaves who shirked the obligations incumbent upon them (thus Zelnick-Abramovitz Reference Zelnick-Abramovitz2005, 284–90)? Or are these cases instead a form of manumission carried out in the law-courts, perhaps through some form of collusive litigation? I support the latter view.

Central to this debate is a fact drawn from Harpocration's definition of apostasiou (s.v.): freed slaves acquitted in apostasiou cases become ‘absolutely free’ (τελέως ἤδη ἐλευθέρους). That the phialai inscriptions mention acquittal but never conviction has led many to accept the notion that apostasiou could function as a purely administrative procedure, either to set slaves free outright or to release freed slaves from certain legal obligations. This reasoning deserves to be met with some scepticism. Why should we make a speculative leap beyond the simplest explanation of the legal terminology in these documents? When we encounter references to juries, law-courts, acquittal, and even the standard panel of five men charged with overseeing the water clock and the votes, it seems reasonable enough to envision full-fledged trials underlying the phialai inscriptions (IG II² 1560, ll. 4–5; SEG XXV 180, ll. 12–19; cf. [Aristotle] Athenian Constitution 66.2).

Nonetheless, several features of the inscriptions elude explanation by appeal to genuine apostasiou prosecutions. Foremost among them are the 100-drachma phialai. Understood as a kind of manumission fee or tax, the phialai dedications are consistent with the revenue-generating policies of the Lycurgan period, and such a tax finds numerous parallels in other quarters of the Greek world (Zelnick-Abramovitz Reference Zelnick-Abramovitz2013, 71–107; Migeotte Reference Migeotte2014, 273–5). But if the dedications were the outcome of genuine prosecutions, how could the state reasonably impose such a great expenditure on freed slaves, regardless of their ability to pay?Footnote ³² Whereas we can envision masters occasionally helping their slaves finance a required fee upon manumission, either through outright contributions or by organising friendly eranos loans, this would surely not occur in an adversarial context – and yet there is at least one example of a prosecutor who also leads the koinon eranistōn involved in the case.Footnote ³³ Alternatively, if the phialē was not mandatory but only an optional fee to publicise one's new legal status following apostasiou acquittal, it remains difficult to envision freed slaves embracing this expenditure with such enthusiasm. One-hundred drachmas would be a steep price to pay for the likes of textile workers, donkey drivers and fishmongers.

Moreover, associating the phialai exeleutherikai with cases of manumission in the law-courts lets us avoid several startling conclusions: that Proxenos and Pherekleides prosecuted their secretarial assistants; that apostasiou prosecutions were ubiquitous in Lycurgan Athens; that ten or more genuine private suits could occur in the same law-court on the same day (SEG XXV 180, ll. 41–68, note ll. 12–19; cf. [Aristotle] Athenian Constitution 67.1); and that individual children (even young children) were regularly prosecuted for apostasiou. This solution also accounts for the families that appear to be split apart through slavery – while manumitted parents would naturally strive to secure their children's freedom, it is unclear how a context of genuine apostasiou (or any other type of trial) could explain such examples (see Appendix). The notion of apostasiou as manumission is fully consistent with the phialai inscriptions’ standard entry format, ‘NN [slave] acquitted in a prosecution by NN [manumitter].’ But even more compelling, it offers a good explanation for the ‘reversed’ entry format that places prosecutors’ names in the nominative case, followed by defendants in the accusative, while omitting the formulaic word ‘acquitted’ (ἀποφυγών) (Lewis Reference Lewis1959, 237–8; Reference Lewis1968, 372; ‘Typ B’ in Kränzlein Reference Kränzlein and Wolff1975). What verb should be understood in its place? A reasonable guess is to supply something like ἀφῆκε ἐλεύθερον, a phrase pertaining to manumission.Footnote ³⁴ This is consistent with the presence of ἀ̣φιεμένου ἐλ̣[-] at IG II² 1560, l. 2 (as read by Rocca Reference Rocca2011). The reversed formula would thus signify ‘NN [manumitter] has set free NN [slave].’

The hypothesis of apostasiou as manumission thus has broad explanatory power, and merits serious consideration. But we need not settle for supposition; at least two literary sources seem to attest to the practice.Footnote ³⁵ First, a fragment of Isaeus, quoted by Dionysius of Halicarnassus, plainly reveals the existence of manumission in court in mid-fourth-century Athens (Isaeus 5 [= Isaeus fr. 15 Thalheim]):

This case concerns an incident of aphairesis eis eleutherian, with Dionysios claiming Eumathes as a slave and the speaker (apparently named Xenokles [Isaeus fr. 16 Thalheim]) asserting Eumathes's freedom. Though the speaker makes no explicit reference to apostasiou, he justifies his position by stating that Eumathes had been manumitted in court (ἐν τῷ δικαστηρίῳ). This courtroom manumission by Epigenes is not itself a matter aphairesis eis eleutherian, but rather a prior action which the speaker invokes to defend his assertion of Eumathes's freedom (pace Meyer Reference Meyer2010, 21 n. 42). As Dionysius of Halicarnassus writes in preface to the quotation, the heir of the manumitter seeks to claim Eumathes as his slave. Epigenes, then, is the original manumitter from whom Dionysios claims to have inherited Eumathes – a claim the speaker disputes, given his knowledge of the courtroom manumission.Footnote ³⁶

The fragment provides no specific date for the manumission of Eumathes, but it plausibly coincided with Eumathes opening his own bank.Footnote ³⁷ Probably no more than a few years had passed since the speaker's trierarchy in 358/357, and this comes as something of a surprise, as it well predates Lycurgus's oversight of financial affairs beginning in the 330s. But an attractive solution presents itself: that mandatory phialai dedications were a later addition to a pre-existing form of manumission in court. This, in fact, solves two problems at once, for it was in precisely these terms that David Lewis explained the handful of phialai inscriptions that omit the phialē from each entry (Lewis Reference Lewis1959, 235–7; cf. n. 3 above). The emergence of ‘apostasiou as manumission’ in the mid-350s seems a good contextual fit with the policies of the time. Following the Social War there was some desire among Athenians to improve the lot of metics (Meyer Reference Meyer2010, 47–50). The clearest example of this attitude is Xenophon's Poroi of c. 355/354, proposing to ‘remove whatever seems to dishonour metics while according no benefit to the city’ (εἰ ἀφέλοιμεν μὲν ὅσα μηδὲν ὠφελοῦντα τὴν πόλιν ἀτιμίας δοκεῖ τοῖς μετοίκοις παρέχειν; 2.2). The ultimate goal, of course, was to augment revenues from the metic-tax, and instituting a new type of manumission would lure no new immigrants to Attica. But freed slaves were metoikion-payers in their own right, and providing a form of manumission that stripped away various triggers for re-enslavement (as apostasiou acquittal naturally accomplished) would help ensure that freed slaves remained within the Athenian tax base. Such a policy may have also sought to reduce the emigration of freed slaves, as it entailed granting freedom on comparably favourable terms. I doubt that any regulation existed at this time to compel masters to use the courtroom for manumission (perhaps later Lycurgus would have introduced such a requirement in order to maximise income from the phialai). But the social pressures of a civic effort to increase revenues from the metic tax, combined with masters’ willingness (as shown by a fourth-century reference to theatre proclamations [Aeschines 3.41, 3.44]) to embrace public forms of manumission, may have been enough to make the law-courts a popular venue for freeing slaves in the 350s–340s.

While the Isaeus fragment speaks of manumission in court without expressly referring to apostasiou, a passage of Aristotle helps make the case that apostasiou was in fact used as a mechanism for freeing slaves in fourth-century Athens. One consequence of acquittal in apostasiou trials was the right to choose one's own prostatēs (Harrison Reference Harrison1968, 185). If manumission through apostasiou really existed, it would have granted the same freedom of choice. Aristotle's Rhetoric seems to allude to this very freedom (3.8.1):

The idea here seems to be that Athenian children, overhearing proclamations of manumission, were accustomed to utter the name ‘Cleon’ as a joke, apparently ridiculing the fifth-century Athenian politician known for his populism; just as the children await the herald's calls, Aristotle says, the audience of an oration in verse fixates on when certain metrical structures will recur (cf. Chiron Reference Chiron2008, 27 n. 1; Burkett Reference Burkett2011, 188–90; Dimopoulou-Piliouni Reference Dimopoulou-Piliouni2008, 33 n. 25). Thus Aristotle reveals in passing that freed slaves in his day were regularly offered the right to choose their own patron, exactly as one would expect following manumission through apostasiou.Footnote ³⁸ The passage's use of the term epitropos, though surprising, tends to be understood as a reference to the prostatēs (Youni Reference Youni and Gonzales2008, 167; Mactoux Reference Mactoux and Gonzales2008, 441; implied by Harrison Reference Harrison1968, 183 n. 2, 185), and the two words are intuitive enough as synonyms. The choice of patron Aristotle describes surely cannot refer to ordinary apostasiou cases, which could hardly be common enough to generate a well-known childhood ritual of this nature. Proclamations of manumission, however, were so common in Athens that they were banned from the theatre at some point before 337/336.Footnote ³⁹

Chronologically, this all fits well with the idea that manumission in court emerged in the mid-to-late 350s. Apart from some later revisions, Rhetoric 3.2–12 appears to date to Aristotle's first stay in Athens, before his departure following the death of Plato in 348/347 (Kennedy Reference Kennedy and Rorty1996, 418–21; Fortenbaugh Reference Fortenbaugh2006, 396, 411 [3.8 written ‘toward 347 B.C.’]). Perhaps the children who regularly disrupted manumission proclamations were present at the courtroom; their presence would be unsurprising among the well-known bystanders (περιεστηκότες) of Athenian legal proceedings, a notably heterogeneous group (Lanni Reference Lanni1997, 186–7). But the reference to heralds – and the apparent pervasiveness of their cries – makes one think of the frequent theatre proclamations. The attendance of children at dramatic productions is firmly attested (Henderson Reference Henderson1991, 135, 140 n. 36). Thus, by the time Aristotle became aware of this childhood ritual and incorporated it into the Rhetoric, the practice of manumission in court was in full swing, and the corresponding theatre proclamations were becoming an annoyance – due to their sheer number, and to the raucous replies of children, in unison, mocking an infamous demagogue of years past.

Taking these disparate pieces of evidence into account, we can, I believe, construct a satisfying chronology of the Athenian phenomenon of manumission in court.