Slaves

The number of slaves in Athens has attracted more attention and controversy than that of the metics. While this has not obviously brought definitive answers any closer, at least the terms of the debate have become clearer. It will be easier, therefore, to start with this part of the non-citizen population. The direct literary evidence is scanty and can be summarised very briefly.

There are only two total figures surviving from antiquity, both of which we have already met in the discussion of citizen numbers in Chapter 3. First, there is the passage from Ctesicles in Athenaeus 272c, where it is claimed that in the late fourth century there were 21,000 Athenians, 10,000 metics, and 400,000 oiketai. Whatever one makes of the figures for citizens and metics, it is now generally accepted that the figure of 400,000 for slaves is simply too high to be credible as an accurate figure.Footnote ⁶ The arguments go back to Hume. The oiketai figure could be salvaged by translating it as ‘household members’. That in itself is not an obviously implausible suggestion. In the first place, however, that does not make it any more informative about the number of slaves. In the second place, the fact that it is plausible need not make us believe that Demetrius’ census-takers could actually (even if they had wanted to) have reached an accurate figure for the total population of Attica at the time. Even Van Wees, who has recently argued that the number deserves to be taken seriously, concedes that it is ‘uncertain’ whether the number can be taken as accurate, and that it seems ‘unfeasibly high’.Footnote ⁷

The figure provided by Hyperides fr. 29, is barely, if at all, more believable. After the crushing defeat of Chaeronea, the orator proposed conscripting ‘the 150,000 or more slaves from the silver mines and from the rest of the countryside’. At least it is tolerably clear here that only adult males are being counted, but the total figure is still very high – indeed it implies a total figure (including female slaves and males who were not suitable for military service) of the same magnitude as Ctesicles’.

What these passages tell us, if they accurately reflect fourth-century Athenian views at all (not necessarily a foregone conclusion), is that there was a belief, at least in the later part of the century, that the free residents of Attica were hugely outnumbered by the unfree. This is an important observation in itself, but it is of limited help to us here.

The only figure from the fifth century is provided by Thucydides at 7.27.5, where he claims that more than 20,000 slaves fled from Attica during the Decelean War. Perhaps just because it is in Thucydides, this passage has sometimes been taken very seriously.Footnote ⁸ Although I have argued that Thucydides’ numbers at 2.13 and 3.87 should be treated with respect, it is much more difficult to be as confident about this figure. In the first place, by this stage in the war, Thucydides had been in exile for some years, and can no longer be considered a first-hand observer of what was going on in Attica. In the second, even if he had been on the spot, it is hard to see how he, or anyone else, could possibly have known how many slaves ran away or were captured by Athens’ enemies during the course of the war. The vagueness that comes over him when he talks about the numbers of poorer citizens is hardly likely to have been lifted when it came to non-citizens. It seems unlikely too that the garrison at Decelea kept detailed records of the numbers of slaves that passed through their hands one way or another – or even that all the slaves lost to the Athenians actually ended up in the fort at all. The 20,000 figure should be treated in a similar manner to the figures of 20,000 or 30,000 given in literary sources for the size of the citizen population – as a conventional ‘big number’ that means nothing more precise than ‘lots’. Having said which, it may be of some relevance that this is the number that Thucydides – who did after all have some idea of the size of the citizen population, and was in a position to judge the relative proportion of the slave and free populations – actually chose to represent ‘lots’. There is also some interest in asking who these slaves were, or rather how they were employed, and this is a question to which I return below.

Given how few data there are, it is hardly surprising, then, that there is wide variation in the estimates for slave numbers provided by modern historians. The most minimal view was famously articulated by Jones, who argued for a total number of slaves in Attica of around 20,000 to 30,000.Footnote ⁹ At the other end of the scale, Gomme suggested a total number of slaves in 431 of 115,000.Footnote ¹⁰ Most other estimates have fallen between Jones’ and Gomme’s positions. Sargent concluded that there were perhaps 70,000 to 100,000 slaves in the period when ‘the slave population reached its greatest expansion’, that is ‘in the fifth century, after the Persian Wars, and before the disasters of the Peloponnesian War’.Footnote ¹¹ Rhodes has suggested a round figure of 100,000.Footnote ¹² At the lower end of the scale, Hanson followed Jones and suggested 20,000 to 30,000.Footnote ¹³ Scheidel, in an article concentrating on the importance of women’s labour in Greek households, and so rhetorically disposed to play down the importance of slaves, suggested that 30,000 to 50,000 should be considered a maximum. Jameson again responded that a figure of 50,000 is more likely to have been a minimum.Footnote ¹⁴ Hansen suggested a moderate-to-high range of 66,000 to 93,000.Footnote ¹⁵ Wood, while concerned to play down as far as possible the importance of slaves in agriculture, is less concerned about global figures. In fact, she argues that even if Gomme’s high estimate were allowed to stand, the number of agricultural slaves would still be very small.Footnote ¹⁶ In his general account of Greek slavery, Fisher avoids committing himself to any numbers in his own voice (reasonably enough in what is essentially a textbook), but implies support for a moderate maximalism. More recently, Bissa was happy to follow Hansen, and Moreno to follow Sargent.Footnote ¹⁷

It is worth noting that in his ‘conservative’ estimate of the population in 431, Moreno adopts Sargent’s higher maximum number of slaves of 97,000 (higher of two: the other one is derived from Beloch’s estimate of the free population, which entailed smaller figures for slaves engaged in household service). This in turn was based on Meyer’s estimate of the size of the free population.Footnote ¹⁸ This figure included 700 to 1,000 ‘slaves owned by the state’, 29,000 to 30,000 ‘employed in household service’, 10,000 to 12,000 ‘employed in agriculture’; 15,000 to 20,000 ‘employed in mining’; 28,000 to 30,000 ‘employed in other industry’, and 9,000 to 10,000 ‘children under nine years of age’. Although Moreno’s comment that Sargent’s ‘methodology of proportionality and of differentiation between individual sectors of ownership (public, household, agriculture, mining and other industry) is … sound’ is fair enough, it is less obvious that it ‘yields conservative results, useful in achieving minimum estimates of population and consumption’.

Moreno has committed himself earlier in his account to the traditional picture of Greek (and by extension Attic) agriculture defended by Isager and Skydsgaard; presumably this makes him sympathetic to the relatively small fraction of Sargent’s total represented by ‘slaves employed in agriculture’. Sargent’s account remains valuable; it is thorough, sober, and sensible. Moreno’s evident admiration for Meyer is not at all unjustified. Nonetheless, the Sargent–Meyer model is reflective of the times in which it was produced, and of the modernising orthodoxy that prevailed (and which Meyer himself had, of course, done so much to establish). That there were 30,000 slaves in domestic service, and another 30,000 in ‘other industry’ would not now necessarily appear to be ‘conservative’ estimates. In the case of ‘other industry’, one might wonder what these industries were, and whether they were as fully dominated by slaves to the extent implied by such a figure. The most impressive remaining products of imperial Athens, its painted pottery and its monumental buildings, were each probably the work of a few hundreds of men at a time, and the majority were not slaves. As for the ‘domestics’, this figure would be most easily explained by fairly deep penetration through the citizen group of slave-ownership – and one would expect many of them to have been engaged at least some of the time in some kind of agricultural labour, in what would have been quite ordinary farmers’ households. It is also possible that views about the indispensability of domestic servants were rather different in the early twentieth century from those after the Second World War. The figure for mining slaves here is not especially low, although it is not at the top of the range of modern estimates. Peter Acton has argued for quite large numbers of slaves involved in manufacturing, but is reluctant (understandably) to suggest definitive total numbers.Footnote ¹⁹ The ‘representative’ figures he deploys in his appendix are derived from existing scholarship, including Sargent.Footnote ²⁰

In the absence of good direct evidence, all of these arguments depend on the same two basic approaches. In the first place, there are inferences about the extent of slave-ownership in the citizen population, based on what the literary sources seem to think is normal. In the second there are attempts to look at the tasks in which slaves were certainly employed in classical Athens, and to try to work out how many slaves it would have taken to carry out those tasks.Footnote ²¹ While some progress has been made with variations on the second approach since Sargent, it is important to note that while she ended up with quite high (maximum) estimates, Jones developed his extreme minimalist position by arguing that there were really very few places where we can be sure that significant numbers of slaves were employed.

The overriding impression given by the literary evidence, however, is that slave-ownership was normal and widespread, if not universal, among the whole citizen body, and so the natural inference is that the slave population was indeed large. Beyond the very large totals given by the Hyperides and Ctesicles passages, the usual assumption in oratory and comedy seems to be that most citizens could routinely be expected to own at least one slave. Most clearly, there is Lysias 5.5, delivered in the aftermath of the restoration of democracy:

Todd’s comment in his commentary is worth noting: ‘it is extremely dangerous to draw demographic conclusions from a remark like this, not least because Lysias is seeking to magnify the argument from social consequences by generalising the threat’.Footnote ²² We might compare Lysias 24.6: ‘My trade, at which I can now only work with difficulty, can only help a little, and I cannot yet acquire someone to take it over from me.’ The implication here is that even the most destitute of citizens expected and aimed to be able to buy slaves to assist them in their work. The problem with this particular (and rather inept) speech is that it may well not have been delivered in an actual case, and also may not accurately reflect the views of ordinary Athenian citizens.Footnote ²³ On the other hand, it strengthens the impression that at least some people in fourth-century Athens thought that slave-ownership was very widespread.

Likewise, Aristophanes gives us a consistent picture of men who are apparently to be taken as ‘ordinary’ citizens – men like Dikaiopolis in Acharnians and Chremylus in Wealth – who possess not one but several slaves.

In detail, this kind of evidence is open to attack, and Jones exploited the openings with some vehemence:

While admitting that ‘slaves were employed in many capacities – as domestic servants, as clerks and agents in commerce and banking, in agriculture, and in industry and mining’ and that ‘all well-to-do Athenian families had several servants, and no doubt wealthy men kept large households of a dozen or more’ – Jones also contended that ‘the domestic servant probably did not go very far down the social scale’. The point about the well-to-do authors of comedy is potentially a telling one, but Jones seems to have overstated the strength of his case: such a consistent picture cannot just be explained away in the terms he uses. Still, reference to the direct literary evidence alone is clearly going to remain indecisive. Most estimates, and certainly the more generous ones, rely on the identification of the roles that slaves played in classical Athens. The principal areas of slave employment that need discussion are public slaves, agriculture, warfare, mining and other industry, and ‘domestic’ slaves.Footnote ²⁵

Two of these categories are relatively unproblematic. The polis of Athens clearly owned a certain number of slaves who performed a variety of functions. While we surely only hear about some of them in the sources, there are unlikely to be huge numbers of which we neither know anything nor can reasonably infer the existence. Estimates here are all in the hundreds (the largest single body may have been the Scythian archers of the fifth century), with about 1,000 being a plausible round estimate for the total. In the mines, estimates for the numbers of men employed at periods of peak production vary from 11,000 to 30,000. The lower estimate is that of Conophagos, whose detailed and knowledgeable account has strong claims to be taken seriously. His argument centres on the physical remains of the Laurion district, but it is also entirely compatible with such written evidence as we have. His total compares interestingly with what Xenophon recommends in the Poroi at 4.23–4; although what Xenophon goes on to say implies that he believed that far larger holdings existed in the fifth century. Still, the figures that Xenophon provides at 4.14–15 for the (presumably unusually large) holdings of Nicias, Hipponicus, and Philemonides, if we can have any confidence in them at all, seem to be at least consistent with a total labour force that only just gets into five figures at times of particularly intensive exploitation.Footnote ²⁶

The other categories are much harder to get to grips with, not least because they are less clearly defined. Public slaves and mining slaves only had one job. A slave owned by a private household might perform a range of tasks according to the needs of that household and the time of year. In a household with a small farm, a single slave might help in the fields at times of peak demand for labour, but at others might perform general domestic duties and/or be employed in petty manufacture or retail to supplement the household’s cash income; a male slave might attend his master when the latter went to war, whether as hoplite or as rower. In wealthier households, there would undoubtedly have been at least some greater specification of slave tasks (such as the overseers referred to by Xenophon in the Oikonomikos, and perhaps also nurses for childcare tasks) and, defining ‘household’ more widely, the workshops owned by the wealthy contained slaves who were engaged permanently in the relevant artisanal activities.Footnote ²⁷ We can, however, at least make a start at quantification. As with the citizens, even if there is no possibility of reaching absolutely firm numbers, it is worth establishing the limits of the possible (or at least the plausible), and the exercise itself can generate some interest by making us confront our assumptions.

With ‘other industry’ (possibly something of a misnomer, since the silver mines and the dockyards were perhaps the only sectors of the Athenian economy that can meaningfully be said to have been industrialised), there is a limit to what we can say at present. The most familiar and enduring physical remains of classical Athens – its monumental architecture and its painted pottery – were certainly in part the products of slave labour. But both probably only required the work of at most a few hundreds of men at a time, the majority of whom were free men in any case.Footnote ²⁸ Clearly, there were some quite large manufacturing businesses, each employing dozens of slaves, but it is very difficult to know how many there were beyond the ones we hear about.Footnote ²⁹ The workshops belonging to Demosthenes’ father and to Polemarchus are not referred to as if they were utterly exceptional, but they also clearly belonged to some of the very wealthiest men in Athens, and ones who may in fact have been unusual in their focus on manufacture rather than agriculture. How many more workshops were there on this kind of scale? And how many more that were smaller, like the perfumery we hear about in Hyperides 3? To supply the needs of a city the size of Athens, not to mention the other sizeable towns of Attica, there must have been very many people making all kinds of goods.Footnote ³⁰ But how many of them, and how many were slaves, is virtually impossible to establish with more precision than ‘thousands’.

With the central concerns of polis existence – warfare and agriculture – there is, however, some more scope for assessing the scale of the contribution of slaves.

Metics

The metics – free, but non-citizen residents of Attica – constitute the third and last significant portion of the population that we need to discuss. They present some of the same sorts of problems for us as do the slaves. There is very little in the way of direct evidence for their numbers. Such evidence as we do have suggests that they were, throughout the classical period, present in substantial numbers, but going beyond this observation is much more difficult. Although their numbers were of some interest in antiquity, just because of their ubiquity – and what seems to have been their general reliability – the Athenians seem for much of the time just to have taken them for granted. From our perspective, it is difficult, if not impossible, to get to grips with the question of their numbers separately from that of the number of citizens. To make matters worse, the relationship between the number of metics and the number of citizens is potentially more complex than the relationship between the number of slaves and the number of citizens. It is also important to raise the question of what the metics actually did in classical Athens, just as Jones did with the slaves – even if that question is not a straightforward one to answer at all definitively.Footnote ⁸⁴

In other ways, however, the metics seem to be a rather less troublesome category of people for us to understand. The definition of a metic in Athens was, in at least some senses, pretty clear-cut – a metic was someone who was subject to a particular tax, the metoikion. The people who had to pay this tax were non-Athenians who were resident in Attica for some period that was extended beyond a certain minimum – usually, and plausibly, taken to be 30 days. The situation in classical Athens should in fact be especially clear, and not just for the familiar reason that we are much better informed about metics in Athens than we are about their counterparts in other Greek poleis. Athens also presents a clear case because under the democracy (or at the very latest from 451/0 BC onwards), where there was no property qualification for what amounted to full citizenship, we can be sure that almost everyone who was not either a citizen (or a citizen’s direct kin) on the one hand, or a slave on the other, was going to be a metic. The only exceptions would be visitors who were merely passing through.Footnote ⁸⁵ This is obviously not the case in those other cities where there were other non-citizen but free groups in the population. The clearest example is provided by Sparta with its perioikoi, but the overwhelming majority of other poleis would have fallen into this category too.

Furthermore, we know, as surely as we know anything about classical Athens, that metics were, and were sometimes explicitly recognised as being, essential to the continued well-being and even existence of the city. This importance is clearest in the operation and maintenance of the fleet (as Pseudo-Xenophon makes explicit, but which we could reasonably infer in any case on consideration of the logistical realities), but we also find metics playing important roles in land warfare and involved in almost every sector of Athenian economic and intellectual life.

Finally, and not least importantly, there is the fact that we are very well informed about some individual metics. These are figures as familiar to us as any Athenian citizen, and have done as much as any of them to shape our picture of classical Athens: Aristotle can be and is cited for almost every subject in Greek history, and the speeches of Lysias – quite apart from providing many students’ first extended encounters with genuine Attic prose – are central to our understanding of early fourth-century Athenian society.

This clarity and familiarity, however, comes with a cost, which is that we are often blinded to or complacent about the metics and their roles. Full-length treatments of the Athenian metics have tended to concentrate either on the legal status of metics and their relationships with the formal institutions of the Athenian polis or with what Whitehead famously discussed as the ‘ideology’ of the metics – which is, of course, the ideology constructed by the Athenians about the metics, not one constructed by the metics themselves. While these are, undeniably, important issues, further points of interest emerge from consideration of what may seem to be the more basic question of how many metics there actually were in Athens to make them so important. This question in turn should inspire us to wonder from where these non-Athenians were coming to live in Athens. Exactly as with the slaves, we should also ask what they were doing: unlike the slaves, however, they had at least some choice about where they did it, and so there is also a question about what it was that made them stay.

We can start with the ‘how many’ question. Again, the best evidence we have, just as with the slave population, gives us only an impression of the size of the metic population. Having said that, the indications are that it was quite large, and a significant fraction of the size of the citizen population. The only explicit numbers we have are those given by Thucydides (in the passages in book two previously discussed, which deal with the metic hoplites alone), and Ctesicles (the figures reported, including one for the metics, for Demetrius of Phaleron’s census).

An attempt to get to grips with the numbers question is complicated by the reluctance of some scholars to accept that there really might have been very large numbers of them, both in absolute terms and relative to the size of the citizen population. This reluctance is well illustrated by Whitehead, who in his fundamental and influential treatment of the Athenian metics, relegates discussion of their numbers to little more than a page.Footnote ⁸⁶

Whitehead was willing to accept the figure Ctesicles gave for Demetrius of Phaleron’s census – that is, 10,000 metics compared to 21,000 citizens (and of course the 400,000 slaves) – but seemed less prepared to accept the implications of Thucydides 2.13, although his reasons are not clear. Whitehead cites Clerc’s arguments simply to show (what he considers to be) their obvious weakness:

It is not unreasonable, I think, to focus on Whitehead’s own reasoning exempli gratia here, given the influence of his account. It is implicit, of course, that what has gone wrong is in Clerc’s argument, rather than in the opinions of those who are not his adherents. Clerc’s numbers, while speculative, are however perhaps not quite as obviously silly as Whitehead suggests. The 11,750 figure, as we shall see, seems to be an entirely reasonable inference from Thucydides – although not the only possible one. The suggestion that metics should be half the number of citizens does, of course, contradict the results of Clerc’s calculations, but to be fair to him it is what is implied by Demetrius’ census, which suggests at least that this is what was believed in Athens at the end of the fourth century.Footnote ⁸⁸

Gomme suggested a figure of just under 30,000 in 431, and this is the ‘consensus’ figure alluded to by Whitehead.Footnote ⁸⁹ Rhodes, in the CAH suggested about 50,000;Footnote ⁹⁰ Implicit in Duncan-Jones’ approach is a figure nearer 100,000, which actually reaches a similar conclusion about Thucydides to Clerc’s.Footnote ⁹¹

Clearly we do need to look at Thucydides 2.13 again, but in the light of the discussion in Chapter 3. If, in our interpretation of the figure given there for the main hoplite army of 13,000, we make assumptions that minimise the number of citizens, then that does seem to imply a figure of 10,000 to 12,000 metic hoplites. Such minimising assumptions – that the 13,000 figure is actually a population figure for all the men with hoplite equipment between the ages of 20 and 50 – would mean that relatively few of the 16,000 figure for hoplites on guard duty could have been citizens, and so the number of metic hoplites that is implied is really very large.

If, however, we make a different (and perhaps more plausible) set of assumptions about the 13,000 figure – that it is meant to be a realistic army figure, and it represents a narrower range of age groups – then there might only have been 3,000 to 6,000 metic hoplites. We could further assume that, as seems likely, Hansen was right to argue that there must have been a total of at least 60,000 citizens in 431. In that case, if we minimise the number of citizen hoplites, then we seem to be required to accept that Athens was overflowing with citizen manpower in groups other than the hoplites. But if we were to make these other, different assumptions, then the hoplites would have made up a much larger proportion of the citizen population. But the assumptions which are less minimising about citizen hoplites should in turn imply that there was a smaller supply of citizens to row in the fleet.

Now, there is a consistent assumption in the contemporary sources that the metics were crucial in some way for the fleet. Their importance is made explicit by the Old Oligarch, who claims (at 1.12) that Athens ‘needs the metics because of the multitude of their skills and because of the fleet’. Unfortunately the reason for this importance is left frustratingly vague. If we assume that more citizens were fighting on land, there was (at least potentially) greater scope for that metic contribution to be a quantitative one, with the metics filling many places on the rowing benches of manpower-intensive triremes. But if we return for a moment to the minimising assumptions about citizen hoplite numbers, the metic contribution to rowing manpower might not have been so great; in this case, the crucial contribution could have been qualitative, with the metics providing not rowing muscle but more specialised technical skills. Which of these situations is more plausible – whether the contribution of the metics was important quantitatively or qualitatively – will therefore depend partly on what we end up concluding about the citizen population (including the distribution of wealth within it, and how much of a socio-economic elite the hoplites constituted), but also on what kind of people we think the metics were. Putting this issue this way does perhaps overemphasise the distinction, since it is entirely possible that the metics were important in both respects. It is, however, important to emphasise that there is a distinction. Were all the metics like Polemarchus, at ease among the wealthiest Athenians, or were they mainly poorer men? We shall turn to this question in a moment, but it is also worth returning briefly to the question of what the ratio of citizens to metics was.

As we have seen, the results of Demetrius of Phaleron’s census imply that there were roughly two citizens to every metic (at a time when there was a property qualification for citizenship, as there had not been earlier in the classical period). But even if we want to trust these figures, the situation may not have been quite so straightforward as that suggests. For example, Hansen takes the 21,000 citizens to be essentially an army figure. That is not unlikely in itself, and it is possible that the figure for metics was an army figure too. However, it is easier to imagine that the number of metics was simply derived from the amount of the metoikion tax that was collected in a year. In fact, the metics, because of this tax, were probably the one part of the population whose numbers could at least roughly be quantified in antiquity – even if the figure was probably too low, as presumably there was a significant degree of evasion of the tax; however, in terms of assessing the ratio of citizen to metic, this probably does not matter all that much, as military service was clearly evaded too.Footnote ⁹² In that case, the number of metics would actually represent a population figure, albeit a rather odd one, as it would include not just all the adult males but single adult females too. On that basis, Hansen ended up suggesting about 40,000 non-citizen long-term residents in total, at a time when the number of people in citizen families would have been in the range of roughly 100,000 to 120,000, giving a ratio between two-and-a-half and three to one.Footnote ⁹³

It is worth considering briefly if there was a similar pattern in the 430s. If there was, a figure of 60,000 citizens might be taken to imply a population of 20,000 to 30,000 (adult) metics. On their own, the army figures provided by Thucydides, while they do not rule out such a figure, would also be consistent with a population of 10,000 to 15,000 metics. That is a similar absolute number to that of Demetrius’ census, but one smaller relative to the size of the citizen population, with a ratio of citizens to metics of between four and six to one.Footnote ⁹⁴ Again, deciding whether either the absolute numbers or the relative numbers of metics changed between 431 and 312 requires us to think about what kind of people the metics were, and what they were doing in Athens.

There is still a widespread, and not inherently unreasonable, assumption that the bulk of the metics were people like Lysias and Polemarchus, even if most of them were not nearly as successful or as wealthy. Because metics could not own land, then surely, this line of reasoning goes, they must have been making their living by other means than agriculture. Therefore, they should mainly have been working in trade and manufacture. This is fair enough as far as it goes, but this is not as far as it might appear. Being unable to own land does not prevent one from farming. Farmland could be (and at least sometimes was) rented by metics, and it was also possible to work for people who needed additional labour, at least at times of peak labour demand like harvest.Footnote ⁹⁵ It is also worth remembering that some of the views that remain common about metic employment were formed first in a period when it was more widely believed that citizens would have despised any employment in trade and manufacture, and that therefore these fields would have been left more or less completely to non-citizens. While this picture may not be wholly misleading, it is nonetheless clear that citizens were engaged in the full range of non-agricultural economic activities, and at all levels.

This last point is accepted now almost as a truism, but it bears repeating explicitly – because the usual explanation for the presence of large numbers of metics in classical Athens is that they were drawn there by the opportunities provided by a large urban centre and major port. There must be some truth in this, and it is part of the way to explain the presence of men like Lysias and Aristotle. On the other hand, it might reasonably be asked how many opportunities there really were for non-Athenians of lesser means and/or talents, given that they would have had to compete, on generally unfavourable terms, with the citizens, who were themselves present in numbers far too large to be supported by agriculture alone. The large number of metics was not inevitable; urban centres do not attract economic migrants, or grow in size generally, just because they are urban. Something has positively to attract voluntary migrants: rural poverty and landlessness can act as ‘push’ factors (where they are present), but for a given urban centre the ‘pull’ factors have also to be explained. If the migration is economically motivated, then there has to be something for those migrants to do; otherwise, they tend to migrate elsewhere (or just go back where they came from, when the city proves not to be what they had hoped for).Footnote ⁹⁶ James Watson has suggested that Athens’ political stability was a factor in encouraging immigration.Footnote ⁹⁷ There were extensive periods in the fifth and fourth centuries when Athens genuinely appeared politically stable, and this may well be relevant: it is easy to imagine that this played into Cephalus’ calculations, for example. On the other hand, Athens was not always predictably stable, and may have looked a bit different to those without the benefit of hindsight. If this was a factor, then the way in which the Athenians were able to take the loyalty of the metics for granted becomes even more surprising (more on this below). If stability was what they were after, then one would expect them to have left when that stability was threatened, as it was on many occasions.

One source of those opportunities was provided by the fleet. Even if Athens’ warships were not necessarily all built in the Piraeus itself (since it may have been more practical to build them in the areas where both timber and the skills to work it were abundant), there would have been a market for specialist labour in the shipyards and military harbours for maintenance, as well as a need for trireme crews. Bissa suggests that the difficulties of transporting ship-building timber would have led to most of Athens’ warships being built outside Attica. These difficulties were real, but they may not have been insurmountable: although the convenience of building near sources of supply was recognised, local resources could often be overwhelmed by demand, and so later wooden navies were often obliged to look far afield for their raw materials.Footnote ⁹⁸ We might expect non-citizens to play an important role here anyway, but the Old Oligarch’s comment compels us to consider it. Neither demand is easy to quantify (as we have seen in the case of ship crews); comparative data (from later and better documented times and places where large wooden fleets were operated) might be some help with assessing at least the likely scale of the demand.

One obvious source of a comparison is another city-state with a maritime empire dependent on a substantial galley fleet, and about which we are reasonably well informed: the republic of Venice, especially in the fifteenth to seventeenth centuries. In that period, the republic’s naval strength was concentrated in a fleet of galleys not so very different from those of classical Athens.

The basis of the fleet, as the name implies, was the galia sotil or ordinary galley, the very distant descendant of the classical warship via the dromones of the Byzantine fleet: early fifteenth-century examples tended to be about 38 metres long, with a beam of just over 5 metres (very similar in fact to the reconstructed Olympias, which is slightly shorter and beamier). Later galleys were built rather longer (41 m) but with the same beam: the addition of heavy cannon armament at the prow with a counterweight at the stern contributed to a significant increase in displacement but also to the increase in length. Before the battle of Lepanto in 1571, the ships were overwhelmingly rowed at three levels, like Athenian triremes. A shift to the alternative system, where three men pulled a single oar, took place in the Venetian navy only in the decade after the battle, when crews were increasingly dominated by convicts chained to the oars. There were 25 to 30 rowing benches each side, with three men to a bench (allowing for 150 to 180 oarsmen). The important difference was in armament: Venetian galleys (and their contemporaries in other navies) did not carry the heavy bronze rams at the water line of their ancient counterparts. Their projecting prow spurs were metal-tipped, but were meant to function as boarding bridges rather than weapons in their own right.Footnote ⁹⁹

The home base of the fleet of these vessels, Venice’s Arsenale, was a centre for both production and maintenance. In the 1560s, when in the face of imminent threat the Arsenale was at its busiest, the number of men employed there seems to have fluctuated between a minimum of around 1,000 and an ‘emergency’ peak of around 3,000, with 2,000 being fairly typical. At this time the republic aimed to maintain a reserve of 100 seaworthy galleys, while it actually deployed fleets in the order of 40 to 60 ships.Footnote ¹⁰⁰

These are smaller resources than those maintained by Athens at the height of its power, of course, when hundreds of galleys were not only available but might actually be sent out in very large numbers. In the fourth century, however, the number of Athenian galleys manned and at sea was also likely to be around 40 to 60, even though there were still hundreds of hulls on hand.Footnote ¹⁰¹ Presumably Athens would not have needed significantly fewer shipyard workers than Venice to maintain a fleet of similar size: probably a figure of about 2,000 men in the fourth century would not be too far off. The figure in the fifth century would certainly have been higher, although it is difficult to tell how much; as with later dockyards, the number of men employed would have varied according to need.

There are two possibly relevant factors to bear in mind here. First, Venetian galleys seem not to have had such long service lives as their Athenian predecessors, only lasting about 10 years as opposed to the 20 of the latter. They were however considered at the time to be of good quality (although this may not have been a straightforward reflection of reality). Second, the hull-first method of constructing ancient galleys would have made them harder to maintain than later vessels that were built frame first. Very many of the men employed in the Arsenale were caulkers, whose work was necessary to keep vessels seaworthy throughout their service lives. Older vessels (and those captured from the enemy) would have required a great deal of work to keep (or make) them seaworthy. Because of the labour demand generated by maintenance (which could have taken place within the Piraeus’ ship sheds), the question of where new ships were built may not be so important for our current purposes, where it is the overall size of the establishment we are interested in, not the specification of the tasks undertaken there.

At times when Athens regularly sent a hundred or even more ships to sea, perhaps a figure of 4,000 or 5,000 men would not be excessively high. This kind of figure would have made the naval shipyards of the Piraeus a very large centre of industry indeed – about which our relative ignorance is therefore all the more frustrating. To provide a little more context, in the seventeenth century, the English Royal Dockyard at Chatham never employed more than 1,000 men, even at moments of extreme crisis, and usually fewer.Footnote ¹⁰² In their eighteenth-century heydays, the Chatham, Portsmouth, and Plymouth dockyards each employed about 2,000 men, of whom about a third were engaged in clerical activities.Footnote ¹⁰³ In this context, it is also well worth noting that in the earlier period a great deal of the work in English dockyards was seasonal. Much the same may have been true at the Piraeus, with the migration it induced being ‘circular’ in many cases.

Chatham was, of course, in many ways a quite different kind of establishment from the Piraeus; uniquely for an English royal dockyard, however, it acquired some specialised facilities for galley operations, including a dry dock built in 1571.Footnote ¹⁰⁴ The later history of the dockyard also illustrates what should be obvious (but in fact is not always to ancient historians): that a naval dockyard may not always also be a naval base.Footnote ¹⁰⁵ Chatham’s increasing unsuitability as one of the latter led to its replacement in that role over the course of the eighteenth century by Portsmouth and Plymouth. Although these ports on the south coast also had ship-building facilities, this was not their main focus, which instead was on the maintenance of operational fleets. However, this in turn meant that they had spare building capacity, which was taken up in emergencies when new ships were needed in a hurry. It may not be too far-fetched to imagine that the Piraeus worked in similar fashion, so that although some ships were built there, most new Athenian triremes need not have been – except in exceptional circumstances, such as the construction of the fleet that fought at Arginusae.

Not everyone in the dockyard could have been a metic. We should expect that both citizens and slaves worked there too: even in the fifth century, we can hardly believe that more than 2,000 metics could have found regular employment in the naval yards. Beyond this, though, how much scope really was there for men of limited means to make a living in classical Athens? Clearly there were not 10,000 Lysiases, each living off the proceeds of substantial armaments workshops. The city provided a large potential market of consumers, but presumably it was possible to make money selling things to the Athenians without having to live with them, let alone having to fight for them. And what would have forced the metics to stay in times of war and hardship, as during the Peloponnesian War? Armaments manufacturers and food merchants would have had fairly strong financial incentives (at least while there was money to pay for their goods), but again, how many metics can these activities explain, given that citizens were clearly involved here too? Have we been too influenced by later examples of the pull exerted by cities over migrant workers? Athens was not in every way like imperial Rome; nor was it necessarily like cities in the modern industrial and industrialising worlds.Footnote ¹⁰⁶

At this point, it is worth coming back to the fact that at least some metics were not voluntary immigrants, but instead were freed slaves. This is generally accepted, but its potential significance has not always been fully appreciated. It is, for example, a surprising feature of Whitehead’s study that although he was determined to debunk what he saw as a prevailing orthodoxy that being a metic at Athens was in some sense a privileged status (an idea which, in spite of the strength and cogency of Whitehead’s arguments, has proved remarkably resilient),Footnote ¹⁰⁷ he was almost as determined to deny that metics as a body were tinged with the negative connotations of servility. Whitehead rightly pointed out that having the status of ‘metic’ was not inconsistent with retaining one’s status as a citizen of another polis (although he also rightly observes that after too prolonged a stay in Attica, and after the first generation, it may have become harder actually to get one’s entitlement recognised at home). There is no reason to doubt that there must have been many metics who fell into this category. But that still does not affect the extent to which the term ‘metic’ was in fact associated with servility, and so in my view Whitehead underplayed the number of metics who were freedmen and women.

Whitehead must be right when he says:

I would, however, question whether the Old Oligarch can really be cited as an ‘ordinary observer’. The evidence of his problematic text can cut both ways on this issue. Slaves, metics, and freedmen are lumped together (and apparently equally reviled) by this author as much as they are distinguished. The text does not seem necessarily to support Whitehead’s claim that ‘the inclusion does not appear to have been felt to characterise or contaminate the whole class’.Footnote ¹⁰⁹

The crucial point, I think, is one that is not quite explicitly addressed by Whitehead. He notes that the metoikion ‘was … unique in Athens in being not merely a direct tax but a poll-tax, levied on the person rather than his (or her) property of activities’,Footnote ¹¹⁰ but he does not address the obvious implication that this in itself made metics look closer to slaves than to citizens. The tax was unique precisely because it was unthinkable to impose such a levy on citizens. It is also highly suggestive that (as Whitehead does note) the punishment for failing to pay the metoikion was enslavement. Whitehead goes on to consider the extent to which the tax was a financial burden. How much of a burden it was must have depended on individual circumstances: for the poorest metics, it could not have been trivial.Footnote ¹¹¹ But to consider the tax only in this way is, surely, to miss precisely the ideological point.

That metics were subject to (and to a large extent defined by) the payment of a tax not on their property but on their bodies would on its own have made them vulnerable to identification with slaves. Even more strikingly, however, they were (theoretically) liable to torture in judicial settings. The need for a prostates, whatever their precise role, also smacks of servility. It is further worth noting that the passage (Lysias 31.9) where this is explicitly sneered at concerns an Athenian citizen who was a metic in another city.Footnote ¹¹² The role of the metic women in the Panathenaic procession can hardly have been seen as anything other than servile.Footnote ¹¹³ These surely served to align them, at least in the official view of the Athenian polis, much more closely with slaves than with citizens.

Whitehead was, of course, absolutely right to point out that metics rarely if ever described themselves as metics (and the city in certain contexts avoided the word too), preferring to use patronymics and city ethnics, except when they were able to describe themselves as isoteleis. Whitehead’s response to his own observation is curiously half-hearted, however. On the use of city ethnics, he wonders briefly whether describing oneself as, for example, ‘Samios’ (not a neutral claim in fourth-century Athens, of course) entails a realistic claim to citizen status in one’s home city. His conclusion is essentially ‘why not?’ On the preference for the use of isoteles as an identifier, he notes that it shows how keen metics were to show that their adoptive city was honouring them. It strikes me that there are alternative ways to read this pattern. If ‘metic’ did carry a tinge of servility, then an assertion of citizen status somewhere (irrespective of whether or not it was justified) would be one way to counter it. Being isoteles, however, was a matter of public record in Athens itself, and was a status that could only be achieved by someone who was wealthy enough to have done significant service to the city; it could have lifted a metic unmistakably out of the ranks of those who were subject to suspiciously servile obligations and treatment.

If we take many of the metics to have been freedmen and -women, then many of them, still largely dependent on their former masters, will have had few realistic alternatives to remaining in Athens, even in the midst of plague and military defeat. This might have reduced the numbers who left (although it is also obviously true that many metics who could have left did not – again, Lysias is a familiar but not necessarily typical example). From the perspective of military history, it might also influence how we interpret the significance of Thucydides’ mentions of metic troops when he seems so interested in the Spartans’ use of freed slaves (like Brasidas’ neodamodeis).

Whitehead also observes that there was a well-developed vocabulary of abuse in servile terms for those who wanted to employ it, whereas no one is ever explicitly denigrated by being called a metic. This is true, but it need not be as decisive as Whitehead thinks – if metic status was linked to servility, then it would be at least as effective, and probably more so, to label an opponent straightforwardly as a slave rather than labelling them as a bit like a slave. After all, we know so much about this servile vocabulary of abuse precisely because it is frequently directed at citizens.

Whitehead thinks that where metics are referred to as such, there is in fact a certain amount of sympathy and even affection for them. This, I think, is entirely in the eye of the beholder. The two key passages for him are Thucydides 1.143.1–2:

and Aristophanes Acharnians 507–8:

Thucydides has Pericles acknowledge the necessity of the metics, along with a breezy assumption that they are absolutely reliable. In the Aristophanes passage, which, as Whitehead puts it, characterises the metics as the citizens’ ‘inescapable (if nutritionally inferior) companions’, Whitehead sees sympathy; the patronising element to which he also refers seems rather more apparent. Nothing in these passages, even if we were to accept that there is something favourable, necessarily makes implausible the idea that the metics were considered, as a body, to be somewhat servile. Whitehead takes them as ‘perhaps elevat[ing], if only momentarily, the social status of the freedman qua metic, as perceived by the citizen’. However, some moral qualities and standards of behaviour could perfectly well be expected of, or at least recognised in, slaves.

The clearest message from these texts seems actually to be the inseparability from the citizens. This would make more sense if the metics were substantially composed of freedmen and -women than if they were economic migrants. Freedpersons had no other homes to go to in hard times, and clearly owed continuing obligations to their former masters.

Xenophon’s well-known discussion of the metics in Poroi 2 is consistent with this picture. It is often taken to confirm the view that the majority of metics were economic migrants. This is because Xenophon suggests a variety of possible methods for enticing this kind of immigrant, to the economic benefit of Athens. Austin and Vidal-Naquet suggested in this context that there may really have been a distinction between a fifth-century metic population that was made up mainly of immigrants from other Greek cities (and who might settle permanently, and be more easily assimilated) and a fourth-century one that was more ‘barbarian’ and, partly for that reason, more transitory, and unassimilated even when they chose to remain.Footnote ¹¹⁴ Miller, however, presented reasons for doubting this – for example, the existence of a ‘Little Phrygia’ in Athens in Thucydides 2.22.2, and the introduction of new cults in the fifth century.Footnote ¹¹⁵ However, it is important to note that nowhere does Xenophon actually claim that there were few metics in contemporary (mid-fourth-century) Athens. Rather, he seems to think that the problem is that there are not enough of the right kind of metics. Those that are in Athens (and fighting alongside the Athenians) are ‘Lydians, Phrygians, Syrians, and all kinds of other barbarians’,Footnote ¹¹⁶ of whom there seems to be no shortage. The named peoples are, of course, some of those that provided Athens with many of its slaves. Clearly the wealthier economic migrant-type metics (including, albeit not exclusively, those of Greek origin) could and did leave Athens when they saw no particular advantage in staying, or indeed they could choose to stay at home, or go elsewhere in the first place. It would, surely, be extremely odd if these were the sort of men of whose loyalty the Athenians could be so complacently sure.

Instead, we must see at least two quite distinct types of metics. In the first place, there was a relatively small elite of wealthy men, many of whom were probably citizens of other Greek cities who could take advantage of the opportunities offered by Athens, but whose first loyalty might not always be to Athens. In the second place, there would have been a much larger, but to us less visible, number of freed slaves. A few of the latter (such as former banking slaves) might be wealthy, but the majority would have been a good deal poorer; alienated from their natal communities and still with ties of dependency to their former masters, they would have been short of alternative places to go.

If there is any truth to this suggestion, it has implications for how we think about the structure of the metic population. The usual assumption that metics were economic migrants tends to encourage a further assumption that they would mostly therefore have been young (and mainly single) men – so that, for example, the thousands of metic hoplites in Thucydides have not usually been taken to imply necessarily similar numbers of women or of dependent children.Footnote ¹¹⁷ But if many metics were freedmen, they would have had every incentive to marry, even if their choice of partners was restricted, and to have children.

Moreover, the proportion of women in the metoikia may have been rather higher than is usually assumed. In all slave societies where it is possible to tell, women tend to be manumitted at a higher rate than men. The reasons are as obvious as they are depressing. Women are generally more likely to be employed in domestic household situations which lead to manumission more often than employment in distant farms (or silver mines); in paternalistic societies, the continued dependence of a freedwoman on her master is easier to assure; and prostitution is often the most common way to earn the money to buy one’s freedom.Footnote ¹¹⁸

Freedmen and -women appear to be a good source of some of the thousands of individuals on whose continued presence the Athenians seemed absolutely able to rely in the fifth century. However, we know very little about rates of manumission in the fifth century. It need not in fact have been practised very much at all. Labour was cheap and the economy was expanding rapidly; for most of the period, Attica was militarily secure, and most of the time there was relatively little disunity in the citizen slave-owning group (part of the 460s and the oligarchic episodes towards the end of the century notwithstanding). These are not the circumstances where one would necessarily expect high rates of manumission. On the other hand, a low rate of manumission in a large population of slaves could still produce a sizeable number of freedmen in absolute terms. This might, therefore, incline us towards stability in the absolute size of the metic population, with change over time being seen in the number of metics relative to the number of citizens. At which point, it is time to turn to the issue of population change more generally if we are to make any progress.