Slavery and Slave trade in Antiquity I: background

I checked the conditions of use on JSTOR; there doesn't seem to be anything which precludes my quoting a whole article on a public site. The reason I do it is that its view on the extent of slavery in the civilisations of antiquity provides the background for the proposal of an explanation of certain roots spread over several language families which I bring in my next posting. So please bear with me on this one. I know it's long.







Believing the ancients: quantitative
and qualitative dimensions of slavery
and the slave trade in later
prehistoric Eurasia

Timothy Taylor

Abstract
This paper briefly examines two types of slavery in the first millennium Aegean, Carpatho-Balkan and Pontic regions - branded silver-mine slaves and blinded milk-processing slaves. The first is examined primarily in quantitative terms, to attach economically sensible order-of-magnitude figures to the trade in people. The second is examined qualitatively, to show how indigenous forms of dependence and subordination were caught up in the emergence of Graeco-Roman chattel slavery. In both cases I have taken the rather unusual step of trusting what the classical authors tell us they knew. Finally, a symbolic connection is made between shackles and torcs.

Keywords
Ancient economy; Celts; freedom; Scythians; silver mining; slavery; torcs.

Introduction

That the classical world used people whom we call slaves is beyond dispute, as is the understanding that the Iron Age in non-classical Europe is a period characterized by a complex of developments which are ultimately, and often intimately, linked to the city state developments of the south and east (Wiedemann 1980; Nash 1985; Alfoldy 1988; Cunliffe 1988a; Kristiansen 1998). However, the terms 'core' and 'periphery' underestimate the significance of the Achaemenid Persian model of empire as a political aspiration for all those - Ateus, Alexander, or Ariovistus, Burebista or Caesar - who sought great power in the second half of the first millennium BC. The Scythians, Celts, Germans and Dacians each strove for dominion as hard as Greece, Macedonia and Rome. Only hindsight projects Rome as inevitably core, and Germania and Dacia as naturally peripheral. That the latter ultimately lacked Rome's integrative mechanisms, notably disembedded institutions for property transfer and credit, is of course significant in relation to their failure to be adjudged central by history. But on every side there was both wealth and poverty, freedom and slavery, and the bargains by which these things were created were Faustian, with profound implications for the psychological texture of the societies involved.

Slavery is materially elusive in prehistory. Diodorus's famous statement that Italian merchants could purchase a slave from the Gauls in exchange for wine has provided fuel for inferences concerning the source areas of Gaulish slaves traded through Massalia on the basis of the archaeological distribution of amphorae up the Rhone valley and beyond. In fact, despite persistent mistranslation, Diodorus indicates (V.26), slightly unspecifically, a crock of wine (δίνου κεράμιον) for a young slave boy (διάκονος) - not necessarily a whole amphora then, which might have bought several. And the reality in many places, as at Cetăţenii din Vale in the lower Danube basin, is that there was mass transfer of amphora contents into wineskins in order to facilitate the journey onward, beyond the Carpathians (Taylor 1994: 400). Such a wineskin is depicted, with attendant drunkenness, on a gilded Scythian silver wine cup from Gaimanova Mogila (Taylor 1996).

If the wine flowing one way becomes untraceable as it is drunk, so are the sad streams of captives that may have paid for it. Something of the order of fifty iron shackles and gang-chains are known from the later pre-Roman Iron Age and Roman period from Europe, like those from Llyn Cerrig Bach, Anglesey, and - illustrated here (Plate 1) - Hay Hill, Lord's Bridge, Cambridge (Clark 1821; Cunliffe 1988b; Gollop 1994), but these are exceptional. Iron was re-forgeable, and, where slaves were transported using such chains, they may usually have been struck off and made into other objects. Such slave chains may have been a useful 'self-transporting' form of iron, just as East African ivory tusks were tied to slaves for the convenient transport of both. Chains are referred to in texts, and clearly depicted in iconography where we have no ferrous evidence, as on the Trajanic war memorial at Adamklisi on the lower Danube. Such chains indicate the standard way that very substantial numbers of Dacians (perhaps half a million; see below) were exported after the fall of Sarmizegethusa, but no single example survives.
Plate 1:
http://tech.dir.groups.yahoo.com/group/cybalist/files/Six-slave iron neck ring chain.jpg

In an important paper of 1962 entitled 'The Black Sea and Danubian regions and the slave trade in antiquity', Moses Finley set out, against a background of longstanding scholarly scepticism on the issue, the principal historical arguments for believing that a systematic, large-scale and long-lived slave trade had existed in classical times in eastern Europe. It was not that the existence of slaves in Greece and Rome had previously been doubted, but that their occurrence had been seen, essentially, as epiphenomenal - a sporadic function of piracy, for example - rather than systemic. Slaves on this score were of the butlering or high-class concubine sort. No mass slave-force could be envisaged in agriculture, mining and industrial production, because slaves were few, and obtained more or less informally.

Although the 'informal model' runs in the face of much textual and archaeological evidence it has not been completely superseded: witness the typically fleeting acknowledgement of slaving in Derek Williams's Romans and Barbarians (1998: 12f.). In large part this may be because Finley was essentially pessimistic about quantification ('No very significant or reliable quantitative results can be expected': 1962: 56), and only a few order-of-magnitude studies have been attempted. Among these, Madden (1996) argues that directly captured prisoners of war and, subsequently, foundlings (abandoned infants) contributed significantly to the basic requirements of the classical world, reproduction among slaves themselves being low. Scheidel (1997), on the other hand, rejects the idea of a significant foundling contribution and makes up numbers by assuming a high rate of slave reproduction (vernae - children born by slaves into slavery). Neither seems able to envisage a fully integrated mass slave trade as a principal mechanism (e.g. Scheidel 1997: 164, n. 34). Crawford's interesting 1977 quantifications for Dacia are discussed below. That ancient figures for slavery are rare may, paradoxically, be precisely because the phenomenon was so widespread and highly organized - just as one might vainly scour newspaper after newspaper today, searching for figures relating to total national electricity or water consumption, the ready availability of which is simply assumed. Atheneus preserved a fragment of Ctesicles reporting a census by Demetrius of Phaleron taken some time between 317 and 307 BC (Atheneus 272C). According to this, Attica had 21,000 citizens, 10,000 resident aliens (metics) and 400,000 slaves. This provides us with both absolute and relative information. Although the census was taken at a time of demographic upheaval (Garnsey 1988: 136), and thus cannot be generalized without reflection, it is clear that Atheneus does not expect his audience to find the idea of a society in which just over 7 per cent are unenslaved implausible. Yet most modern commentators find it hard to imagine slavery running at much more than 20 per cent in ancient society (e.g. Scheidel, for the early Roman Empire, goes for 10 per cent: 1997: 158).

The absolute values are rejected too. Fisher (1998) believes that, while there were large chattel slave populations engaged in Greek agriculture, 'the only surviving ancient figures are totally incredible: 400,000 slaves in Attica, 470,000 slaves in Aegina, 460,000 slaves in Corinth, all of which imply impossibly high population densities for those states' (1998: 199f.). Finley accepted 80-100,000 as the total for fifth-fourth century BC Attic slaves. Thus the best numbers we have, census figures that the Greeks themselves provide, are dismissed - 'generally considered worthless' by Garnsey (1988: 136) who (note 3) also directs us to Hyperides, fragment 29 'for another implausible figure'. This is done without careful reflection on what 400,000 might mean when all the slaves in all the Athenian-owned silver-mining concessions are added to those engaged in quarrying stone, powering trade galleys on the seaways, felling and transporting timber, and sowing and reaping on landed estates. (It should not be forgotten that this landscape had a rather higher subsistence carrying capacity than its highly eroded and degraded present-day incarnation might immediately suggest.) It remains the case that carefully weighed general statements, such as Cunliffe's 'There is no reason why the slave trade should not have been well organized before 125 BC' (1988a: 78), lack agreed quantitative support.


Quantitative extrapolations and inferences

Two investigative modes for estimating the scale of eastern European Iron Age slaving and slavery are: extrapolations from known figures and value ratios; and inferences from other information sources that may be construed as more or less proportional indices of scale.

The sorts of figures that could be useful to extrapolate order-of-magnitude figures for slave numbers are slave costs in terms of silver, silver output rates and tribute sizes. For fifth- and fourth-century Thrace we know that slaves were purchased by Greeks; that silver was a monetary metal in Greece and 'proto-monetary' in Thrace (being frequently worked and deposited in units of known value); that tributes in silver were levied by local dynasts; and that slaves worked Greek- and Thracian-owned silver mines in Thrace. The profit margins for a Greek of running a silver mine must connect in some way with the cost of using silver to buy and maintain slaves.

Xenophon, writing around 355 BC when Athenian finances were in disarray, gives advice on how to revitalize silver mining using public-private partnerships (advice that appears to have been successful for the mines at Lavrion: Xenophon, Ways and Means IV: Marchant 1925: xxvi). These are some salient things that he says, with commentary: [W]e all agree that the mines have been worked for many generations ... it is continually being found that, so far from shrinking, the silver yielding area extends further and further ... even at the present day [i.e. during an economic crisis] no owner of slaves employed in the mines reduces the number of his men; on the contrary, every master obtains as many more as he can. (Xenophon IV.2-4)

That there is plenty of silver to mine, in Lavrion, Thasos and elsewhere, is clear. Profits can be made in proportion to the available workforce, who are slaves.

[N]either is silver like furniture, of which a man never buys more when once he has got enough for his house. No one ever yet possessed so much silver as to want no more; if a man finds himself with a huge amount of it, he takes as much pleasure in burying the surplus as in using it.
(Xenophon IV.7)

Silver was in plentiful circulation at this time (Vickers and Gill 1996) and was deposited in hoards. The archaeologically uncovered silver hoards of Thrace represent a tiny fraction of the total.

Nicias son of Niceratus once owned a thousand men in the mines, and let them out to Socias the Thracian, on condition that Socias paid him an obol a day per man net and filled all vacancies as they occurred.... Hipponicus, again, had six hundred slaves let out on the same terms and received a rent of a mina a day net. Philemonides had three hundred and received half a mina. There were others too, owning numbers in proportion, I presume, to their capital.
(Xenophon IV.14-15)

The fifth-century leasing of slaves was well organized and large-scale: a total of 1,900 slaves are cited here as examples of only some among many lease agreements. There are 600 obols in a mina, so the obol a day rate is standard in the three examples Xenophon cites. Total employment of mine slaves in the Aegean world in the later fifth century must have been at least an order of magnitude higher (19,000) and possibly much greater. An obol a day per man must be a fraction of total net profits (Sosias must have been making a profit), and represents less than a gram of silver (one obol = 0.72 gm). The per-slave production of finished metal must have varied considerably, but would have outstripped this figure in a smooth-running silver mine by at least a factor of a hundred (e.g. 72gm per day of silver; I shall round this estimate down to 50gm to make subsequent calculations easier to follow). There were, obviously, costs and risks involved: buying and developing the mine in the first place, housing and feeding mine workers, obtaining charcoal for smelting and producing silver in a finished, negotiable form. A mine could be struck by the bad luck of seams running out and shafts falling in. 'Vacancies' is probably a euphemistic reference to mortality rates, which, in the absence of specific evidence, must be considered relatively high. As slaves were wholly owned and relatively cheap (see below), there was no economic imperative for, or even nudge towards, health-and-safety measures. Some skilled workers and foremen may have been prized, but conditions may well have been such that the first year of a mining slave's life was the most profitable from the owner's point of view.

[On the potential state acquisition of slaves to supply to lessees for mining] Assume ... that the total number of slaves to begin with is twelve hundred. By using the revenue derived from these, the number might in all probability be raised to six thousand at the least in the course of five or six years. Further, if each man brings in a clear obol a day, the annual revenue derived from that number of men is sixty talents.
(Xenophon IV.23)

A talent is c.26 kg of silver. The weight scheme drifted and varied slightly over space and time; the figures I shall use here are 1 talent (at 25,920gm), equivalent to 60 minai (at 432gm), or 6000 drachmai (at 4.32gm), or 36,000 obols (at 0.72gm). Thus, 1,200 slaves will bring lease fees of 12 talents (in a 360-day year) and six thousand will produce 60 talents.

Out of this sum [the 60 talents above], if twenty talents are invested in additional slaves, the state will have forty talents available for any other necessary purchase. And when a total of ten thousand men is reached, the revenue will be a hundred talents.... But the state will have received far more than that, as anyone will testify who is old enough to remember how much the charge for slave labour brought in before the trouble at Decelea [413 BC, when slave labour in the mines was hit by mass desertion].
(Xenophon IV.24-5)

Xenophon sets a target of ten thousand slaves to be state owned. He also suggests a conservative factor in his estimate. Slaves can be bought cheap and leased at good profit. Fortunately, we have a clear record of what some slaves cost the year before the desertion at Decelea. Surviving epigraphic fragments listing confiscated property sold at Athens in 414 BC refer to forty-five slaves, of varied origin, with prices preserved for twenty-four of them. Three of them are expensive: a Carian goldsmith (male) at 360 drachmai (2160 obols or 3.6 minai), a Syrian man (301 drachmai), and a Macedonian woman (310 drachmai). The median price was 157 drachmai, and many could be had for around 100 drachmai or less (Meiggs and Lewis 1975: 247). What the difference in price was between, on the one hand, a proven and trained-up house slave in Athens and, on the other hand, basic mine fodder, brought down from inland Thrace to Thasos or by sea directly to the Lavrion peninsula, is unclear. If we stick with a 100-drachmai figure, which is probably too expensive for a mine slave even more than half century later, we can see that Xenophon's projected 20 talents would buy 1,200 slaves to add to the 6,000 already invested in. Given that the lessee replaces 'wastage', using this percentage rate the state-owned slaves could be built up by 20 per cent a year.

The projected revenue from Xenophon's scheme would obviously not in itself solve the economic crisis he was addressing, but it must have been intended to demonstrate that a significant contribution to Athenian finances could be made in this way. At the end of the fifth century a single gold monument, celebrating the victory of Selinus, cost 60 talents (Meiggs and Lewis 1975: 82), which sets the sum in context.

It would be useful to see how this projected revenue might compare with actual revenues drawn in by those who controlled the slave supply, and to whom a proportion of the silver went, in payments and tribute. Thucydides, himself a silver mine owner on Thasos and thus precise and reliable in money matters (as in much else), tells us of the king of the Odrysian Thracians, Seuthes, who, in the later fifth century, had raised his tribute (implied as annual) to around 400 talents (Thuc. 2.97). Thucydides does not express surprise at this figure; for him it was representative of Odrysian economic power, concerning which he endeavoured to paint a realistic picture. He adds that 'at least an equal amount of gold and silver was contributed in presents', on top of which again were non-precious metal tributes and gifts.

Seuthes' tribute as a percentage of the total production of wealth by his tributaries can be assumed to have been of the order of 10 per cent and no more than 25 per cent (taking a higher cut than this, historically speaking, might have been militarily, politically and economically untenable). A glimpse of what form such tribute might have taken is provided, if we move forward in time to the mid-fourth century BC, by the hoard find from Rogozen (Cook 1989) in north-western Bulgaria. This comprised 165 silver and silver-gilt drinking vessels weighing a fraction under 20kg, probably assembled as part of a tribute payment and buried in two portions for safe keeping, apparently at the same time. Metrological analysis by Vickers (1989) suggests that it was made up of units of known value, totalling exactly 3,600 Persian sigloi (equivalent to four-fifths of a talent).

It is interesting to see what Seuthes could have done with his tribute. If we imagine that Seuthes was to use just 10 per cent of it to buy mine slaves at Greek prices (a preposterous notion, given that he was indubitably a supplier of Thracian slaves as well as a controller of local silver mines in the Sredna gora), he could buy 2,400 of them. The slave-worth of his precisely accounted 400 talents alone is thus 24,000 slaves. Owned and leased out, they would produce a revenue of 240 talents in one year and double Seuthes' money in less than two. A further order-of-magnitude calculation can be made to show how many days of mine-slave work Seuthes' tribute represented on the basis of the 50gm per day of finished silver production estimate. Here the calculation is 400 talents = 10,368kg. This is over ten tonnes of silver (and well over twenty if we add the presents) - a lot, but a drop in the ocean compared with the amount Trajan was much later to plunder out of Dacia (Taylor 1994: 406f.). This equals 207,360 slave days or 576 slaves working a mine for a year (numbers to at least double with the presents added).

I suggested that the percentage of silver paid as tribute to Seuthes by the Greek colonies he was connected with, and by indigenous vassal kings, can be estimated reasonably as of the order of 10 per cent. Thus, the mine activity required to support the local economy of Odrysian Thrace would have involved of the order of 5,000 mine slaves (10,000 for Thucydides's total figure). Hypothetical though this is, it is satisfying that this number falls within an order of magnitude similar to the numbers Xenophon directly refers to for the Greek-controlled mines he discusses. The big picture, including Thasos and Lavrion to the south, Triballian-controlled silver in northern Bulgaria and the precious metal sources of the Apuseni mountains in the (Agathyrsian-controlled?) Carpathian basin, and so on, compels us to envisage a further ten-fold multiplication of miners (although the legal status of those beyond the Greek world is not directly known).

Seuthes' 10,000 kg of silver can be compared to what has been archaeologically accessioned, of which there is no more than 100kg, i.e. 1 per cent. But this is catalogued Thracian silver for the whole period of the fifth and fourth centuries BC and across the entire Carpatho-Balkan region - it is not the rump of Seuthes' local, limited-percentage, annual tribute. Questions of gift-exchange, use-lives and recycling over several centuries have to be taken into account in calculations of the total amount of silver in circulation at any one time, and its percentage relationship to the annual production of new indigenous silver within the region and what came in as payment from the Greek colonies. Nevertheless, based on these, and on other, similar quantitative data, it can be demonstrated that, archaeologically speaking, there is at best of the order of a 0.01 per cent representativity for silverwork in the archaeological record (Taylor 1999). That is, of what was once in circulation, no more than a one hundred-thousandth part is now known. The same ratio holds for reusable and valuable copper metal in earlier periods, where ancient mine extraction tonnages can be compared to archaeological material. It is, in part, an inability to grasp this massive under-representation in what has survived through two-and-a-half millennia to be accessioned by museums and published, that has led to such a profound underestimation of the scale of Iron Age Eurasian economies.

Crawford (1977) provides useful comparator figures in his discussion of the influx of silver denarii to Dacia in relation to the Roman slave trade. His calculations support the idea of an export of some 30,000 Dacian slaves per year in the period 65-30 BC ('a substantial part of the annual requirement of Italy, if one assumes a total slave population of 2,000,000': 1977:123). He assumes a slave to cost 50 denarii at source and that the 25,000 archaeologically recorded denarii from Romania are representative at a 0.1 per cent level. But this is, as I have indicated above, likely to be a factor of ten too high; thus the total export figure might be better reckoned at 300,000 slaves per year (if no cattle, honey, wax and fish were exported, which, of course they were). This estimate would allow us to accept the figures given by John the Lydian (De Mag. 2.28) that 500,000 Dacians were enslaved at the conclusion of Trajan's campaign - a figure otherwise rejected out of hand by Madden (1996).

We could of course reject all the original classical figures and ratios, along with any extrapolations based on their comparison with what is archaeologically known. But the consistent picture that arises when one takes them seriously suggests that this is unwise. Just because the Greeks were proudly literate and liked art there is no reason to suppose them innumerate and uncovetous of wealth.

The figures demonstrate that silver mining in the Balkan-Aegean world in the mid-first millennium BC, at least periodically and according to macro-economic cycles, involved the contemporaneous utilization and maintenance of tens of thousands of mine-slaves. This could not have been achieved without the establishment of reliable supply zones, the existence of primary acquisition strategies, organized secure transportation, trans-shipment nodes, frameworks of custom and law governing ownership of humans and transfers of property in them, and so on - in short a large-scale, organized slave trade. This is evident inferentially from the ethnic diversity of slaves sold in Athens. Buyers wanted complements of slaves of diverse origin so as to inhibit solidarity and the chance of trouble (Finley 1962: 55; Ps.-Arist. Oeconomica 1.5), and competition between different local and regional suppliers must have driven prices down. Slaves from Asia Minor, Thrace and Scythia figure most commonly; in the surviving lists from 414 BC, the largest proportion of slaves whose nationality is known were Thracians (twelve out of thirty-six or 30 per cent; there were also seven Carians and three Scythians). But it is clear that there was further ethnic differentiation under the surface (see Braund and Tsetskhladze 1989: 119ff.; Arnold 1988: 182). That is, 'Scythian' slaves - flagged by Polybius (IV.38) as being of good quality - were simply slaves sold out of Scythia through the Pontic colonies. We are not to think of them as having once belonged to the royal Scythian elites as described by Herodotus, but as having come from their tributary peoples, and perhaps typically only via them.

Herodotus's ethnic mapping of Thrace and Scythia is in its bare essentials a commodities digest (Taylor 1994: 390). All the ethnonyms Herodotus records may in fact relate to regional and inter-regional elites alone - those peoples the Greek traders dealt with on a more or less equal footing. The locals who were preyed upon to supply the majority of market slaves were, most probably, the unnamed 'people without history' (in Eric Wolf's memorable phrase). Jack Goody has written that there is:

no slavery without the slave trade. But the social consequences begin further back, because in order to supply the trade, violence has to be applied by someone, in many cases far removed from the eventual slave owner. So that the consequence of the sliding status of the slave is the continual violence required to replenish the ever-emptying pool of forced labour.
(Goody 1980: 41)

The violence is something Herodotus mapped as well.


Indigenous institutions and responses

Classical slavery has been understood through the lenses of European medieval slavery (our word deriving from the ethnonym 'Slav', the people who constituted the majority of the target or 'resource' population for the Viking-Arab trade) and the most recent mass slave-trading episode involving the European and American enslavement of East and West African populations. This historical alchemy has produced a synoptic view or Gestalt of 'the slave', a term that can be sensibly used to translate Herodotean doulos. But to what extent were the Greeks responsible for the emergence of slavery? Goody's definition does not require a specifically monetary economy; trade is to be understood here anthropologically, for it is clear that slave-owning and the existence of agreed terms for exchanging people with (other) material items occurs in the majority of pre-state formations that have been ethnographically documented, and can be reasonably inferred for prehistoric Europe as well (Gronenborn 1999). Indeed, one basic value unit may be the slave, as in the term cumal, known from the Irish early medieval texts and designating a female slave. One cumal was equivalent to six heifers or three milch cows, and could be used to express the worth of a chariot or a piece of land (Powell 1981: 119f.).

Slavery begins, perhaps, with women, and with children. Engels drew attention to what he termed 'the slavery latent in the family' and it is salutary to see, under the newspaper headline 'Auctions for sex: Europe's thriving slavery industry' (Carroll 2000), a description of a mass trade in young Balkan women of a kind that Herodotus might well have recognized (Hdt 5.6: 'There is a custom among the ... Thracians that they sell their children for export, and as far as young maidens are concerned, they keep no watch over them but let them couple with whom they will'). It is as socially embedded now as it appears to have been then: the contemporary women, who are sold and resold among pimps for sums of the order of ?1,000 (and perform oral sex for ?15), do not try to escape as 'they are inhibited by fear of retaliation against their families back home'.

The roots of slavery are in the non-consensual and habitual extraction of value from one person and its transfer and conversion to benefit others with a controlling interest, and in their sovereign right to dispose of that interest as and when they see fit. This usually occurs where a person is already weaker than another, in certain specified ways (hence Engels's telling phrase). Chattel slavery indicates a named status where this tends to be absolute and without limit: the enslaved person has no rights at all, no means of production, and is tradable as an item (Patterson 1982).

Gronenborn (1999) considers that some form of slavery was in existence in central Europe from at least the beginning of the Neolithic, inferable from more general archaeological indices of inter-personal conflict, warfare and human sacrifice. Similar indirect methods have been used by Arnold (1988), while two fascinating, largely text-based studies investigate indigenous categories of 'slave' type among the Germans (Grünert 1969) and the Gauls (Daubigney 1979). Daubigney argues that between the later Iron Age LTI and LTIII domestic servants of various kinds, such as the ambacti (termed comites for the Germans) who work for, eat with and die for their masters according to various embedded and customary forms of subordination (dependance), and who are distinguished from more servile ranks (such as adolescent magu-), ultimately become anonymous objects of commerce when they are sold to Rome (1979: 175).

For the Balkan and Pontic regions Herodotus also provides information on many types of indigenous servitude and slavery among the Thracians and the Scythians (Khazanov 1975). The mass funerary sacrifice of dependent servants in royal tombs was described in detail by him, and has been clearly attested archaeologically (e.g. Rolle et al. 1998). Other information concerns: the sale of Thracian children by the Thracians (Hdt 5.6, noted already); the sexual and reproductive relationship that sprung up between the Scythian women and their slaves while the male Scythian warriors were on a twenty-eight-year campaign south of the Caucasus (Hdt 4.1); and how the Scythian king has 'no purchased slaves' but is served only by those of Scythian birth (Hdt 4.72). To these descriptions of types of dependance we must add the initially enigmatic passage reproduced below, in which Herodotus tells us that the Scythians blind all their slaves. It occurs near the beginning of Book IV, and thus introduces us to Scythia. It might be that it contains something Herodotus felt was essential or particularly representative, and that is why I want to give it careful consideration:

Now the Scythians make all their slaves blind [το`υς δε δούλος ´οι σκυθαι παντας τυφλου~σι] and this custom is because of the milk - for the Scythians are milk drinkers. This is how they do the milking: they take a kind of bellows, with a flute-like bone pipe, and insert it into the female orifice of their mares. They blow down this by mouth as others milk the mare. They say that blowing makes the mare's veins swell and her udder descend. When they have taken the milk, they pour it into wooden vats, about which they range the blind whose job it is to churn. The top part of the milk they draw off and think the most valuable; that which sinks to the bottom, less so. This then, is why they blind all their prisoners of war. The Scythians are, you see, not cultivators at all, but nomads.
(History, Book 4.2; translation based on Grene 1987 with reference to the standard Greek text)

This is considered opaque and problematic by modern commentators and translators (e.g. Grene 1987: 279f., n.2: 'most insoluble .... No editor has any definitive answer to the difficulties'). De Selincourt, in his translation for the Penguin Classics series (1970: 273), editorializes so far as to have his English version say 'The Scythians blind their slaves, a practice in some way connected to the milk which they prepare for drinking' - as if Herodotus were himself uncertain of the reason why, when it is plain from the original that he in fact considers it so obvious as to be beneath mention. De Selincourt also twice ignores Herodotus's categorical assertion that all the slaves were blinded. It is as if De Selincourt himself, unable to comprehend what is really being described, has sought to minimize it. (Herodotus has, as noted above, referred in other places, to other kinds of slavery in Scythia, and Scythian slaves sold in Athens were not habitually blind; but the all is very clear and I take it to mean all of this, as I shall argue, predominant type of indigenous slave.)

A different approach is taken by Stephanie West. She believes Herodotus has made the milking method up: 'The use of any apparatus other than the human lips is not easily paralleled' (West 1999: 79), and that Herodotus's description of the flute-like pipe is 'bizarre'. She argues that his description of processing milk into fermented koumiss (the latter being an extrapolation from what Herodotus actually says) is 'simply wrong' (West 2000: 31f.) and concludes 'That all those whom the Scythians took captive are supposed to be engaged in an imaginary occupation enhances the improbability of these opening chapters [of Book 4]'. West does not think Herodotus ever actually visited the northern Black Sea region in person, and her arguments are detailed and interesting. The deficiencies of Herodotus's description, as she sees them, 'firmly discourage us from supposing that he had enjoyed the hospitality of a nomad encampment'.

There is no space here to make a detailed defence of the idea of Herodotus having actually visited Scythia. He implies that he did, and I believe him (my views are clearly stated elsewhere: e.g. Sulimirski and Taylor 1992; Taylor 1994). The milking techniques he describes that West doubts are in fact well supported anthropologically, and tubes for gently inflating rectum or vulva are known from early pastoralist archaeological sites in southern Russia (Sherratt 1997: 177, with reference to Galkin 1975). I do agree with West that Herodotus may never have been a guest at a nomad encampment. Not, however, because he was never in Scythia, but because he was not some kind of Aurel Stein figure, wandering the landscape, and the elite Scythians were not rough Kazakh herders. Olbia, home to many, many Greeks, had become, by the fifth century BC, the centre of a truly massive trade enterprise with the steppe and forest steppe regions, attested both by imports and defended production centres (Rolle 1985). Belsk on the Vorskla, with its 33km of imposing ramparts - equal in length to the Boulevard Peripherique that encloses modern Paris, has been identified by its excavator (Boris Shramko) with Herodotus's Gelonus (Taylor 1994: 387, A.ii), but it is one of several and perhaps not the largest. Herodotus would have stayed in comfort in Olbia, and met traders and members of the Scythian elite in high social circles. Indeed, whether such an elite would have known how to construct a 'nomad encampment' is a moot point. They had by this time become so aggrandized, with several social strata beneath them in a complex mesh of lateral and vertical ethnicity, that they were insulated from pastoral realities. Their wealth was based on slaves, iron and grain, none of which they dirtied their own hands producing. Greek whitesmiths worked in the colonial centres, producing toreutic for the Scythian elite, which is eloquently (and rather uniquely) uncondescending in its narrative portrayal of native lifeways (Taylor 1996).

I read Herodotus' passage as follows:

1 The slaves are those whom the Scythians have caught, bred or bought and chosen not to trade through the colonies. Endemic warfare (as evidenced by both texts and archaeology) provides the Scythians with many tens of thousands of captives. Unlike those who serve the king in life and death (therapontes, like ambacti or comites), such slaves are from Scythia but not 'Scythians'.

2 The blind slaves are the principal workforce in horse dairying, which is not a small-scale business, but something involving large wooden vats and truly massive herds (that is, unlike what is ethnographically known from the steppe in recent times, with its skin koumiss-makers and relatively small volumetric output). Some idea of the size of Scythian herds is suggested by the fifth-century BC burial of Ulski Aul, where the articulated skeletons of 360 contemporaneously sacrificed riding horses were discovered. The excavator, Veselovsky, simply gave up counting further horse bones in the subsequent sacrifices of the upper layers of the kurgan mound (Taylor 1994: 392, ii), and it may be that over a thousand horses were provided for this particular noble's entourage to ride in the afterlife.

3 The reason the slaves are blinded is to stop them deserting across the steppe, riding the horses they work with. Stephanie West notes that Shalmaneser I seems to have blinded 14,400 prisoners of war so that they could be employed in dire drudgery without risk of escape or insurrection (West 1999: 78, citing Gelb). Blakesley (in contrast to modern commentators) was quite clear on this: 'Of course their blindness prevented the possibility of their ever absconding, which would otherwise be rendered very easy by the nomad life their masters led' (Blakesley 1854: 437, note 9).

4 The Scythian churning by the blind slaves is indeed hard drudgery. In relation to it Herodotus fortuitously signals to us something that, if it was so, his audience would have known anyway: that they do this because they are not cultivators, i.e. not Greek. That is, they do not brand their slaves for sighted use in the field as the Greeks do; from which we may understand that it was previously understood that Greek agricultural labourers were indeed typically slaves - something many scholars have suspected (though many have questioned it too).

5 If the ratio of free Scythians to slaves was similar to that adduced for Attica by Demetrius of Phaleron (that is, if the elite and their families were outnumbered ten to one by those forced to fuel their massive economy) then, given the ready availability of swords, spears and, most pertinently, vicious composite reflex bows, blinding would have been a prudent security measure.

This passage dramatically qualifies Homer's epithet from c700 BC, the 'mare-milking Scythians', to inform us of relations of dependance among them, arising from the scale of their pastoral economy and - which is almost the same thing - the scale of their war-making. Stronger than Greece, they had threatened the Persian Empire at home in the sixth century, and Herodotus's audience would have no illusions that this is a characterization of a great and powerful nation, not a quaint local tribe. To be able to conduct the blinding of prisoners of war, en masse, effectively (without undue wastage from septicaemia for example) implies a particular specialism and, alongside it, a particular system for classifying human beings - or, to put it perhaps more precisely, the absence of a universalizing category 'human being'. It is a truism that those who conducted the blinding must have been exposed to behavioural models of a type that were compatible with a capacity for developing and maintaining such a custom, and thus the practice must have emerged over a period of several generations. It seems possible that this form of pastoral enslavement pre-dated Greek colonial activity, emerging with earlier pastoral elites on the north Pontic region some time in the preceding Bronze Age (for example, with the Seima-Turbino group). It is hard, in any case, to see this class of slaves primarily as a reflex of the establishment of a Greek slave-trading operation in the Black Sea.

What the Greeks were, perhaps, partly responsible for was the sharpening of the conceptual distinction between slavery and freedom in the regions from which their slaves were sourced. The legal and philosophical clarity of Greek chattel slavery must have had an effect on indigenous forms of subordination in Eurasia, beyond the point of sale where magu- became doulos, just as the symbolic grandeur of Persian imperial command inspired imitation (Taylor 1988). Thus, the emergence of a new, more inclusive (though obviously selective) cemetery style among the La Tène Celts, and their military, Männerbund expansionism, have suggested to many an entrepreneurial spirit and a particular valorization of 'freedom'. The provocative and compelling idea that 'freedom' may originally have emerged as an explicit value in meaningful contradistinction to the institution of slavery proposed by Finley and examined by Patterson (1991).

It is in this connexion that the elite neck torcs of the La Tene Celts and the Dacians (see Plate 2), inspired in part by those of Persia (the Trichtigen ring being the earliest example in a central European context), are a reflex of the existence of slave chains. Torcs in some way symbolize enslavement, not to a superior human but to a superior power. To wear a gold or silver neckpiece signalled enslavement to deity and, by that very token, freedom on earth. The Celtic invasion of the Transylvanian and lower Danube basins, culminating in the establishment of the kingdom of Tylis in the Balkans, was intimately connected to an economy of trade and raid, in which the capture of precious metal booty was at least balanced by the capture of prisoners to sell and the gaining of slave source zones. In this context, it is hard not to see their voluntary adoption of a heavy precious metal neck ring as a classic item of personal adornment as embodying a statement about their position vis-a-vis those whose necks involuntarily wore iron or rope.

We do, that is, have to explain the appearance of torcs as indicators of rank, both visually and chronologically. In Poseidonius's description of a Gaulish banquet, the chief sits (or squats) with the person he is receiving next to him, with others in the hierarchy tailing off, according to rank on either side. The shield bearers of these warriors are behind them, but the spear carriers, ambactes of a higher rank, close the circle in front of the warriors. The drinking vessels are brought around by servants, who pass from right to left in front of the feasters and Poseidonius notes, in this context, that prayers are offered to their gods in the same way, turning (as the servants have to do) to the right (Poseidonius in Atheneus IV, 36, 152; discussed by Daubigney 1979: 164f.). No mention of torcs here, but what is telling is the way that status is choreographed, and the orientation of service to chiefs is explicitly equated with the orientation of chiefs towards deities. It is just such a relational, symbolic, grammar which, I believe, we can see being activated in the appearance of precious metal neck torcs as an elite signifier. This begins no earlier than the fifth century BC, when classical trade with the late Hallstatt is already well established, and when we know from textual references that slave chains are already in use (though none has been archaeologically recorded).

There are many ways of enslaving and being enslaved, some more permanent than others. The Scythians' milking slaves each lost their sight for ever, irrespective of whether they escaped or were manumitted. There are times when an extreme form, often termed chattel slavery, achieves such widespread economic importance that it becomes iconic, blinding us as much to the myriad other nascent and evolved forms of unfreedom in its shadow, as to the significant nuances of the major phenomenon itself.


Acknowledgements

I wish to thank Michael Vickers for help with metrological and fiscal issues, Irina Levinskaya and Tom Braun for help, at various times, with the Greek textual material, and Detlef Gronenborn for sight of the text of his unpublished 1999 paper. I should also like to thank Bettina Arnold, David Braund, Barry Cunliffe, Adrian Gollop, John Madden, Stephen Mac-Namara, Orlando Patterson, Gerhard Trnka, Stephanie West and Sarah Wright. It should be clear that responsibility for the argument and the calculations above is my own.

Department of Archaeological Sciences
University of Bradford


References

Alfoldy, G. 1988.
Antike Skaverei. Widerspruche, Sonderformen, Grundstrukturen.
Bamberg: Buchner.

Arnold, B. 1988.
Slavery in late prehistoric Europe: recovering evidence for social structure in Iron Age society.
In Tribe and Polity in Late Prehistoric Europe: Demography, Production, and Exchange in the Evolution of Complex Social Systems
(eds D. B. Gibson and M. N. Geselowitz).
New York: Plenum, pp. 179-92.

Blakesley, J. W. 1854.
Herodotus, with a Commentary.
London: Whittaker.

Braund, D. C. and Tsetskhladze, G. R. 1989.
The export of slaves from Colchis. Classical Quarterly, 39(i): 114-25.

Carroll, R. 2000.
Auctions for sex: Europe's thriving slavery industry.
Guardian, 23 May: 19.

Clark, E. D. 1821.
Letter.
Archaeologia, 19: 61.

Cook, B. F. (ed.) 1989.
The Rogozen Treasure.
London: British Museum Publications.

Crawford, M. H. 1977.
Republican denarii in Romania: the suppression of piracy and the slave trade.
Journal of Roman Studies, 67: 117-25.

Cunliffe, B. 1988a.
Greeks, Romans and Barbarians: Spheres of Interaction.
London: Batsford.

Cunliffe, B. 1988b.
Iron Age Communities in Britain.
London: Routledge.

Daubigney, A. 1979.
Reconnaissance des formes de la dependence Gauloise.
Dialogues d'Histoire Ancienne 5: 145-93.

De Selincourt, A. (trans.) 1970.
Herodotus: The Histories (with an introduction by A. R. Burn).
Harmondsworth: Penguin.

Finley, M. I. 1962.
The Black Sea and Danubian regions and the slave trade in antiquity.
Klio, 40: 51-9.

Fisher, N. 1998.
Work and leisure.
In Ancient Greece (Cambridge Illustrated History) (ed. P. Cartledge).
Cambridge: Cambridge University Press, pp. 193-218.

Galkin, L. L. 1975.
Odno iz drevneishikh prakticheskikh prisposoblenii skotovodov.
Sovetskaya Arkheologiya, 3: 186-92. 42

Garnsey, P. 1988.
Famine and Food Supply in the Graeco-Roman World: Responses to Risk and Crisis.
Cambridge: Cambridge University Press.

Gollop, A. 1994.
'Slave chains, slavery and slave trade in Iron Age Europe: a conceptual overview'.
Unpublished UG dissertation,
Department of Archaeological Sciences, University of Bradford.

Goody, J. 1980.
Slavery in time and space.
In Asian and African Systems of Slavery (ed. J. L.Watson).
Oxford: Blackwell, pp. 16-42.

Grene, D. (trans. and intro.) 1987.
The History. Herodotus.
Chicago: University of Chicago Press.

Gronenborn, D. 1999.
Zum Nachweis von Sklaven/Unfreien in prähistorischen Gesellschaften Mitteleuropas.
Paper given at the 3rd German Congress of Archaeologists,
Heidelberg, 27 May.

Grünert, H. 1969.
Zu den Anfängen und zur Rolle der Sklaverei und des Sklavenhandels im ur- und frühgeschichtlichen Europa, speziell bei den germanischen Stämmen.
Ethnographisch-archaologische Zeitschrift, 10: 501-15.

Khazanov, A. M. 1975.
Sotsialnaya istoriya skifov.
Moscow: Nauka.

Kristiansen, K. 1998.
Europe Before History.
Cambridge: Cambridge University Press.

Madden, J. 1996.
Slavery in the Roman Empire: numbers and origins.
Classics Ireland, 3: 109-28.

Marchant, E. C. (trans.) 1925.
Xenophon in Seven Volumes VII: Scripta Minora.
Cambridge, MA: Harvard University Press, Loeb Classical Library.

Meiggs, R. and Lewis, D. (eds) 1975.
A Selection of Greek Historical Inscriptions to the End of the Fifth Century B.C.
Oxford: Clarendon.

Nash, D. 1985. Celtic territorial expansion and the Mediterrranean world.
In Settlement and Society: Aspects of West European Prehistory in the First Millennium B.C. (eds T. Champion and J. V. S. Megaw).
Leicester: Leicester University Press, pp. 45-67.

Patterson, O. 1982.
Slavery and Social Death.
Cambridge, MA: Harvard University Press.

Patterson, O. 1991. Freedom I:
Freedom in the Making of Western Culture.
New York: Basic Books.

Powell, T. G. E. 1981.
The Celts, new edn.
London: Thames & Hudson.

Rolle, R. 1985 Der griechische Handel der Antike zu den osteuropäischen Reiternomaden aufgrund archaologischer Zeugnisse.
In Untersuchungen zu Handel und verkehr der vor- und frühgeschichtlichen Zeit in Mittel- und Nordeuropa 1:
Methodische Grundlagen und Darstellungen zum Handel in vorgeschichtlicher Zeit und in der Antike (eds K. Diiwel, H. Jankuhn, H. Siems and S.Timpe).
Göttingen: Abhandlungen der Akademie der Wissenschaften in Göttingen, Phil.-Hist. Klasse 3(143): 460-90.

Rolle, R., Murzin, V. Ju. and Alekseev, A. Ju. 1998.
Königskurgan Certomlyk: ein skythischer Grabhügel des 4. vorchristlichen Jahrhunderts, 2 vols,
Hamburger Forschungen zur Archaologie.
Mainz: Philipp von Zabern.

Scheidel, W. 1997.
Quantifying the sources of slaves in the Early Roman Empire.
Journal of Roman Studies, 87: 156-69.

Sherratt, A. 1997.
Economy and Society in Prehistoric Europe: Changing Perspectives.
Edinburgh: Edinburgh University Press.

Sulimirski, T. and Taylor, T. 1991.
The Scythians.
In The Cambridge Ancient History, 2nd edn, Vol. III, part 2 (eds J. Boardman, I. Edwards, N. G. L. Hammond and E. Sollberger).
Cambridge: Cambridge University Press, pp. 547-90.

Taylor, T. 1988.
The Persian Empire, 6(d) International craft traditions (ii): In Europe.
In The Cambridge Ancient History plates to Vol. IV, new edn (ed. J. Boardman).
Cambridge: Cambridge University Press, pp. 78-94.

Taylor, T. 1994.
Thracians, Scythians, and Dacians.
In The Oxford Illustrated Prehistory of Europe (ed. B. Cunliffe).
Oxford: Oxford University Press, pp. 373-410.

Taylor, T. 1996.
Scythian and Sarmatian art.
In The Dictionary of Art 28 (ed. J. Turner).
London and New York: Macmillan/Grove, pp. 319-26.

Taylor, T. 1999.
Envaluing metal: theorizing the Eneolithic 'hiatus'.
In Metals in Antiquity (eds S. M. M. Young, A. M. Pollard, P. Budd and R. A. Ixer).
Oxford: Archaeopress, pp. 22-32.

Vickers, M. 1989.
Panagyurischte, Dalboki, Loukovit and Rogozen: questions of metrology and status.
In The Rogozen Treasure (ed. B. F. Cook).
London: British Museum Publications, pp. 101-11.

Vickers, M. and Gill, D. 1996.
Artful Crafts: Ancient Greek Silverware and Pottery.
Oxford: Clarendon Press.

Wiedemann, T. 1980.
Greek and Roman Slavery.
London: Routledge.

West, S. 1999.
Introducing the Scythians: Herodotus on koumiss (4.2).
Museum Helveticum, 56: 76-86.

West, S. 2000.
Herodotus in the north? Reflections on a colossal cauldron (4.18).
Scripta Classica Israelica, 19: 15-34

Williams, D. 1998.
Romans and Barbarians: Four Views from the Empire's Edge, 1st Century A.D.
London: Constable.