Bryn Mawr Classical Review

Bryn Mawr Classical Review 2011.04.09

Ulrike Roth (ed.), By the Sweat of your Brow: Roman Slavery in its Socio-economic Setting. BICS Supplement 109.   London:  Institute of Classical Studies, School of Advanced Study, University of London, 2010.  Pp. ix, 121.  ISBN 9781905670291.  £21.00 (pb).  

Reviewed by Kyle Harper, University of Oklahoma (

By the sweat of your brow is a collection of seven papers on Roman slavery. Most of the contributions have their origins in a conference held in September of 2007 in Edinburgh. Anyone familiar with the institutional politics of ancient slave studies will recognize that the Edinburgh meeting was a virtual peace conference, gathering representatives from research traditions which have been mutually suspicious when not simply indifferent to one another. Such collaboration sans frontières is encouraging, even if the results are uneven.

The first contribution is Benet Salway’s discussion of Diocletian’s Edict on Maximum Prices. The editor was wise to place this piece at the head of the volume, for it is supremely valuable. Salway offers a guided tour through the modern history of the edict’s chapter on slave prices, tracing the “incremental” progress which gradually allowed its reconstruction. Small fragments were recovered from Megara and Ptolemais in Cyrenaica, but the major breakthrough waited until the publication of fragments from Aezani in 1973. Then in 1989 Reynolds published a number of fragments from Aphrodisias which suggested significant variations from the price figures at Aezani.

At this impasse Salway intervenes. In 2000 he was commissioned to track down a fragment from Aphrodisias that had been discovered in 1970, photographed in 1972, and ignored ever since. Salway convincingly reconstructs the first entry: mancipium rusticum sive urbanum virilis sexus ab annis sedecim ad annos quadraginta. This requires the second line from Aezani to have read SIVE rather than SPADO. More importantly, the incorporation of the new fragment gave Salway the “opportunity for checking” all the readings from Aphrodisias. He is able to confirm that the top price for prime-age males was triginta milibus; readers can see from the photograph that tri]ginta quinque m[ilibus was an error. Similarly, the price of 15,000 denarii should belong only to the entry for males over sixty years old or under eight. Salway is to be thanked for setting this important data on such firm foundations.

The second chapter is valuable in a different way, for here Elio Lo Cascio offers a sweeping interpretation of the demography of Roman slavery. After a brief review of the question, Lo Cascio seeks answers by trying to situate slavery within the broader parameters of the population as a whole. He argues that the overall proportion of slaves in the Roman imperial period was declining even as the free population was growing. Central to his case is a graph, which also serves as the book’s cover image, plotting the marginal product of labor against the cost of a slave (“subsistence” here includes the purchase price divided by years of activity; it would be preferable to speak of the amortized purchase price and, separately, maintenance and supervision costs). The graph is built on a number of assumptions, the most important of which is that the population has reached a point of rapidly diminishing returns on units of labor. The graph only represents the population after it has reached maximum average productivity (necessarily wherever the marginal and average curves intersect, here shown as the y-axis). The situation is dire indeed if, as this graph shows, the average product of labor goes to zero — a state of affairs that is thankfully impossible. It is an open question whether the imperial population was within the band represented on the graph. More importantly, the graph cannot tell us about the profitability of slavery. If there was anything like a functional labor market in the Roman empire, then the price of slaves would have gone down with the marginal product of labor. If one wants to argue that the marginal product of labor was approaching or lower than the bare maintenance and supervision cost of a slave, then the market price of slaves should have been approaching zero. The empirical evidence for slave prices, which shows relative stability across the first four centuries of our era, in fact is a strong argument that the Roman population was not in the band suggested here.

The virtue of Lo Cascio’s model is that it explicitly recognizes the many interrelated problems implicated in the demography of slavery. He presents a strong case for a “high” count of the overall population, in combination with a traditional story according to which the slave system is created in the period of conquest and thereafter enters demographic decline due to an unbalanced sex ratio and frequent manumission. Many of the assumptions remain open for debate, but this is what makes the topic continually interesting.1 Here is an internally coherent model, vigorously presented.

Leonhard Schumacher’s essay argues that slave status was “not just typical” but “in principle necessary” for occupations which required agency and management. Schumacher does notice some of the ambiguous (e.g. CIL X 5081) and countervailing evidence (e.g. CIL VIII 25817, caveat lector on the use of the late antique legal sources), before he turns to ask why slaves were so useful as managers and agents. Dismissing explanations which emphasize the ability to train slaves or the potential to discipline them, Schumacher claims that institutional factors explain why slaves “became a sine qua non for business transactions.” Readers might wish for more engagement with the ample literature on forms of direct and indirect agency in Roman law and on the use of freedmen as agents (compare Roth’s contribution).

In “The slave in the window,” Paul du Plessis explores the use of slaves as implied pledges in urban rental arrangements. He opens with a vignette from the Digest at 20.2.9, where the jurist Paul distinguished between slaves who were used as security in a contract and those who were implicitly pledged as part of the invecta et illata of an urban rental property. The former could not be manumitted, while the latter might be freed up to the point when the defaulting tenant was shut out and his property seized. The jurist Nerva, however, argued that the slave who had been seized (and apparently locked in the house!) could still be manumitted, so long as the master could point at him through a window. There are problems in both the text and translation presented here.2 One might question, too, the author’s doubts that Nerva meant the slave could be manumitted; the entire discussion is about manumission. Aside from these quibbles, du Plessis presents a fascinating and sensitive exploration of the rules governing how seized slaves were to be managed and maintained. The jurists were keenly aware of the fact that the slave’s value might appreciate or depreciate. The discussion neatly proves that even when slaves were not literally in the market, their bodies were inscribed with an exchange value.

Michael Crawford’s essay compares the slave prices in the Attic stelai (415-413 BC) with those in the Price Edict (AD 301). Crawford’s treatment rightly insists that the data are “fraught with multiple uncertainties.” He reminds us that the equivalence between a phormos and a medimnos is not certain. Crawford is willing to work through these uncertainties in search of the best ascertainable data. Nevertheless, his analysis, which sets out to undermine Scheidel’s long-range study of slave prices, is not fully convincing. First, the calculations, whose transparency could be improved, are flawed. He lists the maximum price for a male in AD 301 as the equivalent of 2.25 metric tons of wheat (one slave = 300 modii castrenses and 1 modius castrensis = 13 litres, so 1 ton = 1733.3 litres). He claims “the maximum paid for a ‘normal’ male slave” in the stelai was equivalent to 1.70 tons of wheat. This might refer to the gold-worker bought for 360 drachmae; at the stelai’s average of 6.28 drachmae per phormos this would be the equivalent of 1.74 tons of wheat. It would be strange, though, to consider this slave “normal,” or not to notice the 100 percent premium for skilled workers allowed by Diocletian (thus, 4.5 tons would be the appropriate comparandum). The absolute highest sale price for a male slave in the stelai otherwise is 301 drachmae, the equivalent of 1.45 tons of wheat (the only way this value could equal 1.70 tons is if, for some unexplained reason, a wheat price of 5.36 drachmae/phormos is assumed). More meaningfully, the average market value of male slaves in the stelai was around .86 tons of wheat. Crawford objects to Scheidel’s comparison of the “maximum” prices for adult male slaves from the edict and the “mean” prices in the stelai. But the objection misses the mark. With the Price Edict, we are at the mercy of the bureaucrats’ desired maxima for slaves and for wheat, and the ratio between them is alone meaningful; the edict’s figures are useful only as relative normative values. It is a misunderstanding to construe the highest outlying slave price in the stelai as though this were a “maximum” in the same sense as the price ceilings in the edict. Even more decisively, Scheidel and others have collected a rather robust set of sale prices showing that the edict, if anything, actually undervalues slaves.3 Thus, Scheidel’s analysis is based on broader and better data, and it rests on a more plausible understanding of what the various prices mean; his demonstration that slaves in classical Athens were cheap relative to slaves in Diocletian’s empire still stands.

In “Recruitment and training of Roman estate managers in a comparative perspective,” Jesper Carlsen asks how vilici were selected and trained to become specialized farm managers. Agricultural management is no job for amateurs. Yet, there is very little concrete information about how vilici were trained, and most of what we do have is couched in highly moralized terms. These anxieties about the manager’s virtue are echoed, as Carlsen shows, by American and Russian landowners. The landowner-manager relationship, in all three contexts, was a classic principal-agent problem. Whereas a recent comparative exploration by Dal Lago and Katsari emphasized differences between ancient and modern slave management, Carlsen emphasizes similarities.4 Both have their points, and the comparative dimension is to be welcomed.

The last and longest contribution is Ulrike Roth’s treatment of manumission, which follows in the line of Hopkins, Bradley, and more recently Scheidel and Temin. In Roman law slaves manumitted by testament lost their peculium unless it was explicitly given to them; slaves manumitted while the master was alive kept the peculium unless it was explicitly taken away. Roth first argues that the peculium, or some significant part of it, was typically taken by living masters as the price of manumission. Then she extends the foundational work of Sirks on informal manumission and Junian Latins. Roth argues that slaves may have paid for iteration — in other words, they could have paid twice, first for informal, then for formal manumission. One has to admit the evidence is exiguous. The next frontier will be to treat manumission as part of an incentive system which could be finely modulated, in just the ways Roth describes, within a broader analysis of the occupational structure of Roman slavery.

These seven papers belong to a wave of recent studies on slavery that employ methodologies rooted in comparative analysis and quantitative modeling. This is only to be welcomed. The book does leave one wishing for a broader sense of the overall dimensions of the Roman slave system and its chronological span: the “Roman slavery” of the title is taken for granted. Roth has elsewhere done a great service for our understanding of the early development of Roman slavery, and three of the seven chapters here bear significantly on the later phases of Roman slavery, so the investigation would benefit from more attention to when and why the exceptional institution we call “Roman slavery” begins and ends.

List of Contributions:

Benet Salway, “MANCIPIVM RVSTICVM SIVE VRBANVM: the slave chapter of Diocletian’s Edict on Maximum Prices”

Elio Lo Cascio, “Thinking slave and free in coordinates”

Leonhard Schumacher, “On the status of private actores, dispensatores, and vilici

Paul J. de Plessis, “The slave in the window”

Michael H. Crawford, “From Alcibiades to Diocletian: slavery and the economy in the longue durée

Jesper Carlsen, “Recruitment and training of Roman estate managers in a comparative perspective”

Ulrike Roth, “Peculium, freedom, citizenship: golden triangle or vicious circle? An act in two parts”


1.   See K. Harper, Slavery in the Late Roman World, AD 275-425 (Cambridge, 2011): Chapter 2 for discussion about sex ratios, Chapter 5 for frequency of manumission.
2.   Manifesta is preferable to manifestati; percludamur should probably be praecludamur and rendered “we are locked out,” just as the author translated it in “The Interdictum de migrando revisited,” RIDA 54 (2007) 219-44 at 229. For a convincing discussion of all of this, see B. Frier, Landlords and Tenants in Imperial Rome (Princeton, 1980), 119-22.
3.   K. Ruffing, H.-J. Drexhage, “Antike Sklavenpreise,” in Antike Lebenswelten. Konstanz — Wandel — Wirkungsmacht, ed. P. Mauritsch et al. (Wiesbaden, 2008), 321-51. W. Scheidel, “Real Slave Prices and the Relative Cost of Slave Labour in the Greco-Roman World,” Ancient Society 35 (2005): 1-17. K. Harper, “Slave Prices in Late Antiquity (And in the Very Long Term),” Historia 59 (2010): 206-38.
4.   E. Dal Lago and C. Katsari, “Ideal Models of Slave Management in the Roman World and in the Ante-bellum American South,” in Slave Systems: Ancient and Modern, eds. E. Dal Lago and C. Katsari (Cambridge, 2008), 187-213.

Comment on this review in the BMCR blog
Read Latest
Index for 2011
Change Greek Display
Books Available for Review
Bryn Mawr Classical Commentaries

BMCR, Bryn Mawr College, 101 N. Merion Ave., Bryn Mawr, PA 19010
HTML generated at 02:44:40, Tuesday, 05 April 2011