Staple Provisions for Rome. Problems of Quantification


The quantity of grain imported to Rome is evidence so basic that all historians with an interest in the Roman economy have had something to say about it. They have all drawn on the same sources, which are few in number, and their arguments have always been similar in nature. Either they take as their starting point the four texts that may allude to Rome’s total consumption and discuss how to interpret them (this I call Approach A); or they start from a figure for the population of Rome, which they multiply by a number representing per capita consumption, arrived at from the sparse data available on rations and from computation of a plausible intake of calories (Approach B). In the latter case, care should be taken not to use a precalculated figure for the population devised by dividing a sum of total consumption by an individual ration presumed to be normal. This was one of the three methods used by Beloch (1886: 392-412) in his calculations of Rome’s population. With Rome, the intersection of demography and consumption is one that affords great scope for circular reasoning.

Approach A

The textual evidence is so meagre that it occupies little space.

  • 1. Flavius Josephus, Bell. Iud. II. 386: Egypt supplies Rome with enough grain for four months—in other words, a third of its needs;
  • 2. Epit. de Caesaribus, I. 6: Egypt supplies Rome with 20 million modii of grain per annum.

First published in C. Nicolet, R. Ilbert, and J.-C. Depaule (eds), Megapoles mediterraneennes (Actes du colloque de Rome, 8-11 May 1996), Paris: MMSH, Maisonneuve et Larose, and Ecole fran^aise de Rome, 2000), 751-60. Draft partially revised.

  • 3. SHA, Sev. 23, 2: moriens septem annorum canonem, ita ut cottidiana septuaginta quinque milia modium expendi possent, reliquit.
  • 4. a scholium to Lucan, Pharsalia, I. 318: daily consumption of 80,000 modii in the last years of the Republic.

Sources 1 and 2

The figure for supplies derived from Egypt (Source 2) multiplied by three (in accordance with Source 1) amounts to annual deliveries to Rome of 60 million modii—that is to say, 527 million litres or about 390,000-420,000 tonnes, depending on the density of the grain.[1]

This way of doing things was dismissed by Beloch, with every semblance of good sense and in such a devastating way (Ein so unmethodisches Verfahren bedarfkeiner Widerlegung (‘Such an unmethodical approach is beneath refutation’)) that good scholars have tended to eschew it.[2] The fact is that it brings together two texts, one of which may be reliable, since it is by an author from the Flavian period who was a contemporary of the events he recounts, while the other one is by an epitomator writing three centuries later. Beloch adds that the total of 60 million modii would correspond to a population of 2 million, which he finds preposterous. These criticisms could be strengthened by an argument drawn from Flavius Josephus, Bell. Iud. II. 383: Africa (not counting Egypt) feeds Rome’s plebs for eight months. Taken in conjunction with Source 1, this passage was once taken to exclude any supply of grain unless it came from Egypt or Africa, which was implausible. What it really dealt with, though, according to Federico De Romanis, was eight months’ worth of the free grain doled out to the plebs and not eight months of the total grain supply.[3] Hence, that text is no longer of relevance.

Sources 3 and 4

Beloch put more faith in Source 3, the text from the SHA, from which he took it that canon was a reference to Rome’s total consumption, which is not the usual meaning of it. The figure of 75,000 modii per day, equivalent to 27,375,000 modii per year, is conveniently close to that given in the scholium to Lucan (Source 4). Beloch extrapolated from it to estimate the population of Rome at about 760,000, allowing for an average consumption of 36 modii per person per year. But the notion of stocking up in advance seven years of

Rome’s consumption is impossibly absurd. The first part of the sentence cannot be construed as having a real meaning.

Federico De Romanis has recently suggested a quite different reading of Source 3, one that makes the whole text clear and coherent. Instead of the traditional interpretation of ita ut (given in Piganiol’s paraphrase[4] as ‘Severus was said to have left grain enough for seven years at the rate of 75,000 modii per day’), De Romanis, following B. Sirks,[5] sees each element of the sentence as using a different system of calculation, the correspondence between them being indicated by ita ut. From that it would follow that the second element must be referring to a daily amount over the course of a year: the amount that could have been given out over a year if the amount specified in the first part of the sentence had been distributed during that time. This, however, raises a difficulty: the figure for the possible distribution, 27,375,000 modii, implied by the second element of the sentence, equating to seven times the annual canon, ought to be divisible by seven, which is not the case. De Romanis gets round this by suggesting that LXXV should be changed to CLXXV, a figure that is divisible by seven and also lets us read the extremely high canon in the first element of the sentence as referring to a contribution from the provinces and the imperial estates that was seven times greater than the requirements of the free dole-out, five modii per month for each of the 150,000 people entitled to it. The total, 63 million modii, comes out as roughly equivalent to Rome’s total consumption.

In justifying this approach, De Romanis brings this figure of 63 million modii closer to the sum we arrive at if we take Sources 1 and 2 together. This would mean that either the population or their consumption per capita was far in excess of Beloch’s estimates. The recipients of the grain dole were allotted 5 modii per month, which is generally assumed to have been intended for their whole household. It is this same amount that Seneca says in a letter[6] was given to a slave who acted in tragedies and who also received 4 denarii a month. In my view, he must have been a slave with a very generous peculium, even if we compare him just with Seneca's second example, a few lines later, of another slave who was also an actor but who received only a diurnum. In fact, there is, to my mind, nothing to prove that the first actor’s 5 modii per month were the sum total of his daily allowance, rather than being a complement in kind of his 4 denarii. Consumption of 5 modii per month, which is equivalent to 35 kilograms, would provide 3,900 calories per day,[7] an extremely high figure for even a modern adult that cannot be extrapolated to the whole population, including women and children, even if we make allowance, as De Romanis rightly does, for the privileged food supply enjoyed by the inhabitants of the capital. Their diet, after all, did not consist solely of grain; and they also drank a fair amount of wine.

In fact, if we accept De Romanis’s subtle argument and the consistency of the figures given by the adjustment he proposes, nothing obliges us to infer from the second element of the sentence anything more than a mere possibility, with no necessarily close bearing on the real world, a virtual consumption implicitly seen as particularly high, which nothing suggests represented real consumption. This is exactly what is expressed in the Latin: expendi possent means there was a possibility but says nothing about that possibility ever being realized. There is no reason to doubt that tax grain was for sale in other cities where demand for it existed or that some was stockpiled for the following year. All in all, De Romanis’s interpretation is tantamount to eliminating the SHA text (Source 3) from the evidence relevant to Rome’s total consumption, if not from the evidence on the Empire’s stocks.[8]

Approach B

In recent years, differing estimates of the population of Rome have substantially converged; nowadays few scholars would dispute a figure in the range of 800,000-1,200,000. The difference between these two extremes (amounting to 50 per cent of the first of them) is still quite large. The most reliable figure does appear to be the number of recipients of the grain dole, which may be supposed to mean all adult male citizens, in other words 320,000, until Caesar revised the lists in 46 bc. One question this raises is whether the number of citizens of Rome was not particularly inflated at that time, the grain dole itself being the very reason for this, whether many non-citizens might not have infiltrated the ranks of those entitled to grain, in short whether the same figure turns up under Augustus. This last point depends on our knowing who were the 320,000 recipients of a congiarium in 5 bc (only adult males, or all males including children?), and on the parallel question whether the 200,000 who received public grain in 2 bc represented the whole population of plebs or only a part of it previously limited by Augustus.[9] This is a significant difference. The next thing is to imagine the structure of Rome’s population, which probably had a large preponderance of men, given the preference for males at birth, a consequence of the accepted practice of exposure of newborn children.[10] Then we would require an average ration of grain per inhabitant;

and, on that matter, too, there are wide differences of opinion: P. Garnsey sets the minimum calories for the Roman population at 1,745 calories per day, which would not be very nourishing;[11] L. Cracco Ruggini arrives at a more reasonable figure of 2,500 calories per day;[12] and, as we have seen, De Romanis would set it a fair bit higher.[13]

We do have evidence on grain consumption in the Roman world, but not for Rome itself, if we accept that the grain dole of 5 modii per month was not intended solely for individuals. In the second century bc, a Roman soldier’s ration was 3 modii per month.[14] At roughly the same period, Cato’s slaves were fed according to the work they did: 4-4.5 modii for labourers; 3 for foremen, overseers, and shepherds.[15] During the Empire, the traditional ration for a labourer in Egypt was 1 artaba, equivalent to 39.5 litres or about 4.5 modii.[16] But was this meant only for his personal consumption? Such relatively consistent data show the disparity there could be among different rations for different people. The type of work they did is not the only point to consider; there were possible differences in other sources of food to which they might have had access. However, a monthly ration of 4-4.5 modii, corresponding to 3,100-3,500 calories per day, does seem a plausible amount for adult male labourers, who were the greatest consumers of grain.

Clearly, this figure cannot be extrapolated to a whole population. Here, too, though, any calculation of average required calories would also require some knowledge of how the population was structured. And one would have to take account, as Lo Cascio does,17 of other foods consumed. In fact, estimates proposed for Rome’s grain requirements in recent years derive more from impressions of what seems plausible (while being careful not to depart too much from Beloch’s canonical 36 modii per year) than from calculations that, if taken to their logical result, might merely mask the same uncertainty.18 Nevertheless, I transcribe in Table 10.1, for the interest of readers, two modern estimates of what constitutes a desirable level of nourishment, one from a French organization, the other from India. If we assume, as I do, that there was an imbalance in the population (females 40 per cent, males 60 per cent), with adults accounting for about 65 per cent of it, and if we take the highest Indian figures, we arrive at a potential total consumption of up to 2,900 calories per day. This falls close enough to the oft-quoted canonical figure of a litre of grain per person per day, giving roughly 2,700 calories. But it also adds up to 42 modii per annum,

Table 10.1. Recommended daily calories, according to European and Indian research

Fondation frarnjaise pour la nutrition Indian Council of Medical Research (1988) (1981)






Adult male


Adult male

Very active (heavy) Moderately active Sedentary

  • 3,800
  • 2,875
  • 2,425

Adult female


Adult female

Very active Moderately active Sedentary

  • 2,825
  • 2,225
  • 1,875

Child 1-3 years


Child 1-3 years

Child 4-6 years


Child 4-6 years


Child 7-9 years


Child 7-9 years


Child 10-12 years

2,600 (M) 2,350 (F)

Child 10-12 years Child 13-15 years

2,190 (M) 1,970 (F) 2,450 (M) 2,060 (F)

Child 12-19 years

3,000 (M) 2,400 (f)

Child 16-18 years

2,640 (M) 2,060 (F)

which is significantly higher than the figures given by Beloch, Garnsey, and Lo Cascio, and somewhat higher than Rickman’s and Cracco Ruggini’s.

In an area fraught with such indeterminate variables (which we would be mistaken to assume had remained unchanged over a period of two and a half centuries), I am a great believer in using ranges of figures. That way, we can be almost certain of bracketing the target of Rome’s grain consumption. At the lower end of the scale, there is Garnsey’s calculation of consumption and Beloch’s cautious estimate of population at 800,000, giving altogether about 24 million modii per annum (meaning some 170,000 tons). And, at the other end, we have a calculation based on the 42 modii per annum, which I have just considered to be a maximum, and on a high population in the order of 1,200,000. This would give about 50 million modii per annum—that is to say, more than double the other figure. Wide as this discrepancy may be, we can still work with such evidence.

  • [1] One modius = approximately 8.8 litres, or 6.5-7 kilograms, of grain.
  • [2] See discussion in Rickman (1980b: 231-5).
  • [3] De Romanis (1996c). This interpretation was accepted and defended by Virlouvet (1995b: n.27. [De Romanis reprinted his demonstration (2003a: 700-2); and his translation of то ката т-qv'? ш^-qv пХцво; should be seen as authoritative.]
  • [4] Piganiol (1972: 8-9, n. 4). 2 Sirks (1991b). 3 Ep. 80, 7.
  • [5] 7 It should be noted that Duncan-Jones (1974: 147), with a different calculation based on the
  • [6] calorific value of bread, arrives at 3,000-3,500 calories per day, which he sees as close to ‘modern
  • [7] ideals of 3,300 calories per day for male adults'.
  • [8] As for the scholium to Lucan, De Romanis suspects it might have derived from the SHAmanuscripts.
  • [9] Lo Cascio (1997) prefers interpretations resulting in the lowest figures, Virlouvet (1995a)those resulting in the highest figures.
  • [10] See the important article by Harris (1994).
  • [11] Garnsey (1983; 1988:231). 2 Cracco Ruggini (1985).
  • [12] 13 On these difficulties, see Foxhall and Forbes (1982); Gallant (1991: 62-8).
  • [13] 14 Polybius (VI. 39, 12-14) puts it differently: two-thirds of a medimnus per foot soldier.
  • [14] Depending on what capacity is accepted for the medimnus, conversions vary; some say 4 modii.Walbank (1957: i. 722) I find reliable.
  • [15] Cato, Agr.56. 6 Cuvigny (1996: 141). 17 Lo Cascio (1990: 232-8).
  • [16] 18 See Rickman (1980b: 10); Garnsey (1983: 118 and n. 3); Cracco Ruggini (1985: 229).
< Prev   CONTENTS   Source   Next >