Accurate Mixed Use Farming for ACKS
Despite Mr. Macris’s best efforts, the mixed use farming families depicted in AXIOMS 3 and the ACKS II Judge’s Journal do not perfectly reflect the hardscrabble reality of premodern farming. This essay is an attempt to create a system that more closely matches reality and is useable for ACKS domain-level play. No criticism of Mr. Macris is implied by these proposed revisions, as this work heavily leans on his initial work. The primary secondary source used to reality-check the numbers herein is Professor Devereaux’s A Collection of Unmitigated Pedantry blog, specifically his multi-part series on farming. Prof Devereaux, in turn, relies heavily on Columella’s De re rustica. As Mr. Macris also uses Columella as a source, it is interesting, from a historiographical perspective, to note where the two diverge.
How Much Wheat Per Day?
First of all, we must determine the “poverty floor” below which a family is actually starving. This is nontrivial. If we use ACKS’s standard assumption of a 5-person family, we must feed an adult male 3,500 calories (sailors in the Age of Sail got closer to 4,200, but they did not have seasonal slack periods), an adult female 2,500 (a totally sedentary women generally needs 2,000, but a pseudo-Roman peasant matron would not be sedentary), a senior woman 2,000 (the mother of the adult male, effectively sedentary but contributing to the household production, as will be seen), and two children, one ~6 and eating 1,500 and one 12-15 and eating 2,500 (these numbers are very rough, and taken in part from this source, as well as Prof D). Total daily food consumption: 12,000 food calories, slightly higher than Mr. Macris’s estimate of 11,000. While I do not pretend to know how Mr. Macris generated his estimate, I suspect he budgeted for an infant eating 1,000 calories instead of the widow eating 2,000, however, a family in that stage of life (the widow still married, and contributing to a different household’s needs), would also likely have more than 3 children (say, an 8 year old, a 6 year old, a 4 year old, a 2 year old, and an infant; the 6 and 4 year olds and the infant dying later), bringing back up the food requirement. These jumbo-families would be balanced by older families of 2 adults (the widow having died) and one child of ~14, and younger still families of 2 adults and 3 under-5 children. Each family would have different productive capabilities as well, which is why I’m sticking with the 12,000 calories per family idealized number (12 is a nice, divisible number. Sue me).
Now, how does the family fulfill its calories need? ACKS II suggests that they consume >95% of their calories as wheat bread. That, bluntly, is absurd. Substantially greater amounts of barley could be grown on similar soils; in addition, barley would be grown where wheat could not and required less labor. Likewise, cabbages, onions, turnips, rapeseed (put by Pliny the Elder as third most important plant after ”corn” and “bean”, source), chickpeas, lupine, peas, lentils, kidney-beans, spelt, millet, leeks, and oats were all known and grown by the Romans, and all (it seems) were cheaper per calorie than wheat. Wheat was a luxury good. That being said, it was one affordable to the masses, and consumed in substantial quantity. It is important to note that, just because the majority of peasants’ money was spent on food, it does not follow that peasants were only ever inches away from starvation (people, even poor people, like good food!). ACKS does not even attempt to calculate the wholesale price of any non-wheat food except chickpeas, which have a price set to be the same, by weight, as wheat. However, I cannot be too critical of this decision, as ancient sources are notoriously concerned with what nobles did (and ate), and strongly incline, when a price is given at all, give one of a foodstuff delivered to troops (usually wheat, since transportation costs make the “wheat premium” a smaller fraction of the total), or grown for sale on latifundae (again, usually wheat, for similar reasons). A insight into the relative cheapness of barley can be seen in Revelation 6:6, valuing barley at 1/3rd the price of wheat, although this literary reference is meant to illustrate a severe famine and, in such conditions, barley is likely in increase less in price than wheat (due to its relative hardiness as a crop). Medieval sources, meanwhile, give barley prices in England of a little over half of contemporaneous wheat prices, beans as similar, oats at less than half, peas at half, and rye at 3/4s wheat’s price. For the purposes of this paper, I will set 1 quarter of “barley” at 2 gp.
The rarity of wheat can partially be seen with the Roman annona (the “bread” of the “bread and circuses”). Professor Devereaux treats the subject at length here. Suffice to say: 5 modii per family per month. From here, citing Rickman’s The Corn Supply of Ancient Rome, we learn that the ~8.73 liter modius came out to around ~6.75 kilograms (dry weight is not volume, a point that the ACKS JJ p. 434 misses). As a kilogram of wheat contains roughly ~3400 calories per kilogram (source), we can determine that the Roman family received ~114,750 calories per month. Given our previous estimates (ACKS’s and mine) of 11,000-12,000 calories per day per family, we can see that the annona provided only 1/3rd of a Roman families’ caloric need. Unlike America’s modern SNAP program, the annona was not intended to supply even the minimum caloric need, but to supplement private purchase of other foodstuffs.
This assumes that the wheat would be eaten as a porridge, maximizing extraction of calories. Calculating the calories in a loaf of bread is more difficult. ACKS JJ p. 434 fails in this estimate in several ways, although its errors mostly cancel each other out. A bushel of wheat weighs 60 pounds, not 72, and flour must be milled before baking. Roughly 25-30% of the weight will be lost, but the remaining flour will now have an average of 3,600 calories per kilogram (source), due to the bran containing disproportionate amounts of fiber (and also micronutrients). As such, given that one imperial pound weighs 453.59237 grams, the bushel will produce around 20,000 grams of coarse flour (at 25% loss), containing ~72,000 calories. Notably, this is the same caloric value arrived at by ACKS, despite a vaguer method of derivation. If we use our estimate of 12,000 calories per family per day, we find that the bushel is enough to feed a family for 6 days, assuming that calories lost due to yeast fermentation are roughly equal to the increased bioavailability of calories due to the Maillard reaction (a giant assumption, but tracking down just this amount of data is exhausting). Professor Devereaux arrives at 1,746 liters or 49.5 bushels per Roman family of five, which, under our estimate of caloric need, would suffice for 297 days. Given that the Professor, unlike us, most likely is not assuming an all-wheat diet, but rather some supplementation, this is acceptably close to our estimate.
In conclusion, if the peasant family eats solely wheat bread, they must buy 5 bushels per month, at ACKS’s cost of 0.5gp per bushel, for a cost of living (not including clothing, to be addressed later, or firewood, likewise) of 2.5gp per month. ACKS, by contrast, arrives at a total of 2.6gp, including a very small amount of animal product. However, the amount of protein is absurdly low (375 calories per five-person family per day; as a comparison, American slaves in the antebellum south generally got around 3 pounds of salt pork per week per field slave, or ~8 oz per day, source). Likewise, the peasants would have eaten a substantial fraction of their calories as the above-mentioned legumes, barley, and root vegetables. Therefore, 2.5gp of foodstuffs per family per day can be viewed as an average, with daily caloric intake consisting of, say, ¼ wheat bread, ¼ barley porridge, ¼ miscellaneous vegetable/legume, and the remainder dairy, eggs, and some meat. Since animal products cost so much more than plant matter (a fact which ACKS models correctly), the average expenditure should be roughly similar.
Effort of Growing Wheat
ACKS’s estimates of the time that growing wheat requires are completely wrong. This is bizarre, because the “correct” measurements I will introduce come from Columella, who ACKS quotes extensively in the section on olives and vineyards. Per ACKS, growing an acre of wheat requires 1.1 days plowing, 1.5 days reaping, 1.4 days harvesting, and 1 day threshing, for a total of 5 days per acre. None of this is correct. For a start, “reaping” and “harvesting” are synonyms. More pertinently, Columella gives 10.5 days of labor per iugerum, or 17.5 days per acre, not including threshing (source and source). The confusion seems to arise in part over the definition of an acre; one acre could be plowed by one man in one day, with a team of eight oxen, not just one ox (source).
Wheat Yield Per Acre
ACKS makes a notable mistake in its discussion of seed ratios: seed ratios, conventionally, are presented in the form X:1, where X is the number of seeds harvested, and 1 is the number of seeds sown. ACKS, however, refers to a 4:1 seed ratio as 2 bushels of seeds producing 10 bushels, with 2 bushels of seed kept back for the next year, resulting in a net yield of 4 seeds eatable for each 2 sown. The correct notation for this pattern of yield is 5:1.
Actual seed ratios are notably variable in our sources. One estimate gives a typical Roman yield of 5-6:1, in keeping with ACKS’s estimate. However, this arises from literary sources, many of them offhanded or lacking in rigor. Columella, for his part, gives a 4:1 seed ratio; Columella is highly pessimistic of wheat production, but this estimate tracks well with estimates from detailed records remaining from Middle Ages abbeys (source, gives 4.34:1; additional source). Columella gives 4-5 modii sown per iugerum, or approximately one bushel. However, the iugerum was only roughly 0.6 of an acre. Therefore, if we assume a 30-acre field (one Medieval English Virgate), we would sow 50 bushels and, at a 4:1 seed ratio, reap 200, for a net yield of 150 bushels or 5 bushels per acre. Our medieval sources seem to indicate gross yields of around 7-15 bushels per acre (vide supra), so Columella is indeed a pessimist when it comes to wheat, but not unreasonably so. ACKS assumes a high density of seed sown and higher seed ratio, but not outside the bounds of historical authenticity. For the purposes of this paper, we will leave ACKS assumptions as-is.
How Much Barley?
ACKS does not address in its simplified mixed-use farming any crop other than wheat. In my view, this is a mistake. While the three-field crop rotation system is usually associated with the later Roman period, Revelations 6:6 seems to indicate the common man viewing wheat and barley prices with similar salience, implying similar production quantities, while Columella discusses at length both “middle” crops such as barley and oats, and cover crops such as vetch and clover, a three-field system seems an appropriate approximation. This can be further seen because, at 17.5 labor-days per acre, 20 acres of wheat would take 350 days of labor, straining the man of the family to his limit. However, per Columella, barley takes only 6.5 days of labor per iugera, or roughly 10 days per acre. 10 acres of wheat and 10 acres of barley would take 275 days to manage, a more reasonable workload, leaving room for other miscellaneous tasks.
Additionally, because the planting and harvest seasons of wheat and barley are not quite the same, the “crunch” around these seasons would be lessened. Further, seeding 2/3rd of a set of fields with wheat will rapidly deplete the soil. Historical examples typically cropped half-and-half, or 1/3rd 1/3rd 1/3rd, the middle crop being a less nutrient-intensive crop such as barley.
As for barley yield, a similar amount of seed is planted, so I will assume an equal output (barley can produce more seed than wheat, but this barley is being planted in the exhausted soil, after the wheat, to absorb the last of the nutrients before the cover crop regenerates them). At the 2gp per quarter (see above), barley produces slightly less income per man-day than wheat (2.3 silver per day for wheat, 2 silver per day for barley) but is more resilient to crop failure. Sowing the fields half fallow, half wheat would yield similar cash value, slightly less labor, but substantially fewer calories and would compress the harvest season very tightly. It would also be more exposed to crop failure due to weather, as wheat has higher variance in yields.
Threshing
Columella, in the above sources, only counts the labor needed to bring the grain to the threshing floor, not the labor of threshing. Some exploration by myself shows that wheat requires approximately one day of labor per quarter (source) and barley one day per two quarters (source, sadly uncited Wikipedia; precise scholarship is lacking, but it is noted often that barley is easier to thresh by an unquantified amount). As the farm yields 10 quarters of each, this will take another 15 days, for a running total of 290 days
Cover Crops
The standard ACKS mixed-use farm assumes no farm labor is spent on the land that is left fallow, accounting only for the farm labor spent rearing and harvesting the animals pastured there. However, this is not accurate. Previous-ploughed farmland is generally planted with a cover crop; this crop is later reaped and the hay stored, with the hay being fed to animals and the stubble grazed. Additionally, the animals would be fed the hay from the grain producing fields and allowed to graze their stubble. ACKS handles these issues well implicitly, but, as mentioned, fails to account for the effort growing the cover crop. Columella (ibid) gives 3 to 4 days of labor per iugerum of vetch, lupine, or clover, and an anomalously high 6 days per iugerum of chickpea vetch (this crop is noted among moderns for its quality as a feed; perhaps this accounts for its continued use despite its high labor cost?). We will simplify this to 5 days of labor per acre of cover crop. This adds another 50 days of labor to the labor in the fields, bringing the total to 340 days of labor.
How Much Labor Per Year?
This feeds into a larger question, which is the above: how many days of labor would peasants spend per year? ACKS sets a total of 500 per family, which is too low. This is an understandable error; memes of a supposed medieval 150-day work year infest the Internet. However, they are untrue (source, source, source). To summarize: 150 days was the labor owed to the feudal lord, the equivalent of ACKS’s Service Revenue. We know from medieval sources that a peasant male would work for an equal or greater amount of time for himself, plus the labor of his wife and other family members (recall our discussion on the multigenerational nature of the peasant family). So, the question stands: how much time? Columella is oddly little help. It is often reported that Romans had holidays for one-third of the year, however, not all families would practice all holidays, and Columella notes that many sorts of labor (e.g. making candles, spreading dung, etc.) were licit on holidays (De re rustica, 2-22). As such, these should be interpreted as perhaps half-days. The Judeo-Christian notion of regular days of total cessation of work should not be imported into our understanding of Roman holidays! We can estimate perhaps 300 working days per worker per year, either trusting the middle source above, doubling the 150 days of corvee labor, or taking every seventh day off and adding two weeks of holidays.
This in turn forces us to grapple with “how many workers per family?” In this case, ACKS II’s rules for slavery give us insight, as one needs three slave laborers to equal one peasant family. This tracks with our initial discussion of family structure above, consisting of husband and wife, and either an elderly mother or teenage child. As such, we assume 900 labor days per family per year.
It should be noted that our above estimate requires 340 days of labor in the fields. This is not a deviation from our assumptions, but rather, a result of the need to bring in extra labor for the harvest. Women and children would participate in the harvest; this is present in all our sources.
Animal Population
The idealized 30-acre farm in ACKS has only one ox. I argue that it should have two. A medieval English oxgang was fifteen acres, less than the 20 acres under the plough. Further, because the oxgang, virgate, and carucate were used as units of tax estimates, there is reason to believe that they are conceptualized as the total fields own by the farmer with that many oxen, including land under cover crops. That is, a virgate is defined by the number of oxen, but not the crop planted. I don’t have a source for this, just my intuition. However, it is backed up by one ox being plainly not enough to plow all of ACKS’s 20 acres, let alone the cover crop, so two oxen it is.
This reduces the families’ cows to 2 from 3. This saves 5 days of labor from the cow (the labor tending of the oxen being subsumed in the estimates for the labor of the fields the oxen plow) and reduces income by 10.41gp per year. Note that I am accepting ACKS’s numbers on labor per animal and revenue per animal unquestioningly; they feel low and high respectively to me, but I am not qualified to critique them.
Price of Wheat Per Workday
We must now establish the wage-rate that the peasant family will earn in their remaining labor-days. Until now, we have not directly addressed the cost of labor, only the costs of various goods in relation to each other. ACKS establishes 1sp per labor-day, but is this plausible with our established price of 4gp per quarter of wheat? Obviously, this estimate arises from the proverbial denarius paid to hired hands in the Matthew 20:1-16, but this source must be used with care, since a) the context is likely one of harvesttime, when prices could rise substantially higher than normal, and b) the landowner is specifically a generous person, although the generosity is primary highlighter in his treatment of the later hires. Professor Devereaux is similarly cautious, noting that Athenian rowers were paid a drachma per day, but voted themselves that wage (and, I will note, faced a risk premium of death), and Roman Legionaries in the early imperial era were paid 225 denarii per year under Julius Caesar and “ten asses” (in the period, the originally ten-ass denarius was valued at 16 asses; a sestertius was 4 asses) per day by the time of Tacitus (Annales, I, 17).
In a prior work, I went into more depth on this, and my conclusion was that 4 modii equals 8 asses equal one bushel of wheat. Roman soldiers seem to be underpaid (rather, paid with the glory of Rome), with the Praetorians paid a more typical wage. With feedback from others on the Autarch Forums (same thread), I feel comfortable agreeing with ACKS’s definition of 1sp per day unskilled labor.
Gross Annual Revenue
Thus, we can at last calculate the annual income of the farming family. They earn 4gp for each of the 10 acres of wheat, plus 2gp per each of their 10 acres of barley, for a total grain income of 60 gp. They earn an\ additional 72.5gp from the animals raised on the farm (see ACKS II JJ p. 434 & p. 436, as well as the Animal Population section above). This costs 340 days of labor in the fields and 55.5 days of labor with the animals (ibid). The family has 505 labor days remaining (the odd remainer having been rounded up in the spirit of generosity). At 1 sp per day, that earns them 50.5gp. This is likely in the form of the women of the household spinning and weaving (mostly spinning) a total of 100 hours per complete change of clothes (source). In total, the family earns 183gp per year, or 15.25gp per month, slightly less than ACKS’s 16.25gp.
At least, before the landlord and taxman get involved.
Cost of Land Capital
We can estimate how much rent ought to be paid to the landlord by recalling from ACKS II RR p. 383, which defines “investing in civilized farms” as a Cautious investment that earns 0.5% monthly. This passes the sniff test, since the land is its own collateral and cannot lose value except by wildly reckless farming practices. Since civilized land costs 15gp per acre, and the farming family has 30 acres, they must pay 2.25gp to the landlord per month.
Cost of Livestock Capital
The total capital cost of the livestock on the farm can be calculated as follows: two oxen at 40gp each, total 80gp, one eighth of a standard herd of cattle (JJ p. 436) worth 120gp, or 15gp, two flocks of sheep worth 64gp each, for 128gp. Total cost of livestock: 328gp. Since the animals might die or be stolen, the risk here is higher, but not too high, as they are their own collateral. I assign it a risk of Balanced, 1% per month. The family must pay 3.28gp per month to the owner of the livestock they rent.
Cost of Operating Capital (seed)
The family must also pay for the cost of the seed they plant. The cost of the wheat seed is 1gp per acre, or 10gp, the cost of the barley seed is 0.5gp per acre, or 5gp, the cost of the clover-seed is incidental. Since the seed goes into the ground, and a bad harvest might lead to a total loss, this seems a classic example of a Risky investment, paying 3% per month. The family must pay 0.35gp per month.
Earnings After Capital Cost
Thus, the family earns 15.25-2.25-3.28-0.35=9.37gp per month. As it happens, this is approximately the income of a peasant family in ACKS that owns the land they work on (JJ p. 405). However, in our calculation, this is the before-tax income of the family.
Capital Cost of Slaves
Estimating the return on investment on slave ownership calculates for us the amount of income that free families receive over and above the minimum standard provide to slaves, as “free” peasants treated as poorly as slaves would surely quit. In ACKS, slaves cost 40 per and three are equivalent to one peasant family. They have an upkeep cost of 2gp per month per head, or 6gp per peasant family. Presumably, the extra two members of the standard peasant family come along via “stork” after purchase, or “120gp per family” is a cost that includes discounts for the younger slaves.
Slave-owning inherently comes with a high risk of runaways, but this is counterbalanced by (in societies with legal slavery) the risk to the slave from slavecatchers, which moderate the desire to escape. I thus estimate slavery as a “Risky” investment; one that must earn 3% per month to be worthwhile. Thus, we estimate that free peasants likely earn 3.6gp per month more than their enslaved brethren, or 9.6gp total. This is within sneezing distance of our after-landowner income estimate of 9.37gp, giving us reason to conclude that free peasants should earn an income in the 9-10gp range.
trimmed for size