As an addendum on to our four-part look at the general structures of the farming of cereal grains (I, II, III, IV) this post is going to briefly discuss some of the key ways that the structures of rice farming differ from the structures of wheat and barley farming. We’ll start with some of the key differences in the mechanics of rice farming itself, before moving into how those differences might motivate different social and economic organization in the countryside, before finally discussing some of the theories as to how rice farming might – or might not – impact larger structures like the state.
As always, if you like what you are reading here, please share it. If you want updates whenever a new post appears, you can click the button below for email updates or follow me on twitter (@BretDevereaux) for updates as to new posts, as well as my occasional ancient history, foreign policy or pop-culture thoughts. And if you want to ensure that I can continue to get my rice, you can support me on Patreon.
(Quick bibliographic aside: I relied primary for this on Hsu, Han Agriculture: The Formation of Early Chinese Agrarian Economy (206 B.C. – A.D. 220) (1980) and F. Bray, The Rice Economies: Technology and Development in Asian Societies (1986) which were recommended to me by specialists in the field. The latter is a wealth of technical details on rice cultivation, although it is as focused on the transition to mechanization and modern agriculture as to the conditions of pre-modern rice cultivation.)
We want to start with the rice cultivation process. There are a lot of varieties of rice out there, but the key divide we want to make early is between dry-rice and wet-rice. When we’re talking about ‘rice cultures’ or ‘rice agriculture,’ generally, we mean wet-rice farming, where the rice is partially submerged during its growing. Wild rice, as far as we can tell, began as a swamp-grass and thus likes to have quite a lot of water around, although precisely controlling the water availability can lead the rice to be a lot more productive than it would be in its natural habitat. While there are varieties of rice which can be (and are) farmed ‘dry’ (that is, in unflooded fields much like wheat and barley are farmed), the vast majority of rice farming is ‘wet.’ As with grains, this is not merely a matter of different methods of farming, but of different varieties of rice that have been adapted to that farming; varieties of dry-rice and wet-rice have been selectively bred over millennia to perform best in those environments.
Wet-rice is farmed in paddies, small fields (often very small – some Chinese agronomists write that the ideal size for an individual rice field is around 0.1 hectare, which is just 0.24 acres) surrounded by low ‘bunds’ (small earthwork walls or dykes) to keep in the water, typically around two feet high. Because controlling the water level is crucial, rice paddies must be very precisely flat, leading to even relatively gentle slopes often being terraced to create a series of flat fields. Each of these rice paddies (and there will be many because they are so small) are then connected by irrigation canals which channel and control the water in what is often a quite complex system.
The exact timing of rice production is more complex than wheat because a single paddy often sees two crops in a year and the exact planting times vary between areas; one common cycle on the Yangtze is for a February planting (with a June harvest) followed by a June planting (with a November harvest). In other areas, paddies planted with rice during the first planting might be drained and sown with a different plant entirely (sometimes including wheat) in the intervening time.
The cycle runs thusly: after the heavy rains of the monsoons (if available), the field is tilled (or plowed, but as we’ll see, manual tillage is often more common). The seed is then sown (or transplanted) and the field is, using the irrigation system, lightly flooded, so that the young seedlings grow in standing water. Sometimes the seed is initially planted in a dedicated seed-bed and then transferred to the field, rather than being sown there directly; doing so has a positive impact on yields, but is substantially more labor intensive. The water level is raised as the plant grows; agian this is labor intensive, but increases yields. Just before the harvest the fields are drained out and allowed to dry out, before the crop is harvested and then goes into processing.
Rice is threshed much like grain (more often manually threshed and generally not threshed with flails) to release the seeds, the individual rice grains, from the plant. That is going to free the endosperm of the speed, along with a hull around it and a layer of bran between the two. Hulling was traditionally done by hand-pounding, which frees the seed from the hull, leaving just the endosperm and some of the bran; this is how you get brown rice, which is essentially ‘whole-grain’ rice. While it is generally less tasty, the bran actually has quite a lot of nutrients not present in the calorie-rich endosperm. Whereas white rice is produced by then milling or polishing away the bran to produce a pure, white kernal of the endosperm; it is very tasty, but lacks many of the vitamins that brown rice has.
Consequently, while a diet of mostly brown rice can be healthy, a diet overwhelmingly of white rice leads to Thiamine deficiency, known colloquially as beriberi. My impression from the literature is that this wasn’t as much an issue prior to the introduction of mechanical milling processes for rice. Mechanical milling made producing white rice in quantity cheap and so it came to dominate the diet to the exclusion of brown rice, producing negative health effects for the poor who could not afford to supplement their rice-and-millet diet with other foods, or for soldiers whose ration was in rice. But prior to that mechanical milling, brown rice was all that was available for the poor, which in turn meant less Thiamine deficiency among the lower classes of society.
Because rice is such a different crop than wheat or barley, there are a lot of differences in the way that rice cultivation shapes the countryside. We’ll move here from the relatively direct impacts on the organization of farmers and then discuss the more speculative impacts on the organization of whole societies.
The thing to note about rice is that it is both much more productive on a per-acre basis than wheat or barley, but also much more labor intensive; it also relies on different forms of capital to be productive. Whole-grain wheat and brown rice have similar calorie and nutritional value (brown rice is somewhat better in most categories) on a unit-weight basis (so, per pound or ton), but the yield difference is fairly large: rice is typically around (very roughly) 50% more productive per acre than wheat. Moreover, rice plants have a more favorable ratio of seeds-to-plants, meaning that the demand to put away seeds for the next harvest is easier – whereas crop-to-seed ratios on pre-modern wheat range from 3:1 to 10:1, rice can achieve figures as high as 100:1. As a result, not only is the gross yield higher (that is, more tons of seed per field) but a lower percentage of that seed has to be saved for the next planting.
At the same time, the irrigation demands for effective production of wet-rice requires a lot of labor to build and maintain. Fields need to be flooded and drained; in some cases (particularly pre-modern terrace farming) this may involve moving the water manually, in buckets, from lower fields to higher ones. Irrigation canals connecting paddies can make this job somewhat easier, as can bucket-lifts (there’s a picture of a simple one below), but that still demands moving quite a lot of water. In any irrigation system, the bunds need to be maintained and the water level carefully controlled, with also involves potentially quite a lot of labor.
The consequence of all of this is that while the rice farming household seems to be roughly the same size as the wheat-farming household (that is, an extended family unit of variable size, but typically around 8 or so members), the farm is much smaller, with common household farm sizes, even in the modern period, clustering around 1 hectare (2.47 acres) in comparison to the standard household wheat farms clustered around 4-6 acres (which, you may note with the yield figures above, lands us right back at around the same subsistence standard).
Moreover, rice cultivation is less soil dependent (but more water dependent) because wet-rice farming both encourages nitrogen fixation in the soil (maintaining the fertility of it generally without expensive manure use) and because rice farming leads naturally to a process known as pozdolisation, slowly converting the underlying soil over a few years to a set of characteristics which are more favorable for more rice cultivation. So whereas with wheat cultivation, where you often have clumps of marginal land (soil that is too wet, too dry, too rocky, too acidic, too uneven, too heavily forested, and so on), rice cultivation tends to be able to make use of almost any land where there is sufficient water (although terracing may be needed to level out the land). The reliance on the rice itself to ‘terraform’ its own fields does mean that new rice fields tended to under-produce for the first few years.
The result of this, so far as I can tell, is that in well-watered areas, like much of South China, the human landscape that is created by pre-modern rice cultivation is both more dense and more uniform in its density; large zones of very dense rice cultivation rather than pockets of villages separated by sparsely inhabited forests or pasture. Indeed, pasture in particular seems in most cases almost entirely pushed out by rice cultivation. That has very significant implications for warfare and I have to admit that in reading about rice farming for this post, I had one of those “oh!” moments of sudden understanding – in this case, how armies in pre-modern China could be so large and achieve such massive local concentrations. But as we’ve discussed, the size of an army is mainly constrained by logistics and the key factor here is the ability to forage food locally, which is in turn a product of local population density. If you effectively double (or more!) the population density, the maximum size of a local army also dramatically increases (and at the same time, a society which is even more concentrated around rivers is also likelier to allow for riverine logistics, which further improves the logistical situation for mass armies).
But it also goes to the difficulty many Chinese states experienced in maintaining large and effective cavalry arms without becoming reliant on Steppe peoples for horses. Unlike Europe or the Near East, where there are spots of good horse country here and there, often less suited to intensive wheat cultivation, most horse-pasturage in the rice-farming zone could have – and was – turned over to far more productive rice cultivation. Indeed, rice cultivation seems to have been so productive and suitable to a sufficient range of lands that it could push out a lot of other kinds of land-use, somewhat flattening the ‘ideal city‘ model that assumed wheat and barley cultivation.
Moving to the next level of abstraction, pre-modern rice cultivation also relies to different degrees on labor as compared to capital inputs (like work animals) compared to wheat cultivation, which in turn seems to lead to different patterns in the status of farmers and their relationship to large landholders (I say seems because I’d like to see more studies, but I should note that what I am laying out here is a key argument of Bray, op. cit.).
We’ve discussed how wheat (and barley) cultivation is pretty dependent on the availability of work animals to draw plows and provide manure, both of which can significantly increase the productivity of the farm. As discussed above, rice cultivation, while it can benefit from fertilizer (especially more modern ‘fast growing’ rice varieties), but isn’t nearly as dependent on manuring as intensive wheat cultivation can be. At the same time, deep plowing could actually be counter-productive in a rice paddy, because the plow could break up or cut through the hardpan at the base of the soil which prevents the water from simply draining out into the water-table. Consequently, while effective plows emerge in China in the fourth century AD, they tend to be shallow and thus less reliant on animal traction (which consequently caused Westerners seeing Chinese agriculture to assume it was ‘primitive,’ which seems more than a little silly given that those shallow plows were managing higher yields than contemporary pre-modern or early-modern Western agriculture); hoe-farming is also common, especially because the extensiveness of rice farming discussed above often means that pasture for traction animals was scarce, but manpower was abundant.
On top of this, while both rice and cereal grains have to be threshed and winnowed in fairly similar processes, the subsequent processing is quite different. Where wheat and barley are milled down to a powder (flour), rice is generally pounded and sometimes polished. Rice processing of this sort seems generally to have been done with much smaller rice pounders made of common materials available to poorer farmers. Consequently while the pre-modern wheat farmer might be reliant on a mill owned by a local miller, large landholder or even the state, the rice farmer was less reliant on this form of outside capital.
At the same time, the high productivity of rice production meant that labor was relatively abundant in the countryside, as households could subsist on much smaller farmers. But this in turn seems to have shaped the development of rice farming techniques. Wheat- and barley-farming technologies (plows, mills, mechanization) tend to act as labor substitutes, allowing a given amount of labor to farm more land and thus produce more food. In effect, the goal here is to substitute scarce labor and sometimes scarce land with capital. In contrast, increasingly sophisticated rice-farming technologies (and I do want to stress – rice farming gets quite sophisticated in the pre-modern, especially with complex irrigation systems) focused on ways that more labor can be employed to produce more food out of the same amount of land and capital.
As a consequence of this, Bray argues, we see different patterns of relationships between the landholding elites and the small farmers. Because the small rice farmer was less reliant on the capital of the large landholder (in the form of plow-teams, manure and mills), while the large landholder was more reliant on the labor of the small farmer (because rice cultivation was so labor intensive), rice farmers tended to get a somewhat better deal. While we see the same range of farming relationships, from smallholding free-farmers to non-free farmers, Bray notes that systems of tenancy dominate the rice-farming countryside in the great majority of places where rice cultivation has long predominated. That’s not to say that being a small rice farmer was fantastic – fewer rice farmers seem to have been serfs or slaves, but at the same time, fewer rice farmers seem to have owned their own farms. And landlords in rice economies were still in a commanding position, in no small part because all of that labor went into what were essentially capital improvements (irrigation and the paddies) to land that they, not the farmer, owned.
Nevertheless, tenancy conditions tended to be more favorable in rice-farming areas than in wheat-farming ones, with a lower portion of the total harvest going to the landlord. Thus the irony that precisely because labor was so abundant, rice farming tended towards labor-intensive methods and solutions, which in turn improved returns to labor (compared to returns to capital), putting the small farmers, despite their abundance, in a marginally better bargaining position.
Now we reach the more speculative end of the discussion and I know that at least some of you reading that section title cringed a little bit. Let’s indulge in some intellectual history (or the fancy term ‘historiography’ which is the history of the history, as it were).
Writing in 1957, Karl August Wittfogel published Oriental Despotism, which argued that the common thread between ‘oriental’ states (which was to include Egypt, Mesopotamia, Alexander’s Empire, the Abbasids, all of freakin’ China, among others) was that they were systems of “total” state power and control (thus ‘despotic’) and that this total power was a consequence of “hydraulic despotism” with certain definable characteristics. Moreover, he argued that these systems of power were functionally impervious to internal disruption or fragmentation (only to outside invasion). it is a basic thesis that still seems to crop up from time to time in the popular imagination and even among intellectuals who should know better. Naturally, the assumption here is that if irrigation-oriented agriculture leads to ‘hydraulic despotism’ then rice (particularly wet-rice, being by far the most common and most productive form), the quintessential irrigation-only-crop, ought also to be associated with ‘despotism.’
This thesis, influential in its day, has not stood up well at all to improved study of many of these places. Perhaps most notably, scholarship on almost all of the supposed ‘hydraulic despotisms’ has tended to draw their despotic nature into question. For one example, Achaemenid rule, often presented as the paradigmatic example of ‘Eastern Despotism’ (which is certainly how the Greeks viewed it, but no one thinks well of their imperial neighbors), on closer inspection seems if anything to have allowed quite a significant degree of local autonomy. The Great King was often a less ‘despotic’ overlord than the Athenian Assembly, if one had to choose who to be a subject to. Likewise, the study of pre-Achaemenid economic systems in Egypt and Mesopotamia over the last 50 years have tended to peel back the idea that the state dominated the economy and society to the exclusion of all other factors; in many ways this is a steady series of victories for the archaeologists whose efforts allow us to ‘see’ the commoners and their activities more clearly. And in almost all of the supposed ‘hydraulic’ societies, the last several decades have seen quiet revolutions in the scholarship brought on by increased exposure to the local sources, rather than reliance on what European-language sources said about these societies from the outside.
Moreover, our better understanding of the chronology of these states has tended to undermine Wittfogel’s assertion that state formation itself was caused by the demands of canal-building and irrigation (I discuss what I consider a far more compelling model of state-formation-through-agrarian-warfare here). And the supposition that supposed hydraulic empires never crack up from the inside or fragment is one that can only be made at great remove; close inspection of almost any of these societies show them as doing exactly that thing, often several times. China, in particular, is both Wittfogel’s key paradigmatic example and also a real problem for his model: no dominating established priesthood, frequent peasant rebellions and multiple periods of significant fragmentation. And to top it off, not clearly ‘despotic’ in any particular sense (in many cases the supposed essential ‘despotism’ of China relies on assuming the great antiquity of things which were peculiar to the Qing Dynasty – themselves not an indigenous ‘hydraulic’ dynasty but an ethnically distinct foreign one – or the modern Chinese state). Thus key parts of Wittfogel’s argument about supposed ‘hydraulic societies’ engaging in ‘the Asiatic mode of production’ (as an aside – any theory which asserts an ‘Asiatic’ mode of anything is pretty much guaranteed to be nonsense; Asia is, as veteran Risk players will know, very big and consequently extremely diverse) seem more premised on outdated exoticizing stereotypes about ‘despotic easterners’ and less on the actual history of the many societies of Asia.
Which isn’t to say that I think the changes brought on by rice cultivation have no effect at all. Once we dispense with simplistic, outdated stereotypes about ‘despotism’ (which has more to do with state behavior than state capacity), it is still possible to note that rice cultivation can create areas of much higher population density with a greater degree of agricultural surplus. For pre-modern states, which are essentially extractive ventures that feed off of the surplus of the countryside, it is not hard to see how a compact, effectively immobile surplus-generating population positioned along often navigable waterways would provide a very strong starting ‘base’ compared to a more diffused population (as in much of Europe). Though even that needs caveats, given that the earliest centralized governments in China emerged not in the rice-farming south, but in the wheat-farming north.
The issue may be less about control of water than about density, distance and communication; we’ve already seen the scale of difficulties of trying to move over a large land-based empire (unless you are Steppe nomads with a completely different subsistence system). It’s hard not to notice how, even after periods of fragmentation, states in places like Egypt, Mesopotamia and China keep forming around the large, navigable rivers that bind the region together (but note that while sloppy scholarship often presents this as if the same state is reemerging over and over again, these are in fact often very different states occupying the same place).
It does seem to be the case, at least from my limited reading, that rice cultivation does tend to shift power away from the large landholding class as compared to wheat farming. Sometimes that power went to the officials of the state who might be directly managing large-scale irrigation and water networks; sometimes, as Bray notes, it went down to the farmers, who might be communally managing water supply (often through long-established understandings with local elites which carried the force of tradition and custom) and whose labor commanded a greater portion of the value of their production. Rice cultivation may contribute to more centralized states (which is not the same this as more despotic ones), but honestly I haven’t seen what I’d consider ‘slam dunk’ scholarship on this point yet.
All of which is to say that while there are some quite importance differences in pre-modern subsistence agriculture with rice rather than with wheat, many of the basic patterns reassert themselves. Poor farmers, often trapped in tenancy or effective debt peonage are common, they appear to frequently rely heavily on horizontal social networks and vertical ties instead of heavy involvement in the market economy as a way of minimizing risk and the living standards, judged by nutrition, are broadly similar. I haven’t gotten into it here, but even in economies where rice dominates, farmers diversify both between different crops and even different varieties of rice to minimize risk, just as our wheat-farmers do.