The conventional view: quartering soldiers

Theodor Gaupp (Reference Gaupp1844) and Ferdinand Lot (Reference Lot1928) are credited with popularizing the idea that Roman laws for quartering soldiers provided the framework for settlement. Evidence comes from 5^th- and 6th-century law codes of the Visigothic and Burgundian kingdoms, and also from royal correspondence of the Ostrogothic kingdom.

Two important pieces of evidence supporting the conventional view are found in titles 54 and 55 of the early 6th-century Burgundian Code:

Similar references can be found in title 10 of the Visigothic Law (‘about the division of lands made between a Goth and a Roman[:] two parts of the Goth [from the] third of the Roman’) and title 277 of the (Visigothic) Code of Euric (‘Gothic allotments (sortes) and the third (tertia) of the Romans’); quoted in Goffart (Reference Goffart1980: 118–19).

We also have the testimony of Cassiodorus Senator, a Roman aristocrat who served as quaestor palatii, consul, and then magister officiorum under Theoderic the Great, Ostrogothic Italy's first king (493–526). A large number of Ostrogothic state papers composed by Cassiodorus – his Variae Epistolae – have come down to us (Cassiodorus, Reference Mommsen1884). This paper opens with an excerpt of Cassiodorus’ panegyric to Liberius:

Liberius was a patrician tasked with administering the Gothic sortes. Here we see an ostensible reference to each Goth receiving an allotment of one-third of a Roman's land.Footnote ⁵

Gaupp and Lott argued that references to hospitalitas and tertiae correspond to a law issued by the Eastern emperor Arcadius in 398 and found in the Theodosian Code (book 7, 8, 5):

If Arcadius’ law provided the framework for settlement, then barbarian sortes were a fixed fraction of the land of their Roman ‘hosts’.

Goffart's alternative: Late Roman tax law

There are problems with the conventional view of hospitalitas. The quartering of soldiers was a temporary expedient during a campaign; its correspondence to a perpetual occupation is tenuous (Sivan, Reference Sivan1987). Furthermore, Arcadius’ law specifies how a host's house (not land) is to be divided. This limitation of hospitalitas to a host's house is consistent with a different title (38) of the Burgundian Code, ‘Of the Refusal of Hospitality Toward Legates of Foreign Tribes and Travellers’: ‘Whoever refuses his roof or hearth to a guest on arrival, let him be fined three solidi for the neglect’ (Drew, Reference Drew1976: 47).

Walter Goffart (Reference Goffart1980: 42–5) emphasizes the irrelevance of Arcadius’ law to lands rather than houses; also the fact that the actual one-third of Arcadius’ law is consistent only with evidence regarding the Ostrogoths; not with that of the Visigoths or Burgundians. Goffart instead argues that barbarians were allotted units of tax assessment on Roman land.Footnote ⁷ He draws on evidence regarding the Ostrogothic settlement in Italy. For example, a certain Faustius complained on behalf of his district about the irregular timing of tax collections. Theoderic (through Cassiodorus) responded:

The above implies that a barbarian's tertia had to be collected from the land's possessor. Faustius was complaining that the barbarians’ tertiae and ‘ordinary tribute’ were being inconveniently collected at different times.

Goffart also emphasizes terminology that Theoderic uses in book 5, ch. 27 of the Variae (Cassiodorus, Reference Mommsen1884). The conventional translation of the relevant text is: ‘Order all the captains of thousands of Picenum and Samnium to come to our court, that we may bestow the wonted largesse on our Goths’ (book 5, ch. 27, loc. 4983; emphasis added). The ‘thousands’ is a translation of the Latin millenarios. Goffart (Reference Goffart1980: 80–7) argues that here millenarios is best translated as ‘holders of millenae’. Millena was a unit of Roman tax assessment dating back to Diocletian's reforms:

According to Goffart, Theoderic's Goths were holders of these tax assessment units: ‘[T]he state-designated recipient of the millena – either a person or an institution – could hereafter collect at source the tax payments of his “unit”’ (ibid., 82). A Goth receiving a sorte became essentially a tax farmer for his own account. He ‘bought’ the privilege through his military service in Theoderic's confederacy.

Goffart (Reference Goffart1980) also marshals evidence concerning barbarian settlements in Gaul. For example, historians generally agree that Visigoths were not taxed within their own kingdom (e.g. Heather, Reference Heather1998: 195; King, Reference King1972: 130–1; Sivan, Reference Sivan1987: 759; Thompson, Reference Thompson1969: 72–7). If this is true, then Gothic recipients of actual land allotments ‘would seem to have been decidedly over-advantaged: they would have pocketed both the private and public revenues of [two-thirds] whereas the barbarian king was limited to collecting only the tax revenues of [one-third]’ (Goffart, Reference Goffart1980: 117). Goffart (Reference Goffart2006, Reference Goffart2010) strengthens the Visigothic case, pointing to references in the Code of Euric to tertias Romanorum and sortes Gothorum. The later Visigothic Code (7th century) makes it clear that tertia was land on whatever taxes were paid to the royal fisc: ‘It say that a Goth was forbidden to have tertia Romanorum unless the king had given it to him. This royal gift could consist not of ownership, which Romans retained [hence still “Romanorum”], but of property assessment yielding revenue’ (Goffart, Reference Goffart2010: 71).

Goffart (Reference Goffart2008) extends his argument regarding the granting of tax assessment shares to the case of the Frankish kingdoms. This extension is particularly relevant given the formative role that historians have ascribed to the late Merovingian and early Carolingian periods in the establishment of West European feudalism. Furthermore, northern Gaul was relatively sparsely populated and the imperial military had largely withdrawn by the 5th century. Therefore, direct imperial involvement in the Frankish settlements was likely minimal. All free Franks had military obligations to Merovingian and Carolingian kings and, like the Visigoths, all free Franks had military obligations to Merovingian and Carolingian kings. Goffart (Reference Goffart2008) points to a number of examples from the annals and capitularies indicating that those obligations were based in property: ‘The assets on whose military was owed appear to have originated from the inherited Roman tax system. [. . .] That exchange of taxes for obligatory, unpaid military service is the system whose traces are still discernible under Charlemagne and his successors’ (p. 167).Footnote ⁸ In addition to the documentary evidence, this argument is appealing to the extent that it accounts for the hereditary nature of military service. If Frankish war leaders simply granted their warriors lands after the fact of conquest, then it is unclear why obligations would persist. Alternatively, if tax obligations on land (which would persist across generations) were exchanged for military obligations then ‘[p]aid soldiers turned into a privileged but not leisured landed class’ (Goffart Reference Goffart2008: 185).

An appealing feature of Goffart's theory is its compatibility with the absence of evidenced outrage and resistance on the part of Romans. Furthermore, Goffart produces evidence that Romans were, if not pleased, relatively unperturbed by the Ostrogothic allotments. He points (1980: 70–1) to Cassiodorus’ panegyric to Liberius (Reference Hodgkin1886) as well as a letter written by the senator Ennodius to Liberius:

While Ennodius clearly aims to flatter, Goffart agrees with Jones (Reference Jones1986: 251) that he ‘would have hardly introduced the topic at all if it had been a painful one’.

Goffart (Reference Goffart1980: 77–9) also points to Theoderic's instructions to the city of Tridentum regarding a particular sors (77–9):

First, the above characterization of a sors is more consistent with a tax assessment share than a land grant. Roman payment of the tertiae is offset by the deduction of the sortes from their (regular) tax calculations. Second, if Goths had actually dispossessed Romans of one-third of their land, then it is difficult to see how Theoderic's ‘generosity’ could have come at no ‘expense to others’.

Veterans’ rights

Hagith Sivan (Reference Sivan1987) finds neither the conventional view nor Goffart's theory of hospitalitas convincing. She argues:

Sivan is most concerned with evaluating what sort of ‘legal fiction’ the empire employed ‘to conceal a radical departure from past practices’ in dealing barbarian groups. However, her theory has implications for the types of lands involved.

According to Sivan, using a framework related to soldier quartering would have been politically problematic. Sivan (ibid., 767) notes that the Theodosian Code includes laws devoted to curbing abuses associated with quartering: ‘[t]he number of abuses that the various laws tried to curtail does not exactly point to the popularity of military billeting’ (ibid., 768). Furthermore, Romans would associate hospitalitas with only temporary settlement.

Alternatively, the Theodosian code provides for veterans ‘exemption from compulsory municipal services, public works and market tax’ (book 7, 20, 2) and ‘assured grant of vacant lands’ (book 7, 20, 3); ‘enabled veterans to choose their legal residence [. . .] and conferred on them perpetual exemption from taxes and public burdens’ (book 7, 20, 8) and ‘encouraged veterans to cultivate properties of absent owners and lands which had not been tilled for a while’ (book 7, 20, 11) (Sivan, Reference Sivan1987: 769–70). These laws would have accounted for ‘the permanent division and occupation of Roman territory and a tax exemption status’ (Sivan, Reference Sivan1987: 759).

The settled barbarian groups (or large elements of them) had all been federates (foederati) of the empire at one time or another. Foederati were granted annual subsidies in exchange for maintaining peace with Rome and providing military services against other barbarian groups and imperial usurpers. A framework of laws pertaining to veterans would have seemed to Romans plausibly applicable to the federate barbarians. Also, it would not have been associated with the abuse of Roman hosts. Sivan's theory suggests that barbarian sortes consisted of vacant lands or lands that were owned but not being currently used in production.

How did hospitalitas actually work?

How hospitalitas actually played out is clearly a matter of controversy and it may not have played out uniformly in all regions of the Western Empire. To the extent that land allotments were the case, whether or not they were orchestrated within a legal framework for quartering of soldiers or remunerating veterans is not particularly relevant from a political economy perspective. For example, Wickham (Reference Wickham1984: 20) accepts that the ‘Germans took from one-third to two-thirds of the land’ without invoking any particular Roman laws.Footnote ¹⁰ Goffart (Reference Goffart1980) provides the only serious alternative in the form of his tax allotment theory of hospitalitas.

Is the Goffart theory plausible? He points to multiple statements from extant sources suggesting allotments of tax assessments rather than of actual land. However, most of that evidence speaks directly only to the regions of southern Gaul and northern Italy.

Furthermore, some scholars have suggested that Goffart's efforts to read into terms such as millenarios furthers his tax allotment theory at the expense of more straightforward interpretations. For example, Wickham (Reference Wickham1984: 20, n. 21) states that ‘the trouble is that rather few texts even have a prima facie reading to support [tax rather than land allotments], no text says it explicitly (and surely some ought to), and several explicitly oppose it’. Barnish (Reference Barnish1986: 173) adds: ‘Indeed, while the units [e.g. millenae] discussed by Goffart can sometimes have a purely abstract force, it is clear that they often denote actual pieces of land’.

Alternatively, Goffart's interpretation makes intelligible the lack of documented Roman outrage. While Roman aristocrats were by no means warriors, they commanded considerable wealth and therefore de facto political power. They could afford to buy military services, quite possibly from competing barbarian groups. If their land were being expropriated and they perceived no offsetting benefits in return, the Roman aristocrats would have been able to mount a resistance that was, if not ultimately successful, certainly costly to the barbarians.Footnote ¹¹

Furthermore, Roman aristocrats held a monopoly on human capital that barbarian groups required to effectively transition from roving bandits to stationary states (Young, Reference Young2016). Effective administration and jurisprudence required literacy and bureaucratic experience. Roman aristocrats associated their elite status with imperial office holding. And Roman aristocrats were embraced by the fledgling barbarian administrations. For example, we have seen the important role that Cassiodorus played in Theoderic's Ostrogothic administration. Furthermore, Heather (Reference Heather1998: 193) remarks upon a portrayal of the mid-5th century Visigothic court provided by the Gallo-Roman senator, Sidonius Apollinaris:

Implied is the contribution of aristocratic Gallo-Romans to the nascent Visigothic legal tradition.Footnote ¹² Undoubtedly, literate Romans aristocrats were critical contributors to the proliferation of barbarian law codes in the 5th and 6th centuries.

Accepting Goffart's (Reference Goffart1980) interpretation of hospitalitas as generally applicable does not mean that actual land allotments to barbarian settlers never occurred. Most of the evidence marshalled by Goffart is associated with the relatively populous regions of southern Gaul and northern Italy. The Franks, alternatively, were settled in the northern parts of Gaul where Roman landowners were relatively few and far between. In northern Gaul allotments of vacant land would have been possible without upsetting established landowners; furthermore, without many Roman landowners the associated tax assessments to provide shares of would also have been scarce. Sivan's (Reference Sivan1987) interpretation of hospitalitas as a framework for settling retired foederati may make more sense in such sparsely populated areas.

The stationary bandit alternative

Under the stationary bandit alternative the incentives of the barbarian rank and file warriors and their elite leadership would have been misaligned. As members of a roving armed retinue, barbarian warriors operated on a relatively small scale (around 200 men; see Heather Reference Heather1998: 66–8) and had a direct role in the extraction of the plunder to be distributed to them. Residual claimancy aligned warriors’ incentives with those of their commander. Furthermore, unlike in kin-based tribal associations, a warrior's relationship to his retinue was based on shared material interests (Young, Reference Young2015). If a warrior did not perceive his interests as being served as part of a particular retinue then he would find another commander to follow.

Considering settlement under the stationary bandit alternative, two factors would have weakened the incentive alignment between barbarian warriors and their elite. First, individual retinues would have been consolidated permanently in a confederacy. Confederacies generally operated towards specific ends under extraordinary circumstances (e.g. war) and they disbanded following the achievement of those ends or the passing of those circumstances. While the confederacy existed, warriors had to follow an unfamiliar commander while their efforts became an additional step removed from the ultimate rewards.

Second, if the confederacy settled down to become a state, the warriors would be expected to refrain from plundering and become tax collectors instead. They may also have been called upon to provide law and order to encourage a larger tax base. The warriors would then no longer have been residual claimants but, rather, administrative agents. The collection of taxes and provision of law and order would also have been activities in which warriors had relatively little experience. Furthermore, the latter activity would require that warriors refrain from independent plundering. The discouragement of ‘moonlighting’ outside of the collective endeavours of the state would have been a new and likely unpleasant development from the perspective of barbarian warriors.

Alternatively, the barbarian elite would have become rulers. They would have gained an encompassing interest in their realm (Olson, Reference Olson1982). This would have aligned their incentives with those of Roman landowners. This would likely have further antagonized the barbarian warriors. Even under the hospitalitas framework, settlement created the perception among some warriors that their elites were acting in ways more consistent with Roman interests than their own (Thompson, Reference Thompson1982 [1930]: 48).

The subjugation alternative

Alternatively, the barbarians could have thoroughly subjugated the Romans and taken over the imperial administration, including the provision of law and order and the collection of taxes. The Romans would have been rendered a subservient (perhaps enslaved) class and the barbarian rank and file would have become the free citizenry. But the subjugation alternative would have obviously left the incentives of Roman and barbarian elites misaligned, and this would have been problematic for the successor states.

In particular, the collection of taxes and the provision of inclusive public goods require political and legal structures more sophisticated than anything present in 5^th-century barbarian societies. In reality, the barbarian elite turned to the Roman aristocrats. Members of the senatorial class possessed the human capital necessary to provide effective government administration and bureaucracy. This level of cooperation between barbarian and Roman elites would have been difficult to reconcile with subjugation of the latter. Romans whose lands had been wholly expropriated would not have skin in the game being played by the new barbarian rulers. Furthermore, relegating proud Roman aristocrats to subservient status (especially to barbarians!) would itself have discouraged cooperation with the new leadership. Roman aristocrats still had exit options. They could depart for parts of the Empire where barbarians were not currently being settled, or seek positions in the Church hierarchy (Mathisen, Reference Mathisen2011). Alternatively, since Roman elites were the barbarians’ only source of organizational/administrative capital, the barbarians had poor outside options in choosing how to deal with their new subjects (Libman et al., Reference Libman, Kolzov and Schultz2012).

Hospitalitas

The settlement of barbarian groups in Gaul and Italy involved allotments of claims to returns on productive assets. The barbarian sortes were formulaic: they were spread broadly and based on predetermined, uniform shares of each individual Roman possessor’s lands.

Not only did the Roman elite and the barbarian recipients of sortes share interests in productive assets, but the allotments also worked to realign the incentives of barbarian warriors and barbarian elites at a time when their relations may have been strained by the roving-to-stationary bandit transformation. The sortes established a link between the barbarian elites’ interests in a growing tax base and warriors’ interests in the productive yields of the land. Actual allotments of land would have been effective in establishing this link. However, to the extent that sortes were tax assessment shares they were even better suited to align warrior and elite incentives. Barbarian warriors were not farmers. They may not have been able to realize returns on land holdings comparable to those of their Roman neighbours. Alternatively, tax assessment shares ‘yielded immediate and reliable returns’ to the barbarian warriors (Goffart, Reference Goffart1980: 54).

Tax assessment shares also mitigated antagonism of the Roman elite. Romans maintained ownership of their lands. The annoyances faced by Roman landowners were confined to the inconveniences of unfamiliar methods of tax collection. If the allotments were tax assessment shares then the actual level of taxes did not change for Roman landowners; only the number of tax collectors.

To the extent that sortes were not tax assessment shares but, rather, actual land allotments, most of the above analysis is unaffected. Barbarian warriors would have had to deal with the actual management of their allotted lands. This task may have been allocated to family members. Alternatively, a barbarian warrior with a sors may have reached an arrangement with his Roman ‘host’ to continue managing the entire landholding. In the case of actual land allotments Romans would indeed have suffered expropriation, but they would also have held onto a meaningful portion of their landholdings. This may have kept antagonism to a minimum.

The second estate

In the barbarian hospitalitas settlements one finds foundations of the medieval second estate. The second estate was the landed and militarized aristocracy that characterized feudal Western Europe. Monarchs granted benefices in the form of lands to the principle magnates of their realms. In doing so monarchs established reciprocal relationships involving their protection and justice in exchange for the loyalty and service of these magnates. In turn, the latter granted portions of their holdings to lesser nobles in return for their homage (i.e. so-called subinfeudation).Footnote ¹⁵

The second estate differed in important ways relative to what characterized medieval China and the Islamic world. In the latter, Sultans inherited more intact bureaucracies from their conquered Byzantine and Sassanid lands; they staffed those bureaucracies with slaves (Mamluks) who swore allegiance to the Sultans (Blaydes and Chaney, Reference Blaydes and Chaney2013: 17). China, alternatively, was a feudal society but one where the feudal ties were between lords and their kinship groups. Hence tribal relationships were predominant. This was quite different from the European relationships between lords and vassals that were based on impersonal, contractual relationships involving the exchange of land and protection for loyalty and military services (Fukuyama, Reference Fukuyama2011: 106–7). Unlike in China, in Europe ‘[a]uthority was vested in the lord himself and not in the lord's clan[;] [t]he fief was a possession of his family but not of a larger corporate descent group’ (Fukuyama, Reference Fukuyama2011: 107).

The roots of feudalism in Western Europe have been a matter of debate amongst historians and the contributing factors were undoubtedly multiple. Among these factors, the most generally recognized is the secularization of Church properties initiated by Charles Martel in the 8th century and continued by his son Pippin III, the first Carolingian king. These properties were distributed as benefices to favoured nobles who served as the military and administrative structure of the Carolingian Empire. The spread of subinfeudation has also been linked to this development (e.g. Cronne, Reference Cronne1939: 356–257; Ganshof, Reference Ganshof1939: 157–9; Stephenson, Reference Stephenson1941).Footnote ¹⁶

However, as noted in section 2 above, Goffart (Reference Goffart2008) argues that one can trace the legal basis for Carolingian feudal ties back to hospitalitas. Without embracing Goffart's theory of tax assessment shares, Wickham (Reference Wickham2016: 28–30) acknowledges that the exchange of taxes for military service was characteristic of the earliest barbarian states. More generally, reciprocal relationships that resembled Carolingian vassalage predated Charles Martel and had precedents in the institutions of loyalty within Germanic war bands (e.g. Bisson, Reference Bisson1995: 746–7; Riché, Reference Riché and Allen1993 [1983]: 37–9; Stephenson, Reference Stephenson1941; Wickham, Reference Wickham1984: 25).Footnote ¹⁷ The Carolingian enterprise, then, exploited well-established institutions. As Rio (Reference Rio2008: 44) notes: ‘the Carolingian era [is interpreted] as a period of unprecedented aristocratic growth, coupled with an acknowledgement [. . .] that Carolingian military dependence cannot be said to constitute an origin-point for the inexorable progression towards feudalism’.

The military might of the Western Empire's successor states was decentralized in the process of the barbarian settlements. This decentralization contributed to an important change in how armies were organized and funded relative to the imperial era. ‘Beginning in the fifth century, there was a steady trend away from supporting armies by public taxation and towards supporting them by rents derived from private landowning’ (Wickham, Reference Wickham2009: 102). The settled barbarians became shareholders in the realm with de facto political power. They had this power by virtue of their landed wealth and the fact that they were the source of military power for the kings of the successor states. This is reminiscent of military organization in medieval Western Europe, based on the second estate.

To the extent that barbarian hospitalitas settlements contributed to a nascent second estate, this is of interest to students of the evolution of Western governance. The estates system has been recognized widely as an important component of the medieval constitution and has been linked to the evolution of traditions in representative and limited government. For examples: ‘Historically, the separation of powers in Europe developed out of the old system of estates’ (Weber, Reference Weber, Roth and Wittich1978 [1922]: 283); it ‘today gives the political life of the whole civilized work its distinctive character [and] traces its historical origins to the system of Estates of the Middle Ages’ (Hintz, Reference Hintz and Gilbert1975 [1931]: 305).

The first estate (the Church) and the second estate were well-defined groups with considerable political authorities. Their authorities were bundled with rights to revenues from their lands (Salter, Reference Salter2015a). (These included the lands of their vassals in the case of the nobility.)

The rough balance of power between monarchs and the estates was a fundamental part of the polycentric governance that characterized Western Europe. The above-cited scholars (along with others cited in section 1 above) view this polycentric governance as being critical to the development of Western traditions in governance. A number of scholars (e.g. Leeson and Suarez, Reference Leeson and Suarez2016; Rajagopalan, Reference Rajagopalan2016; Williams, Reference Williams2016) have argued, in particular, that the competition between the powers of estates and monarchs promoted the development of rule of law in the way suggested by Olson (Reference Olson1993).Footnote ¹⁸

Is this view of the roots of ‘good’ governance in the West accurate? One can start with the observation that the representative assemblies that characterized medieval Europe had no equivalent in any other region of the world.Footnote ¹⁹ But is the exceptionalism attributable, at least in important part, to a militarized and landed second estate? There are good reasons to think so. Medieval monarchs relied on the principle magnates of their realms for the means to pursue their military campaigns.Footnote ²⁰ As such, these magnates were de facto politically powerful and monarchs had to strike bargains with them towards those ends. This is consistent with the fact that the monarchs of medieval Western Europe travelled incessantly to gain support for their policies from the leading men of their realms (Bernhardt, Reference Bernhardt1993; de Jouvenel, Reference De Jouvenel1993 [1947]: 7; Ertman, Reference Ertman1997: 44, 236; Heather, Reference Heather2009: 529; Marongiu, Reference Marongiu1968: 23).

The second estate was a fundamental component of medieval governance hierarchies. Throughout the feudal hierarchy, lords were obligated to provide governance in the form of security and justice to their vassals. Considered from the perspective of Congleton's (Reference Congleton2011b, Reference Congleton2007) theory of constitutional bargaining, these governance hierarchies are consistent with and the fact that medieval monarchs took the initiative in instituting representative assemblies (Russell, Reference Russell1982; North and Weingast, Reference North and Weingast1989; Van Zanden et al., Reference Van Zanden, Buringh and Bosker2012; Salter and Young, Reference Salter and Young2017b). Monarchs wanted members of the second estate to provide them with military services. However, as Congleton notes, insisting on burdens that were too large or policies that the nobles otherwise disliked would have led to low effort, avoidance, or outright revolt. To gain assent for their policies monarchs, then, had to make credible commitments. By providing the second estate with representation in an assembly, a monarch could solve the credible commitment problem by ceding some veto authority in exchange for nobles agreeing to provide services for specific policies put forth by the monarch.

As stated above, there are good reasons to think that a landed and militarized second estate contributed to medieval Western Europe fostering traditions of limited and representative government. Blaydes and Chaney (Reference Blaydes and Chaney2013) also provide empirical evidence that the feudal system created checks on European monarchs that led to greater political stability (in terms of longer monarchical/ruler tenure) relative to that observed in the Islamic world. However, there could certainly have been downsides to the political fragmentation of Western Europe. This is not to say that the estates system did not have drawbacks. Feudal society was characterized by considerable dependence on potentially costly intra-estate economic transactions (Demsetz, Reference Demsetz2002). The fragmentation of the Roman Empire was initially associated with a decline in trade and wealth-creation, especially in the once-vibrant Mediterranean areas (Hodges and Whitehouse Reference Hodges and Whitehouse1983; McCormick, Reference McCormick2001). Furthermore, a number of scholars have argued that the decentralized provision of coercion is likely to be associated with wasteful competition over resources (e.g. Bates, Reference Bates2008; Bates et al., Reference Bates, Greif and Singh2002; North et al., Reference North, Wallis and Weingast2009). This view is consistent with Bisson's association of subinfeudation with the ‘proliferation of coercive (or “banal”) lay lordships’ (Reference Bisson2009: 56) that led to an increasingly violent Western Europe; also Ko et al.’s (Reference Ko, Koyama and Sng2016) evidence that decentralized violence in Western Europe was associated with higher tax rates than those of politically centralized China.