Disruption and Contention During State Centralization

While stronger states can effectively monopolize violence, enforce rules, and centrally administer territory and social groups (Weber Reference Weber1968), the process of state-building itself can be contested and contentious (Boone Reference Boone2003; Garfias and Sellars Reference Garfias and Sellars2022). We build on seminal theories of state formation that emphasize the development and consolidation of a centralized administrative network across the territory. Transportation and communication infrastructures that connect localities more directly to central government authorities enhance the ability of the state to penetrate society through the control and administration of local activity and rules, extraction in the form of taxation and military service, and the disruption of existing social relations and expectations. In France, as throughout Europe, this historic process entailed the extension of authority over—and even the absorption of—disparate regional nobles, lords, church authorities, and other elites (Elias Reference Elias[1939] 1994).

Our focus, however, highlights how this process was characterized by frictions with society more broadly, and to a lesser degree, by frictions with regional elites. Indeed, the material consequences of increased but incomplete state power can generate social contestation and popular resistance stemming from the greater local capacity of the state to enforce taxes, carry out rules of justice, and conscript civilians into military service. It can also result from the crowding out of private interests and activities in spaces that the state comes to dominate. These consequences of state expansion disrupt status-quo activities and expectations, generating grievances.Footnote ⁵ At the same time, the state may lack sufficient coercive power and legitimacy to deter popular challenges, a dynamic that has been documented in other contexts of state-building (e.g., Pruett Reference Pruett2024).

The extent of contestation stirred by state efforts at expansion and centralization may also vary across the territory in accordance with pre-existing fragmentation in authority and administration. Centralizing states routinely encounter regional divisions linked to historical competing authorities and power centers across their territorial expanse, which can, as Migdal (Reference Migdal1988) highlights, condition the interactions between society and the state. That poses a challenge to the state’s ability to effectively penetrate and order society in a consistent fashion across space and can raise regionalized grievances and backlash. This raises the potential for the link between state-building and unrest to be characterized at least in part by heterogeneity across space in accordance with fragmentation.

We explore these dynamics in pre-Revolutionary France, where increased connectivity and penetration of society associated with state centralization fostered disruption and social unrest. While this relationship prevailed across the entire French territory, it also varied to some extent with pre-existing administrative fragmentation: where fragmentation mapped onto historic disparities in the mode of ruling and extent of intended extraction by the state, opposition to the state’s imposition of central authority through new horse-post relays varied accordingly.

The Functioning of the Horse-Post Relay Network

From its inception under Louis XI in the late 1470s, the horse post served as a pivotal state-building instrument of the royal authority to consolidate its informational capacity across the kingdom, particularly in its outlying provinces (Marchand Reference Marchand2006, 183–221).Footnote ⁶ This infrastructure consisted of a nationwide network of relays where horse-post messengers (postillons) could exchange their horses for fresh ones in order to ensure the rapid transportation of directives, reports, and other communications of the royal administration (Jamaux-Gohier Reference Jamaux-Gohier2001; Marchand Reference Marchand2006; Gazagnadou Reference Gazagnadou2013, 109–16). Relays were spaced at regular intervals of 10–15 kilometers, allowing riders to switch horses after an hour or two of travel. This system enabled postillons to maintain high speeds while minimizing the risk of injury to the horse, in line with their physiological limits (Minetti Reference Minetti2003).Footnote ⁷ The horse-post relay network was connected by a road infrastructure maintained by the Bridges and Roads (Ponts et Chaussées) administration, which operated independently of the horse post (Arbellot Reference Arbellot1973).

Located at the entrances to towns or at rural crossroads, horse-post relays were relatively large buildings—comprising stables, lodging quarters, and often a tavern—that stood out prominently in the rural landscape (see Panel a of Figure 1). The postillons who galloped postal roads day and night were typically young local laborers, easily identifiable by their distinctive uniforms: a royal-blue vest, a round leather hat, leather breeches, heavy black boots, and an arm badge bearing the name of their relay (Jamaux-Gohier Reference Jamaux-Gohier2001, 237–40; Marchand Reference Marchand2006, 258–64). Their horses, bred in royal stud farms, were superior to those of private horse renters and bore distinctive markings, such as a knotted or trimmed tail (Jamaux-Gohier Reference Jamaux-Gohier2001, 262–7; Marchand Reference Marchand2006, 270–5). Together, postillons’ uniforms and horses made them instantly recognizable in the public space (see Panel b of Figure 1).

Figure 1. Depictions of the Horse Post

Note: This figure displays contemporary depictions of an eighteenth-century horse-post relay in Panel a and a postillon galloping two horses back to a relay in Panel b. Sources: in Panel a, lithography by Victor-Jean Adam and Louis-Philippe-Alphonse Bichebois, based on a painting of Jean-Antoine Duclaux, La malle au relai [sic], 1817; in Panel b, painting of the French School, Postillon en livrée ramenant deux chevaux au galop, mid-nineteenth century. © Musée de La Poste, La Poste, photographs by Thierry Debonnaire.

In addition to the short distance between relays—which allowed for frequent changes of horses—a series of regulations enabled the horse-post system to achieve a higher speed of information transmission than alternative postal systems available to the general public, such as the letter post or private messenger companies (Jamaux-Gohier Reference Jamaux-Gohier2001, 100–8).Footnote ⁸ Indeed, postillons held a monopoly on galloping and on traveling at night on roads connecting relays, which were generally better maintained by the Bridges and Roads administration than other roads and directly connected to administrative centers (Arbellot Reference Arbellot1973, 766). The horse post also held a monopoly on horse rentals along these roads, effectively preventing private travelers from switching horses when traveling along them.Footnote ⁹ In addition, it had priority access to hay for feeding horses and the right of passage on congested roads. Postmasters (maîtres de postes) profited substantially from these privileges and were quick to report violations to their intendant—the provincial representative of the monarchy—who usually swiftly enforced these regulations (Vaillé Reference Vaillé1953, 166–70; Jamaux-Gohier Reference Jamaux-Gohier2001, 101–4).

The establishment of horse-post relays provided the opportunity for the royal authority to form alliances with and co-opt local elites who came to serve as postmasters. Postmasters were typically recruited among wealthy local notables—bourgeois farmers and innkeepers aspiring to noble status (Jamaux-Gohier Reference Jamaux-Gohier2001, 189–94; Marchand Reference Marchand2006, 223–8). Each held a postal certificate (brevet de poste) that granted them the exclusive right to operate a given relay. The certificate imposed several obligations on postmasters: they were required to reside at the relay, to manage their relay directly (faire-valoir direct) and not farm out its operations, and to maintain its proper functioning—such as keeping at least five or six healthy horses available at all times. In addition, they were expected to serve as intelligence agents and auxiliary police on behalf of the central administration (Vaillé Reference Vaillé1953, 268–71; Marchand Reference Marchand2006, 210–4). Their duties included monitoring passengers passing through the relay—by the late eighteenth century, this involved keeping records of names, occupations, and destinations—as well as reporting notable local political events to their intendant.

Overall, the horse-post relay network positioned the state as the sole authority controlling rapid communications in France, significantly expanding its ability to monitor society and exert control over its territory.

Consequences of the Horse Post for Local Order and Rebellion

The development of the horse-post relay network had several distinct consequences. At the national level, it enabled the rapid communication essential to the effective functioning of the state. For instance, it allowed central officials, tax farmers (fermiers généraux), and provincial intendants and local tax officials (élus) to coordinate swiftly through directives, reports, and fiscal orders on matters of tax collection and enforcement. That included efforts to suppress salt smuggling (faux-saunage) across the kingdom’s disparate and consequential salt-tax zones (gabelles). This coordination required frequent and timely exchanges between central administrators and local agents, who were responsible for estimating tax yields and implementing their collection (Caillou Reference Caillou2005, 177–293; Deharbe Reference de2010, 243–344). The speed of communication was especially critical in times of crisis, such as in the aftermath of natural disasters (Favier Reference Favier and Legay2007). The horse-post system also facilitated the transmission and enforcement of military recruitment orders, enabling directives to travel efficiently from central authorities to intendants and onward to local recruitment and conscription officers (Corvisier Reference Corvisier1964, 147–61).

Beyond these nationwide implications, the expansion of the horse-post relay network disrupted local social equilibria. A range of local actors benefited from these developments. Chief among them were postmasters. In addition to the profits they derived from their exclusive right to operate a relay, postmasters enjoyed a wide range of privileges, including substantial exemptions from taxation—most notably, from the onerous taille—as well as relief from military obligations, from accommodating passing soldiers, and from churchwarden responsibilities (Demangeon Reference Demangeon1995, 44–8; Tonnaire Reference Tonnaire2005, 48–59). They were also exempt from taxes on ancillary activities tied to relay operations, such as the sale of wine in the relay’s tavern. These material advantages fostered broad loyalty to the monarchy among postmasters (Vaillé Reference Vaillé1953, 233–48; Jamaux-Gohier Reference Jamaux-Gohier2001, 189–235; Marchand Reference Marchand2006, 200–46).Footnote ¹⁰ Other local groups also derived material benefits from the introduction of horse-post relays: innkeepers benefited from the increased traffic along roads connecting relays (Tonnaire Reference Tonnaire2005, 59–63), and families allied with postmasters often enhanced their local influence through closer ties to representatives of the royal authority (Tonnaire Reference Tonnaire2005, 48).

Local elites also likely benefited from the establishment of horse-post relays, as improved communication with provincial and central authorities may have enhanced their influence and elevated their status. For instance, historical accounts suggest that tax agents often relied on the knowledge and assistance of local notables as they traveled along the relay network in preparation for their annual round of tax collection (Caillou Reference Caillou2005, 177–293; Deharbe Reference de2010, 243–344). These interactions may have fostered mutually beneficial relationships, reinforcing the authority of local elites and further integrating them into the state’s administrative and logistical apparatus. Local notables may also have engaged with military officials and royal agents passing through or lodging at relays, for instance, by sharing local information, though such ties were likely kept discreet and remain difficult to corroborate in historical records.

By contrast, the various monopolies associated with the horse post crowded out competing private interests. First and foremost, horse-rental businesses were severely disrupted by the ban on renting their horses along roads connecting horse-post relays. Indeed, examples of horse renters attempting to violate this ban abound (e.g., Jamaux-Gohier Reference Jamaux-Gohier2001, 101–4).Footnote ¹¹ Local farmers and innkeepers situated off postal roads also lost out, as postmasters had priority access to hay and animal fodder for their relays—a shortage potentially worsened by the increased demand brought on by rising traffic along postal roads (Tonnaire Reference Tonnaire2005, 61–4). More broadly, the monopolization of these roads by the monarchy through the horse post imposed constraints on travelers as privately owned horses were restricted to trotting during daytime and could not be exchanged at horse-rental establishments.Footnote ¹²

Additional social groups, such as commoners, peasants, and merchants, may have also lost out with the establishment of horse-post relays as they brought increased surveillance and stricter enforcement of state regulations. Indeed, enhanced state communication through the horse-post relay network likely made it easier for authorities to monitor and enforce tax collection, military conscription, and judicial decisions. In addition, local residents could have been subjected to new burdens, including maintaining postal roads through the unpaid labor system of the corvée (Tonnaire Reference Caillou2005, 65–86; Conchon Reference Conchon2016).

We hypothesize that these disruptions to local society fueled grievances, and eventually rebellion, though the material consequences of a new horse-post relay likely took time to fully materialize and for discontent to boil over. Adverse events such as repeated food shortages, tax enforcement campaigns, labor requisitions, or conscription orders, combined with policy changes such as tax increases or stricter judicial procedures or penal codes, could have tipped simmering discontent into open rebellion. State agents who contributed to—or were perceived as contributing to—these pressures could have become targets of local hostility. Counterfactual parishes that did not host new horse-post relays were less exposed to such state intrusion and thus avoided the same stratifying pressures that divided communities into winners and losers of state centralization. Hence, residents of these parishes may not have experienced these adverse events to the same degree, and where they did experience them, they might have been less inclined to attribute them to the presence or actions of state agents.

Several rebellious events in our data exhibit just such reactions against state agents and activities. One example occurred in 1781 in the town of Mirande, in the southwestern généralité of Auch. A decade earlier, the establishment of a new horse-post relay marked a deeper state presence in this formerly isolated region of the kingdom, as before the 1770s, this généralité had only two relays, in Pau and Orthez. Military recruitment in the area had long been among the lowest in France, but the increased needs resulting from the Anglo-French War (1778–83) brought renewed scrutiny to local enlistment.Footnote ¹³ In this context, the new horse-post relay facilitated information gathering on draft-eligible men. In 1781, as the subdélégué (the local representative of the monarchy), military officers, and local notables assembled in Mirande to conduct a conscription draw, draft-eligible peasants and farmers emerged from the forest with sticks and knives, shouting at the authorities, “They should all be killed!” A confrontation over draft resistance followed, resulting in several arrests before a local notable managed to defuse the situation.

A rebellion in the coastal town of Blaye in Gironde in 1783, just three years after the introduction of a horse-post relay, offers another revealing example. Members of a patache brigade—widely despised mobile tax-farming agents charged with collecting gabelles and seizing contraband—attempted to inspect the load of a Breton ship.Footnote ¹⁴ Following the establishment of a series of new relays connecting La Rochelle to Bordeaux along the eastern shore of the Gironde Estuary in the early 1780s, patache patrol boats began to more closely monitor maritime commerce in the region, including in Blaye. When the patache approached, the ship’s crew resisted inspection and joined forces with local residents to stone the patache agents once ashore at the encouragement of the ship’s captain and the owner of the merchandise. Members of the crew denounced the patache brigade as an “enemy of society” that had come to “outrage citizens” who were merely going about their usual business. In short, heightened tax enforcement facilitated by the recent expansion of the horse-post relay network sparked a wave of local resentment that quickly escalated into open resistance.

Other examples illustrate how the introduction of horse-post relays could have crowded out private interests. About a decade after a series of relays was established along the route between Dijon and Auxerre, members of the local bourgeoisie complained that inns near the new relays were monopolizing supplies of food and animal fodder (Tonnaire Reference Tonnaire2005, 63–4). In response, the subdélégué issued a ban preventing innkeepers from making early purchases at local markets to preserve access for other buyers—setting, for instance, an 8 a.m. cutoff in Avallon. Tensions between innkeepers and the bourgeoisie escalated, and although the restriction remained in place despite numerous complaints from innkeepers, many of the underlying grievances persisted. In another such example in 1775 in Picardy, in the context of the Flour War, a crowd of townspeople in Chauny attacked and ransacked the local horse-post relay on rumors that the postmaster had stockpiled wheat despite the local shortage. A detachment of troops was unable to stop the crowd.

Of course, horse-post relays could have impacted rebellion through other channels beyond those illustrated in these examples. For instance, they could have generated local income shocks, disrupted traditional social hierarchies, introduced more policing, and called attention to the recording and registration of rebellious activities. We examine these and other potential alternative mechanisms below.

The Horse-Post Relay Network

To document the development of the horse-post relay network over the eighteenth century, we construct an original historical geographic information system (GIS) based on administrative sources.Footnote ¹⁶ Beginning in 1708, the General Superintendent of Posts had a list of horse-post relays printed and updated annually: the Liste générale des postes de France (Arbellot Reference Arbellot1980, 100–2; Reference Arbellot1992, 110–4). This publication was designed to provide state messengers with the locations of horse-post relays together with distances between them. Each Liste des postes was organized by itinerary, with departure and destination relays clearly marked.Footnote ¹⁷ Distances were reported in postes—a measure of distance equivalent to 8–12 kilometers (Bretagnolle and Verdier Reference Bretagnolle and Verdier2006, 68–9).Footnote ¹⁸

We digitized one edition of the Liste des postes per decade from the 1710s through the 1790s.Footnote ¹⁹ From each edition, we extracted all the segments between relays as listed by itinerary and matched each location to the corresponding parish. We then reconstructed itineraries by drawing straight lines between successive relays and combining the resulting segments. Figure 2 displays the spatial distribution of the horse-post relay network at the beginning and end of our period of analysis.Footnote ²⁰ We also report the number of horse-post relays along with the length of associated postal roads for each edition of the Liste des postes included in our sample in Panel a of Figure 3.

Figure 2. The Horse-Post Relay Network

Note: This figure displays the horse-post relay network based on the 1714 and 1790 editions of the Liste des postes. Horse-post relays correspond to black dots while horse-post roads, to red segments. The underlying shapefile of the kingdom of France as of 1789 is based on Gay, Gobbi, and Goñi’s (Reference Gay, Gobbi and Goñi2024a) jurisdictions database.

The spatial configuration of the horse-post relay network at the beginning of the eighteenth century reflected the monarchy’s long-standing strategic and military concerns vis-à-vis neighboring adversaries, consistent with accounts of subnational state-building efforts that emphasize responses to prevailing geopolitical and bellicist threats (Boone Reference Boone2003; Tilly Reference Tilly1990). In the late fifteenth century, relays were established in Bordeaux, Dijon, Lyon, Arras, and Amiens due to their strategic positions relative to the Kingdoms of Spain and England, the Duchies of Burgundy and Savoy, the County of Flanders, and the Holy Roman Empire (Gazagnadou Reference Gazagnadou2013, 112). During the sixteenth century, postal routes expanded toward the kingdom’s borders, with relays established in Bayonne under Louis XII and in Nantes under Henry III. This strategic expansion continued throughout the seventeenth century, following territorial conquests: northeastward to Strasbourg after the Peace of Westphalia (1648), southwestward to Narbonne after the Treaty of the Pyrenees (1659), and eastward to Besançon after the Treaties of Nijmegen (1678–79) (Marchand Reference Marchand2006, 186–8). These strategic and military concerns led to a center-periphery logic resulting in a star-shaped network radiating from Paris and reinforced by key nodes along the kingdom’s borders.Footnote ²¹

In contrast, the eighteenth century was a period of internal densification of the horse-post relay network (Bretagnolle and Verdier Reference Bretagnolle and Verdier2005; Reference Bretagnolle, Verdier and Roux2007; Verdier Reference Verdier2009). By then, the horse post had become a key instrument in the monarchy’s strategy to assert a more uniform and centralized control over the territory and to project its power into the remotest parts of the kingdom (Marchand Reference Marchand2006, 189).Footnote ²² The network first densified in the north following the permanent establishment of the royal court at Versailles in 1682 and the consolidation of the kingdom’s territorial boundaries. From the mid-eighteenth century onward, the network expanded around regional centers, such as Bordeaux, Lyon, Dijon, Toulouse, and Rennes. By the end of the century, the horse post had evolved into a national network structured around the political center in the north and connected to secondary centers, creating the conditions for the emergence of an integrated city system (Bretagnolle and Franc Reference Bretagnolle and Franc2017).

Figure 3. Evolution of the Horse-Post Relay Network

Note: Panel a displays the evolution of the number of horse-post relays and the length of postal roads in kilometers from 1714 to 1790. Panel b displays the mean and standard deviation of the distance from the 35 thousand parishes in our sample to the nearest horse-post relay in kilometers. Horse-post relay data are based on various editions of the Liste des postes.

We quantify the expansion of the horse-post relay network over the eighteenth century in Panel a of Figure 3. The network grew steadily, essentially doubling both in length and density over the period. Whereas in 1714, it comprised 841 relays spanning 11 thousand kilometers of postal roads, by 1790, it had expanded to 1,403 relays covering 24 thousand kilometers.Footnote ²³ This expansion substantially improved local accessibility to the network, as the average distance between a parish and the nearest relay decreased from 22 kilometers in 1714 to 13 kilometers in 1790, as shown in Panel b of Figure 3. These gains were spread across all parishes, as the standard deviation of this distance also fell from 26 to 13 kilometers.Footnote ²⁴ More broadly, the proportion of parishes located near a horse-post relay rose substantially, from 18% in 1714 to 31% in 1790.Footnote ²⁵

However, the development of the horse-post relay network was not entirely linear as it was periodically reconfigured at the local level (Bretagnolle and Verdier Reference Bretagnolle, Verdier and Roux2007; Verdier Reference Verdier2009). While net gains amounted to 562 relays over the eighteenth century, this reflected the creation of 1,001 relays and the retirement of 439 of them.Footnote ²⁶ In particular, substantial reconfigurations occurred in the interior of the country in the 1740s and 1750s—between Bordeaux and Limoges, Bordeaux and Poitiers, Poitiers and Tours, and Lyon and Clermont-Ferrand. These reconfigurations were generally driven by local practical considerations, such as the retirement of a postmaster or the development of nearby roads in better condition (Arbellot Reference Arbellot1980, 106).Footnote ²⁷ A comparison of the spatial distribution of newly established and retired relays by decade reveals that retired relays were, on average, 11 kilometers away from the nearest new relay.

While the initial configuration and expansion of the horse-post relay network were largely shaped by national strategic considerations, its local development over the eighteenth century was driven by practical constraints. Beyond the quality of the road network, the specific placement of relays was primarily determined by the need to maintain regular 10–15 kilometer intervals—reflecting the physiological limits of horses (Minetti Reference Minetti2003). As a result, many relays were located in small towns and villages situated along routes between administrative centers, often on riverbanks or on foothills.Footnote ²⁸ Accordingly, about one-third of relays were located in cities with more than two thousand inhabitants—typically administrative centers—while nearly half were in towns with fewer than a thousand inhabitants, and 19% in villages with fewer than five hundred inhabitants.Footnote ²⁹ Horse-post relays established after 1714 were even more likely to be located in smaller towns and villages.

Summary statistics on the horse-post relay network at the parish-decade level are reported in Panel A of Table 1.Footnote ³⁰

Table 1. Summary Statistics across Parish-Decades

Note: This table reports summary statistics for the main variables used in the analysis, excluding the city of Paris. The unit of observation is a parish-decade from the 1710s to the 1780s. There are 282,288 parish-decade observations in the dataset. State authorities include the fiscal, military, and judiciary authorities. Non-state authorities include the nobility, seigneurial and municipal authorities, and the Church.

Rebellions

We collect data on eighteenth-century rebellions from the Jean Nicolas survey (Nicolas Reference Nicolas and Nicolas1985; Reference Nicolas2002).Footnote ³¹ Compiled from extensive archival research conducted from the early 1980s to the late 1990s, this survey documents 8,516 rebellions that occurred between 1661 and 1789 (Gay Reference Gay2025, 7–11). It focuses on popular uprisings aimed at challenging authorities, protesting subsistence crises, or resisting state and local policies and actions—it excludes brigandage and strictly criminal activities. The survey sets an inclusive threshold, recording any uprising that lasted only a few hours or involved as few as four individuals. These criteria thus cast a wide net, making the Jean Nicolas survey an unparalleled thermometer of the rebellious fever that swept through France in the run-up to the Revolution.

Figure 4 depicts trends in rebellious activity between 1714 and 1789, our period of analysis. Rebellions began to intensify in the 1740s before accelerating sharply in the 1760s, with more than a hundred events per year from that point onward. This escalation coincided with the growing presence of the state across the territory and rising levels of taxation (Burguière and Revel Reference Burguière and Revel1989).Footnote ³² Rebellious activities further intensified until the outbreak of the Revolution. Their territorial spread—over about three thousand locations—is shown in Figure 5.Footnote ³³

Figure 4. Annual Number of Rebellions

Note: This figure displays the annual number of rebellions from 1714 to 1789 based on Nicolas’s (Reference Nicolas2002) original survey along with a decade-moving average.

We digitized all forty thousand original coding sheets of Nicolas’s (Reference Nicolas2002) survey to construct a comprehensive database of the 6,000 rebellions that took place in 3,172 parishes within the boundaries of the kingdom of France between 1714 and 1789.Footnote ³⁴ In addition to recording the date and location of each event, we also collected information on its motives, size, and the characteristics of its participants.Footnote ³⁵ To measure rebellious activity, we create a variable that counts the number of rebellions per parish and decade. We report summary statistics in Panel B of Table 1.Footnote ³⁶

Figure 5. Spatial Distribution of Rebellions (1714–89)

Note: This figure displays the spatial distribution of rebellions from 1714 to 1789 based on Nicolas’s (Reference Nicolas2002) original survey. Each point corresponds to a parish that experienced at least one rebellion over the period—6,000 rebellions across 3,172 parishes in total. The underlying shapefile of the kingdom of France as of 1789 is based on Gay, Gobbi, and Goñi’s (Reference Gay, Gobbi and Goñi2024a) jurisdictions database.

Nearly half of all rebellions (46%) that occurred from 1714 to 1789 were of moderate size, involving between 11 and 50 participants. Another 21% were relatively large, with more than 50 participants. State authorities were targeted in 60% of rebellions, while 11% were directed against non-state actors—the nobility, seigneurial and municipal authorities, and the Church. Among rebellions against the state, most were aimed at taxation (66%), followed by the military (19%) and the judiciary (23%).Footnote ³⁷ Finally, 69% of rebellions in our sample include information on the social categories of their participants. Among these, 39% involved laborers; 32%, craftsmen; 23%, notables; and 22%, socially marginalized individuals.Footnote ³⁸ The rebellions in our dataset thus span a broad range of the social strata of French society.

While Nicolas’s (Reference Nicolas2002) survey remains the most comprehensive source of rebellious activity in France prior to the Revolution, its use requires careful attention to potential gaps and reporting biases common to historical data (Gay Reference Gay2025, 29–46). These include not only the archival recording of events, but also the possibility that the establishment of a horse-post relay may have increased the capacity of state or local authorities to record rebellions. We address these concerns below by assessing the robustness of our results to these potential sources of bias, showing that they are unlikely to account for our findings.

Administrative and Territorial Divisions

Central for our empirical strategy, parishes were embedded within a hierarchical system of administrative and territorial divisions. The largest of these were généralités. Created in the fourteenth century, these broad territorial divisions served as the basis for the administration of the territory under the authority of intendants (Gorry Reference Gorry and Zadora-Rio2014). On the eve of the Revolution, there were 32 généralités.Footnote ³⁹ Subdélégations were territorial divisions established in the mid-seventeenth century as subdivisions of généralités to extend the territorial reach of the royal authority at the local level. The prerogatives of subdélégués—the royal agents who administered subdélégations under the authority of the intendant of their généralité—increased during the eighteenth century to encompass the fiscal, military, economic, social, and sanitary administration of their territory (Goubert, Arbellot, and Laclau Reference Goubert, Arbellot, Laclau, Nordman, Ozouf-Marignier and Laclau1989). By 1789, there were nearly 700 subdélégations. A more localized territorial unit was the canton, which held various electoral and judicial functions (Ozouf-Marignier, Graziani, and Bonin Reference Ozouf-Marignier, Graziani, Bonin, Nordman, Ozouf-Marignier and Laclau1989). Alongside parishes, cantons represented the most granular level of administration, totaling over 4,500.Footnote ⁴⁰ Although they were only formally created in 1790 along with départements and districts, their territorial scope closely reflected pre-existing patterns of local interaction and daily exchange between parishes (Tanguy, Le Bihan, and Lagadec Reference Tanguy, Bihan and Lagadec2009). At the median, each généralité had 988 parishes, 782 thousand inhabitants (as of 1793), and covered a radius of 70 kilometers; each subdélégation had 39 parishes, 31 thousand inhabitants, and a radius of 15 kilometers; and each canton had 7 parishes, 5 thousand inhabitants, and a radius of 6 kilometers.

Other Data

We supplement our parish-level data with additional sources, although data for this time period remain relatively sparse. Our research design explicitly accounts for this constraint. In particular, we construct a parish-level measure of decade-average wheat prices based on data from Ridolfi (Reference Ridolfi2019), so as to control for shifting local economic conditions and for potential income shocks associated with the local development of the horse-post relay network—subsistence crises and broader changes to the price of staple foods were common triggers of collective action in eighteenth-century France (Bouton Reference Bouton1993; Chambru Reference Chambru2019; Markoff Reference Markoff1985).Footnote ⁴¹ In addition, we measure the size of each parish by its population in the 1793 census—the first census conducted in France. We also assess the connectivity of parishes to the road network based on Cassini’s maps published in 1756–90. Furthermore, we determine whether a parish served as the seat of an administrative division, and we collect information on the type of gabelle taxation regime to which it was subject. Finally, we construct a historical GIS of both the letter-post system and maréchaussée brigades, which we describe in more detail below.

Robustness

Table 3 assesses the robustness of our main results. These tests build on the full specification reported in column 6 of Table 2—reported here in column 1—which includes controls for wheat prices and decade-by-canton fixed effects.

Sample Composition

Next, we hold constant the sample of switchers used for identification to assess whether changes in the effective sample drive our results. The estimate in Column 3 of 0.035 suggests that they do not. Column 4 further restricts the sample to parishes that switch into treatment—that is, parishes that receive a new horse-post relay—as opposed to using both switchers in and switchers out—that is, parishes that also lose a relay. The estimate declines slightly to 0.022 but remains significant at the 1% level and consistent with the baseline.

Table 3. Robustness

Note: This table reports the average total effect of non-normalized event-study effects $ {\delta}_{\ell } $ estimated through Equation 1. All regressions include parish and decade fixed effects, as well as wheat prices and decade-by-canton fixed effects. The estimation is restricted to two leads and three lags. Column 2 removes parish-decades of the 1710s. Column 6 includes an additional set of decade-by-rural status fixed effects. Column 8 restricts horse-post relays to those that are at least 15 kilometers from a retired relay in the same decade, and column 9, to those that are within 15 kilometers, where Reconf. denotes reconfigured relays. The outcome is the number of rebellions in columns 1–9, and an indicator variable for the occurrence of at least one rebellion in column 10. Parish-decades and Parishes denote the number of observations effectively used in the estimation. Switchers (overall) denotes the number of parish-decade observations that switched treatment over the period, while Switchers (at least once) denotes the number of parishes that switched at least once. Outcome residual s.d. denotes the outcome standard deviation once decade, parish, and canton fixed effects, as well as wheat prices are accounted for. Estimates are calculated using de Chaisemartin et al.’s (Reference de Chaisemartin, Ciccia, D’Haultfœuille, Knau, Malézieux and Sow2023) did_multiplegt_dyn Stata command. Standard errors are clustered at the parish level and reported in brackets. $ {}^{***}p\le 0.01.\hskip2.21384pt {\hskip2.21384pt }^{**}p\le 0.05.\hskip2.21384pt {\hskip2.21384pt }^{*}p\le 0.10 $ .

The Role of Pre-Existing State and Administrative Fragmentation

The causal relationship we have identified between horse-post relays and rebellions thus far was part of a broader process of state-building—one in which persistent state fragmentation and enduring subnational divisions continued to undergird local unrest. For instance, boundaries between different tax regimes, particularly those associated with the gabelles, became hotspots for tax evasion and smuggling that disproportionately resisted the state’s centralization efforts (Davoine, Enguehard, and Kolesnikov Reference Davoine, Enguehard and Kolesnikov2025). More broadly, responses to the establishment of horse-post relays may have varied with the historical relationship to the central state, as territories incorporated earlier into the kingdom (pays d’Élection) were under direct monarchic authority—especially in fiscal matters—whereas peripheral territories incorporated later (pays d’États and pays d’Imposition) retained some degree of autonomous ruling. Indeed, these latter regions had historically offered stronger opposition to the central state and were more likely to oppose efforts at state centralization and its penetration into local society (Degrave Reference Degrave2023).

Table 5. Heterogeneity Across State and Administrative Fragmentation

Note: This table reports the average total effect of non-normalized event-study effects $ {\delta}_{\ell } $ estimated through Equation 1. All regressions include parish and decade fixed effects, as well as wheat prices and decade-by-canton fixed effects. The estimation is restricted to two leads and three lags. Columns 1 and 2 distinguish between parishes that are in cantons adjacent to gabelles frontiers versus not, and focus on rebellions targeting taxation authorities, and Columns 3 and 4 distinguish between parishes that are in pays d’États or pays d’Imposition versus pays d’Élection, and focus on all rebellions. Parish-decades and Parishes denote the number of observations effectively used in the estimation. Switchers (overall) denotes the number of parish-decade observations that switched treatment over the period, while Switchers (at least once) denotes the number of parishes that switched at least once. Outcome residual s.d. denotes the outcome standard deviation once decade, parish, and canton fixed effects, as well as wheat prices are accounted for. Estimates are calculated using de Chaisemartin et al.’s (Reference de Chaisemartin, Ciccia, D’Haultfœuille, Knau, Malézieux and Sow2023) did_multiplegt_dyn Stata command. Standard errors are clustered at the parish level and reported in brackets. $ {}^{***}p\le 0.01.\hskip2.21384pt {\hskip2.21384pt }^{**}p\le 0.05.\hskip2.21384pt {\hskip2.21384pt }^{*}p\le 0.10 $ .

We provide evidence for the conditional role of pre-existing state and administrative fragmentation along these dimensions. To test for the role of fiscal fragmentation, we split the sample between parishes located near gabelles frontiers—defined as those in cantons directly adjacent to these fiscal boundaries—and those farther away. We then estimate the treatment effect of the introduction of horse-post relays on rebellions specifically targeting taxation authorities.Footnote ⁶¹ The results, reported in Table 5, indicate a markedly stronger response to state expansion in fiscally fragmented areas. Indeed, the estimate in column 1 represents a little more than the average incidence of rebellions against taxation, while the column 2 estimate represents only one fifth. Next, to test for the role of the mode of ruling more broadly, we split the sample between parishes in pays d’États and pays d’Imposition relative to the more historically centralized pays d’Élection. Footnote ⁶² We then estimate the treatment effect of the introduction of horse-post relays on all rebellions. Consistent with our hypothesis, we find a stronger response to state expansion in areas that were historically more remote from the state’s direct authority, with the estimate in column 3 representing about three times the average incidence of rebellions relative to a little less than this average in column 4.Footnote ⁶³

Gathering Effects

Next we investigate whether the establishment of new horse-post relays impacted rebellious activity in nearby parishes. If new relays brought material changes that extended beyond their host parish—as state agents sought to reach into surrounding areas as well—they could have “gathered” rebellions from nearby, drawing people to express grievances at the locales of this new representation of state authority. The built environment and increased traffic associated with new horse-post relays may have opened new spaces for contention, taking a step in the direction of Tilly’s (Reference Tilly1986) theory of evolving repertoires of contention from dispersed, parochial revolts to more clustered and cosmopolitan forms of collective action. Contributing to this shift, new horse-post relays and postillons clearly identified with the markings of increased state visibility and may have represented a disruptive threat to the existing social order. This dynamic constitutes a complementary potential mechanism that extends beyond the parishes where new relays were established.

To test this hypothesis, we estimate three versions of Equation 1, redefining the treatment as an indicator variable that equals one for the presence of a horse-post relay in neighboring parishes within 5 kilometers, 5–10 kilometers, and 10–15 kilometers.Footnote ⁶⁴ Results in Figure 8 imply that the number of rebellions decreased by 0.007 in parishes located within 5 kilometers of a new horse-post relay and by 0.005 in those 5–10 kilometers away, with both estimates significant at the 5% level. Rebellions further decreased by 0.002 in parishes located 10–15 kilometers away from a new relay, although this estimate is not statistically significant.

Figure 8. Average Total Effect of New Horse-Post Relays on Rebellions across Distance to Nearest Horse-Post Relay

Note: This figure reports the average total effect of a new horse-post relay on rebellion when various treatment variables are input into Equation 1, along with 95% confidence intervals: a horse-post in a given parish and decade, up to 5 kilometers away from a new horse-post (excluding parishes with a relay), 5–10 kilometers away, and 10–15 kilometers away. All regressions include parish and decade fixed effects as well as wheat prices, decade-by-canton fixed effects, and distance to the nearest horse-post relay. Standard errors are clustered at the parish level. The full set of results is available in Table A15 in the Supplementary Material. $ {}^{***}p\le 0.01.\hskip2.77695pt {\hskip2.77695pt }^{**}p\le 0.05.\hskip2.77695pt {\hskip2.77695pt }^{*}p\le 0.10. $

We interpret these results as suggestive evidence that part of the main effect we identify may reflect a local reconfiguration of rebellious activity from surrounding parishes toward those hosting new horse-post relays. Given the small size of parishes and their close proximity, it would have been relatively easy for people to coordinate resistance locally and direct it toward nearby symbols of state authority. Importantly, however, even after accounting for this “gathering effect” of rebellion, the overall net effect of a new relay on rebellion remains positive and twice as large as the cumulated spillover effects.

Disruption of Local Social Hierarchies

Research in other contexts suggests that the erosion of traditional social hierarchies and networks that once structured local society can lead to disorder and to reprisals against declining authorities. For instance, Scott (Reference Scott1976) and Lecce, Ogliari, and Orlando (Reference Lecce, Ogliari and Orlando2022) show that when the disruption of traditional social relations weakens the monopolistic grip of local elites and social networks, it can leave populations vulnerable to severe risks and enable everyday grievances to metastasize rather than be absorbed or resolved locally. These grievances may be directed at both declining elites and the state. Hence, if the expansion of the state near horse-post relays undermined traditional authorities—such as the Church, the nobility, and seigneurial authorities—without eliminating the rents and tithes they extracted, resulting discontent against these declining figures could generate the observed link between new horse-post relays and rebellion. Grievances may also have been targeted at the state itself, especially if it failed to replace the functions previously carried out by these traditional institutions with equivalent state-provided services.

Several pieces of evidence cast doubt on this competing explanation. The establishment of new horse-post relays typically involved forging alliances with select local elites rather than their weakening, suggesting little disruption to traditional hierarchies (Jamaux-Gohier Reference Jamaux-Gohier2001; Marchand Reference Marchand2006). Accordingly, we find little evidence that new horse-post relays triggered rebellions against traditional non-state authorities. Unpacking the null coefficient in column 2 of Table 4, we find no significant impact on rebellions against the Church or municipal authorities, and if anything, a slightly negative effect on rebellions against seigneurial authorities and nobles. We also find no effect on the broader set of local civilian elites, such as notables and labor authorities.Footnote ⁶⁵

War Zones and Wartime

War considerations may also have impinged on the horse-post relay network and rebellion by fueling both the network’s expansion and grievances against the military in border areas prone to or exposed to war. While most changes to the network during the eighteenth century occurred in the interior of the country, substantial extensions took place in the 1740s and 1750s along the northeastern border with the Holy Roman Empire and in the 1760s along the Swiss border. However, the treatment effect is broadly homogeneous across regions and not driven by a single one. Moreover, these border areas were not necessarily those with the highest levels of conscription, nor were rebellions against the military concentrated there.Footnote ⁶⁶ Furthermore, the supply routes and stations for troops were organized through the routes d’étape system, which was distinct from the horse post and followed a different spatial logic (Kroener Reference Kroener1980).Footnote ⁶⁷ Nonetheless, to ensure that the military geography does not confound our results, we exclude subdélégations that experienced battles or were along the borders of the kingdom.Footnote ⁶⁸ The results remain in line with our baseline estimates. We also assess the role of wartime involvement directly by excluding all rebellions that occurred during periods of war, where we define wartime as any year in which France was engaged in a military conflict in Europe.Footnote ⁶⁹ The results are again similar.

Information and Collective Action Through the Letter-Post System

Another potential alternative explanation is the simultaneous development of the letter-post system, which may have increased the likelihood of rebellion by lowering barriers to collective action through the wider dissemination of information.Footnote ⁷⁰ Indeed, such dynamics linking improved communication infrastructure to collective mobilization have been observed with the expansion of railways and telegraphs in nineteenth- and twentieth-century United States (García-Jimeno, Iglesias, and Yildirim Reference García-Jimeno, Iglesias and Yildirim2022) and Europe (Melander Reference Melander2020; Pengl et al. Reference Pengl, Müller-Crepon, Valli, Cederman and Girardin2025). In eighteenth-century France, the letter post operated as a private service for collecting, forwarding, and delivering letters via an infrastructure of post offices (Marchand Reference Marchand2006, 95–180). If the establishment of letter-post offices coincided with that of horse-post relays—and if those offices facilitated collective action—this correlation could confound our results.

To address this potential concern, we collected information on the locations of letter-post offices for the same years as our data on the horse-post relay network.Footnote ⁷¹ Critically, the spatial configuration of letter-post offices followed a different logic from that of horse-post relays, as their establishment followed economic rather than strategic and logistical considerations (Cauvin, Lepetit, and Reymond Reference Cauvin, Lepetit and Reymond1987; Verdier and Chalonge Reference Verdier and Chalonge2018). Letter-post offices were as numerous as horse-post relays and expanded at a comparable pace throughout the eighteenth century, so that they covered most of the territory by the early eighteenth century.Footnote ⁷² However, unlike the horse post, the letter post was not designed as a network requiring offices at regular intervals. Instead, letter-post offices were generally located in populous towns, in sharp contrast with horse-post relays.Footnote ⁷³ Their buildings were also much smaller, as they handled only letter distribution and did not serve as rest stops or provide fresh horses. Mail delivery in the letter-post system relied primarily on private couriers (entrepreneurs de dépêches), who rode their own horses, and on mail coaches (malles-poste), which consisted of two-wheeled carts operated by employees of the letter-post administration.Footnote ⁷⁴

To empirically assess whether the development of letter-post offices confounds our results, we apply de Chaisemartin and D’Haultfœuille’s (Reference de Chaisemartin and D’Haultfœuille2023) estimator, which accommodates staggered designs with binary, non-absorbing multiple treatments that may exhibit heterogeneous and dynamic effects. Specifically, we augment Equation 1 by including a set of event-study coefficients on an indicator variable that equals one if a letter-post office was present in a given parish and decade. The results closely mirror our baseline estimates: the average total effect of the horse post on rebellion, accounting for the concurrent development of the letter post, is 0.027 (standard error: 0.011), suggesting that this alternative mechanism does not drive our findings.Footnote ⁷⁵

Transit Infrastructure and the Road Network

Another potential confounder is the underlying road network on which the horse-post relay system depended, although this infrastructure was managed separately by the Bridges and Roads administration (Arbellot Reference Arbellot1973). If improvements to the road network coincided with the establishment of new horse-post relays, it is possible that the effects we observe were instead driven by changes in the transit infrastructure itself—whether by fostering economic development, disrupting non-state local authorities, or facilitating popular coordination.

In the early eighteenth century, French roads were in a deplorable state (Arbellot Reference Arbellot1973, 766). These conditions had been a longstanding concern and lamented since the Colbert administration as a major obstacle to the expansion of internal trade (Blond Reference Blond2013, 65). However, it was not until the 1730s—by which point the situation had become critical—that the Controller General of Finance Philibert Orry launched an ambitious program of road repair and construction (Massounie Reference Massounie2013).Footnote ⁷⁶ Spearheaded by Daniel-Charles Trudaine, the program ultimately resulted in the construction of nearly 26 thousand kilometers of road by the 1780s—although only two-thirds of this network remained in passable condition by that time (Arbellot Reference Arbellot1973, 772–3).

The order of priority for new road construction, however, was not directly linked to the location of horse-post relays, as roads connecting relays were considered only after those linking Paris to ports, provincial capitals, and the kingdom’s borders (Blond Reference Blond2007, 69–70). While the upkeep of existing roads may have occasionally favored those used by the horse post (Arbellot Reference Arbellot1973, 766), it was exceedingly rare for the creation of a horse-post relay to prompt a request for road construction or maintenance from the Bridges and Roads administration, or that such requests would be granted (Demangeon Reference Demangeon1995, 66–71; Verdier Reference Verdier2009, 13). Overall, historical accounts suggest that the horse post was not tied to the road network and followed its own logic, and that the actual routes taken by postillons between relays often varied seasonally depending on weather and road conditions (Jamaux-Gohier Reference Jamaux-Gohier2001, 125–45).

A comprehensive empirical test of this alternative mechanism is however not feasible due to the severe limitations of available quantitative information on road construction and maintenance at the local level in the eighteenth century.Footnote ⁷⁷ Nevertheless, we show in Tables B1 and B2 in the Supplementary Material that there is little heterogeneity in the treatment effect of horse-post relays across parishes located at different distances from a road. This suggests that our results are unlikely to be driven by connectivity to the road network.

Policing by the Maréchaussée Brigades

A final potential alternative mechanism is the presence of maréchaussée brigades, the state-sponsored security force that policed the countryside. These men-at-arms formed part of the security apparatus of the royal authority and maintained public order along the kingdom’s roads (Fressin Reference Fressin2022). Following their consolidation in 1720, brigades consisting of four to five men were stationed in barracks located in the center of towns and urban areas (Fressin and Georges Reference Fressin and Georges2022). Their spatial distribution closely resembled that of letter-post offices: they covered the entire territory and were generally located in populous towns.Footnote ⁷⁸ However, the expansion of the maréchaussée was relatively limited: the number of brigades remained stable at five hundred units during the first half of the eighteenth century, increased to about eight hundred in the 1750s and 1760s, and then plateaued until the Revolution.Footnote ⁷⁹ While these brigades embodied the symbolic presence of the state, they never mustered more than three thousand men and primarily operated in the vicinity of towns and major roads.Footnote ⁸⁰

Maréchaussée brigades could potentially account for our findings through two distinct channels. First, if brigades systematically accompanied the establishment of new horse-post relays, the effect we identify might reflect the presence of the maréchaussée rather than that of the horse post itself. Second, if the presence of brigades made rebellions more likely to be recorded by these state actors, our results could be driven by selection bias in the archival sources used to document rebellious activity.

We address these concerns in a number of ways. First, we show that although parishes that received a horse-post relay were more likely to eventually host a maréchaussée brigade, pretreatment trends indicate that they were also more likely to have received a brigade prior to the establishment of the relay.Footnote ⁸¹ In other words, while the spatial development of the two institutions was correlated, there is no evidence of a causal link between them. Second, we extend Equation 1 to control for the potential incidence of maréchaussée brigades using de Chaisemartin and D’Haultfœuille’s (Reference de Chaisemartin and D’Haultfœuille2023) estimator for multiple treatments. The results closely resemble our baseline estimates, suggesting that the presence of maréchaussée brigades does not account for our findings.Footnote ⁸² Third, we restrict the analysis to parishes that never hosted either a letter-post office or a maréchaussée brigade. Results in Table A18 in the Supplementary Material suggest that these “uncontaminated” parishes still exhibit a positive causal relationship between the establishment of horse-post relays and rebellion. Finally, we test the robustness of our findings to removing rebellions recorded in the archives of the maréchaussée, finding similar results (see column 6 of Table A13 in the Supplementary Material).