ASRXXX10.1177/0003122415618991American Sociological ReviewHechter et al. 6189912015

American Sociological Review 25–­1 Grievances and the Genesis © American Sociological Association 2015 DOI: 10.1177/0003122415618991 of Rebellion: Mutiny in the http://asr.sagepub.com Royal Navy, 1740 to 1820

Michael Hechter,a Steven Pfaff,b and Patrick Underwoodb

Abstract Rebellious collective action is rare, but it can occur when subordinates are severely discontented and other circumstances are favorable. The possibility of rebellion is a check—sometimes the only check—on authoritarian rule. Although mutinies in which crews seized control of their vessels were rare events, they occurred throughout the Age of Sail. To explain the occurrence of this form of high-risk collective action, this article holds that shipboard grievances were the principal cause of mutiny. However, not all grievances are equal in this respect. We distinguish between structural grievances that flow from incumbency in a subordinate social position and incidental grievances that incumbents have no expectation of suffering. Based on a case- control analysis of incidents of mutiny compared with controls drawn from a unique database of Royal Navy voyages from 1740 to 1820, in addition to a wealth of qualitative evidence, we find that mutiny was most likely to occur when structural grievances were combined with incidental ones. This finding has implications for understanding the causes of rebellion and the attainment of legitimate social order more generally.

Keywords social movements, collective action, insurgency, conflict, military authority

Since the 1970s, grievances have had a roller grievances that are situational and unlikely to coaster career in studies of insurgency and appear in standard datasets, together with the collective action. Once regarded as prime structural grievances traditionally regarded as causes of conflict, grievances were subse- causes of protest. This conclusion has impor- quently shelved in favor of factors like tant policy implications. resources, political opportunities, and state capacity. This article makes the case that the literature has thrown the baby out with the bath water. Based on an analysis of mutiny in aArizona State University and University of the Royal Navy (hereafter, RN), which averts Copenhagen the selection bias that so often dooms studies bUniversity of Washington of collective action, we reinstate the causal priority of shared grievances in explaining Corresponding Author: Michael Hechter, Arizona State University, insurgency. At the same time, we contend that School of Politics and Global Studies, PO Box the concept of grievances is too capacious. 873902, Tempe, AZ 85287 Insurgency is best explained by incidental E-mail: [email protected]

Downloaded from asr.sagepub.com by guest on December 19, 2015 American Sociological Review 2

Before 1970, the major perspectives on (Cederman, Gleditsch, and Buhaug 2013). collective action and social movements Grievances cannot generate collective action emphasized grievances as principal causes on their own, however. The immediate impe- (McAdam, McCarthy, and Zald 1996; Pinard tus to collective action is cognitive. Successful 2011). Whereas collective behavior was collective action proceeds from a significant regarded as driven by emergent norms and transformation in the collective consciousness values, later research emphasized the ration- of the actors involved (Gamson, Fireman, and ality of participants. Gurr (1970) famously Rytina 1982). Activists use extant grievances argued that individuals who perceived they to construct legitimating accounts—or were disadvantaged relative to some refer- frames—to support their activism (Benford ence group were motivated to take action as a and Snow 2000; cf. Walder 2009). result. He claimed that psychological tension How can shared grievances be measured? was relieved by participation in protest and They can be measured directly in sample sur- argued that survey research could be deployed veys, but most such studies are geographi- to assess the theory. A host of studies tested cally and historically limited and liable to the theory, but it received little empirical sup- suffer from selection bias. Indirect meas- port (Brush 1996). Because individual griev- ures—based on the assumption that griev- ances are idiosyncratic, it is unsurprising that ances can be inferred from individuals’ one’s attitudes about such things as career subordinate positions in a social structure— satisfaction, economic well-being, living con- can be used to study grievances across greater ditions, and children’s welfare do not turn out geographic and temporal space. For example, to be significant determinants of collective one can assume that political and economic action. The focus of ensuing research there- inequalities affecting entire groups are likely fore shifted from social psychological to more to fuel resentment and justify attempts to macro-sociological accounts. fight perceived injustice. Moreover, shared Following Olson (1965), who claimed that grievances likely have an emotional valence grievances were insufficient to motivate col- (Kemper 2001) that makes collective identi- lective action, resource mobilization and ties more salient (Hechter 1978; Østby 2008; political process theorists argued that griev- Stewart 2008). ances essentially were constant and thus If it turns out that such structural variables could not account for variations in collective fail to determine collective action, does this action. Instead, resources and opportunities mean grievances can be ruled out as a cause? available to potential insurgents and their There is no warrant to believe that structural targets were what varied (Collier and Hoeffler variables provide the only measures of griev- 2004; Fearon and Laitin 2003). In this view, ances. Grievances can also emerge from a any small group combining collective identity quite different source. Theorists as diverse as with rudimentary forms of organization can Weber, Durkheim, and Mead all held that potentially mobilize for collective action. incumbency in a social role entails a set of This led to the expectation that prior ties to a behaviors that individuals are obligated to movement member, membership in organiza- enact. For Weber ([1918–21] 1978:227), pat- tions, a history of prior activism, and low rimonial rulers were obliged not to exceed the opportunity costs would foster participation traditional limits to their power. Serfs’ exist- in collective action. ence was grim, but masters also had certain Now the tide is turning once again. The obligations toward serfs—providing access to revival of grievance-based explanations of the commons and sharing food during fam- collective action holds that people who share ines—and any lord who failed to do so grievances due to their subordinate positions courted trouble. Likewise, Durkheim ([1897] in class and political structures are more 1951:249–50) explained that individuals in likely to mobilize to enhance their welfare nineteenth-century European societies had

Downloaded from asr.sagepub.com by guest on December 19, 2015 Hechter et al. 3 little expectation that their fortunes could typically skeptical about their collective effi- exceed those of their parents. In consequence, cacy, they are difficult to mobilize: the fear they accepted (with resignation) norms about and shame associated with subordination are their place in the stratification system. Finally, demobilizing. Mead (1934) insisted that social roles make When incidental grievances are severe, individuals act as if they were characters in a they are manifest to all affected. It is thus easy play or athletic contest. Given a set of rules of for victims of these unforeseen events to the game, people’s actions are normally ori- develop a collective identity (Klandermans ented toward that of the other participants in 1997). So long as the group has some shared predictable ways. The failure to honor these conception of fairness, injustice framing is expectations is jarring: shortstops should not relatively straightforward. The simultaneous field a hit and throw it to the right fielder experience of the same threat, injury, or insult rather than the first baseman. improves coordination by providing a com- This classical insight has been elaborated mon focal point among disparate actors, a in recent social psychological research. For substitute for formal organization that is espe- social psychologists, no compelling causal cially powerful in socially dense settings. As account of mobilization can exclude griev- Schelling (1960:90) notes, “The role of ‘inci- ances (Klandermans 1997). But there are at dents’ can thus be seen as a substitute for least two different kinds of grievances. Struc- overt leadership and communication.” For tural grievances derive from a group’s disad- groups that lack many of the resources that vantaged position in a social structure, enable collective action, incidental griev- whereas incidental grievances arise from ances help trigger spontaneous mobilization. wholly unanticipated situations that put groups at risk. Unlike structural grievances, incidental ones—like unexpected disasters, Mutiny in the Royal Navy major court decisions, and state repression— We argue that the combination of structural enhance a group’s capacity to coordinate and incidental grievances better explains the (Turner and Killian 1972; Useem 1998; Walsh genesis of insurgent collective action—in this 1981). case, mutiny in the RN in the Age of Sail. All These two types of grievances have differ- seamen (that is, sailors and marines) in the ent psychological implications (Bergstrand RN were subject to harsh conditions—as 2014; Tausch et al. 2011; Van Stekelenburg Samuel Johnson (Boswell [1791] 1934– and Klandermans 2013; Van Zomeren, Post- 50:438) famously described their lot: mes, and Spears 2008). Incumbents confined to subordinate positions in a social structure No man will be a sailor who has contrivance tend to tolerate grievances like poor wages, enough to get himself into jail; for being in difficult working conditions, and political a ship is being in a jail, with the chance of exclusion so long as they remain at routine being drowned. . . . A ship is worse than a and predictable levels. If people understand jail. There is, in a jail, better air, better com- that bearing such indignities is their lot in life, pany, better conveniency of every kind; and they are not likely to seek redress unless the a ship has the additional disadvantage of magnitude of those grievances increases being in danger. sharply. For example, graduate students suf- fer anxiety about their comprehensive exams Yet, most seamen never took part in a mutiny. but generally do not mount campaigns to What distinguishes the fate of ships that abolish them. Under normal conditions, struc- experience mutinies from that of other ships? tural grievances foster stable expectations Although dozens of mutinies occurred over that tend to discourage protest. Moreover, our study period, mutiny was a relatively rare because members of subordinated groups are event. The rarity of mutiny, we will show,

Downloaded from asr.sagepub.com by guest on December 19, 2015 American Sociological Review 4 owes in part to individuals’ tendency to toler- notable examples (Dening 1992; Frykman ate structural grievances under normal condi- 2010; Neale 1985; Pack 1964; Pope 1987), tions. Harsh as it was, life aboard ships every previous study has selected on the ordinarily met seamen’s expectations as dependent variable. This liability plagues determined by settled customs and predicta- analyses of protest generally (McAdam and ble levels of reward and punishment. Such Boudet 2012). beliefs establish baseline expectations about This study compares a sample of mutinies the treatment that subordinates will encounter that occurred on RN ships from 1740 to 1820 (Kahneman 2011). When these expectations with a random sample drawn from the much are violated, individuals are more likely to larger population of ships that had no such seek redress. rebellion. Scholars of mutiny have endorsed Mutiny is the byproduct of a contest such a design (Lammers 2003), but no previ- between the two principal collectivities ous study has adopted it. Fortunately, RN onboard ship. On one side stands command documentation was excellent, providing origi- (the captain and his officers); on the other nal records that allowed us to analyze system- side stands the crew. The contest is shaped by atic quantitative data as well as rich historical officers’ ability to provide good governance, evidence about mutiny and its causes. and by the crew’s grievances and its capacity Grievances motivate radical protest to the to coordinate collective action. Mutiny is thus extent that they inspire indignation among the outcome of the conjunction of three sets group members and focus their discontent. of factors. The first two factors refer to the We contend that the potential for radical pro- demand for mutiny: poor governance (the test should be greatest when governing provision of insufficient collective goods like authorities are seen as the immediate cause of safety, health, and welfare, and the cultural grievances or as failing in their responses to inappropriateness of rule), and the crew’s challenging situations. Members of subordi- perception that the situation confronting them nated groups assess legitimacy in terms of the requires concerted action to avoid a worsen- integrity of the authorities and the degree to ing of their welfare. which authorities address their interests (Car- The third refers to the supply side of rabine 2005; Useem and Kimball 1989). They mutiny, which varies with a crew’s capacity expect authorities to act in ways they regard to coordinate collective action in response to as reliable, reasonably effective, and cultur- grievances. Although the crews of RN ships ally appropriate (Hechter 2013). Incidents had varying capacities to launch a mutiny, that convince seamen that the authorities are coordination generally was enabled by threats insensitive to their security and welfare put to safety and welfare, for in these circum- commanders at risk. Indignation should be stances free riding does not pay (Goldstone greatest where commanders are perceived as and Useem 1999; Kalyvas and Kocher 2007). acting unfairly and in ways that violate the To evaluate these ideas, we conducted the customs of the sea. first systematic study of naval mutiny. Previ- We expect structural grievances to increase ous research on mutinies shares many of the the odds of mutiny. The testable hypothesis is limitations of studies of other kinds of rebel- straightforward: lion. Most accounts focus on personalities and the idiosyncratic features of famous cases Hypothesis 1: The greater the magnitude of structural grievances, the greater the odds of (e.g., the Bounty). Although scholars have mutiny on an RN ship. surveyed commonalities among mutinies (Frykman 2009; Gilbert 1983; Guttridge In addition to structural deprivation, we 1992; Haine 1992; Hathaway 2001; Lammers expect seamen will be sensitive to incidental 1969, 2003; Rediker 1987; Rose 1982; Wood- grievances. If so, incidental grievances should man 2005) and produced detailed accounts of help trigger mutiny:

Downloaded from asr.sagepub.com by guest on December 19, 2015 Hechter et al. 5

Hypothesis 2: The greater the magnitude of in- Naval labor relations have been painted in cidental grievances, the greater the odds of terms of naked exploitation and conflict mutiny on an RN ship. (Frykman 2009; Neale 1985; Rediker 1987) and yet, although the ranks of officers were Collective action tends to be led by individu- thin, rebellion was unusual and most seamen als who are especially interested in the redress went about their duties most of the time. Why of shared grievances. In particular, ethnic was rebellion so rare? Repression surely heterogeneity among seamen would have played its part. Historians characterize the facilitated a critical mass around which to sailing navy as an “authoritarian system based mobilize rebellion (Centola 2013; Marwell on the principles of dominance and domina- and Oliver 1993). In the RN, the largest group tion” (Valle 1980:3). Naval law defined of non-British seamen, the Irish, should have mutiny loosely and without regard to out- been most prone to rebellion. In the eigh- come, and it was punishable by death (Rodger teenth and early-nineteenth centuries, Ireland 1982). Officers were armed and most ships was forcefully subordinated under a brutal carried a complement of marines providing colonial regime. Co-ethnicity not only facili- extra security. Naval law, shipboard organiza- tated coordination, but the Irish were particu- tion, and hierarchy thus combined to establish larly disaffected and had a preexisting the dominance of commanders and the credi- politicized group identity that made them, on bility of the threat of punishment for insubor- average, primed for grievance interpretation. dination (Byrn 1989; Claver 1954). Ethnically diverse ships also might have Yet social order in the RN did not rest on lacked the organizational cohesion that made repression alone. At the very least, the navy for obedient and satisfied crews (Costa and required a cadre of motivated and skilled Kahn 2008). men—not slaves—to sail the ship, ply its armament, and win battles. Obedience was Hypothesis 3: The greater the ethnic diversity of premised on seamen’s belief in the compe- a ship’s crew, the greater the odds of mutiny. tence of the captain and his regard for their welfare (Lavery 1998). Whereas factory Governance, Compliance, workers who evade their responsibilities may and Conflict in the not suffer the consequences, shirking sailors Royal Navy endanger the welfare of all. Given the stakes, seamen preferred the maintenance of disci- Conditions onboard RN ships were notori- pline and good order by a competent com- ously hard. A standard warship of 74 guns mander. Officers understood that morale was (about 160 feet in length) bore up to 500 men, vital; they thus had to provide incentives for resulting in crowding, privation, and the risk compliance, including the implicit commit- of accident and disease. Officers were lords ment to safeguard seamen’s interests and wel- of the ship, and many infractions were pun- fare (Rodger 1986, 2004). ishable by summary flogging or by a blow The governance of a ship combined bureau- from a knotted rope or cane. The quality and cratic and patrimonial elements. Commanders supply of food and drink often deteriorated held a non-purchasable royal commission, during long voyages. Many seamen died from were installed by the Admiralty, and were illness or shipwrecks. Seamen’s pay was subject to extensive regulations (Allen 2002). poor, and general rates had been set in the The RN sought to balance “political loyalty, middle of the seventeenth century. Com- social suitability and professional compe- manders frequently denied seamen customary tence” in its officers (Rodger 2004:112). shore-leave for fear they would abscond. Dur- Patronage and privilege were not enough to ing wartime, the Admiralty was compelled to secure a commission; every candidate was rely on impressments to fill the ranks. obliged to have spent several years at sea,

Downloaded from asr.sagepub.com by guest on December 19, 2015 American Sociological Review 6 including one as a midshipman or master’s nevertheless comprised up to a fifth of the mate, and pass a qualifying examination. This RN’s total manpower (Frykman 2009). yielded an expectation among seamen that Despite constituting about a third of the commanders would be competent in seaman- population of the British Isles, Irishmen com- ship and navigation. Commanders had to keep prised less than 12 percent of all commis- careful records that were submitted to the sioned officers of the RN—and these were Admiralty at the close of a mission. Their only Protestants, as Catholics (at least 80 conduct was subject to review and the loss of percent of the Irish population) were forbid- their ship, for any reason, resulted in a com- den to hold a commission (Lewis 1960). Irish pulsory court martial. The Admiralty tried to seamen were often poorly treated (Gill 1961). enact rules and incentives to discourage cru- One advice manual counseled officers that elty and arbitrary punishment (Byrn 1989; when deciding upon summary punishment for Claver 1954). Captains and masters main- disciplinary infractions, “if two men make tained separate logs so that detailed records complaints about each other, if one is known could be used to evaluate commanders. Offic- to be turbulent or an Irishman, flog him with- ers could scarcely afford a bad reputation; out further question and show the ship’s com- there were always surplus post captains at the pany the value of a good character” (Pope Admiralty’s disposal who could replace them. 1981:250). Not surprisingly, officers com- Commanders themselves were required to monly regarded Irishmen as troublemakers recruit and retain the best seamen (Earle (Frykman 2009; Neale 1985). Irishmen’s 1998). These measures yielded an expectation expectations concerning fair treatment must among seamen that officers would follow nec- have been diminished. essary regulations and customary procedures. Naval officers were at once all-powerful But this social order was fragile. For one and extraordinarily constrained. At sea, their thing, the RN failed to institutionalize the power had fewer effective limits than royal routine and effective use of voice by seamen. power ashore; this is especially conspicuous To make a complaint against officers, seamen in a captain’s power to inflict flogging at will. had to submit a petition to these very same This system of command had enormous officers. Submitted petitions typically were implications for the attribution of blame. By disregarded, so this created an unintended dint of their great power, captains were incentive for mutiny. By necessity, naval responsible for everything and liable to be the authorities allowed commanders broad disci- focus of a variety of incidental grievances. plinary discretion, and there was little direct Nevertheless, as deprived as seamen were, regulation of summary punishment. As a life aboard ship generally met their expecta- result, seamen were sometimes treated very tions in terms of compensation, conditions of badly. Although tyrannical officers were the service, and danger. Skilled and experienced exception, flogging was poorly regulated and seamen usually supported the status quo, sid- little was done officially to stop officers from ing with their officers against disorderly and flogging frequently and severely. incompetent men (Rodger 2004). Although Another source of strain in the RN was the today flogging is considered inhumane, sea- presence of many non-British seamen in the men did not object to it in principle; rather, ranks—in particular Irishmen, who were their ire resulted from its disproportionate heavily recruited and pressed. Although Ire- application (Byrn 1989; Claver 1954; Lavery land was ruled by the same king, British 2010; Pope 1981; Rodger 1986). Discipline national identity excluded the Irish by dint of was accepted so long as it was limited and religion and colonialism (Colley 2003; Hech- directed to real troublemakers and shirkers. ter 1998). Naval authorities were fearful that Most seamen did not spend an entire career in Irishmen were particularly prone to rebellion, the navy; consequently, they could compare but owing to shortages of maritime labor they naval with merchant service. Whereas

Downloaded from asr.sagepub.com by guest on December 19, 2015 Hechter et al. 7 discipline was more severe and pay lower in hopes of escaping RN service (17 cases). The the RN than on civilian ships, the burden of evidence shows that, although audacious, muti- work was lighter and food and drink were bet- neers’ tactics generally were premised on the ter regulated and more plentiful (Earle 1998). redress of grievances, and they achieved at The experience of life at sea informed stable least one of their stated goals in about half the expectations that generally kept structural cases. Mutineers often armed themselves, but grievances—such as flogging—in check. But they usually refrained from deadly violence this equilibrium sometimes tipped toward because they sought to negotiate with naval rebellion. authorities or wanted to avert lethal reprisals. Nevertheless, 25 mutinies (40 percent) did involve deadly conflict (more often initiated by An Anatomy of Mutiny in the authorities than by the mutineers), and even the Royal Navy when seamen refrained from violence and Mutiny is an insurrection against the consti- peacefully returned to duty, this did not indem- tuted order of a ship. The RN defined mutiny nify them from trial and punishment. Hanging as any form of individual or collective defi- or severe corporal punishment (usually by hun- ance of command or any communication or dreds of lashes) was inflicted on convicted planning to that effect. Historians have long mutineers in more than two-thirds of cases. recognized that the official definition lacks Our extensive collection of qualitative clarity (Gilbert 1983). We narrow our focus to data on these mutinies, drawn from primary mutinies that involved seizure of a ship by its and secondary historical sources, allows us to crewmen, or a halt to a ship’s operations by dissect mutiny. Two general features charac- collective action because seamen wished to terize rebellion among RN seamen: the per- compel their commanders or the naval author- ception of grievances through the lens of ities to act in their interest. Full-fledged muti- “ill-usage,” and the attainment of coordina- nies were rare and dangerous events. We tion through native leadership and commit- identified 62 cases of mutiny meeting our ment mechanisms. definition during the period of study (details on sample selection given below). Of Illness and Ill-Usage: The In our sample, mutinies occurred on ships Understanding of Grievances of all kinds, from ships-of-the-line to small armed vessels (the average mutinous ship in Historians with very different orientations our sample was 1,000 tons, carrying about agree that at sea “the ship’s company became 295 men). Mutinies were not simply crimes an effective, efficient collectivity, bound of opportunity that occurred when a crew was together in skill, purpose, courage, and com- remote from possible sanctions (e.g., the munity” (Rediker 1987:134–35; see also Bounty). The majority of our mutinies took Earle 1998; Lavery 2010; Neale 1985; Rod- place in the home waters of the British Isles, ger 1986). All concur that sailors had a sense and most (about 65 percent) occurred in har- of collective identity, distinctive traditions, bors and anchorages. They tended to happen and belief in their rights and dignity (their on ships close to centers of naval authority. “proper usage”). This occupational culture Mutinies occurred around the globe, includ- helped seamen assist one another and endure ing the Mediterranean (12 cases), the West difficult conditions, but it also provided them Indies (7), Africa (4), North America (3), and with potent injustice frames. The forecastle, in the East Indies and the Pacific (4). Nearly the portion of the ship reserved for their habi- all (58) took place during wartime. tation, gave them the pockets of free space in Our sample includes mutinies resembling which they could discuss their grievances armed strikes and work stoppages (45 cases), and coordinate a response (Rao and Dutta and also occasions when vessels were seized in 2012).

Downloaded from asr.sagepub.com by guest on December 19, 2015 American Sociological Review 8

Certainly, seamen resented their low wages Ferret in 1806, mutineers rebelled against (general rates of pay remained unchanged “ill-usage,” taking particular exception to a from 1658 through 1797), the frequency of captain who excessively beat his crew.3 impressment, the indefinite periods of service Ill-usage violated the seaman’s under- during wartime, “starting” with blows from standing of what was morally and culturally canes and ropes, and so on, but they seem appropriate. Thus, mutineers on the King- rarely to have rebelled against these practices fisher declared that “[m]en on this ship are because they were considered to be the sea- treated like Turks and not as Christians.”4 man’s lot. Rather, particular incidents of poor Seamen frequently employed the concept of governance sharpened and focused existing ill-usage to describe disregard for their health grievances. and welfare. At trial in 1764, mutineers on the Seamen strongly opposed what they called Panther were able to prove ill-usage, includ- “ill-usage.” Proper usage meant order, rou- ing being served spoiled and irregular food, tine, and deference to customary practices fatal neglect of sickened seamen, corruption, that protected seamen from cruel and capri- and delinquent wages.5 A similar set of griev- cious treatment, allowed them their small ances was framed as ill-usage by seamen comforts of regular food and grog, and aboard the Tremendous at Table Bay in 1797 ensured reasonable standards of health and (Theal 1898:161–85). On the Santa Monica welfare. Tyranny, overwork, and disregard for in 1781, mutineers cited a tyrannical captain, seamen’s welfare all constituted ill-usage. For the irregular service of meals, the denial of example, in 1799, seamen on the battleship grog, and the shortage of drinking water as Impeteux rioted and refused to go to sea until proof of ill-usage.6 On the Culloden in 1794, they could deliver a letter of grievances to seamen mutinied because their ship, having Admiral Bridport. The letter reads, in part, been run aground, required more extensive repairs than their captain would pursue.7 My Lord, we the Ship’s Company of the In the minds of seamen, illness flourished Impeteux would wish to make known to with ill-usage, hence illness sharply increased your Lordship the usage we have received grievances in a ship’s company. The more since Sir Edward Pellew commanded us is seamen who were debilitated by illness, the such we have never been used to in this war more work had to be done by the remaining by any other Captain we have sailed under portion of the crew. Unchecked sickness was it is unacceptable. Sir Edward has punished a lethal and immediate threat to the well- different people in the ship to the amount of being of a ship’s company. Authority could nine dozen and upwards for very frivolous collapse when health was severely compro- crimes since we left Cawsand Bay [and] . . . mised, as on the Wager in 1741 (Pack 1964). we never deserved the barbarous usage we Yellow fever overwhelmed officers’ capacity now experience at the hands of Sir Edward to care for the sick on the Camilla, and sea- . . . and his humanity to the sick is little bet- men mutinied on the grounds that the epi- ter than to the well.1 demic had left the ship severely undermanned. Beyond the lethal threat of the illness itself, In framing ill-usage, brutality and con- sickness created an undue burden on the sur- tempt for seamen loomed large. On the Beau- vivors and increased the danger that the ship lieu in 1797, mutiny was justified in terms of would founder if beset by a tropical squall.8 “ill-use” and cruelty, with mutineers espe- Captains were expected to be advocates cially condemning their officers for punishing for their men in dealings with naval authori- them for trivial offenses.2 A captain’s reputa- ties. The impending cessation of hostilities tion for ill-usage prompted an uprising on the increased grievances because seamen were Blanche before he could even assume his eager to be released after long mandatory commission (Dann 1988:208–09). On the terms of service. A delay in mustering out

Downloaded from asr.sagepub.com by guest on December 19, 2015 Hechter et al. 9

Table 1. Primary and Secondary Motives Cited in 62 Cases of Mutiny

Motive 1st Cited 2nd Cited Total (%)

Cruelty/Excessive Discipline 15 10 25 (27%) Wage Issues 13 2 15 (16%) Victuals: Food/Water/Spirits 8 4 12 (13%) Violation of Naval Customs 4 6 10 (11%) Denial of Shore-Leave 4 4 8 (9%) Impressments 5 2 7 (7%) Safety of Ship/Competence of Command 5 2 7 (7%) Service in Foreign Waters 4 2 6 (6%) Demand for Dismissal from Service 0 2 2 (2%) Piracy 2 0 2 (2%)

Total = 94

Source: Documentation found in the Admiralty papers, National Archives (UK), Public Records Office (TNA: PRO ADM). meant skilled seamen might not find a berth lamented by the contemporary British public, in a merchant ship. In addition, seamen were they may have been taken-for-granted hard- angered because wages were frequently in ships deemed inescapable. arrears as hostilities ended. Captains who failed to secure their release, payment of One and All: Coordination wages, or prize-monies thus violated an and Credible Commitments implicit pact between a commander and his followers. among Seamen Content analysis of the thousands of pages If seamen were most likely to rebel when of official documents, court-martial tran- grievances violated their expectations about scripts, and witness accounts assembled for proper usage, how did they manage to act col- this study reveal that both structural and inci- lectively? Given our expectations about the dental grievances were implicated in mutiny causal linkage between grievances and mobi- (see Table 1). Governance failures loomed lization, if severe, widely shared grievances large, with seamen citing shortages of food were imposed on seamen, then the initial and water and spoiled rations (often linked to hurdle in organizing a mutiny consists not of illness), and commanders violating accus- overcoming free riding but in attaining cred- tomed naval conventions and failing to ible commitments to rebel (Hardin 1982; address their concerns about the safety or Taylor 1997). navigational competence following an acci- There was little anonymity aboard a ship; dent. Incidents like these were not understood seamen were reasonably well informed about in isolation and appear to have aggravated one another, facilitating monitoring behavior existing structural grievances; over 40 per- and assessing loyalties. The confidence that cent of all primary and secondary motives one’s shipmates were reliable allowed would- given for a mutiny cite excessive discipline or be mutineers to overcome free riding, making wage issues. However, mutineers rarely coordination the greater hurdle to collective claimed to be motivated by other sources of action (Hechter 1987; Lichbach 1995). Coor- structural deprivation usually regarded as dination, in turn, grew out of the capacity of causes of mutiny, such as impressments, ser- some seamen to provide leadership, and some vice in foreign waters, and long periods of mechanism by which shipmates made credi- obligatory service. Whereas all of these seem ble commitments to support a prospective especially grievous and were frequently mutiny.

Downloaded from asr.sagepub.com by guest on December 19, 2015 American Sociological Review 10

Formal organizations and associations did 1983). Signatures were especially binding for a not propel mutinies (Pfaff, Hechter, and Cor- seaman, because he had to comply or explicitly coran forthcoming) or other protests in this break his commitment to his peers, thereby sul- era (Calhoun 1982). Historical accounts are lying his reputation. united in depicting a maritime world knit by Oath-taking, which was strictly forbidden bonds of interdependence and shared occupa- and subject to hanging, was another coordina- tional culture (Earle 1998; Lavery 2010; tion mechanism that reoccurred in our cases. Rediker 1987). The severity of naval sanctions enhanced the Documents reveal that when grievances credibility of commitments by deterring mounted and command was perceived as defection from the cause. If seamen might be incompetent or ineffective, seamen turned to hanged for swearing an oath, defecting from a experienced shipmates accustomed to provid- planned mutiny also posed substantial danger. ing everyday leadership. In our sample of the In circumstances like these, indiscriminate 199 men described by the authorities as “ring- retaliation by the authorities alters the logic of leaders” and “principal” mutineers, about 30 collective action, such that participation percent were petty officers (who constituted entails no collective action problem but non- just 13 percent of mutinous ships’ companies) participation does. On the Culloden, as many and another 50 percent were able seamen seamen as possible were sworn and a ring- (about 35 percent of mutinous ships’ compa- leader declared, so as to deter “skulkers” from nies). Experience and reputation conferred refusing to take sides. When the mutiny was authority. As a key intermediate group, these staged, the great majority took part in seizing seamen were vital to the maintenance of order the ship and holding it for several days. At and the genesis of mutiny (Lavery 1998). trial, seamen remained willing to stand by Coordination also grew from the seamen’s their oaths, even enduring punishment for own institutional innovations. Seamen devel- withholding evidence.9 That oaths would con- oped two important commitment mechanisms tain opportunism by achieving a norm of “one (Schelling 1960): round-robin circular letters and all” was a common refrain in the mutinies and the swearing of illegal oaths that facili- we studied. A petty officer who helped plan tated mobilization by making commitments the mutiny on the Ferret explained that with- credible. Both practices created expectations out an oath, “we would not be able to stick to that an adequate number of seamen would one another.”10 Oath-takers had an incentive seize a ship (Heckathorn 1996). Although to uphold the mutiny in the hope it would seamen had clear incentives to conceal con- prevail, and they would either escape the spiratorial activities after the collapse of a navy’s clutches or be offered amnesty as a mutiny or after their arrest, protest letters and result of successful negotiations. collective oaths occurred in more than half of The qualitative evidence is broadly support- our cases of mutiny, and they probably were ive of our hypotheses concerning the impor- clandestine in many others. tance of grievances and diversity among Protest letters stated the crew’s grievances seamen in generating mutinies. However, the and demands. Signatures provided leaders with odds of mutiny might also have been influ- information about how many and which of their enced by social structural and contextual fac- comrades supported a prospective mutiny, and tors, such as ties among seamen, and periods of taking part invited a reputational investment in military mobilization that may have increased the endeavor (Chong 1991). Round robin letters strains on officers and seamen alike. The odds (Leeson 2010), in which signatures were written of mutiny might also have been greater when radially, were an astute innovation to overcome seamen perceived an opportunity to rebel, win the first-mover problem resulting from naval concessions, or escape punishment. To evaluate authorities’ tendency to prosecute “ringleaders” these possibilities, we now turn to multivariate more zealously than mere followers (Gilbert analysis of mutiny in a larger-N study.

Downloaded from asr.sagepub.com by guest on December 19, 2015 Hechter et al. 11

A Case-Control Analysis participation in those events (Pfaff et al. 11 of Mutiny forthcoming). For ships experiencing a mutiny, the relevant ship-year begins one year We test our explanation by analyzing ships that prior to the date of the mutiny event. experienced mutiny and a larger set of ran- To select the control group, we drew a domly selected non-mutinous control cases. population-based sample from the risk-set of Because standard techniques for drawing sam- ship-years selected at random (without ples and estimating probability do not effi- replacement) from the entire list of RN ships ciently handle the analysis of rare events (King active during the period 1740 to 1820. Our and Zeng 2001), we used case-control methods sampling frame is Lyon’s (1993) Sailing Navy for obtaining a retrospective sample. These List. This contains every ship (about 4,400 in methods are efficient with respect to sample total) and its years of service, permitting us to size and provide analytic leverage that would condition a ship’s chance of being selected on be absent otherwise (Gail et al. 1976; Armenian the number of years it was in service. We 1994; Lasky and Stolley 1994; Newman 2001; sampled ship-years: each potential control Schlesselman 1982). Like epidemiologists ana- appears as a particular ship in a particular lyzing a rare disease, we searched for factors year (e.g., Crescent-1805). A ship-year is that increase the risk that any given ship is defined as the period one year prior to the predisposed to belonging in the case rather than occurrence of mutiny in a positive case and a the control group. Use of this method is grow- given calendar year in a control case. Limit- ing in the social sciences, including in the study ing the observation to one year thus provides of rebellion (Goldstone et al. 2010). a conservative estimate of the effect of inci- We assembled a reliable set of completed dents. Theoretically, incidents that occurred mutinies from authoritative sources. Our far in the past are not expected to trigger sample includes all cases in which seamen contemporary incidental grievances. fully halted the normal operations of the ves- Aiming to achieve a 4 to 1 ratio of controls sel (either by seizing it or by refusing to per- to positive cases, we selected 284 ship-years form their duties) that we could identify in from the resulting list of all valid ship-years. historical accounts (Gossett 1986; Haine However, archival records proved too incom- 1992; Hepper 1994; Lyon 1993; Rodger plete to allow reliable coding for 104 of these 2004) and in an official Admiralty register of ship-years, resulting in a sample of 179 ship- mutinies. Although we may not have captured years (a 63 percent response rate) in our con- every mutiny, our method does not require it. trol group. Although this introduces the One of the obstacles confronting the system- possibility of a biased sample because records atic study of rebellions like mutiny is possible of smaller ships might be less likely to be pre- underreporting (Lammers 2003; Rodger served than those of large, important war- 2004). Nevertheless, there are good reasons ships, the resulting sample does not appear to to suppose that many cases of full-fledged be biased. Our controls have an average rating mutiny did not escape notice. Captains and of 5.43 (on a scale of 1 to 7 by class of ship) masters kept separate logs of daily events, and weigh an average of 710 tons; the selected and the Admiralty was keen to have informa- ships whose records were missing/incomplete tion about unrest. Because of the possibility have an average rating of 5.48 and a weight of of selective prosecution, we did not confine 699 tons. The two groups are indistinguisha- ourselves to court-martial transcripts; our ble by class and size of ship, suggesting that sources also include official documents, cor- absent records are missing at random. respondence, diaries, and memoirs. Our sam- ple excludes ships taking part in the fleet-wide Data Sources and Variables mutinies of 1797 at Spithead and the Nore, as those cases were not independent. In a sepa- The primary data for this study were obtained rate article, we analyzed the dynamics driving from documents collected in the Admiralty

Downloaded from asr.sagepub.com by guest on December 19, 2015 American Sociological Review 12

Table 2. Mutiny Covariates Used in the Analysis

Expectation on Odds of Variable Mean Median SD Interpretation Mutiny

Percent Irish .13 .11 .07 Social structural + Rank Heterogeneity 1.65 1.76 .96 Social structural + Percent Prior Service .31 .22 .03 Social structural + Marines .15 .16 .02 Opportunity − Fleet or Squadron .51 .03 Opportunity − Inflation-Adjusted Wages 353.67 335.36 47.49 Structural grievance − Crowding 28.56 26.00 24.64 Structural grievance + Percent Impressed .07 .033 .02 Structural grievance + Foreign Deployment .67 .03 Structural grievance + Per Capita Flogging Rate 2.99 1.72 3.90 Structural grievance + Percent Sick .03 .0068 .01 Incidental grievance + Reduction in Rations .09 .02 Incidental grievance + Accidents at Sea .12 .02 Incidental grievance + Last Year of War .12 .02 Incidental grievance + Years at War 4.69 4.00 2.93 Contextual variable + Wartime .90 .02 Contextual variable + files of the Public Records Office deposited at 4. Physical characteristics as provided the British National Archives (TNA: PRO by Lyon (1993) for all RN ships. ADM). We coded the variables listed in Table 2 through the systematic evaluation of the The dependent variable is measured as 0 or 1 following: depending on whether we could document a mutiny meeting our definition aboard that 1. Captains’ logs and masters’ logs con- ship during a given year. taining daily entries for location, ship- board occurrences, accidents, Indicators of Incidental Grievances provisions, disciplinary actions, and so on. These logs provide redundant Percent sick is measured as the proportion of the information that reduces the hazards of ship’s company reported as “sick” in the muster reporting bias and missing data. book. The sick were judged incapacitated for 2. Muster books containing information service by the ship’s surgeon; hence this proba- about all the individuals aboard a ship bly provides a conservative estimate that does for each voyage, including surname, not indicate malingering or trivial complaints. place of birth, volunteer or impress, Mortality would be of obvious interest, but it previous ship served (if any), rank, cannot be reliably measured, because the seri- sickness, and nature of discharge. Mus- ously ill were often transferred to naval hospi- ters were updated weekly. For ships experiencing mutiny, we coded the last tals or other sites prior to their death. muster taken before the date of mutiny. Reduced rations measures whether a ship’s For control cases, we coded the muster crew had been placed on reduced or restricted taken in the middle of the month of rations of food or drink by a ship’s com- June. 3. Courts-martial records detailing pros- mander for a period greater than one day over ecution of mutiny or inquiry into the the course of a relevant ship-year, as reported loss of a ship. These records are verba- in ships’ logs. Seamen generally received less tim trial transcripts, including witness than their statutory rations because of short- testimony, depositions, and verdicts. ages or spoilage of provisions.

Downloaded from asr.sagepub.com by guest on December 19, 2015 Hechter et al. 13

Sailing accident measures whether ships’ Wages rose to 402 pence in 1807. The histori- logs reported an accident that endangered the cal CPI data for Britain from 1750 is reported operation of the ship or the crew (e.g., ground- in Twigger (1999). For the preceding years ing, a collision, a lost mast, or a drowning) we use the London grain price as a proxy. over the course of a relevant ship-year. By Rodger (2004) reports wage levels. custom as well as naval law, a ship’s com- Impressments are the share of a ship’s mander was considered responsible for all company reported as being pressed into ser- failures of navigation and ship-handling. vice as recorded in ships’ muster books. If an Last year of a war measures whether a entry read “see former books,” we matched relevant ship-year was during the disruptive the members of our mustered company to the period of demobilization of fleets at the con- last preceding book that noted recruitment. clusion of Britain’s wars. Seamen tended to Foreign station is a variable measuring blame officers for delays in mustering out and whether a ship served in foreign waters for all receiving their back pay. The wars in our or part of a relevant ship-year, as reported in period of study are the Austrian Succession ships’ logs. Foreign deployments were typi- (1740 to 1748), the Seven Years War (1756 to cally lengthier, more dangerous, and more 1763), the American War of Independence arduous than local deployments. For coding (1775 to 1783), and the wars of the French purposes, foreign stations include operations Revolution (1793 to 1802) ending with the anywhere outside the English Channel, the Treaty of Amiens, followed shortly by the eastern North Atlantic (excluding the Ameri- Napoleonic Wars (1803 to 1815). cas), and the North and Baltic Seas. In addition to our grievance measures, we include variables that measure the social- Indicators of Structural Grievances structural characteristics of ships, and varia- We measure punishment as a per capita rate of bles that might have influenced the opportunity floggings for each of our cases and controls by for and context of mutiny. Table 2 provides taking the total number of lashes inflicted on a summary statistics for the variables included crew over the course of a given ship-year and in the model. dividing it by the number of seamen in the ship’s company. Punishment reports were miss- Social Structural Indicators ing for 22 ships; to ensure missing data have not biased our estimates, we attributed missing val- Irish is the proportion of seamen born in Ire- ues by multiple imputations in our final model. land. In cases in which place of birth was not We measure crowding by dividing a ship’s reported in the muster book, it is inferred company as reported in the muster book by from the seaman’s last name based on Neaf- the ship’s total tonnage as reported by Lyon sey’s (2002) surname dictionary. In the event (1993). that a surname’s origin is indeterminate ethni- Inflation-adjusted wages is a dynamic cally (e.g., the surname Green), we coded the measure of the value of seamen’s wages; we name as English. Inferring ethnicity from adjusted the nominal monthly wage of an able surnames has been used in other studies of seaman by the previous year’s change in the military organization (see, e.g., Costa and buying power of the pound (consumer price Kahn 2008). Other non-British seamen were index). Inflation could substantially erode the in RN ships, but they were generally a small value of a seaman’s wages, which were not proportion of the ship’s company. regularly adjusted and remained basically Prior service is the share of the ship’s com- fixed at seventeenth-century levels until a pany reported as having served with at least general wage increase was granted by Parlia- one other seaman on another ship prior to ment in 1797 and again in 1807. The monthly transfer to the current vessel. If the entry read wage of an able seaman was 320 pence until “see former books,” we matched the members 1797, when wages increased to 354 pence. of our mustered company to the last preceding

Downloaded from asr.sagepub.com by guest on December 19, 2015 American Sociological Review 14 book that noted recruitment. Our measure is a the social heterogeneity among seamen proxy for social network structures, as these expected to increase their capacity to coordi- social ties may have facilitated participation in nate collective action. Only the share of a mutiny (Diani and McAdam 2003). ship’s company that is Irish is positively and Rank heterogeneity measures the stratifi- significantly associated with the risk of cation of the ship by rank and qualities as mutiny. In Model 2, we add variables as prox- reported in muster books and calculated ies for opportunities that could have increased according to a Shannon entropy measure. We the odds of mutiny by increasing its perceived calculated the score for each ship’s company efficacy or lowering expected costs. Although using the following groups by rank/quality: the presence of marines did not deter mutiny, officers, warrant officers, petty officers, able ships that were part of a fleet or squadron had seamen, ordinary seaman, landsmen, marines, significantly greater odds of mutiny than did and other (e.g., servants). Lower scores indi- solitary ships. This finding concurs with the cate less rank diversity. qualitative evidence that seamen often seized ships to negotiate with senior naval authori- ties to redress their grievances. Opportunity Variables In Model 3, we include proxies for struc- Marines is the share of a ship’s company tural grievances. Many of the factors usually composed of marine soldiers, if any, as considered to be onerous to seamen— reported in ships’ muster books. Marines impressments, crowding, and foreign deploy- were supposed to engage the enemy in battle ments—are not significantly associated with and serve as an onboard security force. mutiny. However, the value of inflation- Fleet or detached service is a variable adjusted wages significantly decreased the measuring whether a ship was assigned to a odds of mutiny, and high per capita flogging fleet or squadron or engaged in detached ser- rates significantly increased the odds. These vice over the course of a relevant ship-year as associations are consistent with the qualita- reported in ships’ logs. tive evidence, which shows that rebellious seamen frequently cited low wages and exces- sive discipline as motivating grievances.13 Contextual Variables Model 4 includes proxies for incidental Wartime is a variable measuring whether Brit- grievances. All the indicators of incidental ain was at war during a given ship-year. The grievances are positively and significantly danger of mutiny could have been greater dur- associated with the odds of mutiny. Whereas ing wartime. Number of years at war is a con- reduced rations, seafaring accidents, and the tinuous variable measuring the number of years last year of a war are all highly significant Britain had been at war in a given ship-year. predictors of mutiny, the association between Long periods of uninterrupted warfare could mutiny and sickness is especially strong. have increased burdens on seamen, increased Contextual variables related to the pressures tensions and fatigue, and made the command of of warfare and the duration of conflicts warships harder. For each ship-year, we calcu- included in Model 5 are not significantly lated the corresponding number of years of associated with the risk of mutiny.14 hostilities in which the RN was engaged.12 In Model 6, we add terms to explore pos- sible interaction and nonlinear effects. Although we see no significant interaction Findings effects between classes of grievances, we do Given a dichotomous dependent measure, we find an interaction between the share of Irish- estimated logistic regression models (Long men in a ship’s crew and the share of the crew 1997). Table 3 reports model coefficients and that had previously served together.15 More measures of fit. In Model 1, we regress sev- specifically, when holding all other variables eral social structural variables as proxies for in the model at their means, on a ship with a

Downloaded from asr.sagepub.com by guest on December 19, 2015 Hechter et al. 15

Table 3. Logistic Regression Models Predicting Mutiny

Social Structural Incidental Contextual Interaction Structural Opportunity Grievance Grievance Variable Terms

Percent Irish 5.730** 5.578** 6.438** 7.756** 7.680** 15.938*** (2.040) (2.126) (2.310) (2.695) (2.716) (4.655) Percent Prior Service .093 −.047 −.577 −1.119 −1.137 2.226 (.563) (.578) (.700) (.816) (.818) (1.682) Rank Heterogeneity −.465 −.398 −.300 −.137 −.120 −.101 (.359) (.417) (.332) (.274) (.274) (.294) Share Marines −1.628 −.553 .411 .416 −.157 (1.756) (1.982) (2.153) (2.164) (2.296) Fleet or Squadron 1.159*** 1.575*** 1.804*** 1.789*** 1.785*** (.330) (.396) (.469) (.483) (.492) Inflation Adjusted −.012* −.014* −.014* −.013* Wage (.005) (.006) (.006) (.006) Population Density .017 .012 .011 .012 (.013) (.013) (.012) (.012) Percent Impressed .234 .683 .745 1.907 (1.653) (1.833) (1.839) (2.222) Foreign Deployment −.319 −.802 −.814 −.717 (.373) (.447) (.449) (.462) Per Capita Floggings .117* .172** .173** −.216 (.053) (.060) (.061) (.196) Percent Sick 20.917*** 20.858*** 19.484*** (5.400) (5.382) (5.405) Reduction in Rations 2.004** 2.034** 2.017** (.676) (.682) (.687) Accidents at Sea 1.654** 1.640** 1.800** (.614) (.620) (.641) Last Year at War 1.751** 1.916* 1.776* (.661) (.817) (.873) Years at War −.040 .021 (.105) (.113) Wartime .344 −.159 (.819) (.850) Per Capita Flogging2 .022 (.012) Percent Irish × Percent −21.802* Prior Service (9.848) Intercept −1.070 −1.556* 1.244 .567 .348 −.213 (.621) (.661) (1.706) (1.917) (1.999) (2.070)

AIC 270.651 260.911 229.964 194.448 198.228 192.547 Log Likelihood −131.325 −124.455 −103.982 −82.224 −82.114 −77.273 LR Test(X2) 10.977* 13.740** 12.495* 43.378*** .220 9.681** N 239 239 220 219 219 219

Note: Likelihood ratio tests for significance of interaction effects wereX 2 = 3.521, p = .061 for percent Irish x percent prior service and X2 = 4.897, p = .027 for per capita flogging2 (both tests single tailed). *p < .05; **p < .01; ***p < .001 (two-tailed tests).

Downloaded from asr.sagepub.com by guest on December 19, 2015 American Sociological Review 16 mean level of sailors who previously served maintaining order at sea, at very low levels together, increasing the proportion of Irish increasing flogging decreases the probability seamen one standard deviation above the of mutiny. Holding all other variables in the mean is associated with an 8.1 percent final model at their means, increasing the increase in the probability of mutiny.16 On instances of flogging per capita from 0 to 3 ships one standard deviation below the mean reduces the probability of mutiny by 5.9 per- level of sailors previously serving together, cent. Near the mean levels of per capita flog- the same change in the proportion of Irish ging (2.99), the effect of a one-unit increase in seamen is associated with a 15 percent flogging is essentially zero. Seamen expected increase in the probability of mutiny. Finally, discipline and tolerated moderate levels of on ships one standard deviation above the flogging. However, at one standard deviation mean level of sailors previously serving above average flogging levels (approximately together, the same change in proportion of seven floggings per capita), the effect is posi- Irish seamen is associated with only a 3 per- tive: for instance, going from seven to eight cent increase in the probability of mutiny. per capita floggings raises the probability of This sheds light on an interesting issue of mutiny by 1.5 percent. The effect continu- wider scope than our study. Our findings sug- ously increases: at two standard deviations gest that well-integrated crews were less above average (approximately 11 floggings), prone to mutiny than poorly integrated ones. increasing per capita flogging by one point Ships with high shares of Irishmen and sea- raises the expected probability of mutiny by men who were not closely tied to one another 5.9 percent. At three standard deviations through previous experience were at greater above average, raising per capita flogging by risk of mutiny. Research generally empha- one point (from 15 to 16) increases the prob- sizes how social capital enhances the capacity ability of mutiny by 10.9 percent. Figure 1 of protest groups to engage in collective graphically illustrates this relationship, dis- action (Kitts 2000), but our finding points to playing the estimated probability of mutiny a different relationship—one related to the across a range of hypothetical values of per problem of command. On ships with less capita flogging, with all other predictors in the integrated crews, captains probably found it model held at their respective means. harder to perform well and provide accus- Because there are missing data for some tomed levels of security and welfare. Previ- ships on some variables, we also estimate a ous studies have also found that military unit final grievance model using multiple imputa- cohesion is enhanced by ethnic similarity and tions. After checking that data were missing at longstanding ties (Costa and Kahn 2008; random, we used the Multivariate Imputation Shils and Janowitz 1948). with Chained Equations approach, because it Consistent with our argument that struc- allows for separate probability models to be tural grievances motivate rebellion when used in the imputation of different missing experienced at higher magnitudes that violate variables and utilizes the full likelihood for expectations about proper treatment, there is a estimation (Van Buuren and Oudshoorn 1999). non-monotonic effect of flogging on the prob- The results of the analysis are reported in ability of mutiny. Analysis of a semi-parametric Table 4 and are nearly identical to those regression model replacing the traditional reported in Model 5, indicating that our find- parametric estimate with a smoothing spline ings are not biased by missing data. for per capita flogging supports our expecta- tion of an increasingly positive effect on the probability of mutiny at successively higher Discussion levels. Model 6 includes a squared term for This article presents strong evidence for the per capita flogging that significantly improves causal role of grievances in high-risk collec- the fit of the model (X 2 = 4.90, p = .027). tive action. Unlike many studies of rebellion Because flogging was an expected part of that treat grievances as a constant or use a

Downloaded from asr.sagepub.com by guest on December 19, 2015 Hechter et al. 17

Figure 1. Simulated Probability of Mutiny by Per Capita Rate of Flogging single indicator, we used multiple indicators are consistent with this argument. Seamen of structural and incidental grievances. Unlike tolerated flogging so long as it was fair and all previous comparative studies of mutiny, necessary for the maintenance of order. When we used case-control methods to draw the the rate of flogging grew too high, it was per- sample, systematically including positive ceived as cruel, arbitrary, and indiscriminate. occurrences of mutinies as well as controls. Moreover, because flogging was imposed by The main finding is the support for our captains as discretionary summary punish- structural and incidental grievance hypothe- ment, excessive flogging called their author- ses. Wage and punishment grievances were ity into question, whereas other structural important factors in generating the demand grievances (e.g., wage levels) could less read- for mutiny, a pattern also conspicuous in the ily be laid at the captain’s door. qualitative data. This accords with most intui- It is instructive to compare the combined tive understandings of rebellion and rein- effects of structural and incidental grievances. forces the claim that models of collective Comparing Model 3 to previous models, the action and contentious politics should not set of structural grievance variables improves minimize the causal role of grievances. the model’s predictive ability. Adding inci- Structural grievances raise the odds of dental grievance indicators improves predic- mutiny. In particular, wage depreciation and tive ability. Consistent with Hypothesis 2, all severe flogging increase the baseline propen- our measures of incidental grievances achieve sity of seamen to mutiny by violating expec- statistical significance in the expected direc- tations about their treatment. More broadly, tions. Adding them to a model already con- our theory contends that the disadvantaged taining structural grievances provides a tend to tolerate structural grievances so long further jump in explanatory power, particu- as they remain at routine and predictable lev- larly as a result of the strong association els. Our findings concerning rates of flogging between sickness and mutiny.

Downloaded from asr.sagepub.com by guest on December 19, 2015 American Sociological Review 18

Table 4. Final Model with Imputation

βˆ Number Missing F.M.I.

Percent Irish 15.740*** 0 .025 (4.165) Percent Prior Service 2.875* 2 .022 (1.370) Rank Heterogeneity −.129 0 .013 (.297) Share Marines −.625 0 .017 (2.063) Fleet or Squadron 1.274** 0 .014 (.425) Inflation Adjusted Wage −.009 0 .043 (.005) Population Density .012 0 .010 (.009) Percent Impressed 2.533 1 .031 (1.833) Foreign Deployment −.742 0 .024 (.418) Per Capita Floggings −.125 21 .208 (.177) Percent Sick 12.862** 2 .020 (4.042) Reduction in Rations 1.608** 0 .011 (.588) Accidents at Sea 1.723** 0 .012 (.531) Last Year at War 1.330 0 .039 (.725) Years at War .030 0 .027 (.102) Wartime −.092 0 .020 (.797) Percent Irish × Percent Prior Service −21.879** .024 (8.159) Per Capita Flogging2 .015 .150 (.010) Intercept −1.527 .040 (1.881)

Note: Number missing refers to number of missing values for each covariate. F.M.I. is fraction of missing information and indicates what proportion of information used to estimate each coefficient was imputed. *p < .05; **p < .01; ***p < .001 (two-tailed tests).

Our findings clearly implicate the problem that could be framed as ill-usage. As the quali- of authority. The target of a given grievance tative evidence indicates, incidental griev- appears to matter for the propensity to mutiny. ances trigger rebellion by connecting The genesis of mutiny is most likely in the proximate discontents with a long chain of temporal conjunction of existing structural previously tolerated structural grievances grievances and incidents of poor governance (Beissinger 2011; Horowitz 2001). This

Downloaded from asr.sagepub.com by guest on December 19, 2015 Hechter et al. 19 modified interpretation of the causal role of severely affected a tenth or more of a ship’s grievances can shed light on the outbreak of company would have been highly destabiliz- rebellion in a variety of settings, from prisons ing, both by creating anxiety among seamen to authoritarian states, to ethnically divided and by imposing a daunting test of a com- societies in which incidental grievances boost mander’s abilities. Although one could argue the capacity of previously aggrieved popula- that sickness is operating through an opportu- tions to mobilize (Bates 2008; Cederman et al. nity mechanism, officers were not prone to a 2013; Goodwin 2001; Pfaff 2006; Useem and higher rate of sickness than seamen. Rather, Goldstone 2002; Useem and Kimball 1989). widespread sickness and unchecked conta- We also find support for Hypothesis 3. gion operated primarily on the demand side Ethnic diversity is significantly associated by increasing grievances among seamen. with mutiny. As a consequence of their experi- If sickness were seen as inevitable and ences on land as well as at sea, Irishmen were invariable across ships in similar circum- more prone to a grievance interpretation than stances, it would be surprising that it moti- their British counterparts, and they had weaker vated mutiny. However, keeping seamen loyalty to the Crown to boot. Given the advan- healthy and avoiding the spread of infectious tages of co-ethnicity for coordination and, on disease were challenges that naval authorities average, their greater preference for rebellion, could meet. For instance, Horatio Nelson was Irishmen helped constitute a critical mass. RN a scrupulous administrator who ensured the commanders found that achieving social order ships under his command were sanitary and was more difficult in the context of ethnic well-supplied with fresh foods (Rodger 2004). heterogeneity and weak social capital among Health maintained shipboard order, and the subordinates unable to regulate themselves. remarkable operational effectiveness of the The proportion of the crew coded as “sick” RN came about, in large part, because com- is the most important predictor of mutiny. manders generally worked hard to avert Sickness and its attendant threats exemplify health crises (Bowden-Dan 2004; MacDonald the dilemma facing commanders. Compared 2006). By contrast, the health and welfare of with disease, battle was infrequent and a com- seamen in the French, Spanish, and Dutch paratively trivial cause of death (Earle 1998; navies were far worse, as was the problem of Rodger 2004). During the Seven Years War, mutiny (Dull 2009; Frykman 2009). The the RN lost more than 100,000 men to dis- health of a population is a major challenge for ease, whereas somewhat fewer than 200 were effective governance, and future research killed in combat (Allen 2002). Between 1793 should explore the link between health and and 1815, 82 percent of RN mortality resulted social order in other settings. from disease—versus just 2.1 percent from This study presents a corrective to prevail- battle (Pope 1981). Annual mortality rates in ing ideas about mutiny. Mutinies were not the RN could approach 15 percent as a conse- simply crimes of opportunity. Rather, muti- quence of disease outbreaks (Dull 2009; Gill nies occurred in all settings and were espe- 1961; Rodger 2004). Were unchecked sick- cially likely where seamen thought they could ness to threaten the lives of the crew, the get redress of grievances from senior naval potential costs of engaging in mutiny might authorities. Some historians maintain that the seem small compared with inaction. risk of mutiny increased from the 1790s on The association between sickness and because of the diffusion of revolutionary ide- mutiny occurs far below the level of general ologies and the “proletarianization” of sea- debilitation. The rate of sickness aboard muti- men (Frykman 2009, 2010; Neale 1985; nous ships was, on average, about three times Rediker 1987), but our findings call this con- greater than in controls, with the risk of clusion into question. Although the absolute mutiny increasing sharply after sickness number of mutinies increased during the crossed the 10 percent threshold of a ship’s 1790s, so too did the number of ships in the company. An outbreak of sickness that population at risk. In exploratory analysis, we

Downloaded from asr.sagepub.com by guest on December 19, 2015 American Sociological Review 20 found no association between year and risk of observe; for example, civil wars are more likely mutiny net of controls. If the risk of mutiny to occur in mountainous terrain than in lowlands increased over time, it did so because of the (Fearon and Laitin 2003). Mobility is extremely amplification of structural and incidental restricted onboard ship, however, and the author- grievances, not because of a period effect or ities are cheek by jowl with the seamen. widespread ideological diffusion. Undoubt- Finally, unlike most other insurgents, muti- edly some seamen were influenced by revolu- neers cannot profit from selective recruitment. tionary ideas, but in the 1790s, grievances Mutiny occurs among a set of individuals who appear to have intensified and ethnic hetero- are selected by the RN rather by the insurgents geneity increased in ways captured in our themselves. Save in limited instances—con- measures (Pfaff et al. forthcoming). spiracies to seize ships and turn them over to the enemy—mutineers cannot exclude loyalist seamen opposed to insurgency, and hence they Limitations of the Study must involve others who have widely varying Are eighteenth- and nineteenth-century muti- dispositions. This constraint undercuts insur- nies an exceptional type of insurgency? It is gent solidarity. These last two features also well-known that premodern forms of collec- hold for slave rebellions and prison riots, tive action differed from modern forms (Rudé among other kinds of insurrections (Carrabine 1964). Whereas premodern collective action 2005; Goldstone and Useem 1999; Useem and tended to be localized, reactive, and small- Kimball 1989). On balance, mutinies are scale (Tilly 1984), modern movements are probably more difficult to mount than most national and large-scale. Does this limit this other kinds of collective action.17 study’s generalizability? Not necessarily. Our model allows for a substantial element Although most collective action prior to the of contingency in the episodes generating late-eighteenth century was local in nature, this grievances while avoiding case-specific and was not the case for war-connected struggles ad hoc explanations. We can specify the fac- such as mutiny, which confronted state author- tors that ought to increase the risk of mutiny, ity and had national implications (Tilly 1993). but we cannot predict when incidents that can Mutinies are distinctive in at least three be ascribed to poor governance will occur. ways. Evidently, shipboard social structure Another limitation is that the causal interpre- provided some advantages for seamen’s col- tation of grievances is based partly on lective action. Seamen shared an occupational individual-level mechanisms. Although our culture, social segregation, and a community qualitative data provide insight into subjec- of fate—all circumstances shown to facilitate tive understandings and the motives of sea- coordination (Gould 1995; Kerr and Siegal men, and these mechanisms have been 1954; Tilly 1978). If members of subordinate demonstrated in a host of psychological stud- groups have common grievances and an ele- ies, our unit of analysis is the ship’s crew. mentary capacity to coordinate, they can Perhaps officers were not the true cause of mobilize in response to incidents that heighten seamen’s grievances (e.g., officers may not collective identification and perceptions of have been responsible for the outbreak of dis- injustice. In oppressed, deprived populations, ease aboard a ship, spoiled provisions, or run- loosely structured groups are likely to emerge ning aground) and thus should not have been as vehicles of mobilization (Pfaff 1996). blamed by seamen. However, what matters for But in other respects, seamen face greater the genesis of rebellion is the belief that com- obstacles to collective action than many other manders cause losses, not the reality. Cognitive kinds of insurgents. Mutineers, who serve along- bias should be influential precisely in situations side officers in relatively small ships in which in which actors seek to make a causal attribution there is no anonymity, are much more visible of blame but lack sufficient information (e.g., than most other insurgents. Insurgency is aided on the scientific etiology of disease) to deter- when individuals are difficult for authorities to mine the objective truth (Kahneman 2011).

Downloaded from asr.sagepub.com by guest on December 19, 2015 Hechter et al. 21

Conclusions and There would be no case for institutional Implications reforms that might improve organizational performance, make governance more respon- William Bligh, commander of H.M.S. Bounty, sive, or alleviate deprivation. published an account of the famous mutiny on Resource mobilization and rational action his ship. Bligh ([1792] 1962:140) wrote, “It theorists broke with classical grievance mod- will very naturally be asked, what could be the els for good reasons. However, they overre- reason for such a revolt?” He argued that, led acted. The grievances that matter for rebellion by the vain and unsteady officer Fletcher are shared and come in different types. The Christian, and besotted by “female connex- structural grievances that were a large focus ions” made during several months on Tahiti, of prior research do not exhaust the category. the mutineers “flattered themselves with the We show that the combination of shared inci- hopes of a more happy life among the Otahei- dental and structural grievances provides a teans, than they could possibly enjoy in Eng- superior explanation of mutiny. Future land” (p. 140). His readers may have assumed research should determine if this conclusion the mutineers were driven by the strongest of holds for other forms of insurgency. grievances. But Bligh assured them otherwise: The major conclusion of this study is that “Had their mutiny been occasioned by any incidental grievances are more prominent grievances, either real or imaginary, I must than structural ones as causes of high-risk have discovered symptoms of their discon- collective action. This implies that govern- tent” (p. 141). Musing as to whether the ments ought to take grievances into account if mutiny could have been averted, Bligh pointed they seek to mitigate unrest. Beyond this solely to the fact that his command included nostrum, however, what implications does no marines, who could have come to his aid. this lesson have for policymakers? Urban Bligh had good reason to discount the unrest in Ferguson in 2014 and Baltimore in causal importance of grievances. He did not 2015 was informed by longstanding structural wish his governance to be scrutinized, lest his grievances but triggered by specific incidents relatively heavy hand with the lash, his “fru- of police brutality. Even if those incidents gal economy” with rations, and his demean- could not have been predicted, better policing ing treatment of inferiors be blamed for the in these cities, and the impression that author- mutiny (Dening 1992). As self-serving as ities were respectful and mindful of residents’ Bligh’s account is, it has something in com- welfare, might have reduced the odds that mon with the dominant theoretical under- such incidents would spark violence. standing of rebellion since the 1970s. Like Undoubtedly, it is easier to create policies Olson (1965), Bligh emphasized private aimed at mitigating structural rather than inci- incentives in motivating collective action dental grievances, because the former tend to (“female connexions” and life on Tahiti). be more stable—and thus predictable—than Like resource mobilization and political the latter. This is even true of structural griev- opportunity theorists, Bligh emphasized elite ances that occur with some risk, and hence divisions (a rift between the officers) and can be estimated with a known probability. favorable opportunities (the ship lacked Because incidental grievances are by defini- marines and it was alone in the South Seas). tion unpredictable, authorities cannot readily This article contends that grievances have design and enact plans to forestall their been unduly neglected in much of the litera- effects. Even so, planning may mitigate their ture on social movements and collective consequences. An analogy can be drawn to action. If grievances are omnipresent and earthquake preparedness. Although policy- organization and opportunity are really what makers cannot know when a fault will slip, counts, then there can be few policy implica- they do know where the fault lines are; they tions. If grievances are irrelevant, the amelio- can enact policies to minimize loss of life ration of discontent will not avert conflict. (e.g., through building codes), and they can

Downloaded from asr.sagepub.com by guest on December 19, 2015 American Sociological Review 22 ensure an appropriate emergency response in 11. Besides the other contextual variables, we also esti- areas most at risk. In the absence of triggering mated models with an indicator for the post-1797 period to explore the association between our cases incidents, however, there is often too little and the fleet-wide mutinies that occurred at Spit- political will to make sufficient investments head and the Nore in 1797. Because this variable in future security. In social life, policies to fails to achieve significance and is highly correlated redress structural grievances tend to be con- with our wage measure and years at war, we do not sidered only in the wake of unrest: authorities include it in the reported models. 12. The Peace of Amiens (March 1802 to May 1803) sep- often find themselves shutting the stable door arating the French revolutionary wars from the Napo- after the horse has bolted. This was also true leonic War might be portrayed as a brief interlude that in the RN, which did not act to improve sea- does not indicate a real cessation of hostilities. How- men’s structural grievances—such as infla- ever, this argument strikes us as retrospectively biased. tionary pressure on their wages or arbitrary The motives for the peace were driven by a real desire to end the war that, on the British side, derived from and excessive punishment—until after a spate “financial, political and strategic exhaustion” (Rodger of mutinies in the 1790s. 2004:472). In fact, the peace was marked by substan- tial naval demobilization, in terms of both manpower Acknowledgments and expenditure (Rodger 2004). 13. Although one could argue that flogging is endoge- The authors are grateful for the invaluable consultation of nous to mutiny (i.e., flogging occurs because ships’ com- N.A.M. Rodger and the expert research services rendered panies are unruly), the data do not seem to indicate this. by Moira Bracknell. We would like to thank Eliana Hechter, We coded all ships (cases and controls) for whether Charles Causey, Edgar Kiser, Charles Seguin, David Siroky, they had experienced events recorded as collective Ross Matsueda, Luis Fernando Medina, John Wilkerson, defiance by their commanders during the relevant and the ASR editors and reviewers for their comments and ship-year (for positive cases, these incidents did not suggestions. Research assistance was provided by Univer- include the mutiny). Correlation between our pun- sity of Washington students Lindsey Beach, Jayleen Bow- ishment measure and general collective defiance is man, Elizabeth Cady, Dale Coleman, Lauren Ho, very small (.13). The correlation remains small and Christopher Prather, Sarah Reinecke, Heather Reyes, Joce- is insignificant even when we divide the sample lyn Sayler, Kristin Smith, Samuel Stabler, Thomas Stanton, between cases and controls. Timothy Thomas, Meral Tunador, and Rachel Tweet. 14. As a robustness check, we also estimated models predicting mutiny using non-parametric random- forest modeling (Siroky 2009). These models also Funding prominently indicate an association between our This research was supported by grants from the Air Force grievance measures and our outcome measure. Office of Scientific Research (FA9550-09-1-0502) and 15. Adding this interaction effect significantly improves the Harry Frank Guggenheim Foundation. the predictive ability of the model (likelihood ratio X 2 = 3.52, p = .06). 16. Note that, because we have specified a logistic Notes model, the effects of the predictor variables on the 1. Court-martial trial transcript of mutineers (CM), outcome are inherently nonlinear, which is why we Impeteux, 6/19-20/1799, the British National specify the marginal effects at specific levels of the Archive (TNA), Public Records Office (PRO) interacting variables. Admiralty Papers (ADM): 1/5349. 17. Even so, mutinies have not vanished from the con- 2. CM, Beaulieu, 7/3-17/1797, in TNA: PRO ADM temporary world (Guttridge 1992). For example, 1/5340. a 2009 mutiny in the Bangladesh Border Guards 3. CM, Ferret, 10/8/1806, in TNA: PRO ADM 1/5375. left more than 70 dead and resulted in thousands 4. CM, Kingfisher, 7/10-11/1797, in TNA: PRO ADM of arrests and hundreds of capital sentences (http:// 1/5340. www.bbc.com/news/world-south-asia-12123651). 5. CM, Panther, 6/27-7/1/1765, in TNA: PRO ADM 1/5503. 6. CM, Santa Monica, 7/21-24-730/1781 in TNA: References PRO ADM 1/5318. Allen, Douglas W. 2002. “The British Navy Rules: 7. CM, Culloden, 12/14-20/1794, in TNA: PRO ADM Monitoring and Incompatible Incentives in the Age 1/5331. of Fighting Sail.” Explorations in Economic History 8. CM, Camilla, 8/28/1783, in TNA: PRO ADM 1/5523. 39(2):204–231. 9. CM, Culloden, 12/14-20/1794, in TNA: PRO ADM Armenian, Haroutune K., ed. 1994. “Applications of 1/5331. the Case-Control Method.” Epidemiologic Reviews 10. CM, Ferret, 10/8/1806, in TNA: PRO ADM 1/5375. 16(1):1–164.

Downloaded from asr.sagepub.com by guest on December 19, 2015 Hechter et al. 23

Bates, Robert H. 2008. When Things Fell Apart: State Diani, Mario, and Doug McAdam, eds. 2003. Social Failure in Late-Century Africa. New York: Cam- Movements and Networks: Relational Approaches to bridge University Press. Collective Action. New York: Oxford University Press. Beissinger, Mark R. 2011. “Mechanisms of Maidan: Dull, Jonathan. 2009. The Age of the Ship of the Line. The Structure of Contingency in the Making of the Lincoln: University of Nebraska Press. Orange Revolution.” Mobilization 16(1):25–43. Durkheim, Emile. [1897] 1951. Suicide. Glencoe, IL: Benford, Robert, and David Snow. 2000. “Framing Pro- Free Press. cesses and Social Movements.” Annual Review of Earle, Peter. 1998. Sailors: English Merchant Seamen, Sociology 26:611–39. 1656–1775. London, UK: Meuthen. Bergstrand, Kelly. 2014. “The Mobilizing Power of Griev- Fearon, James, and David Laitin. 2003. “Ethnicity, Insur- ances: Applying Loss-Aversion and Omission Biases gency, and Civil War.” American Political Science to Social Movements.” Mobilization 19(2):123–42. Review 97(1):75–90. Bligh, William. [1792] 1962. An Account of the Mutiny Frykman, Niklas. 2009. “Seamen on Late Eighteenth- on Board H.M.S. Bounty. New York: Signet. Century European Warships.” International Review Boswell, James. [1791] 1934–1950. Boswell’s Life of of Social History 54(1):67–93. Johnson. Oxford, UK: Clarendon Press. Frykman, Niklas. 2010. “The Mutiny on the Hermione: Bowden-Dan, Jane. 2004. “Diet, Dirt and Discipline: Warfare, Revolution and Treason in the Royal Navy.” Medical Developments in Nelson’s Navy.” Mariner’s Journal of Social History 44(1):159–87. Mirror 90(3):260–72. Gail, Mitchell, Roger Williams, David P. Byar, and Brush, Stephen G. 1996. “Dynamics of Theory Change Charles Brown. 1976. “How Many Controls?” Jour- in the Social Sciences: Relative Deprivation and nal of Chronic Disease 29(11):723–31. Collective Violence.” Journal of Conflict Resolution Gamson, William, Bruce Fireman, and Steven Rytina. 40(4):523–45. 1982. Encounters with Unjust Authority. Belmont, Byrn, John. 1989. Crime and Punishment in the Royal CA: Dorsey. Navy: Discipline on the Leeward Islands Station, Gilbert, Arthur. 1983. “The Nature of Mutiny in the Brit- 1784–1812. Aldershot, UK: Scolar Press. ish Navy in the Eighteenth Century.” Pp. 111–20 in Calhoun, Craig. 1982. The Question of Class Struggle: Naval History, edited by D. Masterson. Wilmington, Social Foundations of Popular Radicalism during the DE: Scholarly Resources. Industrial Revolution. Chicago: University of Chi- Gill, Conrad. 1961. Merchants and Mariners of the 18th cago Press. Century. London, UK: Arnold. Carrabine, Eamonn. 2005. “Prison Riots, Social Order Goldstone, Jack A., Robert H. Bates, David L. Epstein, and the Problem of Legitimacy.” British Journal of Ted Robert Gurr, Michael B. Lustick, Monty G. Mar- Criminology 45(6):896–913. shall, Jay Ulfelder, and Mark Woodward. 2010. “A Cederman, Lars-Erik, Kristian Skrede Gleditsch, and Global Model for Forecasting Political Instability.” Halvard Buhaug. 2013. Inequality, Grievances and American Journal of Political Science 54(1):190– Civil War. New York: Cambridge University Press. 208. Centola, Damon. 2013. “Homophily, Networks, and Goldstone, Jack A., and Bert Useem. 1999. “Prison Riots Critical Mass: Solving the Start-up Problem in Large as Micro-Revolutions: An Extension of State-Cen- Group Collective Action.” Rationality and Society tered Theories of Revolution.” American Journal of 25(1):3–40. Sociology 104(4):985–1029. Chong, Dennis. 1991. Collective Action and the Civil Rights Goodwin, Jeff. 2001. No Other Way Out: States and Movement. Chicago: University of Chicago Press. Revolutionary Movements, 1945–1991. New York: Claver, Scott. 1954. Under the Lash: A History of Corpo- Cambridge University Press. ral Punishment in the British Armed Forces. London, Gossett, William P. 1986. The Lost Ships of the Royal UK: Torchstream. Navy, 1793–1900. London, UK: Mansell. Colley, Linda. 2003. Britons: Forging the Nation, 1707– Gould, Roger. 1995. Insurgent Identities: Class, Commu- 1837, 2nd ed. London, UK: Pimlico. nity and Protest in Paris from 1848 to the Commune. Collier, Paul, and Anke Hoeffler. 2004. “Greed and Chicago: University of Chicago Press. Grievance in Civil War.” Oxford Economic Papers Gurr, Ted Robert. 1970. Why Men Rebel. Princeton, NJ: 56(4):563–95. Princeton University Press. Costa, Dora, and Matthew Kahn. 2008. Heroes and Cow- Guttridge, Leonard F. 1992. Mutiny: A History of Naval ards: The Social Face of War. Princeton, NJ: Princ- Insurrection. Annapolis, MD: Naval Institute Press. eton University Press. Haine, Edgar. 1992. Mutiny on the High Seas. New York Dann, John, ed. 1988. A Diary of the Life of Jacob Nagle, and London: Cornwall. Sailor, from the Year 1775 to 1841. London, UK: Wei- Hardin, Russell. 1982. Collective Action. Baltimore, MD: denfeld & Nicholson. Johns Hopkins University Press. Dening, Greg. 1992. Mr. Bligh’s Bad Language: Passion, Hathaway, Jane, ed. 2001. Rebellion, Repression and Power and Theatre on the Bounty. New York: Cam- Reinvention: Mutiny in Comparative Perspective. bridge University Press. Westport, CT: Praeger.

Downloaded from asr.sagepub.com by guest on December 19, 2015 American Sociological Review 24

Hechter, Michael. 1978. “Group Formation and the Cul- Lichbach, Mark. 1995. The Rebel’s Dilemma. Ann Arbor: tural Division of Labor.” American Journal of Sociol- University of Michigan Press. ogy 84(2):293–318. Long, J. Scott. 1997. Regression Models for Categorical Hechter, Michael. 1987. Principles of Group Solidarity. and Limited Dependent Variables. Thousand Oaks, Berkeley: University of California Press. CA: Sage. Hechter, Michael. 1998. Internal Colonialism: The Celtic Lyon, David. 1993. The Sailing Navy List: All the Ships Fringe in British National Development, rev. ed. New of the Royal Navy, Built, Purchased and Captured, Brunswick, NJ: Transaction. 1688–1860. London, UK: Conway Maritime Press. Hechter, Michael. 2013. Alien Rule. New York: Cam- MacDonald, Janet. 2006. Feeding Nelson’s Navy: The bridge University Press. True Story of Food at Sea during the Georgian Era. Heckathorn, Douglas D. 1996. “The Dynamics and London, UK: Chatham. Dilemmas of Collective Action.” American Socio- Marwell, Gerald, and Pamela Oliver. 1993. The Critical logical Review 61(2):250–77. Mass in Collective Action. New York: Cambridge Hepper, David J. 1994. British Warship Losses in the Age University Press. of Sail, 1650–1859. Rotherfield, UK: Boudriot. McAdam, Doug, and Hilary Boudet. 2012. Putting Social Horowitz, Donald. 2001. The Deadly Ethnic Riot. Berke- Movements in Their Place. New York: Cambridge ley: University of California Press. University Press. Kahneman, Daniel. 2011. Thinking Fast and Slow. New McAdam, Doug, John D. McCarthy, and Mayer N. York: Farrar, Strauss & Giroux. Zald, eds. 1996. Comparative Perspectives on Social Kalyvas, Stathis N., and Matthew A. Kocher. 2007. Movements. New York: Cambridge. “How ‘Free’ is Free Riding in Civil Wars? Violence, Mead, George Herbert. 1934. Mind, Self and Society. Insurgency and the Collective Action Problem.” Chicago: University of Chicago Press. World Politics 59(2):177–216. Neafsey, Edward. 2002. The All New Surnames of Ire- Kemper, Theodore. 2001. “A Structural Approach to land. Kansas City, MO: Irish Genealogical Society. Social Movement Emotions.” Pp. 58–73 in Passion- Neale, Jonathan. 1985. The Cutlass and the Lash: Mutiny ate Politics, edited by J. Goodwin, J. Jasper, and F. and Discipline in Nelson’s Navy. London, UK: Pluto. Polletta. Chicago: University of Chicago Press. Newman, Stephen C. 2001. Biostatistical Methods in Kerr, Clark, and Abraham Siegal. 1954. “The Interindus- Epidemiology. New York: Wiley. try Propensity to Stike.” Pp. 189–212 in Industrial Olson, Mancur. 1965. The Logic of Collective Action. Conflict, edited by A. Kornhauser, R. Dugan, and A. Cambridge, MA: Harvard University Press. Ross. New York: McGraw-Hill. Østby, Gudrun, 2008. “Polarization, Horizontal Inequali- King, Gary, and Langche Zeng. 2001. “Logistic Regres- ties and Violent Civil Conflict.” Journal of Peace sion in Rare Events Data.” Political Analysis Research 45(2):143–62. 9(2):137–63. Pack, S.W.C. 1964. The Wager Mutiny. London, UK: Kitts, James A. 2000. “Mobilizing in Black Boxes: Social Redman. Networks and Participation in Social Movement Pfaff, Steven. 1996. “Collective Identity and Informal Organizations.” Mobilization 5(2):241–57. Groups in Revolutionary Mobilization.” Social Klandermans, Bert. 1997. The Social Psychology of Pro- Forces 75(1):91–118. test. Oxford, UK: Blackwells. Pfaff, Steven. 2006. Exit-Voice Dynamics and the Col- Lammers, Cornelis J. 1969. “Strikes and Mutinies: A lapse of East Germany. Durham, NC: Duke Univer- Comparative Study of Organizational Conflicts sity Press. between Rulers and Ruled.” Administrative Science Pfaff, Steven, Michael Hechter, and Katie E. Corcoran. Quarterly 14(4):558–72. Forthcoming. “The Problem of Solidarity in Insur- Lammers, Cornelis J. 2003. “Mutiny in Comparative gent Collective Action: The Nore Mutiny of 1797.” Perspective.” International Review of Labor History Social Science History. 48(3):473–82. Pinard, Maurice. 2011. Motivational Dimensions in Lasky, Tamar, and Paul D. Stolley. 1994. “Selection of Social Movements and Contentious Collective Action. Cases and Controls.” Epidemiologic Review 16(1):6–17. Montreal, Canada: McGill-Queen’s University Press. Lavery, Brian. 1998. Shipboard Life and Organisation, Pope, Dudley. 1981. Life in Nelson’s Navy. London, UK: 1731–1815. Aldershot, UK: Ashgate. Unwin Hyman. Lavery, Brian. 2010. Royal Tars: The Lower Deck of the Pope, Dudley. 1987. The Devil Himself: The Mutiny of Royal Navy, 875–1850. Annapolis, MD: Naval Insti- 1800. London, UK: Secker and Warburg. tute Press. Rao, Hayagreeva, and Sunasir Dutta. 2012. “Free Spaces Leeson, Peter. 2010. “Rational Choice, Round Robin, and as Organizational Weapons of the Weak: Religious Rebellion: An Institutional Solution to the Problem Festivals and Regimental Mutinies in the 1857 Ben- of Revolution.” Journal of Economic Behavior and gal Native Army.” Administrative Science Quarterly Organization 73(3):297–307. 57(4):625–68. Lewis, Michael A. 1960. A Social History of the Navy, Rediker, Marcus. 1987. Between the Devil and the 1793–1815. London: Allen &Unwin. Deep Blue Sea: Merchant Seamen, Pirates and the

Downloaded from asr.sagepub.com by guest on December 19, 2015 Hechter et al. 25

Anglo-American Maritime World, 1700–1750. New in U.S. Prisons.” American Sociological Review York: Cambridge University Press. 67(4):499–525. Rodger, N.A.M. 1982. The Articles of War: The Statutes Useem, Bert, and Peter Kimball. 1989. States of Siege: Which Governed Our Fighting Navies, 1661–1749 U.S. Prison Riots, 1971–1986. New York: Oxford and 1866. Havant, UK: Mason. University Press. Rodger, N.A.M. 1986. The Wooden World: An Anatomy Valle, James. 1980. Rocks and Shoals: Naval Discipline of the Georgian Navy. New York: Norton. in the Age of Fighting Sail. Annapolis, MD: Naval Rodger, N.A.M. 2004. The Command of the Ocean: A Institute Press. Naval History of Britain, 1649–1815. London, UK: Van Buuren, Stef, and C. G. Oudshoorn. 1999. Flexible Penguin. Multivariate Imputation by MICE. Leiden, the Neth- Rose, Elihu. 1982. “The Anatomy of Mutiny.” Armed erlands: TNO Prevention and Health, Report PG/ Forces and Society 8(4):561–74. VGZ/99.054. Rudé, George. 1964. The Crowd in History: A Study of Van Stekelenburg, Jacquelien, and Bert Klandermans. Popular Disturbances in France and England, 1730– 2013. “The Social Psychology of Protest.” Current 1848. New York: Wiley. Sociology 6(5–6):886–905. Schelling, Thomas. 1960. The Strategy of Conflict. Cam- Van Zomeren, Martijn, Tom Postmes, and Russell bridge, MA: Harvard University Press. Spears. 2008. “Toward an Integrative Social Identity Schlesselman, James. 1982. Case-Control Studies: Model of Collective Action.” Psychological Bulletin Design, Conduct, Analysis. New York: Oxford. 134(4):504–535. Shils, Edward A., and Morris Janowitz. 1948. “Cohesion Walder, Andrew. 2009. “Political Sociology and Social and Disintegration in the Wehrmacht in World War Movements.” Annual Review of Sociology 35:393– II.” Public Opinion Quarterly 12(2):280–315. 412. Siroky, David S. 2009. “Navigating Random Forests and Walsh, Edward. 1981. “Resource Mobilization and Citi- Advances in Algorithmic Modeling.” Statistics Sur- zen Protest around Three Mile Island.” Social Prob- veys 3:147–63. lems 29(1):1–21. Stewart, Frances. 2008. Horizontal Inequalities and Con- Weber, Max. [1918–21] 1978. Economy and Society. flict: Understanding Group Violence in Multiethnic Berkeley: University of California Press. Societies. New York: Palgrave Macmillan. Woodman, Richard. 2005. A Brief History of Mutiny. Tausch, Nicole, Julia Becker, Russell Spears, Oliver London, UK: Robinson. Christ, Rim Saab, Purnima Singh, and Roomana Sid- diqui. 2011. “Explaining Radical Group Behavior: Developing Emotion and Efficacy Routes to Norma- Michael Hechter is Foundation Professor in the School tive and Nonnormative Collective Action.” Journal of Politics and Global Studies at Arizona State University of Personality and Social Psychology 101(1):129–48. and Professor of Sociology at the University of Copenha- Taylor, Michael. 1997. The Possibility of Cooperation. gen. A fellow of the American Academy of Arts and Sci- New York: Cambridge University Press. ences, he studies contentious collective action—principally Theal, George McCall, ed. 1898. The Records of the Cape in its nationalist and separatist forms—and its converse, Colony from December 1796 to December 1799, Vol. social order. His most recent book is Alien Rule (2013), II. London, UK: Clowes. which argues that this widely denigrated form of gover- Tilly, Charles. 1978. From Mobilization to Revolution. nance may, under certain conditions, attain legitimacy. New York: McGraw-Hill. Tilly, Charles. 1984. “Social Movements and National Politics” Pp. 297–317 in Statemaking and Social Steven Pfaff is Professor of Sociology at the University Movements, edited by C. Bright and S. Harding. Ann of Washington. He conducts research in the fields of Arbor: University of Michigan Press. historical and comparative sociology, collective action Tilly, Charles. 1993. “Contentious Repertoires in and social movements, political sociology, and the soci- Great Britain, 1758–1834.” Social Science History ology of religion. His book Exit-Voice Dynamics and the 17(2):253–80. Collapse of East Germany (Duke 2006) was honored by Turner, Ralph, and Lewis Killian. 1972. Collective the Social Science History Association and the European Behavior, 2nd ed. Englewood Cliffs, NJ: Prentice- Academy of Sociology. Hall. Twigger, Robert. 1999. “Inflation: The Value of the Patrick Underwood is a doctoral candidate in Sociology Pound 1750–1998.” Research Paper 99/20, House of at the University of Washington. His areas of specializa- Commons Library. tion are political sociology, comparative historical sociol- Useem, Bert. 1998. “Breakdown Theories of Collective ogy, and quantitative research methodology. His Action.” Annual Review of Sociology 24:215–38. dissertation project examines the effect of strategic Useem, Bert, and Jack A. Goldstone. 2002. “Forging implementation of neoliberal economic reforms on the Social Order and Its Breakdown: Riot and Reform risk of collective political unrest in urban populations.

Downloaded from asr.sagepub.com by guest on December 19, 2015