Western Expansion, Deputy Marshals, and the State

Scholars of American state formation have recently explored the often counterintuitive ways in which the American state expanded across the continent in the nineteenth century (Frymer Reference Frymer2014; Gerstle Reference Gerstle2017). Though by the end of the nineteenth century, the US state contained strong pockets of rational-bureaucratic institutionalization promoted by reformers, even in those spaces where it seemed weak, links between public and private actors allowed for a hybridized form of state capacity that could prove quite effective.

The DOJ is a case in point. A product of Reconstruction, the department was founded in 1870 to centralize control over enforcement efforts in the South (Shugerman Reference Shugerman2014). Most significantly, this meant embedding the US Marshals, an old and traditionally small enforcement arm responsible for legal process for all branches of government, in a broader bureaucratic system. During Reconstruction, district justices were given authority to appoint USDMs locally, tying enforcement of voting rights in the South to local political networks. Marshals were, like most federal employees, paid on the basis of fees, a portion of which they then allotted to deputies (often relatives or friends) appointed to serve as their proxies in serving process. This made voting rights enforcement fundamentally a function of local networks and linked entrepreneurial activity to the securing of federal aims (Goldman Reference Goldman2001: 189).

Enforcing voting rights was the focus of much political controversy and, ultimately, did not succeed in establishing a strong DOJ presence in the South in the 1870s and 1880s. Opponents (particularly those with aims of “redeeming” former Confederate states) cast the US Marshalcy as a prime nexus for cronyism and corruption (ibid.: 183–95). These charges were quite effective because Republicans in Congress, too, relied heavily on the rhetoric of “frugality” to legitimate their political rule—hence, much of the DOJ’s early activity involved trying to gain control over the costs of the US Marshals Service rather than on attempting to increase their competence (Calhoun Reference Calhoun1990: 138–40). Indeed, Shugerman (Reference Shugerman2014) has recently argued that the founding of the DOJ was a case of federal retrenchment, as it led to the elimination of a cadre of US attorneys and was intended to curtail rather than expand patronage.

As the USDMs began to involve themselves in the day-to-day protection of freedmen in the South, they became targets of local resistance. In the 1870s and 1880s, USDMs were legally vulnerable because those resisting arrest could only be charged in local courts, which provided manifold opportunities for community attachments to constrain marshal efficacy (Miller Reference Miller1991: 107). And even when there was strong local support by African American residents and a fairly effective military presence, marshals had a very difficult time enforcing voting rights.

Louisiana in 1874 presents an exemplary case. In October of that year USDM Edgar Selye had gone to Lincoln Parish in the company of a military unit to arrest several men on suspicion of intimidation. After confronting a hostile crowd and becoming wary of a coordinated plan to help the prisoners escape, Seyle and the lieutenant accompanying him, Benjamin Hodgson, cut the telegraph wires to prevent communication on the part of the local White League units. Several days later, both Seyle and Hodgson were arrested by a local posse for failing to produce the prisoners in a local court as commanded by a writ of habeas corpus; as a result, both spent 10 days in jail for contempt of court (Dawson Reference Dawson1982: 192–94). The case became a cause célèbre for conservative white Democrats in Louisiana and a means for them to link racial solidarity to a narrative of political corruption and overreach (Anonymous 1874). It also demonstrated, along with other similar cases, that DOJ enforcement ultimately depended not merely on coercive capability but also on the acquiescence of the local white community, which retained both the will and the economic and organizational resources to resist (Goldman Reference Goldman2001: 160–67).

In the trans-Mississippi West, however, the marshal’s force was a going concern. Deputy marshals were very often the only legal officers of the peace available in territories, and they possessed jurisdictional power that allowed them to act over large swathes of land (Calhoun Reference Calhoun1990: 150). Moreover, USDMs were also responsible for tasks that took on special importance in the West, such as protecting the US mail (often the only link between communities in the far-flung settlements of the plains and mountain territories) and helping the army protect Indian communities from white encroachment (Cresswell Reference Cresswell2002: 181–83).

The West had become the focus for the expansion of federal state power both directly and through public/private hybrids in the 1870s. A proliferation of new territories in the 1860s and 1870s, for example, provided the federal government with a new pathway to seize administrative control, creating in the process a system of colonial administration under which settlers were to live. Moreover, the government renewed its struggle against the diffuse Native communities in the western plains in the 1860s, expanding a system of military forts and supply routes that paved the way to infrastructural growth. These military activities also linked quasiprivate suppliers like railroads to professional bureaucrats in a chain of mutual dependence, a chain that made “state-building” in the West in part a cooperative effort of industrial and military planners (White Reference White2015: 236–68, 337–51).

At the same time, as Paul Frymer (Reference Frymer2014) notes, the military’s footprint in the West never particularly large. Indeed, the legal apparatus of the federal government as a local resource was a crucial component in the expansion and one that allowed everyday settlers to exploit property qualifications and land policy to help reproduce federal power in everyday commercial and real estate exchanges in the West. As federal officials, USDMs were often engaged in activities arbitrating such claims or serving process from federal courts that in the West were much more quotidian than that found in, say, the Northeast. Moreover, territorial USDMs also very often held local law enforcement roles as deputy sheriffs or town marshals, which linked traditional practices of property protection in local communities directly to federal oversight (Obert Reference Obert2017: 661–64).

These tendencies, arguably, reached their height in the IT and, later, Oklahoma Territory in the 1880s and 1890s. Unlike other territories like Utah or Arizona, IT was never intended to be an official territorial jurisdiction under US rule, but rather was established to provide tribal governments (especially the Five Civilized Tribes from the southeastern states) with a zone of self-government under American protection and tutelage. That autonomy included systems of law and self-policing, in addition to shared property ownership and customary government. At the same time, relations between IT residents and outsiders were heavily policed by the US government. Whiskey-running, range encroachments, and other activities involving the commercial relations between the resident tribes and “intruders” were the direct responsibility of the federal government, which, in turn, relied on the marshals to aid enforcement.

More importantly, because the United States continued to claim legal oversight over all American citizens, jurisdictional conflict was prevalent. In 1872, for instance, a small posse of USDMs made their way to Going Snake in the Cherokee district to arrest a Native man already on trial in Cherokee courts for assaulting a US citizen. As the men tried to rearrest the defendant, a group of indignant Cherokees fired at the deputy marshals and a bloody fight ensued (Calhoun Reference Calhoun1990: 164–65). Animosity over jurisdictional issues compounded a deeper antagonism between some of the tribal factions, which made built alliances with the Confederacy during the War, and the Republicans in Congress.

One result of these conflicts was that federal law made increasingly broader claims concerning its policing ambit in the aftermath of the Civil War. As early as 1834, the district court in Arkansas had been able to claim IT as part of its jurisdiction, a claim inherited by the new Western District Court of Arkansas at Van Buren (later moved to Fort Smith) in 1851 (Shirley Reference Shirley1957: 14). At the same time, the tribes had long organized their own policing forces, usually modeled after the local institutions found elsewhere in the United States. For instance, the chief of the Cherokee appointed a high sheriff, who served as warden, court martial, and town marshal of Talequah, the tribal capitol. Counties, in turn, were each policed by a sheriff and several constables, who similar to these figures elsewhere, had deputization power. At the same time, most of the tribes possessed small units of “light-horse men,” a quasimilitia used to patrol and police tribal territory. By the early 1870s, agencies managing several of the reservations in the Great Plains states had begun recruiting their own police forces, and by 1879 Congress appropriated funds for the police forces (Burton Reference Burton1995: 76–81, 124).

This led to intense kind of jurisdictional conflict unique to the IT. On the one hand, because there were citizens of so many different Indian nations residing in the IT (as well as an ever-increasing number of mixed-blood and Anglo-American noncitizens) and because tribal courts often only had recourse to administering justice to their own citizens, deciding whether or not a criminal matter was the provenance of a tribal or federal court was often a contested matter. On the other hand, unlike the criminal justice process found in other reservations, in the IT, relations among the tribal governments often led to disagreements over extradition requests and difficulties coordinating law enforcement (ibid.: 100–2, 179, 189).

Thus, by the 1880s, the railroad industry, white settlers, and certain cattle interests demanded more legal oversight of Indian legal institutions, albeit frequently for different reasons. Railroads and cattle enterprises susceptible to property crime across multiple jurisdictions wanted more uniformity of enforcement, while the growth of the “Boomer” movement—those predominately Anglo-American residents of midwestern states and border states eagerly eyeing Oklahoma lands for personal acquisition—saw increasing federal jurisdiction over the IT and the other unassigned lands as the first step in a process of facilitating settlement (ibid.: 41–42, 145).

One of the first major federal reforms intended to address these concerns was passage of the Major Crimes Act in 1885, which asserted US legal jurisdiction over a select number of crimes committed by Indians on reservations (even though the law was not extended to the IT until the next year). The larger effort, however, included rolling back some of the authority of the tribal police forces and centralize legal process in the federal courts, in addition to expanding and reorganizing the federal police presence. This not only involved the expansion in federal police forces attached to the agencies (by 1890, there were more than 700 US Indian police officers on the federal payroll) but also in enabling the power of the US Marshal’s office (Hagan Reference Hagan1966: 43). An 1888 law expanded the capacity for USDMs to enter tribal lands and helped expand federal control over and support for the Indian police, for instance, while in 1890 Congress decreed that all legal matters involving residents of different tribes with the IT be adjudicated in federal court (Burton Reference Burton1995: 127–70).

In the aftermath of the opening of large portions of Oklahoma to white settlement in the early 1890s, federal courts in Texas and Kansas began splitting duties in managing both the IT and the new Oklahoma Territory. In addition, new courts were established for the first time in territorial towns like Muskogee (Burton Reference Burton1995: 151–55). These moves made federal law enforcement a matter of day-to-day concern of all the residents of the territories, while creating the same kind of administrative practices other areas had used to convert territories in states.

As a result, Oklahoma quickly became the most important center for DOJ policing. In 1897, one of the few years in which we have complete records, more than 90 percent of deputy marshals were active in southern or western states. Oklahoma alone accounted for almost 20 percent of the force and approximately 13 percent of the total expenditures on criminal apprehension in 1895, despite only constituting around 1 percent of the estimated population of the United States in that year.

Even among territories Oklahoma stands out. Of the six territorial jurisdictions in 1890 (these six territories together counted for approximately 19 percent of the total expenses of the DOJ in the mid-1890s), Oklahoma was the source of a whopping 69 percent of the $236,000 spent on territories in 1895. This was despite the fact that Oklahoma only possessed 42 percent of the territorial population that year.

When we examine the level of USDM presence per 10,000 residents in each state or territory in 1897 (as depicted in figure 1), the federal policing presence in the Southwest becomes even starker. ^{Footnote 3} Three of the four jurisdictions with the highest rate of USDM presence are in the region, while the other—Arkansas—included a large number of USDMs who were active in Oklahoma as well. Given the fact that Oklahoma was also more densely populated than, say, Arizona or New Mexico, this level of federal police penetration meant that it was quite common for Oklahoma residents to come in contact with USDMs to a degree that would have been unusual elsewhere, even in the territorial West.

Figure 1.

Distribution of marshal’s force (1897).

Source: Attorney General (1897: 234–95).

Race, Resistance, and the Making of Outlaws and Heroes

USDMs were not merely representatives of federal law in an abstract sense, they also engaged in day-to-day interactions with residents of the places they were policing. During Reconstruction in the South, these interactions had taken on the form of “meddling” in established racial hierarchies and oppressing social heroes; in the territorial West in general and Oklahoma in particular, they took the form of chasing outlaws. The difference was due not simply to the fact that Oklahoma was a territory (in which federal law had much greater ambit) while southern jurisdictions were states—though that factor undoubtedly played some role, other territories (such as Idaho or Utah) only had very small federal marshal presence. Nor was it due to difference in the kinds of activities in which marshals engaged in the different regions—in both the South and in Oklahoma, for example, managing the liquor trade was a primary activity of the marshal’s service after 1877 (Calhoun Reference Calhoun1990: 166).

Instead, federal police expansion was at least partly due to the ways in which racial boundary-making was linked to local politics and to resistance to the central state. Emirbayer and Desmond (Reference Emirbayer and Desmond2015) argue that social boundaries like race are largely reproduced through institutional and cultural practices such as legal classification and structural factors such as socioeconomic inequality, which position actors in a field of social relations and constitute a “racial order.” Historical racial orders, in turn, led actors to interpret potentially resisting antistate activity and provided a major pathway through which residents assessed the state’s policing efforts.

In practice this meant that in areas where local racial boundaries were relatively clear and unambiguous—in areas, that is, where race served as a means of propping up local social order by clear binary and hierarchical divisions—the activities of an outside policing agent could easily be seen as political and partisan. Since, in the Reconstruction South, much of the activity of the USDMs involved protecting freedmen in exercising their voting rights, they were usually viewed as directly interfering in what southern whites considered their own prerogative to manage race relations as they saw fit (Cresswell Reference Cresswell2002: 44–78).

Behind this system was an emerging legal concept of the “one-drop” rule—the idea that one’s kinship was determined by discrete racial lines and any biological relation to African American kin was reason for assigning that person as African American. The collapse of slavery and the presence of many mixed-race “mulattos” in the 1860s had created a crisis in classificatory hierarchies, in which bondage could no longer be the primary mark of white superiority. As a result, white southerners became increasingly anxious about what Joel Williamson (Reference Williamson1980: 98) has called “invisible blackness,” the notion that skin color alone could no longer adequately indicate racial membership. The end of Reconstruction thus saw a gradual supplanting of the older regime with a set of legal classifications around marriage and family based on ancestry—miscegenation laws—undermining the fine-grained racial distinctions many whites had used in practice to order social life.

Enforcing the one-drop rule was never solely a matter of legal rules about heredity and marriage but also of forging a new white identity and sense of communal solidarity. Whiteness and blackness had become “a matter of inner morality and outward behavior,” and any threat to white sexual, social, and economic dominance was deemed worthy of suppression (Williamson Reference Williamson1980: 108). The net result was the creation of a network of state and vigilante organizations (including Anti-Miscegenation Leagues) to militate against what many considered a kind of sexual outlawry. Such efforts, Williamson (ibid.: 95) points out, tried to halt the “merg[ing]” of whiteness and blackness through sexual reproduction, a process that would have meant “the absolute damnation of Southern civilization.”

The efforts of the federal government and its deputy marshals to protect freedmen in southern regions thus not only had the effect of threatening to create political equality but also to subvert the existing social and sexual hierarchy between the races. Attempting to enforce the Reconstruction Acts tainted all federal policing in the area, particularly as directed against white residents. Traditions of localism combined with a memory of the interference in the “Southern way of life” to produce widespread antipathy to federal policing, even against moonshiners (Miller Reference Miller1991). Resistance, therefore, was often viewed in heroic terms, what Hobsbawm (Reference Hobsbawm2000: 51–55) has identified as a key condition for social banditry. It was explicitly antistatist and popular, at least among those who enjoyed racial identification with the heroes.

However, gradient or ambiguous racial orders, such as those found in Oklahoma, presented the federal government with an opportunity. In such gradient orders, actors could not easily be classified in particular discrete groups (as being “white” or “black,” for instance), and kinship and group membership institutions often provided for complex or hybrid forms of identity. As a result, resistance might not be interpreted as a way of defending a local racial order, but rather as a threat to “all” white citizens. Indeed, from the vantage point of new white residents (and even some of the Natives), the boundaries between groups within the IT were inscrutable, leading residents to view complex local racial orders largely through the lens of “savagery” and “civilization” (Wickett Reference Wickett2000: 15–41). The clash between gradient and “one-drop” racial orders made resistance the work of “outlaws” posing a threat to the community as a whole in a complex and uncertain racial environment rather than defenders of a well-established order of white supremacy. ^{Footnote 4} This, in turn, paved the way toward reforming policing institutions with the support of, rather than opposition by, local white residents.

Gradient Racial Orders and the Growth of Federal Policing

The link between kinship and racial classification in Oklahoma was unique in the context of the late-nineteenth-century United States. The IT was home not only to the larger Cherokee, Choctaw, Chickasaw, Creek, and Seminole tribes (among other smaller eastern Indian groups) but also, by the late 1880s, to scattered groups of Plains and Western Indians who had been relocated there on the recommendation of the Indian Peace Commission of 1867. This had created a racial crisis for the existing occupants of the territories, many of whom had already been riven with factions over how much to assimilate to white society (ibid.: 7, 36).

The presence of freed African Americans in the territories also complicated the racial order of Oklahoma. Slaveholding had taken hold in the civilized tribes and many had supported the Confederacy during the Civil War, making them not only a prime target for retaliation by the federal government during Reconstruction but also inducing many of them to offer citizenship to freed people. Such citizenship enabled African Americans to gain access to the full spectrum of privileges associated with tribal membership, including landownership and participation in government. But, from the standpoint of many racist white newcomers, this created even more questions concerning the racial status and “civilized” nature of the Indian residents (ibid.: 8–9, 28–41).

Most decisive were patterns of intermarriage, both between whites and Natives and between Natives and freed people. Many Seminoles and Creeks had practiced intermarriage with their slaves for generations, and most of the early white residents in the Indian Territories gained access by virtue of having intermarried within a tribe. Indeed, only one-fourth of those listed in the Dawes Census of 1895 were “full-blooded” members of tribes, while approximately 40 percent were considered less than “half-blooded” (ibid.: 36). These patterns created a major conceptual challenge to the “one-drop” biology determining racial membership in the South. Thus, although there was much vicious antiblack racism on the part of many tribal members—and indeed, some tribes, like the Choctaw, passed antiintermarriage statutes—for many whites, the presence of varying degrees of racial “purity” among residents merely compounded the notion that Oklahoma was disorderly and savage.

The importance of race remains even when examining cases of banditry that, on their surface, appear unrelated. For example, Jesse James was one of the most important such outlaws, at least from the standpoint of the Pinkerton Agency and many of the business interests in the West that chased him in the 1870s and 1880s. While the James and Younger Gang was responsible for a number of high-profile attacks on powerful economic actors, this very profile—connected to the fact that the James Gang was forged in the aftermath of the bloody experience of Missouri’s Civil War, marked by high levels of guerrilla and partisan violence—made him a hero to many in the South and Southwest. His carefully cultivated media strategy tried to cast he and his associate’s many robberies (primarily of trains and banks) as reflective of a persistent antinorthern resistance. Indeed, many southern partisans in Missouri and Arkansas (among other places) supported James and his gang as heroes, even if (as T. J. Stiles has correctly pointed out) this heroism was based on activating Civil War–era divisions rather than establishing a communal solidarity. In both style and mission, James was continuing to fight the war after Appomattox (Stiles Reference Stiles2002: 241, 389–95).

But as an artifact of Civil War–era divisions, Jesse James’s heroic status in the divided communities of Missouri was inextricable from race. James’s family had been respectable slave holders and Stiles (ibid.: 215–17, 236–40) has argued that the confluence of localism and racial identity helped animate the continuation of the Civil War James saw himself pursuing. Moreover, as with many bandit gangs during the period, the James and Younger Gang (as it is more accurately called) was organized principally around kinship—groups of brothers or cousins who joined in small units to fight their own “resistance” against the state, as well as earn some money in the process. They were, in other words, white, family men, men whose solidarity with certain values of racial superiority did not need to be asserted explicitly, but was simply assumed by supporters.

Of course, whether Jesse James is outlaw or bandit hero is in the eye of the beholder, which is exactly the point—his heroic status for some was built on illegal activity directed, in part, against the interests of central state-builders. As a sophisticated political actor in a context of stark racial division, James was able to accomplish something many outlaws in Oklahoma tried but failed to do—successfully communicate his robbery as resistance to an oppressive, meddling federal presence (ibid.: 249–71).

Before proceeding to an analysis of patterns of banditry and legal enforcement in nineteenth-century Oklahoma, I first provide some rough diagnostic results to establish the general point about the relationship between race and the distribution of federal policing. The results presented in figure 2 presents the standardized coefficients and confidence intervals for a simple OLS analysis of the ratio of USDM per 10,000 residents in each of the 48 territories or states found in 1897. Two variables should be correlated with higher such ratios are based on racial classification—one is a race fractionalization or diversity score of the state or territory, which measures the rough probability that two individuals living in the territory would come from different racial groups, while the other is the proportion of African Americans in the state or territory’s population. ^{Footnote 5} The basic argument is that higher diversity should, in general, indicate more ambiguous racial classification in a state or territory, while, when controlling for those scores, higher levels of African Americans in a population reflect clearer racial classification in a state or territory. In other words, the process linking race to policing has two parts: diversity is crucial, but particularly so in cases in which the African American population is not overwhelmingly represented because the status of being “black” was much more dichotomous.

Figure 2.

Exploring USDM presence (1897).

Note: Fully specified OLS model with standardized coefficients. See appendix for details.

In addition to the race-based variables, I also include controls for ethnic diversity (e.g., nativity of residents, to test for the possibility that another identity cleavage is explaining USDM presence), manufacturing capital (to capture the possibility that USDMs are more likely in places with more complex economic conditions), the proportion of the population that is urban and the territorial status of the jurisdiction in 1890 (to capture both the probability that highly urbanized states already possessed local policing institutions, as well as to identify whether or not territorial jurisdictions were more likely to have higher USDM presence), and several indicators to measure the demand for federal policing (level of railway development and number of railway robberies reported between 1890 and 1900, both of which should positively correlate with demand for regional or federal policing efforts). Finally, I use two dummies to control for region.

These results suggest several things. First, higher levels of racial diversity are significantly associated with higher ratios of USDM per population. In addition, higher levels of African American membership are significantly associated with lower federal policing presence. These measures are robust across all models and strongly suggest that in cases where the one-drop rule applies—for example, in cases with large African American populations, particularly in the South—USDM presence is lower. Where such a rule does not apply, where diversity indicates racial ambiguity, federal policing is more prevalent.

Second, there is some evidence that institutional context, demand, region, and ethnicity also affect federal policing. Specifically, territories and urbanized regions were significantly more likely to have federal policing presence, as were jurisdictions which experienced more railway robberies. Region, too affects, policing—southern states seem to be negatively associated with federal policing net all other effects. Finally, intriguingly, higher levels of ethnic diversity are negatively correlated with USDM presence in the population, suggesting both the primacy of race as the salient axis around which such presence rotated as well as a likely turn to subnational tools for policing in contexts where nativity rather than race was at issue.

This model is only meant to be diagnostic—it does not account for actual relationships of individuals on the ground, where the “outlaw” versus “hero” distinction matters most. Nevertheless, it strongly suggests an important relationship exists between banditry, kinship/race, and federal policing in key jurisdictions. I now turn to a closer examination of these factors in Oklahoma.

Federal Posses, Indian Law, and Outlaws in Oklahoma

By the 1880s and 1890s, Oklahoma had become known as a refuge for criminality. As a sort of “extrajudicial” territory during much of the pre–Civil War period, the complex factional struggles of native residents of the IT before and during the war had allowed for the development of a well-entrenched system of violence and disorder. The Starr family, for instance—from which both Henry Starr, a famous desperado of the 1890s, and Belle Starr, the so-called Bandit Queen of the 1880s, derived their monikers—were key players in a series of disputes between the “Treaty” (or “Removal”) and the “Patriot” parties in the Cherokee Nation that had turned violent in the territory in the 1840s and 1850s (McLoughlin Reference McLoughlin2014: 34–58).

Several factors contributed to the IT’s reputation as a haven of criminality. First, IT was surrounded by rapidly growing settlements in Texas, Kansas, and Missouri, settlements that brought both money and migrants to the region. Not only was western Oklahoma a key passage on the cattle trails leading north from Texas (making it a popular destination for rowdy groups of young men), but it also became the target of land speculators and entrepreneurs, who saw in local Indian populations a potential market for controlled and illicit goods (Wickett Reference Wickett2000: 138–50). One early settler and deputy marshal described the IT as a “dumping ground for the outlaws and thieves” of states throughout the Southwest (Morse Reference Morse1937: 35).

Second, IT’s legal ambiguity provided smugglers, horse thieves, and especially railroad robbers—many of whom were already attracted to the anonymity and mobility of the trans-Mississippi settlements—with ample opportunity to exploit differing legal procedures and jurisdictional conflicts to protect their business interests and to halt extradition proceedings. It was also easy for criminals to find both political allies and economic opportunity in the complex layered jurisdictional environment of the IT (Burton Reference Burton1995: 170–201). Moreover, the boundary between Oklahoma and IT provided fertile ground for engaging in legally dubious activities, such as selling whiskey; many saloons were set up on the Oklahoma side of the border so that residents of the IT could avail themselves of access to liquor (Ackley Reference Ackley1937: 104).

However, while the IT certainly was notorious, it was by no means the sole preserve of banditry in the West of the 1880s and 1890s. This was true not only with regard to horse thieving but also with crimes that involved federal claims, such as mail robbery. Indeed, several other western and southern states and territories—California, Alabama, Texas, and Arizona—were the scenes of multiple train robberies, many of which involved federal jurisdictional claims. In many of these states and territories, the railroad companies took the lead in arresting and prosecuting train robbers, engaging their own detectives, hiring gunslinging experts on an ad hoc basis, or hiring specialist firms like the Pinkerton’s or Thiel’s agency to identify and break up criminal train robbery syndicates (Patterson Reference Patterson1981: 133–45).

Take Missouri, for example. There were at least a dozen major train robberies in the state during the 1890s, including one allegedly involving Jesse James’s son, who had become friendly with a notorious figure called Jack Kennedy (James Jr. Reference James1906: 141–85). Local sheriffs, railroad police, and private detectives took the lead in investigating all these crimes, forming posses of local residents and using connections to other local law enforcement agencies to arrest suspects. For example, to track down the Marion Hedgepeth Gang, which was involved in a notorious train heist near Glendale, Missouri in 1891, railroad detectives and St. Louis municipal police officers relied on Pinkerton agents to help bring several suspects back from San Francisco (Anonymous 1891).

This was not surprising given the limits placed on federal policing in states. But even in other territories local officers usually took the lead in investigating banditry. From 1890 through 1899, for instance, Arizona witnessed 26 stage-coach robberies or attempted robberies; USDMs seem to have only been involved in four of the arrest efforts (most of which involved crime against the US mail). Instead, a mixture of deputy sheriffs and local police officers coordinated the legal response. Even in higher-profile train robberies, USDMs did not always take the lead. In the late nineteenth century, there were seven major train heists or attempted heists in Arizona Territory. USDMs were only been involved in four of the subsequent posses, even though several involved relatively organized criminal outfits like the Burt Alvord Gang (Wilson Reference Wilson2003). Moreover, Stephen Cresswell (Reference Cresswell2002: 216–31) has argued that when they did act, USDMs in Arizona (in particular) often encountered local resistance.

In Oklahoma, however, the response to banditry was largely federal. Between 1865 and 1914, at least 1,400 men were deputized to serve as marshals in Oklahoma and the IT, while another 70 or so participated in posses. Some of these deputies became figures of public renown (or, to many native residents, notoriety), including Heck Thomas, Bill Tilghman, and Chris Madsen, the so-called Three Guardsmen. In the 1890s alone, these USDMs were involved in 18 high-profile pursuits in the IT, dangerous activities that often brought them into violent contact with suspects. ^{Footnote 6} These pursuits were covered by the local press and the men were often referred to in heroic terms (Anonymous 1900, 1899). One settler put the issue plainly: “no braver set of men ever lived than the old IT deputy United States Marshals” (Rhodes n.d.: 342).

It was Oklahoma’s unique racial order that best explains the willingness of the federal government to extend the use of USDMs in unprecedented levels in Oklahoma. Much of the rhetoric used by western officials and business leaders interested in abolishing the unique status of the tribal governments in Oklahoma relied on the trope of a racialized form of disorder. Responding to a train heist near the small Arkansas town of Olyphant on in November of 1893, the state’s governor, William Fishback, wrote an open letter to the president claiming that the theft had been planned in the Indian Territories and that local autonomy and governance needed to be curtailed. Oklahoma, he wrote, was “school of crime” and an “asylum of criminals” who were actively “stirring up the young Indians to deeds of blood and theft.” The only solution to such a “state of semi-chaos” was “for the federal government to assert its right of eminent domain over this part of the national domain, and to change its political relations with the United States” (Anonymous 1894d).

Many native residents took exception to Fishback’s language. For instance, an editorial in the Muskogee Phoenix claimed that “there is not a more honest people world than then Indians, and the non-citizens who come here … are peers of their fellows in the states.” Indeed, “of the famous bandits who have operated in and from the Indian Territory … single out the Indians and you have Henry Starr, Christie and a few lesser ones to compare with a long array of James, Daltons, Youngers, Creekmores and others of equal fame, who have made western banditry famous” (Anonymous 1894a).

Fishback’s letter, however, struck a chord with some both inside and outside IT. Many white settlers had serious problems with the racial ambiguities of life in the IT; S. C. Brown, who moved to the territory in 1875, recalled in a later interview that “some of them [the Indians] caused quite a lot of trouble, stealing cattle, but it would be mostly the half breeds doing most of the stealing.” Indeed, he argued that he and his family “found that we could get along better with the full-bloods,” a sentiment that echoed the putative concerns of Fishback and other “reformers” who claimed that it was the mixed nature of the population that led so many Natives astray (Duke Reference Duke1937: 173). Arkansas congressman Thomas McRae, the chair of the House Committee on Public Lands, concurred with Fishback’s plan, arguing that he was “now in favor of breaking up and destroying in toto the anomalous, demoralizing and un-American condition of affairs in the Indian country,” a place of “the choicest agricultural lands, capable of sustaining many millions of people, reserved for only about 60,000 persons of Indian descent and used in part as an abiding place for outlaws, murderers and thieves” (Anonymous 1894c). Even native residents resentful of Fishback’s language agreed that in some ways he had “hit the nail on the head and there is no use denying it,” even as the remarks were castigated for their racism and carelessness (Anonymous 1894b).

Those who wanted to abolish the IT used the example of organized banditry to make their case. Unlike in Missouri or Texas, the bulk of the well-known criminal gangs in Oklahoma were not composed solely of white, disaffected former Confederates. Instead, with the important exception of the Dalton Gang/Wild Bunch conglomerate (a group coming closest to resembling the heroic resisters of the Reconstruction era), a somewhat surprising number of the Oklahoma bandit gangs of the 1890s were either made up of racial minorities or were composed of individuals from a variety of racial backgrounds. It was at least partly because these criminals could not be easily mapped onto a clear system of racial classification that allowed the federal government to expand in Oklahoma where it was stymied in the South.

Race, Outlaws, and Officers

To establish this point and explore the racial makeup of gangs and law officers in Oklahoma in the 1890s, I used a variety of census and archival data to identify individuals who participated in the system of law and order in Oklahoma during the 1890s and their racial classification. To analyze the racial politics of those involved in high-profile crime, I then identified a set of criminal events perpetrated by gangs in Oklahoma during the 1890s, generated lists of gang member participants, and then identified law officers involved in the pursuit of the gangs. USDMs were, by far, the most important members of these legal posses and composed the core of my data set.

To identify racial membership of both gang and posse members, I aggregated individuals into groups (“gangs” and “posses”) and then calculated simple diversity indices for each group. Tables 1 and 2 present this information, with diversity (fractionalization) indices calculated across those of known racial membership and with a weighted average of all the scores across groups.

Table 1.

Racial makeup of outlaw gangs (1890s)

Note: Diversity based on Blau index (see appendix).

Table 2.

Racial Makeup of USDMs investigating selected criminal events in Oklahoma (1890s)

Note: Diversity based on Blau index (see appendix).

Table 1 focuses on the racial identities of members of bandit gangs involved in criminal events in the 1890s. Of the 14 gangs in the list, half are interracial. Those gangs that are not—in particular the Dalton Gang and its successor the Wild Bunch—were composed of white men and are the groups coming closest to the cultural archetype of the “social bandit.”

Less remembered, but just as important, were interracial gangs such as the Cook Gang and Henry Starr Gang, both of which also engaged in railroad robbery and were primary targets of the USDM efforts in the territory. Indeed, the Cook Gang, led by the half-Cherokee Bill Cook, was among the most active of all the Oklahoma gangs, and was responsible for at least 20 robberies and attempted robberies in 1894 and 1895.

These interracial gangs are important because they were fairly unusual. In Arizona and New Mexico, the outlaw gangs were largely white (including the southwestern branch of the Oklahoma-based Christian Brothers’ outfit, the High Five Gang). In Oklahoma, the use of violence by minority gang members against state authority was not only prevalent, but Native American and African American members often played important leadership roles within the groups. The capacity to violently and authoritatively engage in banditry in the territory created a novel kind of social threat, one which the federal state was seen as capable of addressing.

The second table unpacks this response by examining the racial composition of the posses of USDMs assembled to chase down these bandits. Though the numbers within each individual case are small, the most obvious finding over all the cases in table 2 is that the level of diversity within each posse of USDMs is considerably lower than that found in the gangs they were chasing. ^{Footnote 7} The weighted average of the diversity score among Oklahoma gangs is 0.29, while in posses the average is only 0.11. Though not a huge difference, the substantive result was that posses were considerably more homogenous and whiter than their outlaw counterparts, even including the all-white Dalton Gang/Wild Bunch.

Of course, posses were not composed solely of USDMs; a number of Indian scouts, local law officers, bounty hunters, relatives, and other young men with a taste for adventure joined posses in pursuit of outlaw gangs in the aftermath of robberies or other gang crimes (Carselowey Reference Carselowey1937: 296–99). When these individuals are added to the posse composition tables, the racial diversity of the groups goes up (though not to the level of the outlaw gangs they were chasing). But the key figures in these posses were USDMs. And these figures were overwhelmingly white, as is evident in table 3, which highlights a significant difference in the racial composition of USDMs to nonmarshal law officers who participated in posses.

Table 3.

Racial makeup of types of officers (1890s)

Note: There were 51 posse members for whom race could not be identified.

Fisher’s Test (Freeman-Halton Extension, two-tailed) = 0.012.

As a federal legal presence penetrated into the management of day-to-day law and order in Oklahoma, the officers responsible for its management were unlike the groups they were chasing. These outlaws were racially more heterogenous than their legal counterparts, which, in the context of Oklahoma’s ambiguous traditions of racial membership, presented an opportunity for the US government to apply coercive force in a manner denied them in the South. To a large extent, the USDM presence in Oklahoma was a means of reasserting a “white” state in a zone in which ambiguous racial presence had taken on the image of permitting and encouraging criminal disorder.

The Reception and Reform of Federal Policing in Oklahoma

The expansion of federal policing in the Indian Territories in 1890s was not a seamless process. A debate broke out in late 1894 and 1895 over how exactly to reorganize the federal court system, with many of Isaac Parker’s supporters attempting to resist a call by inveterate opponents of Indian territorial sovereignty like Senator John Tyler Morgan of Alabama to form new judicial districts in the territories with the aim of stripping existing federal courts and, ultimately, tribal courts of remaining authority. Not only did tribal dignitaries oppose this scheme but also it was unpopular with President Cleveland and many others concerned about the plight of the tribes, which had become targets of an extraordinary level of economic opportunism (Burton Reference Burton1995: 212–15).

Anxiety over banditry, however, helped push reform through (Harring Reference Harring1994: 60–73). The activities of the racially mixed Cook Gang in late 1894, which robbed a series of banks and trains and became subjects of national notoriety, led to a panic. Figures like the mixed-race “desperado” Crawford “Cherokee Bill” Goldsby, a member of the gang described by newspapers as “wolf in human form,” became emblematic to white residents of the disorders of the region and the need for judicial reform (Anonymous 1897). ^{Footnote 8} As Mihesuah (Reference Mihesuah2018: 147–53) has recently demonstrated in her book on Ned Christie, a Cherokee Keetoowah councilman accused of murdering a USDM in 1887, newspapers and political figures used racially coded language of savagery and desperation to frame and interpret the activities of criminality by nonwhites. Indeed, there was even talk of mobilizing the US Army to chase the Cook Gang, treating them more like a renegade Indian tribe than an outlaw outfit (Anonymous 1894e).

The Dawes Commission, initially established to initiate negotiations with the Five Civilized Tribes regarding land claims and present in the region during the crime wave, picked up on this idiom, implying that the Cook Gang in one of its reports was responsible for a “reign of terror” in the territories, in which “barbarous outrages, almost impossible of belief, are enacted, and the perpetrators hardly find it necessary to shun daily intercourse with their victims” (US Senate 1896: 89). This, of course, could only be addressed by a stronger and more integrated federal presence in the region. And the panic ensuing from the crime wave worked: as an editorial from Ardmore noted, “[W]e have no hesitancy in saying the Dawes commission have been helped more by recent outbreaks of lawlessness than by any other means” and that “it is now a question of only a short time until a change must come, and that change will surely be a territorial government” (Anonymous 1894f).

Parker also became a target for those advocating for territorial organization, who viewed his court as an expensive, inept shambles. One supporter of eliminating tribal sovereignty, journalist and attorney Rezin McAdam, wrote that Parker had had “nothing to say of the outlaw reign of terror which has for some months disgraced his jurisdiction,” and was treating IT residents as “subjects of a conquered population.” Under Parker, he and his fellow residents had come to “mistrust and fear the deputy marshal almost to the degree that we do the outlaw” (Congressional Record 1895: 951–52).

Despite these claims, the main contention was not over the presence of federal policing, but instead that the “outside” courts of Paris, Texas and Fort Smith stood in the way of territorial organization. Powerful economic interests like railroads with close relationships to these courts did not favor allowing a reorganization that might be more sympathetic to local settlers, many of whom shared Populist anger with the railroad companies. Judge David Bryant of the Eastern District of Texas court at Paris, for instance, argued that “the Indians and law-abiding white men of your country have confidence in the juries at Paris and Fort Smith, because they are not involved in the local feuds and local interests, not related nor acquainted with the defendants, and consequently can have no other motive or do other than right” (ibid.: 961). So-called Friends of the Indians, advocates for the assimilation of Native peoples into American cultural life, also saw federal legal oversight as a key means of “civilizing” the tribes. Thus, even while recognizing that many of the IT tribes had sophisticated legal systems and could “govern themselves with a good degree of success,” the jurist James Bradley Thayer (Reference Thayer1891: 541), wrote that “the time will probably come when, with or without their consent, there must be a readjustment of our relations with them” and “an ultimate absorption of that region into the Union.” In this sense, while supporting the existing system of courts organized at Paris and Fort Smith, Thayer thus nevertheless agreed with proterritorial advocates like McAdams, who argued that federal courts coterminous with territorial boundaries would protect the integrity of the jury and would make federal justice more available to local residents. In other words, while they disagreed on which federal court should take jurisdiction over Oklahoma and on the pace through which reform should occur, both opponents and advocates of expansion effectively viewed federal police control over the IT as inevitable and desirable. Only members of the tribes resisted the growth of federal policing power to any serious degree (Harring Reference Harring1994: 93–99).

In an environment of perceived lawlessness, those advocating for a transformation of the IT into a more carefully policed “home rule” federal jurisdiction carried the day. While retaining some of the jurisdictional prerogatives of the existing system headquartered at Fort Smith and Paris, the compromise Teller Bill of 1895 reorganized the federal system at Muskogee with the aim of disbanding the oversight of the external courts. This process of expropriation was completed in 1898, when the Curtis Act—which, among other abrogations of tribal sovereignty—largely liquidated tribal courts. The upshot was twofold: on the one hand, the federal imprint on local law and order became even more pronounced, while, on the other hand, the removal of external jurisdictional control paved the way for eventual territorial incorporation of IT and statehood (Burton Reference Burton1995: 217–18, 234).

In other words, white residents in Oklahoma Territory somewhat ironically found federal policing and the consolidation of federal criminal courts in the IT an ally in the push for settlement. Just like in the South during Reconstruction, external actors attempted to expand federal coercion in the region in Oklahoma (albeit in this case those external actors were often prosettlement residents from many states from the former Confederacy). Unlike the South, however, such efforts were supported by white residents, many of whom viewed federal policing as a key means to abolish tribal sovereignty and secure a (white) future for Oklahoma. It was racialized banditry that helped them achieve this agenda, providing a key excuse for the DOJ to expand its reach in a manner it had never been able to do during Reconstruction.

Newspaper Coverage and Social Banditry

Making the case for both statehood and the reorganization of federal policing in Oklahoma thus took place against a backdrop of a widespread perception of the Indian Territories as a place of specifically racialized disorder. But how did this widespread perception emerge and what shape did it take? Was Oklahoma banditry really perceived as somehow “different” from that found in places like Missouri?

To answer these questions, I pursue two strategies. First, to get a sense of the intensity of coverage of Oklahoma banditry, I consulted several databases of digitized US newspapers for the years 1870 through 1910, including almost 14 million pages of scanned material and collected information on newspaper coverage of two of the most important Oklahoma gangs active in the area during the 1890s—the Cook Gang and the Dalton Gang/Wild Bunch—as well as the Jesse James Gang (active in the Midwest and Southeast from the 1860s through 1880s). ^{Footnote 9} This strategy allows me both to get a sense of the overall level of coverage for Oklahoma crime relative to that in Missouri, with its rich tradition of social banditry, as well as whether or not gangs with different racial profiles were covered differently.

These gangs are ideal for exploring how banditry was interpreted. The James Gang, responsible for many stagecoach, train, and bank holdups, has, undoubtedly, entered into collective memory as an almost mythic exemplar of the “Wild West.” The Dalton Gang/Wild Bunch, famous for its high-profile showdowns with law officers at Coffeyville, Kansas and Ingalls, Oklahoma has similarly entered American historical myth. The Cook Gang, however—despite the effect of its activities on the policing reform efforts of the early 1890s—has almost been completely forgotten. Figure 3, for instance, depicts the relative frequency with which gang names appear among all bigrams—or two-chunk phrases—in Google’s enormous data set of published works from 1850 through 2000. As is evident, both the James and Dalton gangs have been mentioned much more frequently than the Cook Gang from late nineteenth century onward. Without a doubt, the Dalton Gang and the James Gang have been memorialized within American culture in a manner the Cook Gang have not (White Reference White1981).

Figure 3.

Frequency of gang name appearances in published works (1850–2000).

Note: Smoothed with five-year moving average.

Source: https://books.google.com/ngrams.

There are several possible reasons for these differences. For instance, the James Gang, active from at least 1866 through 1882, had a longer career than the Oklahoma gangs—the Dalton Gang was, in various iterations, active from 1890 through 1896, while the Cooks only seem to have mobilized themselves in earnest in 1894 and 1895. Moreover, the James Gang and the Dalton Gang were also relatively more successful than the Cook Gang; some estimates place the James Gang’s haul at more than $250,000 dollars (Anonymous 1882), while the Dalton’s probably stole at least $45,000 (some estimates put that number closer to $300,000; see Anonymous 1892.) The Cook Gang, however, seems to have only been able to take less than $5,000. At the same time, the Cook Gang also appears to have been very active; they were responsible for at least 20 successful or attempted holdups, while the Dalton Gang/Wild Bunch are known to have been involved in perhaps 15 robberies, and the James-Youngers with just 22 over a much longer period. Moreover, while the Cooks were in some respects less successful than other outfits, during the mid-1890s, at least, they were just as well-known as the Daltons (see figure 4).

Figure 4.

National media coverage of Oklahoma bandit gangs (1890–1910).

Note: Smoothed with three-year moving average.

Source: www.newspapers.com.

Another possible reason for differences in the how the gangs were remembered might have to do with the political contexts in which they operated. Some scholars have argued that the growth in antifederal and antimonopoly sentiment in the post–Civil War years may have made “social banditry” more acceptable to the public (Thelen Reference Thelen1986: 59–84). Thus, because the James Gang, for instance, was active during the height of the backlash to federal activities in the Reconstruction South, one might imagine that they were able to tap into a romantic narrative of antityrannical “Lost Cause” mythology that later dissipated. But this cannot explain why the Dalton Gang (active nearly 20 years later) achieved such a “heroic” status, while the Cook Gang was forgotten because both largely operated in the same place and time. Indeed, Oklahoma, like other southern and western states, was a hotbed for Populism during the 1890s, a political movement marked by just as much antimonopoly sentiment as Reconstruction (Thompson Reference Thompson1986: 70–96). But if antirailroad sentiment could have been a sufficient explanation for popular support for banditry, why would populist anger lead to the mythologization of the Dalton Gang and not its colleagues in the Cook outfit (White Reference White1981: 396–97)? Moreover, wouldn’t such sentiment have curtailed rather than facilitated the expansion of federal policing?

Instead, it appears as though racial differences were crucial in shaping how these groups were both perceived over time as well as how they served as catalysts for policing reform. Like the James Gang, the Dalton Gang/Wild Bunch were composed of white men (see table 1). The Daltons, however, were not only white they were related by marriage to Jesse James’s affiliates, the Younger brothers, allowing them to claim a kinship connection to the nation’s most infamous “bandit-hero.” Moreover, like the James and Younger families, the Dalton brothers came from a relatively respectable family of homesteaders. One of the elder brothers, Frank, had even served as a USDM in the late 1880s, while several others—Grat, Bob, and Emmett—had also served as officers at various points before turning to crime (An Eye Witness 1892: 20–65). The men they recruited into their efforts—including Bill Doolin, George Newcomb, and Charley Pierce—had backgrounds somewhat similar to their own.

The Cook Gang, however, was the quintessential “mixed-race” gang. It too was founded by a pair of brothers (Bill and Jim Cook), one of whom (Bill) had experience as a federal posse member (Shirley Reference Shirley1994: 6). But unlike the Daltons, the Cooks were partly Cherokee and were fully embedded in the tribal life of the West, spending their youths roaming through both the Cherokee and Creek nations (as well as New Mexico), pursuing odd jobs, and occasionally illegally selling whiskey or engaging in petty theft. In 1894, after deciding to evade several federal warrants, the brothers organized a more serious criminal effort in concert with their childhood friend “Cherokee Bill,” who resided in Catoosa, as well as several other associates from the area. This group, along with the Henry Starr and Rufus Buck gangs, became emblematic of a more aggressive form of mixed-race banditry; Glenn Shirley (ibid.: x), the foremost expert on Oklahoma outlawry, has described the Cook Gang as “perhaps the most vicious band of criminals that scourged the west,” making their absence in the collective memory of the Wild West even more striking.

One way to explore the tie between race-based outlawry and the construction of this public memory is to depict changes in news coverage intensity of Oklahoma banditry over time. ^{Footnote 10} Figure 4 depicts the yearly proportion of total newspaper pages in the data set with references to the Cook or Dalton gangs, revealing that the national media covered Oklahoma banditry with particular attention in the early 1890s. By the early 1900s, such coverage had declined to relatively stable levels. However, attention on the gangs differed between Oklahoma and the rest of the nation as well as between regions. These differences suggest the role the mixed-race Cook Gang played in focusing attention on Oklahoma as site of federal policing.

Between 1890 and 1899, for instance—the years when federal policing began its strongest expansion in the area—newspapers in Oklahoma were significantly more likely than newspapers elsewhere to cover the activities of the Cook Gang rather than Dalton Gang. ^{Footnote 11} This suggests that intense focus on the mixed-race gang by local news outfits may have helped engender more willingness to call for and accept federal policing reform. The regional data in table 4, which presents a contingency table of the number of news pages covering each gang by region, reinforces this point; during the 1890s, coverage of gang activity in Oklahoma and northern states was relatively balanced, while in the South and the West, the Dalton Gang was already becoming better known. By the first decade of the twentieth century, not only did all regions cover the Daltons more extensively but also Oklahoma had the highest relative level of such coverage. It appears as though at the moment the Dalton Gang were becoming enshrined as classic “social bandits” within popular historical memory, the more complex legacy of the Cook Gang was being forgotten precisely in the area in which they had been most active.

Table 4.

Regional coverage of Cook and Dalton gangs (1890–1910)

Source: Data from www.newspapers.com and chroniclingamerica.loc.gov.

Note: Numbers in parentheses indicate percentage deviation from expected value.

N = Number of news pages; regional differences all significant at p < 0.05.

Figure 5 expands this analysis to compare the relative intensity of news coverage of the Dalton and Cook gangs with the James Gang, as well as the ratio of such coverage. ^{Footnote 12} As is evident, while areas in which the gangs were active had the highest average within case coverage intensity (as we might expect), newspapers in the South covered the Dalton and James gangs more intensively than the Cook Gang, suggesting that attention to white bandits was more pronounced in the South than in the North. Further evidence of this appears when we examine the ratio of coverage intensity across gangs. ^{Footnote 13} In the Deep South the press covered white gangs relatively more intensively than they did the Cook Gang, suggesting yet another connection to the legacy of banditry as social resistance.

Figure 5.

Newspaper coverage of bandit gangs.

Sources: www.newspapers.com and chroniclingamerica.loc.gov.

Coverage ratios and intensity do not tell us about what these newspapers said. Thus, the second pathway I use to explore the perception of Oklahoma banditry involves analyzing the content of articles. Drawing on a hand-coded data set of 252 articles covering the activities of the Cook, Dalton, and James gangs from 1870 through 1920, I used a Latent Dirichlet Allocation (LDA) topic model to unpack the semantics of newspaper coverage of nineteenth-century banditry. LDA is a well-regarded algorithm capable of classifying terms within a set of documents (a corpus) into k topics (Blei and Lafferty Reference Blei, Lafferty, Srivastava and Sahami2009). Figure 6 represent the 10 terms with the highest probability (Beta) of appearing in a two-topic model for a data set including articles covering the Cook and Dalton Gangs only, while figure 7 includes articles covering all three gangs, divided into a three-topic model. The goal behind these models is to identify terms that are associated with each other and with some underlying latent topic. ^{Footnote 14}

Figure 6.

Topics in articles regarding the Dalton and Cook gangs (1890–1910).

Note: Terms stemmed and cleaned and duplicate entries removed.

Source: California Digital Newspaper Collection (cdnc.ucr.edu).

Figure 7.

Topics in articles regarding the James, Dalton, and Cook gangs (1870–1910).

Note: Terms stemmed and cleaned and duplicate entries removed.

Source: California Digital Newspaper Collection (cdnc.ucr.edu).

In figure 6, Topic 1 largely addresses documents related to the Dalton Gang, while Topic 2 refers to Cook Gang documents. In addition to terms related to the members of the gangs (e.g., “cook” and “bill” in Topic 2 and “dalton” in Topic 1), the algorithm also detected several other interesting topic associations. Topic 2, for instance, also was most likely to produce references to the federal coercive infrastructure (both “state” and “deputies” pick up on this concept) and to the racial distinctiveness of Oklahoma (“indian”), while Topic 1 is more likely to produce terms related to the actual behaviors of the Daltons (including “robber” and “kill”).

Further evidence of this comes when we expand the analysis to include coverage of the James Gang. Just as with the two-topic model of Dalton and Cook articles, the three-topic model presented in figure 7 reveals that the algorithm partitioned the expanded corpus in interpretable ways; Topic 2 is most associated with the Cook Gang, Topic 3 with the Daltons, and Topic 3 with the James Gang. Once again, Topic 2 also is associated with concepts related to federal coercion and the notion of the “outlaw” and Topics 1 and 3 with active banditry. Given the small size of this corpus, care has to be used in interpreting these results. Nevertheless, white and minority gangs seemed to have been covered quite differently, and the mixed-race Cook Gang was associated in the imaginary of nineteenth-century journalists with federal rather than local forms of policing.

Supplementary material

To view supplementary material for this article, please visit https://doi.org/10.1017/ssh.2021.13