Abstract
Was the American Western Frontier a violent place? While many agree with the popular Wild West conception, others argue for a Workaday, quiet version of the West. The Wild West/Workaday West debate is an important one because some argue that current levels of violence in the United States are a result of the Western history of violence. Unfortunately, establishing such a history of violence is difficult because of methodological issues. We propose here to address these methodological issues and participate in the Wild West/Workaday West debate. We examine two death records sources for Walla Walla and Spokane counties in Eastern Washington between 1903 and 1907. We find that the two Northwestern counties under study were more violent at the beginning of the 20th century than at the beginning of the 21st century, even though Walla Walla was significantly less violent than Spokane.
Introduction
The crime problem in the United States has been firmly established as a violent crime problem (Lynch & Pridemore, 2011; van Dijk, van Kesteren, & Smit, 2008; Zimring & Hawkins, 1997). Interpretations as per the etiology of this violence vary greatly, depending on the analytical lens used (see Nivette, 2011a for an overview). Some of that research has been historical in nature, looking at long-term trends to glean clues as to the explanation of violence (Eisner, 2011; Nivette, 2011b; Pinker, 2011; Roth, 2009). One historical aspect of particular interest has been interpersonal violence in the late-19th century American West. It has been a controversial issue among historians, who remain divided in their interpretations of the facts as supporting or undermining a Wild West hypothesis. Indeed, while some see contemporary violence in the United States as inherited from a history of violence, particularly in the West, others argue to the contrary that far from being chaotic, lawless, and violent, the West was first and foremost a place of quiet rural life. Before we can assess whether the United States’ violent past is to blame for its contemporary violence, we propose here to focus on the Wild West/quite West contradiction in the historical literature. After reviewing the much-debated question surrounding the levels of violence at the American Western Frontier, we will endeavor to add our stone to the edifice. First, we will establish rates of lethal violence in two Eastern Washington counties at the beginning of the 20th century. Second, we will compare those early 20th century rates with contemporary rates for the same counties. Finally, we will propose that because of limitations inherent to mortality data, especially in a historical context, it may be useful to consider lethal violence beyond the realm of homicide, including suicides, and accidents as well. We thus propose to not only replicate in a new geographic area other studies in the Wild West Versus Workaday West debate, but to also rejuvenate the debate using a different set of methods.
The Wild West Hypothesis
Proponents of the Wild West Hypothesis have contended that American contemporary violence reflects the frontier heritage of the 19th century (Brown, 1975; Drago, 1970; Elliott, 1944; Frantz, 1969; McGrath, 1984; McKanna, 1997; Rosa, 1969). Specifically, previous studies argued that much of America’s contemporary criminal behavior may stem from the frontier heritage of lawlessness (Elliott, 1944), which “encouraged the idea and practice of violence” (Frantz, 1969, p. 151).
This culture of lawlessness was framed by the intersection between characters such as gunslingers, vigilante, or legal enforcers, and the various land, labor, racial, and religious wars of the West (Brown, 1983). The literature about Western gunfighters is extensive as well as contradictory (McGrath, 1984), tending toward exaggerating the exploits of gunfighters (Burns, 1926; Connelley, 1933), or describing the gunfighters as less than noble characters (Rosa, 1969; Waters, 1960). Such literature, however, lacks statistical support (McKanna, 1997).
Alongside gunfighters, vigilantism has garnered a sizeable amount of scholarly attention. Most notable among those, Richard Maxwell Brown (1975), in Strain of Violence, has offered one of the seminal studies on vigilantism in the American West. The author notes that the presence of outlaws coupled with ineffective law enforcement and lack of public funds created social problems, particularly three categories of violence: (a) the crime wars of the West, involving the violence of vigilantism, legal law enforcement, personal redress, outlaws, and gunfighters; (b) the land and labor wars of the West; (c) the racial and religious wars of the West (1975, p. 237). Following Brown, Lane (1976) further argues that America’s gun culture was associated with freedom in armed self-defense and the United States was recognized as the most violent society in the Western world in 1876.
These aspects of the Wild West hypothesis, however, remain plagued by their anecdotal nature, and it is questions of statistical estimations of violence in the frontier West that are central to the whole debate. Numerous authors have documented higher rates of violence in the frontier West than either in contemporary Eastern regions, or in late 20th or early 21st century America. These rates, however, are often calculated for small areas (sparsely populated counties or towns), and for short periods of time (because of data availability, trends are difficult to establish). Even though most of the available rates are indeed higher than contemporary eastern rates or late 20th and early 21st century rates, there is a lot of variation, including some extreme numbers—Boessenecker (1999) estimates the homicide rate in Los Angeles county in 1850-1851 to be 1,240/100,000. Other examples include McGrath (1984), who documents all instances of violence in two California mining camps; Aurora, which flourished from 1861 to 1864; and Bodie, which flourished from 1877 to 1882 and had gained a well-deserved reputation for violence in the West. McGrath describes violence involving women, minorities and juveniles, robberies and shootouts, stipulating that the frontier homicide rate was higher than the national rate (116/100,000 per year between 1878 and 1882 for Bodie, California). Such findings are confirmed by Zanjani (1992)’s study in Goldfield, Nevada, where annual homicide rates usually exceeded the levels in present-day American cities. Specifically, Goldfield had 31 homicides between 1904 and 1909, with a population oscillating between the hundreds in 1904, 18,000 in 1907, and close to 5,500 in 1910. Assuming a very unlikely equal distribution of those homicides over the 5 years under study here, and taking the largest population as the denominator, a very rough, yet conservative estimate of the yearly homicide rate in Goldfield at the beginning of the 20th century is 34.4/100,000. It is clear that “a gold rush acted as a magnet for the kind of men most likely to commit violent acts” (Zanjani, 1992, p. 42). High lethal-violence levels in the American West are documented on a larger scale by McKanna (1997) in Homicide, Race, and Justice in the American West, 1800 to 1920. McKanna addresses subcultures, rapid population increase, racism, saloons, weapons, and heavy drinking as dynamic forces and argues that such forces contributed to the prevalence of violence in the West, concluding that the West was far more violent than the East (homicide rates of 152/100,000 for 1880-1884 in Gila county, Arizona, 107/100,000 in the 1850s for San Luis Obispo county, California, 95/100,000 in the 1850s for Tuolomme county, California, for example, as opposed to 5/100,000 in New York City in the 1880s, and 0.8/100,000 in Maine in the 1860s). At the state level, Peterson Del Mar (2002) estimates Oregon’s homicide rate to be at least 30 per 100,000 adults per year between 1850 and 1865 (see Table 1 for recap).
Wild West Hypothesis Homicide Rate Recap Table From Most to Least Violent Places.
Per 100,000 per year.
The Workaday West Hypothesis
Such numbers, however, are not well received by proponents of the Workaday West hypothesis, who argue that Wild West historians in general tend to just assume that violence was pervasive in the frontier West and that those who provide quantitative evidence do so using unsound figures. Workaday West historians, in turn, propose that the West was not a violent region contrary to its popular image in Hollywood films. Much of the literature on this perspective asserts that violence was not a major fact in the development of the West (Dykstra, 1968; Friedman & Percival, 1981; Hollon, 1974; Prassel, 1972; Turner, 1893/2010; Udall, Dykstra, Bellesiles, Marks, & Noble, 2000).
Some of the arguments proponents of the Workaday West hypothesis make in opposition to the Wild West hypothesis pertain to media distortion of violence in the West, hoax imagery, and the U.S. government’s military interventions (DiLorenzo, 2010; Udall et al., 2000). Much of these arguments, however, seem to be suffering from the very same weaknesses imputed to the other side of the debate. For example, Bellesiles (in Udall et al., 2000) argues, following Frederick Jackson Turner, that the violent image of the West was created by the media and is now entrenched in the collective conscience as a “mythology,” which scholars have difficulties debunking. This is particularly true because violent aspects of the West have been overemphasized through a multitude of media such as literature, motion pictures, and television (McGrath, 1984). Yet, this argument is not tested empirically, such a “mythology” possibly having emerged from a reality of violence.
Beside the relation to media overemphasis, Workaday West proponents argue that the U.S. government’s military interventions in the West created the culture of violence of the late 19th century frontier, engaging in wars with the Plain Indians, for example (DiLorenzo, 2010). Yet, it is unclear what about these Western military interventions is different from the similar Eastern or Southern military interventions of the time, creating a geographically unique culture of violence, instead of a generalized culture of violence against native people.
Overall, proponents of the Workaday West hypothesis simply argue that the West was not violent. Some even contend quite the opposite. Prassel (1972) argues that, compared with populous centers of the East, the Far West was actually peaceful, and crime in the American West was more of an urban phenomenon than a derivative of the frontier. Moreover, Hollon (1974) contends that violence on the frontier was the result of violence in our nation’s settled areas, “[…] frontier lawlessness was primarily the result, rather than the cause, of our violent society” (p. ix). His attempt to show that “the Western frontier was a far more civilized, more peaceful, and safer place than American society today” (p. x), however lacks statistical documentation (McKanna, 1997).
Dykstra (1968), though, supports his Workaday West argument with some numbers. Examining the Kansas towns of Dodge City, Abilene, Ellsworth, Wichita, and Caldwall, he reaches the conclusion that they were not especially violent. Dykstra notes that from 1876 to 1885, there were 15 documented homicides in Dodge City, an average of only 1.5 homicide per cattle-trading season, which he interprets as an indication of low violence levels. He also shows how the relatively nonviolent Dodge City became the historic locale for the cultural construction of violence. Dykstra points out that most unfavorable moral judgments stemmed from Dodge City’s sin rather than its occasional violent death. Similarly, Friedman and Percival’s (1981) study of the criminal justice system of Alameda county, California, from 1870 to 1910 shows that murder, armed robbery, violent rape, and burglary were on the decline in Oakland in the late 19th century, noting that “the coroner reported only four homicides in the county in 1893 and people killed each other at a very low rate” (p. 27). These studies point to the heart of the question: How to interpret quantitative historical data (Roth, 2007)?
Methodological Disagreements Fueling the Debate
Whether the West was violent or not is a question that ultimately has to be settled using quantitative empirical evidence. Earlier studies (Dykstra, 1968; Friedman & Percival, 1981; Hollon, 1974; McGrath, 1984; Prassel, 1972) were criticized for lacking statistical sophistication, failing to provide statistical documentation and establish homicide rates, and lacking a quantitative research design. Dykstra (1996, 2003, 2009), however, has argued that all homicide rates tend to be statistically skewed upward in sparsely populated places and therefore appear to be lower in more populated areas—the “fallacy of small numbers.” Nevertheless, Roth (2007) reminds us that despite the sensitive nature of rates calculated with small denominators, the issue at play is not one of body counts but one of risk and that theories of probabilities discredit the “fallacy of small numbers” as long as the unit studied is representative of the population.
Much of the debate also centers on questions of data choice and availability, with some scholars using death certificates, others coroners’ inquests and other court records, or sometimes newspapers articles (Dykstra, 2003; McKanna, 1997; Peterson Del Mar, 2002). Studies in the field of homicide research have examined differences between data sources. For instance, Lyman and colleagues (2004) compare death certificates, medical examiners’ reports, and Uniform Crime Reports and demonstrate that death certificates and medical examiner’s reports provide the most comprehensive child homicide data counts. Wiersema, Loftin, and McDowall (2000) examine agreement between homicide estimates from the Supplementary Homicide Reports and the National Vital Statistics System. They find that the lack of agreement between the two systems is mainly due to “differences in case definition and ambiguous or failed procedures” (p. 335). In a similar vein, comparing information from law enforcement reports and medical examiner data, a recent study found that law enforcement reports complement data from the medical examiner reports, mostly because law enforcement focuses on who caused the death when the medical examiner focuses on what caused the death (Gabor et al., 2008). In the public health field of violent deaths studies, similar research proposes similar findings, ascertaining data limitations from all sources (Bonnie, Fulco, & Liverman, 1999), or specifying situations in which using multiple sources can be helpful—to assess circumstances of lethal events more so than to increase the quality of victim-related variables, for which death certificates provide the most comprehensive information (Goldsmith, Pellmar, Kleinman, & Bunney, 2002; Logan, Karch, & Crosby, 2009).
Current Study
The debate between Wild West and Workaday West proponents rests on data and their understanding. To be sure, historical data can be spotty and of questionable quality (Monkkonen, 2001), nonetheless, we propose here to participate to the Wild West/Workaday West debate from a different angle in hopes to offer added confidence in empirical findings as well as potential for theory. Indeed, while capture-recapture methods used to bolster the accuracy of homicide counts in historical research have offered significant improvements in data quality (Monkkonen, 2001), the very nature of homicide investigations in the past, particularly with regard to the use of forensic medicine in its infancy (Thorwald, 1965), can lead to questions regarding the correct identification of homicides versus other violent deaths such as suicides or accidents. Let us keep in mind that at the beginning of the 20th century, deaths, particularly in less urban areas, were unlikely to be ascertained by medico-legal professionals. Coroners did not have to be physicians, let alone forensic pathologists, and it is therefore very likely that deaths were misclassified quite often. More specifically, it is more likely that homicides and violent deaths in general were undercounted rather than overcounted, particularly in cases involving nonwhites, women, or children victims. Broadening the debate surrounding violence and examining homicides along with suicides and accidents helps partly alleviate concerns regarding death classification at the beginning of the 20th century. By relaxing the standards surrounding definitions of violence, we are more likely to include some misclassified deaths and minimize undercount. While our concern here is methodological and we will not stray from that goal, it is worth noting that such an approach is not without theoretical backing. Homicides, suicides, and accidents can indeed be seen as indicators of deviance and violence in society (Holinger, 1982; Lane, 1979; McKanna, 1997; Neuilly, 2007, 2011), some of it external, some self-imposed (Unnithan, Corzine, Huff-Corzine, & Whitt, 1994), and some of it related to social class, quality of life, and risk taking factors (Gottfredson & Hirschi, 1990; Reiman & Leighton, 2013). In addition, while our strategy is supported by the extant literature, and used here as a methodological safeguard against limitations of historical data, we also present homicide rates independently of other violent death rates. Finally, we should also acknowledge that our approach does not provide a perfect solution, as it is very likely violent deaths were also misclassified as natural deaths. That being said, the reliability and thus comparability of mortality data is a problem that goes beyond historical data (see discussion in previous section) and should therefore be acknowledged and minimized as much as possible, but should not refrain us from our studies.
As such, the project is responding to injunctions by Brown (1983) and McKanna (1997) to further investigate death rates in Western counties in the last part of the 19th and early 20th centuries to reconcile the debate between Wild West theorists and Workaday West theorists. We thus tackle issues of historical data reliability and levels of lethal violence. It is only once these issues are resolved that questions pertaining to the varying meanings and types of violence can be addressed.
Method
The current research casts a broad archival net to retrieve as many violent death records as possible using a strategy similar to Monkkonen’s (2001), triangulating two official sources of death records, a coroner’s ledger listing all cases investigated, and a random sample of death certificates, for the years 1903 through 1907 in Spokane county, Washington. Coroners’ ledgers for Walla Walla county for the corresponding dates are also included. We further describe our research design and logic here.
Unit of Analysis
Based on the review of the literature on the history of violence in the West, counties have emerged as the best unit of analysis. They are indeed the central repository of all information pertaining to administration and government, including death records and coroners’ inquests (Lane, 1997; McKanna, 1997). They also allow for a subtler approach to regional variation than states on variables of interest, differentiating for example between mostly agricultural counties, mining counties, and logging counties (Brown, 1983; Dykstra, 2003). An investigation into territorial and early statehood periods, however, does not afford the researcher much uniformity in terms of available data; the counties chosen are thus those for which most data are available covering the earliest periods.
Sampling
The Wild/Workaday West debate seems to be centered on the Midwest, the Southwest, and the West Coast, the Inland Northwest being, to our knowledge, unrepresented (we are, however, acknowledging Peterson Del Mar’s 2002 study of 1850-1865 Oregon). The data sampling decisions are based on this initial fact. Spokane and Walla Walla Counties were identified as the two counties with the most readily available and earliest data through the Eastern Washington Archives. Data for Spokane County include two coroners’ ledgers starting in 1903 and listing 1,963 coroner-investigated deaths until 1918, and 4,174 digitalized death certificates for the years 1891 through 1907, when vital record responsibilities started being centralized through the state. Walla Walla death records include 1,270 online death certificates from 1891 to 1907, with missing data between 1903 and the beginning of 1907, and two coroners’ ledgers covering the same dates and including information on 2,850 coroner-investigated deaths. We thus decided to focus on the 1903 to 1907 period for which there was the most overlap in data across the two counties and the two sources. While these counties might not be representative of the West as a whole, we argue that what we provide here is another stone onto the edifice of research investigating violence in the West. Unlike much of the literature presented in the review section, we have not selected these two counties because they were known as violent, or because they were known as peaceful. Rather, we have selected them because they had never been studied under such an angle, and there were appropriate data available. We propose that our findings will not settle the question of the Wild West versus Workaday West debate for good, but rather they will solidify the validity of one side or the other of the argument.
Analyses
As previously stated, the goal of this research is twofold: first, to assess historical homicide data reliability in the West, and second, to establish levels of lethal violence in two Northwestern counties at the turn of the 20th century, to assess whether they were “wild” or “workaday.” Considering the limitations of the data accessible to us, we used Spokane county as our data reliability control county and first compared a subpopulation of coroners’ records with a random sample of death certificates for the overlapping period of 1903 through 1907 to assess levels of lethal violence overall—including homicide in all its forms, along with suicides and accidents (Lane, 1979; McKanna, 1997; Neuilly, 2007, 2011)—as well as rates of homicides through triangulation. This also allowed us to establish the level of data reliability for the area and period. In a second step we compared data from Spokane and Walla Walla Counties coroners’ ledgers to ascertain whether both counties were similar in terms of homicides and violent deaths.
Throughout, we use both rates per 100,000 calculated with population counts estimated from historical Census data, and raw body counts. Rates are very useful to make comparisons with other places and times. Raw counts are used for the calculation of various tests of homogeneity of proportions, which establish whether frequency counts on a categorical variable are distributed similarly across different populations. We thus compare distributions of homicides, suicides, accidents, and natural deaths, as well as distributions of aggregated violent deaths between the two counties at the time and the two counties now. Indeed, present day local mortality rates seem the most appropriate to establish whether the frontier West was violent compared with the modern day established West. While mortality data collection processes have undeniably changed from the beginning of the 20th century compared with the beginning of the 21st century, it is important to remember that such processes were no more unified across space at the time, making historical comparisons equally risky. In addition, the debate about the Wild West is not simply focused on whether the West was more violent than the settled East, but more importantly on the fact that those rates of violence were higher then than they are now, which is why comparing our study counties with their modern day selves serves the purpose the best.
Results
Counts and Rates
Comparing Walla Walla and Spokane counties for the early study period, we first notice that the natural death counts and rates identified for Walla Walla county are higher than for Spokane county between 1903 and 1907 despite the fact that Spokane county was about four times as populous as Walla Walla county at the time (see Table 2). This is an important factor because it points to the limitations inherent to the data, and in this particular instance, to the discrepancies between institutional data gathering strategies.
Death Counts and Rates by Manner for Walla Walla and Spokane County Over the Two Study Periods.
A within-county triangulation of death records and coroners’ ledgers is helpful to ascertain the nature of the difference between the two counties. Indeed, looking at coroners’ ledgers and death certificates available for Walla Walla county (see Table 3), one notices how close the two sources are between 1891 and 1899 and how death certificates lack between 1900 and 1906 in comparison with the ledgers. This indicates that the Walla Walla coroners likely recorded every single death in the county, whether suspicious, violent, or neither.
Comparison of Death Counts Using Two Sources for Each County Under Study.
This is different from Spokane county, where the comparison of coroners’ ledgers with available death certificates shows coroners’ practices focusing solely on suspicious and violent deaths, which translates in lower overall death counts using the ledgers. A further comparison of Spokane county coroners’ ledgers with a random sample of Spokane county death certificates for the overlap years of 1903 through 1907 shows underrepresentation of violent deaths in the death certificates sample (see Table 4). Considering the purpose of our study being the assessment of violent deaths in the West, focusing on the most accurate source of violent deaths reporting is paramount. We thus use coroners’ ledgers as our most trusted source of information on violent deaths in Walla Walla and Spokane county between 1903 and 1907.
Comparison of Spokane County Coroners’ Ledgers (CL) and a Random Sample of Death Certificates (DC) Death Counts by Manner of Death.
Rates of each manner of death are calculated for each year and averaged over the study period to avoid instability of low incidence events such as homicides and suicides, especially in small population units (even though our smallest county, Walla Walla, is beyond twice the 10,000 population threshold given by Spierenburg (1996) for meaningful statistical inference from rates per 100,000). Setting natural deaths aside, and before delving into any test of significance, it appears that Walla Walla’s average death rates between 1903 and 1907 were lower than between 2003 and 2007 for accidents (51.8 vs. 53.1) and suicides (6.5 vs. 14.7), but higher for homicide (3.2 vs. 0.7). Spokane county, however, seemed to have been more violent across all three violent death categories between 1903 and 1907 than in recent years (51.2 vs. 43.2 for accidents, 19.6 vs. 15.9 for suicides, and 6.1 vs. 3.3 for homicides; see Table 2).
Tests of Homogeneity of Proportions
To establish the statistical significance of the differences between observed death distributions in the counties under study, we conducted tests of homogeneity of proportions using raw counts of deaths. We compared the distributions along several dimensions. Specifically, we compared Spokane with Walla Walla as well as each county with itself on the following dimensions:
all manners of deaths by year (see Table 5),
all manners of deaths as well as homicides versus all other deaths by aggregated period (old versus new; see Table 6), and
aggregated violent deaths versus natural deaths by aggregated period (old versus new; see Table 7).
Test of Homogeneity of Proportions Between Counties for All Manners of Deaths Distributions by Year.
p < .01. **p < .05. ***p < .005.
Test of Homogeneity of Proportions Between Counties for All Manners of Death Distributions per Time Periods and Within-Counties Between Time Periods.
p < .01. **p < .05. ***p < .005.
Test of Homogeneity of Proportions Between Counties for Violent Versus Natural Deaths Distributions per Time Periods and Within-Counties Between Time Periods.
p < .01. **p < .05. ***p < .005.
The comparison of the two counties by manner of death by year shows that while they exhibited significantly different manner of death distributions between 1903 and 1907, they do not for 2003 through 2007 (except for 2005; see Table 5). Aggregating the manner of death distributions by study periods (1903-1907 and 2003-2007) confirms the differences between Walla Walla and Spokane counties in the early 20th century, and lets more differences emerge between contemporary Walla Walla and Spokane counties (see Table 6), which remains true for the distribution of homicides compared with all other types of deaths (see Table 6). Aggregating violent deaths together (see Table 7) reiterates the findings from Table 2 (different Walla Walla and Spokane for 1903-1907 but similar for 2003-2007). Within-county comparisons exhibit differences between the early 20th century and the early 21st century for each county (see Tables 6 and 7). Finally, the significant differences identified point to Spokane county being more violent than Walla Walla county at the turn of the 20th century, while both counties were more violent then than they are now.
Discussion
It is always reasonable to approach widely held “common sense” beliefs cautiously and to put numbers under the microscope, leaving no data stone unturned. In that respect, the Wild West historical literature often suspiciously resembling a Clint Eastwood Western, the Workaday West criticism seems warrantied. Anecdotal evidence, however, is not always unrepresentative of the reality from which it is plucked, and those Wild West historians describing examples of gunslingers and vigilantes do illustrate a quantifiable reality. Whether it is Bodie, California’s 116/100,000 homicide rate per year between 1878 and 1882 (McGrath, 1984), Goldfield, Nevada’s average 34.4/100,000 in the early 20th century (Zanjani, 1992), Gila county, Arizona’s 152/100,000 rate for 1880-1884, San Luis Obispo county, California’s 107/100,000 rate in the 1850s, Tuolomme county, California’s 95/100,000 homicide rate in the 1850s (McKanna, 1997), or Oregon’s 30/100,000 yearly rates between 1850 and 1865 (Peterson Del Mar, 2002; see Table 1 for a recap), the picture of a violent West does not seem to emerge out of thin air. Yes, those rates have to be taken with a grain of salt, but if anything, it is probably more because they underestimate homicide rather than overestimate it. Indeed, our study shows how historical data sources tend to provide undercounts rather than overcounts because of disparate reporting, missing data series, nonspecialized death professionals, and archaic forensic medicine (Thorwald, 1965). These rates also only provide a partial view of the American Frontier in that cases are often purposely sampled because of presumed or documented histories of violence, and they tend to be limited to certain parts of the frontier (Midwest, Southwest, West Coast). Finally, while the rates listed above would be considered high in any set of circumstances, comparisons established with either much larger contemporary Eastern cities or modern-day homicide rates often miss the point in that they do not provide adequate comparison points. These rates, nevertheless, are coherent in a long-term approach to historical trends in violence, participating in our understanding of an overall decline of violence corresponding to the civilizing process going hand in hand with the establishment of government (Eisner, 2011; Elias, 1939; Nivette, 2011b; Pinker, 2011; Roth, 2009).
We endeavored to add our contribution to the Wild West/Workaday West debate by investigating death records, rates, and distributions in two Eastern Washington counties at the beginning of the 20th century. The Inland Northwest has, to our knowledge, not been studied in the context of frontier violence. Our research thus replicates previous findings from other parts of the West, reinforcing their external validity. Our approach nonetheless is somewhat different, aiming at providing better comparison points and stronger violence estimates.
We indeed did not compare 1903 to 1907 Walla Walla and Spokane counties to 1903-1907 New York City, Chicago, or Boston, nor did we compare them with overall 21st century United States. We instead compared the two counties first with each other, which allows us to investigate questions of historical data reliability, as well as what may be the issue underneath the Wild West/Workaday West debate, which is the idea of a somehow uniform West, in culture, history, economics, demographics, and by extension crime. Second, we compared the two counties with themselves, 100 years later, a more tenable comparison, particularly in the greater context of long-term historical trends in violence.
In addition, we did not stop at examining homicide distributions across- and within-counties, we also aggregated homicides together with accidental and suicidal deaths. The theoretical rational behind this strategy is that nonnatural deaths such as accidents, suicides, and homicides, are different expressions of the same underlying lethal impulse principle (Henry & Short, 1954; Holinger, 1982; Lane, 1979; Neuilly, 2007, 2011; Unnithan et al., 1994, only to cite a few). Furthermore, the limitations identified here in historical mortality data (disparate reporting, missing data series, nonspecialized death professionals, and archaic forensic medicine, see above) make such an aggregation all the more relevant: If violent deaths can arguably be considered as one theoretically, their aggregation makes the estimates stronger methodologically, as it renders null or at least greatly minimizes questions pertaining to the ability of small-jurisdictions early 20th century coroners to adequately determine manner of death and produce reliable death counts. Finally, our two counties of choice yielded overall stronger violence rates because of their larger populations compared with some of the towns and counties on which much of the Wild West hypothesis is founded.
Based on those specifications, our findings indicate that while both counties had lower homicide rates than those quoted from the Wild West literature, they were significantly more violent at the turn of the 20th century than they are now, providing support to the Wild West hypothesis. Furthermore, the higher violence levels were independent of whether we used homicides only, or homicides aggregated with other violent deaths. The coroners’ ledgers we used to establish our death counts were helpful in contextualizing such overall violence. Indeed, these ledgers included qualitative narrative information on the circumstances and cause of the death, as well as on some of the decedents’ characteristics. The amount of details included in such narrative descriptions varied by county and by what appeared to be changes in either coroners or scribes, or both, but what transpired from first impression was the harsh nature of life out West. Even natural deaths seem to corroborate such an existential struggle, including high numbers of infant and children’s deaths. Further research will focus on analyzing such rich death narratives.
The two counties under study may both have been more violent than their 21st century counterparts, they were nonetheless different from each other back then while they are at least partially similar now, showing the complexity behind the notion of a uniform West. Indeed, Spokane and Walla Walla counties were different in at least two dimensions with regard to lethal violence. First, Spokane county was more violent than Walla Walla county with regard to homicides as well as aggregated violent deaths. Just as both counties show violence rates that are lower than some of the violence rates identified in the literature, they are not both equally as violent. This element is important in that it pertains to what may very well be at the heart of the Wild West/Workaday West debate, as previously stated, the idea of a West seen as one. Our study shows the importance of replication in such historical research, to strengthen external validity of findings regarding such a large and varied geographic area. While general patterns of violence may remain unquestioned by such replications (the West was violent), they bring additional subtleties to those patterns (that violence may have been relative), which in turn have to be causally examined. Further research will delve into such variations and their explanations. Second, the two counties are also different in their death recording systems, with Walla Walla county apparently using the coroner’s office or at least its associated remaining ledger to keep track of all deaths, while Spokane county seemed to have used its coroner for the more common goal of investigating nonnatural and suspicious deaths only. This element is important because it reminds us of the limitations inherent to the use of historical—and in truth all—data and the importance of taking all possible and necessary precautions to understand such limitations and adequately select data sources and series. The best data series for the purpose of studying violent deaths in Walla Walla and Spokane counties were thus established to be coroners’ ledgers. This was done through triangulation with death certificate series for overlapping years, limiting the current study to 1903 through 1907. Further research will expand back in time to 1891.
Conclusion
In our endeavor to participate to the Wild West/Workaday West historical debate, we examined death records in Walla Walla and Spokane counties between 1903 and 1907 and established violent death rates to be higher then than they are in those counties at the beginning of the 21st century. Our study lends support to the Wild West hypothesis and adds to the data coverage of the issue while underscoring the complexity of conceptualizing such a wide geographic area as “the West” as monolithic when it comes to violence. While we took great care in selecting our study sites and approaching the topic soundly, on theoretical and methodological grounds, ours is not flawless research. We were limited in the period we were able to cover with adequate reliability, and it is possible that our relatively lower violence rates compared with those identified in the literature might be an artifact of our early 20th century timeframe. In addition, we purposefully refrained from tackling an explanatory goal for this particular project and our theoretical hypotheses remain untested. Nonetheless, the present study provides a sound basis for future research aiming at expanding the period under study, allowing for trends to appear, and for causality to be tested.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: 2010 University of Idaho John Calhoun Smith Memorial Fund: “Violent Roots: Assessing the Levels of Lethal Violence in Eastern Washington and Northern Idaho Counties During the Territorial and Early Statehood Period.” Research resources involved hourly pay for second author Kelly Rae Stout as a research assistant.