Doing Violence, Making Race

Author(s): Mattias Smångs

Source: American Journal of Sociology , March 2016, Vol. 121, No. 5 (March 2016), pp.
1329-1374
Published by: The University of Chicago Press

Stable URL: https://www.jstor.org/stable/10.2307/26545739

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/10.2307/26545739?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access
to American Journal of Sociology

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
Doing Violence, Making Race: Southern Lynching
and White Racial Group Formation1

Mattias Smångs
Fordham University

This article presents a theoretical framework of how intergroup vio-

lence may figure into the activation and maintenance of group cate-
gories, boundaries, and identities, as well as the mediating role played
by organizations in such processes. The framework’s analytical advan-
tages are demonstrated in an application to southern lynchings. Find-
ings from event- and community-level analyses suggest that “public”
lynchings, carried out by larger mobs with ceremonial violence, but not
“private” ones, perpetrated by smaller bands without public or cere-
monial violence, fed off and into the racial group boundaries, catego-
ries, and identities promoted by the southern Democratic Party at the
turn of the 20th century and on which the emerging Jim Crow system
rested. Highlighting that racialized inequalities cannot be properly un-
derstood apart from collective processes of racial group boundary and
identity making, the article offers clues to the mechanisms by which
past racial domination influences contemporary race relations.

INTRODUCTION
The centrality of boundary processes in social life has in recent decades gained
renewed appreciation, leading to important theoretical developments and a
growing body of empirical research ðfor reviews, see Lamont and Molnar
½2002 and Pachucki, Pendergrass, and Lamont ½2007Þ. Scholarship on race
and ethnicity has been particularly central to this literature and provides

1
Earlier versions of the article were presented at the annual meetings of the American
Sociological Association, the Eastern Sociological Society, and the International Network
of Analytical Sociologists. Support for this research was provided by a Fordham Uni-

© 2016 by The University of Chicago. All rights reserved.

0002-9602/2016/12105-0001$10.00

AJS Volume 121 Number 5 ( March 2016): 1329–74 1329

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

new conceptual tools for analyzing the relational and dynamic processes by
which ethnoracial groups define and establish themselves ðe.g., Lamont 2000;
Brubaker 2004; Alba 2005; McDermott 2006; Bail 2008, 2012; Fox and
Guglielmo 2012; Wimmer 2013Þ. Intergroup conflict and violence are in that
respect particularly relevant because they reveal in unusually stark fashion
the conceptions about racial or ethnic differences and affiliations that form
the basis for group formation. In consequence, simply observing that race
and ethnicity have different meanings in different times and places does not
suffice if we are to take their contested, constructed, and contingent nature
seriously. In order to do so, we should consider the configuration and content
of actors’ conceptions of racial or ethnic differences and similarities as well
as the consequences of such conceptions for relations and behaviors among
and between racially and ethnically defined groups ðcf. Lamont and Molnar
2002; McDermott 2006Þ.
This article advances the study of lynchings of African-Americans in the
post-Reconstruction U.S. South and the literature on ethnic and racial bound-
ary formation by analyzing intergroup violence and its role in the construc-
tion of racialized group boundaries and identities. I both explain differences
in the types of lynchings that took place and link the pattern of lynchings
to broader political developments via the consolidation of white unity and
power through the Democratic Party and disfranchisement. This topic is
particularly suited for the study of how racial categories, boundaries, and
identities affect interracial relations for three reasons. First, white-on-black
lynchings stand as one of the most complex and disturbing aspects of the
long and rich history of racial violence in the United States. Second, in the
late 19th and early 20th centuries the South witnessed the rise of Jim Crow,
a system of durable racial inequality of remarkable scope. As I document,
the co-occurrence of the lynching era with the rise of Jim Crow was no
mere coincidence; these developments were symbiotically intertwined. Third,
grounded in Blalock’s ð1967Þ theory of intergroup relations, the existing
sociological literature on lynching emphasizes mainly its role as an instru-
mental means for social control furthering whites’ economic interests ðCor-
zine, Huff-Corzine, and Creech 1988; Beck and Tolnay 1990; Olzak 1992;
Soule 1992; Tolnay and Beck 1995; Stovel 2001; Gullickson 2010Þ. Thus

versity Faculty Fellowship. I thank Peter Bearman, Janet Box-Steffensmeier, Christine

Fountain, Marissa King, Daniel Navon, Andrew Ritchey, Matthew Salganik, Matthew
Weinshenker, the AJS reviewers, and participants in the Comparative Research Work-
shop, Yale University, for helpful comments and suggestions on earlier versions of the
article. I am especially indebted to Emily Erikson for her generous advice on and support
of this research. Direct correspondence to Mattias Smångs, Department of Sociology and
Anthropology, Fordham University, Lowenstein 916, 113 West 60th Street, New York,
New York 10023. E-mail: msmangs@fordham.edu

1330

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

the processes through which racial group boundaries and identities were
culturally encoded and socially enacted in this period have been given in-
sufficient attention. Given that the turn-of-the-20th-century South was in-
tensely preoccupied with the so-called race question, that is, the question of
the nature of white-black relations, this arguably neglects something impor-
tant about the phenomenon we are attempting to describe and explain.

Lynching as an Institutional and Differentiated Phenomenon

This research is not alone in considering the role of lynching in fomenting
the Jim Crow racial cast line. Inverarity ð1976Þ approaches lynching as a
form of “repressive justice” triggered by the “boundary crisis” in white society
caused by Populism in the 1890s. Inverarity’s study has, however, been se-
verely and successfully criticized on theoretical as well as methodological
grounds ðsee Bagozzi 1977; Berk 1977; Bohrnstedt 1977; Pope and Ragin
1977; Wasserman 1977; cf. Bailey and Snedker 2011Þ. Tolnay and Beck ð1995Þ
recognize lynching as a mechanism for maintaining the color line and ce-
menting white solidarity, but they do not develop this line of argument the-
oretically or empirically as fully as they do the role of economic forces in un-
derstanding lynching.
In a recent study, Bailey and Snedker ð2011Þ apply a “moral solidarity”
framework to explore the connection between religion and lynching. One
of their key findings is that the higher the local-level heterogeneity among
white Protestant denominations, taken as signaling the erosion of white
group solidarity, the higher the lynching rate. Bailey and Snedker argue that
this demonstrates that mob violence served to uphold white racial solidarity.
While the present study confirms Bailey and Snedker’s general argument,
it throws doubt on their assumption that lynchings were an undifferentiated
phenomenon. Bailey and Snedker’s argument that religious diversity was
conducive to collective violence that enhanced white group solidarity hinges
on the positive association between religious diversity and mob violence
holding for highly public and ritualized events. However, it is undermined
in case the association applied to mob killings carried out by a smaller number
of whites. In that case, religious diversity may be seen as reflecting anomic
conditions of moral fragmentation inducing wayward white individuals to
kill blacks. Regardless of this uncertainty over what their results actually
show, Bailey and Snedker suggest important directions for moving the lit-
erature forward from its focus on economic conditions in explaining lynch-
ings. The present study may therefore be regarded as an extension and elab-
oration of their approach.
While lynching is perhaps most strongly associated in the popular as well
as scholarly imagination with large-crowd public events coupled with ex-

1331

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

treme violence, not all lynchings featured broad-based participation, sup-

port, or overt brutality. Contrast, for instance, the following two examples.2
Hose lynching.—On April 23, 1899, in front of about two thousand peo-
ple in Coweta County, Georgia, Sam ½Hose, a black man alleged to have
killed his white employer and raped the employer’s wife, was stripped of
his clothes and chained to a tree with kerosene-soaked wood stacked high
around him. Before Hose was burned at the stake, his ears, fingers, and gen-
itals were cut off and his face skinned. On the trunk of a nearby tree someone
hung a sign reading “We Must Protect Our Southern Women.”
Welly lynching.—On the night of September 3, 1900, in Thomas County,
Georgia, a black man named Grant Welly got into a dispute with a white
man who cut Welly badly in the neck. A friend of Welly’s, Joe Fleming,
took him to a physician and then brought him to his house and put him to
bed. Later that same night a group of five or six white men came to and
fired into Fleming’s house, killing Welly and wounding Fleming. Fleming
tried to hide under the house but was pulled out and told by the white
men that if he left the house or told anyone about what had happened, they
would kill him.
In past research these two events have been treated as part of a unitary
phenomenon, lynching. As I show, this approach lacks the complexity to
account for qualitative differences among repressive practices and is thus
insufficient for disentangling the complex motives that drove white lynch
mobs and for explaining the stark contrast between these two events. The
apparent violent excesses in the Hose lynching, far beyond necessary for
causing the victim’s death, suggest that some instances of lynching may
have revolved around concerns other than an instrumental move toward
material ends. “To kill an economic competitor or make an example of a
recalcitrant worker was one thing; to mutilate him . . . appears,” as Holt
ð1995, p. 5Þ observes, “to be something else altogether.” As I shall show,
linking lynching to group boundary and identity processes connected to the
rise of Jim Crow allows us to account for this “something else” and sys-
tematically explain the differences in these two types of intergroup violence.
Additionally I revisit and reconceptualize the relevance of another insti-
tutional arena and type of organization in explaining white mob violence
against blacks in the post-Reconstruction South—politics and political par-
ties. Previous work has explored to what extent lynching was driven by
competitive electoral party politics in the South but has not been able to
establish a firm or consistent link between lynching and factors such as the
strength of the Populist and Republican parties ðBeck, Massey, and Tolnay
1989; Tolnay, Beck, and Massey 1989; Soule 1992; Tolnay and Beck 1995Þ.

2
The descriptions of the Hose and Welly killings are based on accounts in the Atlanta
Constitution on April 24, 1899 ðp. 1–2Þ and Sept. 4, 1900 ðp. 3Þ, respectively.

1332

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

Rather than conceiving the connection between politics and lynching solely
in terms of electoral competition, this research integrates the boundary-
making literature with research on the influence of political, as well as other
types of, institutions and organizations on the alignment and mobilization of
groups in contention to develop and support the argument that the Demo-
cratic Party was crucial in disseminating the symbolic racial boundaries, as
well as in building the in-group solidarity, that motivated and enabled whites
to come together in collective violence against blacks. In this, the research
continues and contributes to a long-standing line of research addressing ques-
tions of how political institutions mediate the relations between and among
groups and affect the concentrations of coercive power that give rise to con-
tentious collective action, including now-canonical studies by Gamson ð1975Þ,
Tilly ð1978Þ, and McAdam ð1982Þ, as well as more recent ones by Redding
ð2003Þ, McVeigh ð2009Þ, and Cunningham ð2013Þ.

THEORETICAL FRAMEWORK
Social groups are founded on symbolic boundaries with two dimensions:
one categorical, separating the in-group from out-groups, and one norma-
tive, prescribing and proscribing appropriate intragroup as well as inter-
group relations and practices ðLamont and Molnar 2002; Wimmer 2013Þ.
Symbolic boundaries are articulated, among other things, through public or
collective identity narratives revolving around questions such as Who are
we? Who are they? What are our rights? What are their rights? What are
our obligations? What are their obligations? What is our fate? What is their
fate? ðSee, e.g., Somers 1994; Tilly 2005.Þ Grounded in symbols, metaphors,
and images that embody the shared meanings people hold regarding what
it means to be who they and others are, such narratives are not abstract
dogma but ideologies connected with people’s expectations and experiences
of group life. They exemplify in that regard the kind of cultural knowledge
and resources—or “tools” ðSwidler 1986Þ—that actors employ in interpret-
ing and acting on the social world.
Symbolic boundaries and categories hold the potential for group for-
mation to the extent that they evoke feelings of similarity, solidarity, and
joint action with those perceived as similar to oneself. Symbolic boundaries
further sustain systems of stratification and inequality insofar as they pro-
duce hierarchy and exclusion. In that regard, public narratives provide
purposes for collective action, which may entail visions of transforming
symbolic boundaries into social boundaries through the institutionaliza-
tion of perceived group differences. A social boundary prevails to the ex-
tent that the categorical and behavioral dimensions of symbolic boundaries
coincide—“when ways of seeing the world correspond to ways of acting in
the world” ðWimmer 2013, p. 9Þ—with important consequences for re-

1333

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

source distribution as well as associative patterns among groups ðLamont

and Molnar 2002Þ.
Social group boundaries may be “bright” or “blurred.” The former are
widely recognized and institutionalized and significantly influence the access
to economic, political, and social resources of as well as the social distance
between groups, whereas the latter are less recognized and institutionalized
and do not direct the access to resources or create social distance to the same
extent ðAlba 2005; Fox and Guglielmo 2012; Saperstein and Gullickson
2013Þ. Bright social boundaries may give rise to durable group inequalities
wherein unequal intergroup relations and interactions are institutionalized
such that social, economic, and political advantages stably and continually
accumulate to members of one rather than another group. Such categorical
inequality does not emerge arbitrarily but depends on extensive collective
mobilization and organization ðTilly 1998Þ.

Group Boundaries, Identities, and Violence

Analyses of how actors use symbolic categories to grasp, create, and main-
tain interactions and relationships within and between groups can be use-
fully figured through an “eventful” approach ðBrubaker 2004Þ, because events
are main markers of the social actions through which the formation and
transformation of social structures transpire ðAbrams 1982; Moore 2011Þ.
Furthermore, if cultural forms are tools in construing meaningful action in
different settings and situations, then events give us empirical access to them,
particularly revealing the symbolic and social group boundaries and iden-
tifications that matter to actors ðGeertz 1973; cf. Weber 1978Þ. In such events,
participants deploy the cultural resources provided by group ideologies, for
example, symbolic boundaries and categories, and concretize themes inter-
pretable within the emplotment of particular collective narratives in ways
accounting for “what is going on in a way that makes an evolving iden-
tity part of the explanation” ðPolletta 1998, p. 141Þ. Such events thus not only
are made possible and comprehensible by their framing within collective iden-
tity narratives but are integral in realizing the visions of such narratives and,
thereby, the production and reproduction of group boundaries, identities,
and, as it may apply, inequalities.
The perhaps most powerful and consequential events are in that regard
dramatic ones involving conflict in general and violence in particular ðCol-
lins 2004Þ, and intergroup violence may be understood as enacting symbolic
group boundaries, categories, and identities. Since the meaning of perfor-
mances turns on their dramaturgical character, we can specify ideal-typical
models of the interactional structure of intergroup violence—that is, the pat-
terns of the violent communicative transaction relating perpetrators, victims,
and their respective peer groups—according to what message is sent, by

1334

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

whom it is sent, on whose behalf it is sent, to whom it is sent, and what is

socially constructed in the process. Figure 1 displays models of what I call
collective identity-building intergroup violence, social identity-building inter-
group violence, and intergroup violence as social control.
In the model of collective identity-building intergroup violence, the per-
petrators and the intended audience converge, involving the in-group as
performer as well as spectator, whereas these roles are separated or absent
in the other two models. This highlights that this type of violence is a per-
formance that the group stages for itself on the basis of its identity narra-
tives. Its effectiveness in generating shared social experiences and under-
standings depends on drawing participants into a strong mutual focus
of attention and high levels of emotional energy, commonly achieved by
various ceremonial elements, for example, defilement and desecration of

F IG . 1.—Ideal-typical models of intergroup violence

1335

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

bodies— torture, mutilation, burning, decapitation, and flogging ðCollins

1974, 2004Þ. Collective identity-building violence is enacted to uphold the
integrity and sovereignty of the group as a whole by dramatizing symbolic
boundaries and the extent of the group’s mutual loyalties. Such violence
defines the relationship among and between perpetrators ðincluding ad-
herentsÞ and victims in terms of, on the one side, similarity, solidarity, and
power, and, on the other, otherness, exclusion, and powerlessness. Thus, it
may be not only expressive but also generative of the very social bound-
aries and collective identities that it invokes.
As models for sociocultural understanding, experience, and action, iden-
tity narratives affect the formation of social identities—conceptions of self
that derive from membership in particular categories or groups ðBurke and
Stets 2009Þ. Since group inclusion is contingent on acceptance by relevant
others, social identity-building violence is a performance for peers to gain
their approval and the group membership implied thereby. This type of vio-
lence is thus not perpetrated to uphold the sanctity of the group itself, but on
behalf of would-be individual group members. It is therefore less dependent
on generating high levels of collective mutual focus or emotional energy and
less likely to involve broad-based participation or ceremonial brutality than
collective identity-building intergroup violence ðcf. Collins 2004Þ.
Abstracting away cultural meanings related to group boundaries, cate-
gories, and identities, the ideal-typical model of intergroup violence as so-
cial control represents violence as an instrumental means for immobilizing
adversaries—for example, by generating fear among them—in order to en-
force their submission and compliance. Its interactional structure, therefore,
in contrast to the previous two, does not feature the perpetrators’ in-group
at all but only the victims and their peer group. In emphasizing instrumental
concerns, this model remains agnostic on the character of violence and ef-
fectively views differences therein as a matter of degree rather than kind.
The Hose and Welly killings described above would thus be considered qual-
itatively similar but quantitatively different: the former sending African-
Americans a stronger message than the latter of their vulnerability and the
price for crossing whites’ interests. As noted earlier, the social control con-
ceptualization of lynching dominates the extant sociological literature. Below
I discuss how lynchings may be conceptualized in terms of collective identity-
and social identity-building intergroup violence.
Lynching as social identity- and collective identity-building intergroup
violence.—In order to gain conceptual purchase on the complexity of the
lynching phenomenon, I distinguish between “private” and “public” lynch-
ings, the former representing social identity-building and the latter collective
identity-building intergroup violence. This distinction draws on the lynch
mob typology developed by Brundage ð1993Þ consisting of private, terror-
ist, posse, and mass mobs, with “private” lynchings making up the former

1336

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

two and “public” the latter two types. The reason for collapsing the fourfold
typology into a twofold one is that significant overlap in size, composition,
and behavior makes it difficult to differentiate private from terrorist mobs
on the one hand and posses from mass mobs on the other either conceptually
or empirically ðfor similar arguments, see Brundage ½1993, Hale ½1998, and
Pfeifer ½2004Þ.
Private lynch mobs were typically small-scale furtive affairs, concerned
more with “secrecy than ceremony” ðBrundage 1993, p. 30Þ, perpetrated by
people united by bonds of kinship or friendship to pay retribution for ðal-
legedÞ harm done to self, family, or friends. The lynching of Grant Welly ex-
emplifies this type of lynching. This was not a communal collective identity-
building event: the victim was killed in a rather straightforward manner,
apparently in order to settle the score once and for all on his earlier altercation
with a white man. While mobs of this type claimed from time to time to defend
or “regulate” community morals by responding to blacks’ ðpurportedÞ viola-
tions of standards of proper interracial conduct, they, like mobs openly moti-
vated by personal grievances, were commonly disguised and carried out their
deeds at night or in secluded places; they comprised family and friends of an
offended white party, and they were often the culmination of interpersonal
conflicts. Such vigilante groups, called “terrorist” mobs in Brundage’s typol-
ogy, thus shared important similarities with mobs admittedly meting out
personal vengeance ðAyers 1984; Brundage 1993; Pfeifer 2004Þ.
Public lynchings represented communal efforts and signify collective
identity-building intergroup violence. While the legitimacy of private mobs
could be publicly questioned, public mobs, marshaling broad-based par-
ticipation and support and often led by the community’s “best citizens,” were
hardly ever challenged. As Wood ð2009, p. 43Þ observes, this attests to their
power to collectively unite whites: “Making a lynching public and spectac-
ular rendered it more legitimate than an act of vigilante violence performed
secretly.” The “spectacular” character of public lynchings refers to their oc-
casional ceremonial elements like the ones displayed in the conspicuously
excessive brutality of the Hose lynching. These elements worked to generate
a strong mutual focus of attention and shared emotional arousal between
the lynchers and their adherent audience, bonding them inside the event in
the way implied by the convergence of perpetrators and spectators in the
interactional model of collective identity-building intergroup violence. Fur-
thermore, groups of white men ð“posses”Þ that in their search for suspected
black criminals ended up killing rather than apprehending the suspects were
similarly a highly emotionally charged form of interracial violence enjoying
wide communal participation and approval among whites. Although lack-
ing ceremonial elements, these killings resembled “ritual warhunts” ðCollins
1974Þ that brought perpetrators together in prekilling frenzies and the fel-
lowship of the hunt, as well as fusing them with their communities in col-

1337

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

lective postkilling celebrations that glorified their civic heroism and con-
tribution to the greater good ðBrundage 1993Þ.
Note that the separation of perpetratorðsÞ, audienceðsÞ, and victimðsÞ
and their stipulated relations in each ideal-typical model of intergroup vio-
lence refers to analytical elements: real-world violence may very well in-
volve multiple perpetrators, multiple audiences, multiple victims, and not
only unilateral violence. The matter in the present case is thus not whether,
say, public lynchings were about imposing control on and suppressing blacks
or expressing a collective identity among whites. It is safe to assume that
public lynchings sent one message of fear and vulnerability to African-
Americans and a different message of solidarity and empowerment to white
perpetrators and sympathetic spectators ðcf. Blee 2005Þ. Further, while the
ideal-types of collective and social identity-building intergroup violence of-
fer conceptual traction to capture, compare, and contrast lynch mobs, typ-
ification is not explanation. Ideal-types define what is to be explained,
whereas explanation requires accounting for why and providing evidence
demonstrating that the phenomena occur to lesser or greater extent in one
form or the other in certain contexts rather than others.

Organizations and Collective Mobilization

One such relevant context is the collective processes taking place within
and through formal organizations. A large literature on social movements
shows that organizations are essential vehicles for mobilizing people in col-
lective action ðe.g., McCarthy and Zald 1977; Tilly 1978; McAdam 1982;
Gould 1995; Redding 2003; McVeigh 2009; Cunningham 2013Þ. But what
makes organizations such powerful mobilizing contexts? One, collective
action does not arise automatically from the aggregation of individual per-
ceptions or interests, but emerges from intersubjective perceptions among
people of a shared commonality and purpose relative to others, that is, a
collective identity. And organizational actors are important for conveying
the symbolic boundaries and categories of public identity narratives, as well
as for encoding, or “framing,” them in ways that transform and align indi-
vidual perceptions and interests with a collective identity that motivate
people to act in the name of and for the sake of the collectivity ðSnow and
Benford 1988; Brubaker 2004; McVeigh 2009; Cunningham 2013Þ. Two,
the relevance of identity narratives and the categories and boundaries they
invoke in their calls for collective unity and action are contingent on people
being able to consider their own situation as homologous to that of others
and thereby see themselves and others as being part of a group. As associ-
ational settings that bring people together in repeated interaction, organi-
zations foster interpersonal ties that let people share experiences, beliefs,

1338

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

and values with others beyond what ties established through the routines
of everyday life alone can sustain, making identity narratives more salient
and convincing to their intended constituencies and potential participants
in collective action ðGould 1995; Redding 2003; McVeigh 2009; Cunning-
ham 2013Þ.
In these respects, organizations play important roles in intergroup con-
flicts. In order to prevail in conflicts with other groups, a group cannot doubt
its strength and cohesion, which depends on generating and displaying group
solidarity inhering in its collective identity. But the solidarity of a group is
observable only in collective action, and unless such action is forthcoming,
there will always be some uncertainty about its solidarity. By accumulating
and coordinating resources as well as mediating frame alignment and social
tie formation processes, organizations enable the mobilization of collective
identities and actions in ways that allow people to understand and represent
themselves as a solidary group ðGould 1995, 2003; Brubaker 2004Þ. Various
types of organizations, for example, political parties, may thus be the most
consequential actors in promoting and shaping intergroup conflicts, and below
I explore how the postbellum southern racial conflict played out through the
southern party system. In particular, I explore how the Democratic Party
served, for one thing, as a conduit for the ideologically-narratively embed-
ded symbolic boundaries and categories whites used to define and distin-
guish themselves from blacks and, for another, to provide associative net-
works that enabled whites to come together against blacks in collective
violence.
During the course of intergroup conflict, what social movement scholars
call “critical events” may affect the dynamics of collective action ðStaggen-
borg 1993; Gould 1995Þ. Although critical events can take many forms,
policy outcomes are of particular interest here because governments are
powerful actors that serve as arbitrators recognizing and validating some
but not other collective actors, for example, organizations, in the polity. Pol-
icy outcomes result from contentious processes in which various contenders
in the polity vie for primacy in defining the “criteria for acceptable political
organization, membership, identity, activity, and claim making” ðMcAdam,
Tarrow, and Tilly 2001, p. 146Þ. By accentuating certain group boundaries,
categories, and identities, the adoption of policies that affirm the beliefs,
claims, and practices of organizational polity actors and their constituencies
may therefore provide occasion for further advancing collective purposes
through collective action. The enactment of public policy thus does not nec-
essarily represent the end of contentious episodes, but a stage therein dy-
namically affecting the conditions of contention as it unfolds ðMcVeigh,
Welch, and Bjarnason 2003Þ. Following this line of argument, I argue and
document below that disfranchisement was a critical policy event that val-
idated the racial group categories, boundaries, and identities channeled

1339

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

through the Democratic Party in the post-Reconstruction South, with con-

sequences for collective white violence against blacks.

HISTORICAL CONTEXT
The end of Reconstruction in the late 1870s put control of southern state
governments back into the hands of southern whites. The return of “home
rule” was, however, not followed by any immediate or drastic changes in
white-black relations, including the civil rights granted African-Americans
in the wake of the Civil War by the passing of the Thirteenth, Fourteenth,
and Fifteenth Amendments. This lack of change can be explained by the
“conservative” white supremacist ideology of the “Redeemers,” that is, the
region’s Democratic Party leadership coalition of elite planters, merchants,
and industrialists that defeated Reconstruction. Conservative white suprem-
acy was a genteel form of racism that in many respects carried on antebellum
slaveholder paternalism and imagined blacks as good-natured but simple-
minded and dependent on the benevolence and forbearance of intellectually
and morally superior whites. While depicting blacks as inferior to whites,
this ideology thus did not posit any natural interracial antipathies but con-
sidered blacks integral to southern society. Taking the dependent nature of
blacks for granted, it was assumed by southern elite and nonelite whites
alike in the late 1870s and early 1880s that left without protection or re-
sources from the federal government, the black masses were helpless and
without much difficulty would soon be resubordinated ðFredrickson 1971;
Williamson 1984Þ.
Public concern about the race question consequently remained for a time
relatively low but soared as the ambiguity of conservative white suprem-
acy became all but apparent. In short, while slavery lay destroyed in the
past, centuries of racialized bondage had left a powerful legacy by endur-
ingly connecting, in the mind of whites, whiteness with superiority and
privilege on the one hand and blackness with inferiority and subjugation
on the other ðDesmond and Emirbayer 2009Þ. But without social bound-
aries institutionalizing the symbolic categories of white superiority as ðwhiteÞ
freedom and ðblackÞ bondage had done during slavery, paternalist white
supremacy for a number of reasons could not meaningfully set the lot of
whites apart from or above blacks in the early post-Reconstruction years. For
one, the right to consider oneself an independent and honorable citizen, with,
in the case of men, the right to vote, was no longer an exclusive white priv-
ilege. For another, blacks were, naturally, less than willing to surrender to
white authority, and the contrast between blacks asserting their autonomy
from notions of their dependency and docility was galling to whites. More-
over, developments in the southern agricultural economy—increasing land-
lessness, plummeting cotton prices, an appreciating dollar, and grinding crop

1340

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

lien debt—exacerbated many whites’ precarious position in society in gen-

eral and relative to blacks in particular. To find themselves competing with
blacks at the bottom rung of society for farm tenancy and laborer opportu-
nities without a clear edge not only meant material deprivation but also put
serious strain on whites’ efforts to uphold a superior social standing above
blacks. Finally, the rising middle-class blacks taking advantage of the eco-
nomic, educational, and political opportunities available during Recon-
struction and the early years thereafter proved false the notions of black
intellectual-moral inferiority and were a powerful source of aggravation
among plain whites ðAyers 1992; Litwack 1998; Dailey 2000Þ.
Thus, although whites had regained control of southern state govern-
ments, interracial interactions and relations in the early post-Reconstruction
years were not as scripted or constrained as they had been in the slavery past
or would become in the imminent Jim Crow future. In these circumstances,
where the symbolic racial boundaries and categories of paternalist white
supremacy did not map onto whites’ expectations or experiences and where
interracial social boundaries where blurred, the meaning and worth—in
other words, the economic, political, social, and psychological “wages” ðDu
Bois 1935Þ—of whiteness remained uncertain and elusive ðAyers 1992;
Dailey 2000Þ.
The widespread dissatisfaction among whites with the absence of sym-
bolic and social boundaries clearly defining and firmly realizing their racial
superiority and privilege toward the second half of the 1880s found ex-
pression in radical ðor extremist or militantÞ white supremacy. Informed
by social Darwinism and scientific racism, the core assumption of extremist
white supremacy was that blacks were innately and hereditarily inferior to
whites and that, without the civilizing influence of slavery, they were in a
process of evolutionary “retrogression.” In viewing blacks as set apart from
and beneath humanity, it rejected notions of blacks as useful members of
society and instead painted a picture of an apocalyptic Manichean racial
struggle that portrayed blacks as imminently endangering whites and their
communities. The notion of black racial degeneration, most powerfully cap-
tured in the image of the “black beast rapist,” was in that respect particularly
consequential. As a result of their supposed degeneration, blacks were rep-
resented as increasingly driven by primordial animalistic instincts toward
sexual gratification, leading to uncontrollable passions—purportedly result-
ing in an epidemic of sexual assaults on white women by black men. This
attribution of collective threat to African-Americans in the myth of the black
rapist invested the interracial sexual boundary with a previously unknown
salience because neither the strong taboo on interracial sexual contacts nor
acute white concern with black-on-white rape had characterized antebellum
times, but emerged as fundamental to white group ideology in the late
19th century ðe.g., Williamson 1984; Sommerville 2004Þ.

1341

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

This symbolic alignment of racial and sexual boundaries in the radical

racist narrative ðfrom the vantage point of whitesÞ, then, had homogenizing
effects on both sides of the color line. First, it indiscriminately cast all blacks,
with little consideration for factors such as education, occupation, or income,
into the category of primal fiends. Second, resonating with traditional cul-
tural values attached to the family and the glorification of white woman-
hood, it fostered a sense of commonality and solidarity, that is to say, a
collective identity, among whites transcending class lines by placing them all
in the same position of being besieged by perceived predatory black men
lurking for their mothers, sisters, wives, and daughters. The rhetoric of
radical white supremacy was accordingly filled with metaphors of whites as
forming an extended family and white women as the keepers of the com-
munal home and hearth. In this, extremist white supremacy bestowed calls
for white unity, superiority, and domination with a credibility, focus, and
reason lacking in conservative racial paternalism ðLitwack 1998; Kan-
trowitz 2000; Downs 2011Þ.
With the Democrats well established as the “white man’s party,” the
radical white supremacists, unlike the Populist movement, did not create a
new party but appropriated, in sharp contention with the incumbent Re-
deemer conservatives, the Democratic Party as their primary organizational
base and vehicle for engaging collective white racial commitments and iden-
tifications. The convergence of racial and partisan communities invested the
act of voting with meaning beyond electing public officials or influencing
public policies and turned elections into communal rituals generating and
affirming racial group commitments and solidarities. In these circumstances,
intensive in-group partisanship policing and sanctioning were important
means of achieving collective racial identity and cohesion among whites
ðcf. Brubaker 2004Þ. Whites voting against the Democratic ticket were seen
as racial traitors and risked ostracism within their communities ðKantrowitz
2000; Redding 2003; Downs 2011Þ.
An important associative setting for communal engagement, the Dem-
ocratic Party not only disseminated a racial narrative rooted in a particular
white supremacist ideology but also provided contact opportunities between
southern whites, bringing and binding together whites of different socio-
economic standings and forging race- rather than class-based collective in-
terests, identities, and solidarities. The rhetoric of extremist white suprem-
acy, furthermore, converged seamlessly with that of the Lost Cause. The
sine qua non of extremist white supremacy as well as the Lost Cause was
the solidarity of all southern whites, and their clear affinities were no coin-
cidence: they both grew out of the social dislocations of the late 1880s and
early 1890s, and the rise of one provided fertile ground for the other. The
Democratic Party so became “a political church with sacred commitments

1342

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

to white supremacy . . . and holy symbolism spiritualizing the Lost Cause”

ðBartley 1990, p. 97; see also Gaines 1987Þ.
As their ideology eclipsed conservative white supremacy and they wrestled
control of the Democratic Party from the Redeemers in the late 1880s and
early 1890s, the racial extremists led the South down a path of rigid, exclusion-
ary, and oppressive racial relations leading to the system of Jim Crow—“the
highest stage of white supremacy” ðCell 1982Þ—including its pillar institu-
tions of disfranchisement and legal segregation. If this outcome seems in-
evitable from our vantage point, it was anything but from the perspective of
contemporaries, because it was one thing to rhetorically posit white suprem-
acy and a very different thing to transform it from slogan into social fact.
Thus, while whites’ racist impulses may have arisen unbidden, in themselves
they did not provide the basis for the type of collective identity and action
necessary for turning the visions of radical white supremacy into reality. As
the architects and advocates of radical white supremacy recognized, trans-
forming its symbolic boundaries into institutionalized social boundaries of
racial domination required that they enjoy widespread recognition and le-
gitimacy, which could be achieved only through considerable agitation and
mobilization uniting whites around a shared understanding of the “true”
nature and stakes of the race question. That the rise of radical white su-
premacy, the move to disfranchise blacks, the onset of legal segregation, and
an increase in both frequency and barbarity of racial mob violence against
blacks co-occurred in the decades straddling the turn of the 20th century was
no coincidence but reflected different aspects of the same process of realiz-
ing the radical white supremacy imperative of reestablishing the connection
between race and domination in all aspects of southern society that slavery
had provided but emancipation disrupted. This involved transforming the
symbolic racial boundaries inherent in radical white supremacy into bright
social boundaries as well as mobilizing the collective white identity envi-
sioned by extremist white supremacists in collective action against blacks.

EMPIRICAL IMPLICATIONS
Event-Level Hypotheses
To the extent that lethal mob violence played into the activation and main-
tenance of the symbolic and social racial group boundaries at the foundation
of the Jim Crow system, we should expect systematic associations between
the situational triggers of lynchings and their violent form—not in every
case, of course, but consistently enough to reveal meaningful patterns. As
per our theoretical framework, perceived black transgressions threatening
the integrity and sovereignty of whites as a group rather than injuring par-

1343

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

ticular whites should have been followed by collective identity-building vi-

olence in the form of public lynchings. In particular, lynchings following on
ðreal or imaginedÞ black sexual assaults ðor attempts thereofÞ against white
women should more commonly have been public rather than private lynch-
ings. This prediction, however, needs further elaboration. As the fearsome
“black rapist” moniker did not become a crucial symbolic racial boundary
marker until the rise of radical white supremacy in the years around 1890, the
hypothesized association between black-on-white sexual assaults and lynch-
ing should hold to a greater extent for the period after than before then. We
should, furthermore, not expect to observe this pattern in white-on-white
lynchings, because their significance should not have lain in enacting and
constructing symbolic or social racial boundaries. This leads to the following
hypotheses.
HYPOTHESIS 1.—Lynchings of blacks triggered by alleged sexual assaults
on white women were more commonly public lynchings in the period 1890–
1915 compared to the period 1882–89.
HYPOTHESIS 2 .—In the period 1890–1915, lynchings of whites triggered
by alleged sexual assaults on white women were not predominantly public
lynchings.
In addition to these main event-level hypotheses, two subsidiary hypoth-
eses follow from present theoretical and historical considerations. At all
times whites perceived lethal or potentially lethal black-on-white attacks as
seriously challenging their communal integrity and supremacy over blacks,
whereas, conversely, they did not consider nonsexual and nonmurderous
incidents, for example, black violations of informal interracial etiquette rules
or property offenses, to carry implications for their communities as a whole
but rather as affronts to particular whites ðBrundage 1993Þ. My third and
fourth hypotheses are thus as follows:
HYPOTHESIS 3 .—Lynchings triggered by alleged lethal or potentially lethal
black attacks on whites were more commonly public than private lynchings.
HYPOTHESIS 4 .—Lynchings triggered by alleged nonsexual and non-
murderous black transgressions against whites were more commonly private
than public lynchings.

Community-Level Hypotheses
As Bailey and Snedker ð2011Þ do not argue that southern Protestant
churches as such orchestrated mob violence against blacks, I do not argue
that the Democratic Party as such carried out lynchings. The focus here is
to what extent and how white collective identity and solidarity mediated
by the Democratic Party shaped the landscape of interracial relations and
thus were conducive and responsive to the prevalence and form of white
lethal mob violence against blacks. Hypotheses about these matters should

1344

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

take into account that during the years 1890–1915, intergroup relations be-
tween southern whites and blacks moved toward the reimposition of insti-
tutionalized black subordination. Among the many aspects of the Jim Crow
order emerging in this period, I will attend here particularly to the disfran-
chisement of blacks for two reasons. First, suffrage restrictions were con-
tentious events that took place within a relatively short period of time with
clear beginnings and ends in each southern state, making it apparent when
they were happening and when they were concluded. Second, the franchise
was the primary marker of citizenship and polity membership at large. The
significance of suffrage thus extended beyond electoral politics to collec-
tively endow blacks with a claim to power and equality with whites. In the
minds of militant white supremacists, true white unity and supremacy could
therefore occur only once the taint of blacks had been eliminated from the
southern body politic. Disfranchisement therefore was not only about elec-
toral politics but was a critical event affecting interracial group relations at
large. We may therefore expect that the association between white collec-
tive identifications forged through the Democratic Party and lynchings in the
pre-disfranchisement period was different than in the post-disfranchisement
period.
What local group-level dynamics should, then, have led whites to enact
their collective identity and solidarity in violent public display in the pre-
disfranchisement period? Given the role of collective identifications and
solidarities in intergroup conflict and given that they are unambiguously
established only in observable action, white communities characterized by
fragile or ambiguous cohesion should have been more likely than ones
characterized by high levels of cohesion to stage collective violent perfor-
mances in order to achieve a sense of unity and empowerment. Thus, before
disfranchisement, conditions of low and intermediate collective white racial
identity should have increased the likelihood of public lynching.3 To the
extent that political organization produced collective identity and solidarity
among southern whites, the relative strength of the Democratic Party should
have influenced the incidence of collective identity-building violence in the
form of public lynching; we can formulate the following hypothesis:
HYPOTHESIS 5 .—The incidence of public lynchings before the implemen-
tation of disfranchisement was higher in localities with lower and inter-
mediate levels of Democratic Party support than in localities with high
levels of Democratic Party support.
If disfranchisement and lynchings are viewed as part of the competitive
electoral party politics of the post-Reconstruction South, then we would ex-
pect the implementation of the former to have reduced the latter because

3
The operationalization of “low,” “intermediate,” and “high” levels of Democratic Party
support is given below.

1345

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

once blacks were politically neutralized, racial mob violence should have
become an unnecessary form of repression ðcf. Beck et al. 1989; Tolnay and
Beck 1995Þ. However, a broader perspective of post-Reconstruction politics
as revolving around the alignment of racial groups leads to different ex-
pectations about the association between disfranchisement and lynching.
The reason is that disfranchisement served not only to eliminate blacks’
influence on southern politics but also to increase the salience of collective
white racial identifications by, for one, legitimizing the beliefs, claims, and
practices of extremist white supremacists and, for another, muting divisions
among whites that could have jeopardized their unity and, in turn, their
domination over blacks ðWoodward 1951; Kousser 1974; Perman 2001;
Redding 2003Þ. To the extent that disfranchisement represented a critical
event boosting race-based collective identification and mobilization among
whites, the following hypothesis should hold true:
HYPOTHESIS 6 .—The incidence of public lynching was higher in places
with than without disfranchisement.
Similarly, if the enactment of disfranchisement affirmed the collective
white beliefs, claims, and practices promoted by the Democratic Party by ex-
ercising the specter of “Negro domination” from southern life, whites would
no longer have been compelled to come together in collective identity-
building violence to dispel doubts about their cohesion. Rather, they would
have used it as a vehicle for further expressing and entrenching their sol-
idarity and power by demonstrating to themselves, as well as to blacks,
their unity and “the ferocity of ½their determination to dominate” ðPerman
2001, p. 269Þ. Accordingly, we can formulate the following hypothesis:
HYPOTHESIS 7 .—The incidence of public lynchings after the implemen-
tation of disfranchisement was higher in localities with higher levels of
Democratic Party support than in localities with lower and intermediate
levels of Democratic Party support.
With the institutionalization of white domination throughout the South,
including the implementation of disfranchisement and legal segregation
throughout the region, the race question had, for all intents and purposes,
been solved by the mid-1910s. Furthermore, with the onset of the Great
Migration, support for the most extreme practices of racial domination
began to wane among southern whites ðTolnay and Beck 1995Þ. Thus, to
the extent that public lynchings figured into the rise of Jim Crow, they
should have done so during its most formative years of 1890–1915, which
leads to the following hypothesis:
HYPOTHESIS 8 .—The incidence of public lynchings in the period 1916–30
was unaffected by the strength of the Democratic Party.4

4
I am indebted to one of the AJS reviewers for suggesting that I pursue this line of
analysis.

1346

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

Assuming that collective identity is a group-level property that does not

emerge from nor influence individual actors in automatic ways and that
disfranchisement affected intergroup rather than interpersonal white-
black relations leads to the following conjectures:
HYPOTHESIS 9 .—The incidence of private lynchings was not affected by
the implementation of disfranchisement or associated with the strength of
the Democratic Party either before or after the implementation of dis-
franchisement.
Finally, the theoretical and substantive arguments developed here charge
that public lynchings not only derived from but also forged the white col-
lective identifications that lynchers invoked in their violent practices dur-
ing the coming of Jim Crow. However, owing to their supposed social rather
than collective identity-building character, private lynchings should not
have had such group-level reverberations. Assuming that Democratic Party
support mirrored collective white identity and solidarity, these arguments
hold if we observe the following association:
HYPOTHESIS 10.—Public but not private lynchings in the period 1888–
1916 were followed by increasing support for the Democratic Party.

DATA, MEASURES, AND METHODS

Setting
This study covers the southern states Georgia and Louisiana for the period
1882–1930. They were chosen because the available data on lynchings in
these states are sufficiently detailed to allow moving beyond a unidimen-
sional lynching concept. The online, full-text ProQuest Historical News-
papers database offers easy access to contemporary newspaper reports on
lynchings in the Atlanta Constitution—the most important and widely read
newspaper in Georgia at the time. Pfeifer’s Rough Justice: Lynching and
American Society, 1874–1947 ð2004Þ lists lynchings in Louisiana according
to the same lynch mob typology on which the present coding of Georgia
lynchings at root rests.5 The geographical scope of the study is also justified
on substantive grounds. While no one state can be seen as representative of
the South as a whole, Georgia is “an obvious choice for a study of lynching:
the character and harshness of white domination . . . became the measure of
race relations in the Deep South. . . . The sheer scale of mob violence in
Georgia alone commands attention” ðBrundage 1993, pp. 15–16Þ. More-
over, as Du Bois pointed out in the early 1900s, “not only is Georgia . . . the

5
To achieve sampling consistency and data comparability across states, the few dis-
crepancies between the Tolnay-Beck inventory and Pfeifer’s listing were resolved in
favor of the former.

1347

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

geographical focus of our Negro population, but in many other respects,

both now and yesterday, the Negro problems ½i.e., the race question have
seemed to be centered in this State” ð2003, p. 112Þ. What is more, Georgia
and Louisiana spanned all the major subregions of the South at the time in
terms of demographic, economic, historical, and physiographic character-
istics ðAyers 1992Þ, which is methodologically important because as race
relations varied across subregions, a study that does not include them all is
liable to misleading results. There is thus little reason to suppose that the
geographical scope of the study drives its results.

Lynching Data
This article draws on primary and secondary data augmenting parts of the
comprehensive lynching inventory constructed by Tolnay and Beck ð1995Þ.6
First, they include 39 white-on-black lynchings and 3 white-on-white lynch-
ings in Georgia confirmed or discovered ðas far as lynching inventories goÞ
by either myself or E. M. Beck subsequent to the construction of the orig-
inal inventory. I have also disconfirmed a couple of events that in light of
newly discovered information should not be considered lynchings.7 Second,
each event is coded either as a “public” or “private” lynching. While relying
on Pfeifer’s ð2004Þ classification of Louisiana lynchings, I coded all Georgia
lynchings in the inventory on the basis of contemporary newspaper reports
in the Atlanta Constitution.8
In view of the conceptual discussion above, all mobs reported as com-
posed of more than 50 participants were coded as public lynchings.9 So
were mobs drawing participants from beyond the family and friends of
the victim of the ðreal or imaginedÞ precipitating black offense, indicated
by descriptors of the mob as composed of the “community,” “citizens,” or
“people” of a locality. Public lynchings also include incidents that involved
ceremonial and ritualistic elements such as torture, mutilation, burning,
collective shooting, or signs left near the site. Events in which the lynch

6
In addition to Georgia and Louisiana, the Tolnay-Beck inventory includes Alabama,
Arkansas, Florida, Kentucky, Mississippi, North Carolina, South Carolina, and Tennessee.
7
A lynching is defined as an extralegal killing of at least one person by a mob of at least
three people.
8
Beck and Tolnay have generously shared their original research notes with me, and
I have occasionally used them to complement the newspaper information. Lynching
events with insufficient information for classification as either public or private were
assigned to a third category of “unknown” and were not included in the analyses. This
category includes 52 ð<9%Þ of the total 598 white-on-black and 36 white-on-white lynch-
ings taking place during the time covered by the analyses in this article.
9
The lower limit of 50 participants was chosen to assure consistency with Brundage’s
ð1993Þ operationalization. Using 30 participants as a cutoff does not affect the results
reported here.

1348

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

victim was brought to and killed at the scene of his alleged crime, brought
before the victim of his alleged crime for public identification, or given the
opportunity to publicly confess his alleged crime before being killed were
likewise coded as public lynchings. Events involving white bands ðpossesÞ
pursuing and killing blacks accused of serious crimes against whites were
designated public lynchings. Events for which the newspaper information
suggests that black victims were killed by smaller white parties in or near
the victims’ homes or other nonpublic settings, for example, forests or
swamps, were coded as private lynchings. Killings of blacks by similarly
sized mobs in more public places but without indication of the kind of
ceremonial or ritualistic elements described above were considered private
lynchings as well.
After transforming the listing and classification of Georgia lynchings in
Brundage ð1993Þ and Louisiana lynchings in Pfeifer ð2004Þ into the present
twofold lynching typology, I assessed the reliability of the coding procedure
by calculating Cohen’s kappa ðkÞ measure of intercoder reliability. The
kappa for Georgia lynchings classified by both me and Brundage ðN 5 250Þ
is k 5 .89 ðP > .00Þ. On the basis of newspaper articles in the Times-Picayune
ðcollected through the database Early American Newspapers, Series IIIÞ I
coded 146 Louisiana lynchings also classified by Pfeifer, yielding a kappa
score of k 5 .84 ðP > .00Þ. These kappa levels represent in terms of con-
ventional standards almost perfect intercoder agreement ðLandis and Koch
1977Þ.

Measurements
Collective white identity.—The Democratic Party was the organiza-
tional backbone of white supremacy at the time, and the community-level
salience of collective white racial identifications is measured here in terms
of the percentage of votes cast in a county in favor of the Democratic can-
didate in presidential elections ðcf. Raper 1933; Pope and Ragin 1977;
Tolnay and Beck 1995; Hagen, Makovi, and Bearman 2013Þ.10 Values for
interelection years were calculated by linear interpolation. Table 1 pre-
sents descriptive statistics for the county-level percentage of Democratic
votes in presidential elections from 1888 to 1928. The table shows that the
Democratic Party generally enjoyed high levels of support in these elec-
tions; at no time did it receive less than an average of 60% of the votes. But
it also shows that its support varied considerably across counties in a given
election as well as across elections. The county-level correlation between
Democratic returns in a given election and the immediately preceding elec-
10
This measure draws on information in the data set “Electoral Data for Counties in the
United States: Presidential and Congressional Races, 1840–1972” ðClubb, Flanigan, and
Zingale 2006Þ.

1349

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

TABLE 1
Descriptive Statistics for County-Level Democratic Vote
Returns in Presidential Elections, 1888–1928

Election Year Mean SD Correlationa

1888 . . . . . . 74.54 18.30 .66
1892 . . . . . . 63.13 19.50 .30
1896 . . . . . . 64.22 19.67 .57
1900 . . . . . . 71.70 16.62 .66
1904 . . . . . . 71.75 18.81 .81
1908 . . . . . . 63.10 20.11 .92
1912 . . . . . . 77.74 14.42 .46
1916 . . . . . . 82.82 14.18 .75
1920 . . . . . . 74.48 18.70 .82
1924 . . . . . . 78.20 13.31 .74
1928 . . . . . . 66.82 17.75 .48
a
Reports correlation with preceding election.

tion particularly illustrates the local-level volatility in Democratic support

in the South in this period. Thus, although fairly stable at high levels in the
aggregate, the Democratic vote displays sufficiently broad local-level varia-
tion within as well as across elections to be a sociologically meaningful in-
dicator as well as to avert concerns regarding adequate model parameter
identification.
As discussed above, disfranchisement was meant not only to eliminate
African-Americans from the polls but also to promote white group identity
and solidarity. In order to assess whether its introduction affected the dy-
namics between collective white racial identity, as measured by the Dem-
ocratic Party vote share, and lynching, I apply a strategy for specifying
statistical models with interaction terms with a categorical modifying var-
iable proposed by Brambor, Clark, and Goldner ð2006Þ. The modifying vari-
able of interest here is disfranchisement, which was implemented in Loui-
siana in 1898 and in Georgia in 1908,11 and assuming a one-year delay in its
potential effect ðTolnay and Beck 1995Þ, I created pre-disfranchisement
and post-disfranchisement period dummy variables for each state. The pre-
disfranchisement dummy takes the value one in 1890–1908 and zero in
1909–15 for Georgia observations and the value one in 1890–98 and zero in
1899–1915 for Louisiana observations. The post-disfranchisement dummy
conversely takes the value zero in 1890–1908 and one in 1909–15 for Georgia

11
Determining the year of disfranchisement in Louisiana is straightforward because the
state passed a poll tax, a literacy test, and a property test in the same year. Determining
the year of disfranchisement in Georgia is somewhat less clear-cut because it introduced
a poll tax already in 1877, whereas a literacy test, a property test, an understanding
clause, and a grandfather clause were introduced in 1908. Following previous work
ðBeck et al. 1989; Tolnay and Beck 1995; cf. Kousser 1974Þ, I find it most reasonable to
consider 1908 the year of disfranchisement in Georgia.

1350

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

observations and the value zero in 1890–98 and one in 1899–1915 for Loui-
siana observations. I thereafter generated interaction terms by multiplying
the pre- and post-disfranchisement dummies with the linear and quadratic
collective white identity measures.
Two things should be noted. First, a model including these four inter-
action terms along with the post-disfranchisement dummy is analogous to
a model including the linear and quadratic measures of white racial iden-
tity, their ðin total twoÞ interaction terms with the post-disfranchisement
dummy, and the post-disfranchisement dummy. The advantage of the present
model specification is that directly estimating linear and quadratic coefficients
for collective white racial identity in the pre- as well as post-disfranchisement
period simplifies the calculation of marginal effects of collective white racial
identity on lynchings in each period. Second, as voting laws were imple-
mented at the state level, the pre- and post-disfranchisement dummy vari-
ables do not vary across counties within the same state. There is consequently
no variation in these variables before Louisiana implemented disfranchise-
ment in 1898 or after Georgia did so in 1908; their only variation is found in
the period 1899–1908. The inclusion of the disfranchisement dummy vari-
able in model estimations thus does not capture temporal but spatial lynch-
ing patterns, that is, whether lynching was more or less likely in places with
than without voting restrictions.
Control variables.—The multivariate analyses ðdescribed in more detail
belowÞ include a number of variables controlling for economic and demo-
graphic conditions that may have influenced white mob violence against
blacks: linear and quadratic terms of black population concentration ðin
percentageÞ, the natural logarithm of absolute black population size ðin
1,000sÞ, average farm size ðin acresÞ, percentage of improved farmland de-
voted to cotton production, and percentage of farm ownership ðvalues in
intercensal years for all continuous control variables were calculated by
linear interpolationÞ.12 The analyses also include the cumulative number
of previous lynchings of opposite type to gauge whether different forms
of interracial violence affected one another. Table 2 presents descriptive
statistics for all control variables in 1890–1915 and 1916–30.

Statistical Methods
Event-history analysis.—The first set of analyses uses simple contin-
gency tables to assess event-level associations between precipitating situ-
ational triggers of lynchings and the character of the ensuing mob violence.
The second set employs Cox event-history models to ascertain the associ-

These variables are based on information in Haines ð2006Þ and U.S. Bureau of the
12

Census ð1913, 1922Þ.

1351

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

TABLE 2
Descriptive Statistics for Control Variables in 1890–1915 and
1916–30

Mean SD
1890–1915:
% black . . . . . . . . . . . . . . . . . . . 45.68 21.80
Ln black population ð1,000sÞ . . . 1.50 1.20
Average farm size . . . . . . . . . . . 130.27 77.72
% farms owned . . . . . . . . . . . . . 44.14 19.83
% farmland devoted to cotton . . . 29.19 15.87
No. prior private lynchings . . . . . .74 1.08
No. prior public lynchings . . . . . .81 1.08
1916–30:
% black . . . . . . . . . . . . . . . . . . . 40.70 20.10
Ln black population ð1,000sÞ . . . 1.62 1.24
Average farm size . . . . . . . . . . . 96.38 51.04
% farms owned . . . . . . . . . . . . . 40.06 19.07
% farmland devoted to cotton . . . 26.52 15.60
No. prior private lynchings . . . . . 1.24 1.41

ation between the local permeability of white racial collective identity and
different lynching types. On the basis of partial likelihood methods, Cox
models require no assumptions regarding the underlying distribution of time
durations until event occurrence. By leaving the particular distributional
form of duration times unspecified, Cox models impose fewer restrictions
on model specification and estimation and are therefore preferable to alter-
native parametric event-history models, which require that assumptions,
which are often of secondary or no substantive importance to the relation-
ship between the predictors and outcome variables under consideration,
about the nature of duration times be made ðBox-Steffensmeier and Jones
2004Þ.
Cox models furthermore allow for model diagnostics currently unavail-
able for parametric models regarding identifying and accommodating vi-
olations of the “proportional hazards” assumption, which requires that
covariate effects remain constant relative to time. A violation of this as-
sumption means that the effect of a covariate differs between time points,
and models incorrectly fitted under conditions of nonproportionality are
liable to misspecification and biased estimates. All models in this study were
therefore tested for violations of the proportional hazards assumption and
evidence thereof remedied by estimation of a subsequent expanded model
including an interaction term between the offending covariateðsÞ and the
natural logarithm of time elapsed from the beginning of the study period
ðfurther technical details are given in Box-Steffensmeier and Zorn ½2002Þ.
Another advantage of the Cox modeling approach for present purposes
is offering a strategy for dealing with unobserved heterogeneity arising from

1352

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

counties experiencing multiple lynchings of a particular type during the study

period. The problem of unobserved heterogeneity due to event reoccurrence
is akin to autocorrelation in traditional regression analysis by overestimating
the amount of information in the data, leading to possibly biased estimates
and incorrect inferences. Cox models can adjust for this problem by strati-
fying estimations according to event sequence rank—that is, first lynching,
second lynching, and so on—and using robust variance estimation of stan-
dard errors by clustering estimations on the county units of analysis ðBox-
Steffensmeier and Zorn 2002Þ. Without taking into account the actual order
of multiple events within a county in such a fashion, model estimations, fur-
thermore, proceed as if a county could have experienced all lynchings in its
sequence at all times. That is, a model that ignores the order of lynchings
is estimated as if later events could have happened before earlier events,
whereas estimation by rank-order stratification preserves the actual event
sequence.
Cox models leverage temporal dynamics by including time-varying co-
variates whose values change during the span of the investigation. As is
conventional and as this study specifies event times in terms of dates, the
lynching event data were “split” to generate one observation per county for
each date a lynching of a particular type took place. As the values of co-
variates are measured on a yearly basis, each observation was then assigned
the covariate values of its associated county in the year the lynching oc-
curred. In other words, the analyses reported below are based on multiple
data sets—one for each lynching type and time period—with multiple ob-
servations per county containing relevant covariate information, as well
as a binary ðoutcomeÞ variable indicating whether or not it was the county
in which the lynching actually took place. Table 3 gives a sense of the data
structure and includes a portion of the data used in model 3 predicting

TABLE 3
Data Excerpt for Clay County, Georgia, 1908–9

PREVIOUS LYNCHINGS
DATE Public Private EVENT INDICATOR
2/26/1908 . . . 2 1 0
7/26/1908 . . . 2 1 0
8/9/1908 . . . . 2 1 0
8/25/1908 . . . 2 1 0
9/22/1908 . . . 2 1 1
10/11/1908 . . . 2 2 0
4/9/1909 . . . . 2 2 0
6/19/1909 . . . 2 2 0
6/24/1909 . . . 2 2 0
7/2/1909 . . . . 2 2 0
7/30/1909 . . . 2 2 0

1353

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

private lynching from 1890 to 1915. The table displays selected data for
Clay County, Georgia, for 1908 and 1909, and, comprising 11 observations,
it shows that private lynchings took place in either Georgia or Louisiana on
11 different days in that two-year period. As the event indicator as well as
the value for “previous private lynchings” for the observation of Septem-
ber 22, 1908, is one, it shows that a private lynching took place in Clay that
day and that the county had experienced one private lynching before then.
At the time of this lynching, Clay County was therefore in the second rank-
order stratum for private lynchings and after that accordingly placed in the
third stratum.
The dynamism of Cox models, furthermore, allows for incorporating
new units into the analysis as they are created. For while most counties
enter into model estimations at the beginning of each study period, some
counties were created at later points and therefore not included until their
creation. This way of handling new counties is preferable to the typical
procedure in longitudinal county-level studies of using “county cluster”
composed of all counties involved in the formation of a particular new
county. This is so because southern life during this period was highly lo-
calistic. Accordingly, there is widespread agreement that counties repre-
sent meaningful sociospatial units whereas county clusters do not and that
using geographical units larger than necessary should be avoided ðcf. Hagen
et al. 2013Þ.
Model estimations use one-year lagged rather than current-year covari-
ate values in order to avoid simultaneity problems between event occur-
rences and temporal changes in independent variables ðBox-Steffensmeier
and Jones 2004Þ. The covariate values of new counties were accordingly
calculated by initially assigning a newly created county the average value
of each covariate across its counties of origin in the year before creation.
Values in years between the year before creation and the next census were
then approximated by linear interpolation. The one exception to this pro-
cedure is absolute black population size, which was calculated by dividing
the total black population in counties of origin by the number of counties
of origin plus one. The addition is necessary as the total black population
in relevant counties after the creation of the new county otherwise would
exceed the total before its creation. In order to preserve the historicity of the
data, a new county was similarly assigned a value indicating the number of
private and public lynchings it had experienced at the time of its creation,
as it were, by averaging the total number of each type ðrounded to the next
integerÞ that had taken place up until then within the counties from which
it was created.
Dynamic panel data analysis.—In order to investigate to what extent dif-
ferent lynching types affected levels of collective white racial identity dur-
ing the rise of Jim Crow, I fit a number of dynamic panel data models using

1354

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

county-election-year observations for the period 1888–1916. The dependent

variable in these models is white racial collective identity as measured by
the Democratic vote percentage in election years, and the key predictors are
the number of public and private lynchings within certain time intervals
before elections. The reason for using dynamic rather than static panel data
models is that the argument under consideration concerns changes in more
than levels of white collective identity as a result of racial mob violence
ðcf. Koçak and Carroll 2008Þ. Furthermore, as indicated in table 1, it is not
unreasonable to assume that the Democratic vote percentage was subjected
to an autoregressive process in the sense that the outcome in one election to
a certain extent was influenced by the outcome in the previous election. In
addition to contemporaneous ðelectionÞ year values of the economic and de-
mographic control variables described earlier, a state dummy variable, and
election year dummy variables, these models accordingly include the first lag
of the dependent variable as a covariate. And to assess whether the impact
of racial mob violence varied across levels of collective white racial identity,
the models include interaction terms between the number of public and pri-
vate lynchings, respectively, and the first lag of the dependent variable.
The inclusion of lagged values of the outcome variable as a covariate for
the purpose of accounting for autoregressive processes makes ordinary
least squares estimates biased and inconsistent because lagged dependent
variable values are correlated with the error term. This correlation can be
purged by exploiting the panel structure of the data to obtain instrumental
variables. Present models employ the system generalized method of mo-
ments ðGMMÞ instrumental variable panel estimator, which is more effi-
cient than the two-stage least-squares estimator ðArellano and Bover 1995;
Blundell and Bond 1998; the advantages of the GMM estimator over other
estimators are discussed in Wawro ½2002Þ, as well as the Arellano-Bond var-
iance estimator, which is robust to cross-section heteroscedasticity ðArellano
and Bond 1991Þ. The models presented below use the following instruments:
all first-difference lags for all independent variables, all lags beyond the first
lag of the dependent variable, and all lagged differences of the dependent
variable. The systems GMM estimator requires that there is no serial cor-
relation in first-differenced errors at the second and higher orders, and di-
agnostics using the Arellano-Bond test of autocorrelation in first-differenced
errors did not indicate violations of this assumption for any model presented
here.

LYNCHING AND THE COLOR LINE

Table 4 cross-classifies lynchings according to situational precipitants and
mob types in 1882–89 and 1890–1915 and shows in line with our expec-
tations in hypotheses 3 and 4 that lethal mob violence following on mur-

1355

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

TABLE 4
Lynchings of Blacks by Type and Alleged Offense, 1882–89 and 1890–1915

1882–89 1890–1915
Private Public Total Private Public Total
ALLEGED OFFENSE ð%Þ ð%Þ ð%Þ ð%Þ ð%Þ ð%Þ
Sexual assault . . . . . . . . . . . . . . . 49 51 100 34 66 100
ð20Þ ð21Þ ð41Þ ð49Þ ð95Þ ð144Þ
Murder and attempted murder . . . 33 67 100 33 67 100
ð4Þ ð8Þ ð12Þ ð51Þ ð105Þ ð156Þ
Other . . . . . . . . . . . . . . . . . . . . . 86 14 100 76 24 100
ð18Þ ð3Þ ð21Þ ð51Þ ð16Þ ð67Þ
Total . . . . . . . . . . . . . . . . . . . . . 57 43 100 41 59 100
ð42Þ ð32Þ ð74Þ ð151Þ ð216Þ ð367Þ
Pearson x2 . . . . . . . . . . . . . . . . . 10.92** 41.46**
df . . . . . . . . . . . . . . . . . . . . . . . . 2 2
NOTE.—Numbers in parentheses are the numbers of observations.
P < .05 ðtwo-tailed testÞ.
P < .01.

ders and attempted murders was more likely public than private lynchings,
whereas the opposite holds for nonsexual and nonlethal assaults, in both
periods. More importantly, the data bear out the prediction of hypothesis 1
that patterns of lethal mob violence responding to black men ðallegedlyÞ
sexually assaulting white females in the period before 1890 diverged from
patterns in the period after.13
As displayed more clearly in table 5, lethal mob violence following on
such allegations in the latter period was significantly more likely to take the
form of public than private lynchings, whereas in the earlier period it was
as likely to take either form. Even though lynchings related to interracial
sex surely occurred before 1890, whites thus did not demonstrate the same
kind of public concern about them before the rise of extremist white su-
premacy. Public white understanding of lynching in the latter period was
so heavily centered on the rape of white women by black men that it mat-
tered little that lethal mob violence against blacks associated with alleged
sexual assaults was as common as, or even slightly less common than, vi-
olence associated with murderous assaults.
While based on a comparatively small sample of events, table 6 confirms
hypothesis 2 that white-on-white lynchings in the period 1890–1915 did
not exhibit the same patterns as white-on-black lynchings but were pre-
dominantly private affairs even when related to sexual assaults, or mur-
derous assaults for that matter. That a white mob killing a white person
accused of sexually assaulting a white woman was more typically composed

13
These results are not sensitive to alternative temporal cutoff points. Using 1888 or
1892 as a cut point instead of 1890 yields substantively similar results.

1356

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

TABLE 5
Public Lynchings of Blacks Precipitated by Allegations
of Sexual Assault, by Time Period

LYNCHING TYPE
TIME PERIOD Private ð%Þ Public ð%Þ TOTAL ð%Þ
1882–89 . . . . . . . . . . . 49 51 100
ð20Þ ð21Þ ð41Þ
1890–1915 . . . . . . . . . . 34 66 100
ð49Þ ð95Þ ð144Þ
Total . . . . . . . . . . . . 37 63 100
ð69Þ ð116Þ ð185Þ
Pearson x2 . . . . . . . . . 2.97**
df . . . . . . . . . . . . . . . . 1
NOTE.—Numbers in parentheses are the numbers of observations.
P < .05 ðtwo-tailed testÞ.
P < .01.

of her kin and kith than the broader white community reflects that such
a mob was driven by different concerns than mobs lynching black men
similarly accused.
Always a powerful mechanism for establishing intergroup social bound-
aries, the regulation of intergroup sexuality takes on further significance in
cases in which races are understood as discrete groups rooted in biology so as
to safeguard racial purity. Indeed, the implementation and maintenance of
sexual boundaries and practices in such circumstances are, to a certain ex-
tent, what makes a racial group a racial group ðJacobson 1998Þ. These event-
level analyses reveal in that regard how whites’ conceptions of and actions
toward blacks changed toward the end of the 19th century as symbolic racial
boundaries and categories contrasting the purity and innocence of white
female sexuality with the impurity and menace of black male sexuality took

TABLE 6
Lynchings of Whites by Type and Alleged Offense, 1890–1915

LYNCHING TYPE
ALLEGED OFFENSE Private ð%Þ Public ð%Þ TOTAL ð%Þ
Sexual assault . . . . . . . . . . . . . . . 83 17 100
ð5Þ ð1Þ ð6Þ
Murder and attempted murder . . . 67 33 100
ð12Þ ð6Þ ð18Þ
Other . . . . . . . . . . . . . . . . . . . . . 100 0 100
ð7Þ ð0Þ ð7Þ
Total . . . . . . . . . . . . . . . . . . . . 77 23 100
ð24Þ ð7Þ ð31Þ
NOTE.—Numbers in parentheses are the numbers of observations. Fisher’s exact
test 5 0.221.

1357

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

hold. A rally point for collectively upholding the sanctity of white woman-
hood, public lynchings thus brought whites together unlike anything else,
and its brute materiality—leaving tortured, burned, mutilated, or otherwise
destroyed black bodies in their wake—forcefully aligned whites’ ways of
seeing with their ways of acting. The hunting down, capturing, and killing
of black men accused of sexually assaulting white females served to palpa-
bly stigmatize blacks, thereby turning symbolic representations of African-
Americans into a social boundary fundamentally shaping racial inequal-
ity in following decades. “Sex,” as Myrdal notes in An American Dilemma,
“½is . . . the principle around which the whole structure of segregation . . . —
down to disfranchisement and denial of equal opportunities on the labor
market—is organized. . . . Sexual segregation is the most pervasive form of
segregation, and the concern about ‘race purity’ is, in a sense, basic” ð1944,
pp. 587, 606Þ.

LYNCHING AND COLLECTIVE WHITE RACIAL IDENTITY

Lynching from Collective White Identity
Table 7 presents a number of Cox models assessing the association be-
tween conditions of collective white racial identity proxied by Democratic
Party support and different lynching types.14 Model 1 evaluates whether, as
hypothesized above, the impact of collective white racial identity on public
lynching was shaped by larger institutional developments in the emerging
Jim Crow order and varied between the pre- and post-disfranchisement
periods. Model 1, at first glance, does not give much reason to think that
collective white racial identity had any impact on public lynchings in the
earlier period as both the linear and quadratic effect coefficients are sta-
tistically insignificant, whereas they are statistically significant in the later
period. This inference is, however, premature because the size and statis-
tical significance of linear and quadratic coefficients cannot be interpreted
separately as unconditional effects but must be evaluated jointly at relevant
values of the covariate in question ðBrambor et al. 2006Þ. Figure 2 therefore
graphically displays marginal effects of low ðone SD below its meanÞ, in-

14
In supplementary analyses, I assessed whether different lynch mob types may have
been driven by the pool of potential white participants rather than collective white racial
identity. I did so by substituting absolute white population size for absolute black
population size in model estimations. The analyses did not produce results substantively
different from what is presented here. Drawing on information in Haines ð2006Þ, at the
suggestion of one AJS reviewer, I also assessed to what extent past racial oppression was
predictive of different lynching types by estimating models including the percentage of
slaves of the total population in 1860. These analyses did not yield substantively or
statistically significant results ðthe results of these supplementary analyses are available
on requestÞ.

1358

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

TABLE 7
Cox Models for Public and Private Lynchings of Blacks

Model 1: Model 2: Model 3:

Public Public Private
Lynchings, Lynchings, Lynchings,
1890–1915 1916–30 1890–1915
Collective white racial identity . . . . . . . .022
ð.015Þ
Collective white racial identity,
pre-disfranchisement . . . . . . . . . . . . . .040 .010
ð.040Þ ð.013Þ
Collective white racial identity squared,
pre-disfranchisement . . . . . . . . . . . . . 2.001
ð.000Þ
Collective white racial identity,
post-disfranchisement . . . . . . . . . . . . 2.101* .013
ð.043Þ ð.022Þ
Collective white racial identity squared,
post-disfranchisement . . . . . . . . . . . . .001*
ð.000Þ
Disfranchisement dummy . . . . . . . . . . . 4.91** 21.74
ð1.76Þ ð1.18Þ
% black . . . . . . . . . . . . . . . . . . . . . . . . . .0391 .136* .008
ð.024Þ ð.065Þ ð.032Þ
% black squared . . . . . . . . . . . . . . . . . . 2.0011 2.001* 2.000
ð.000Þ ð.000Þ ð.000Þ
ln black population ð1,000sÞ . . . . . . . . . .386** .277 .414**
ð.145Þ ð.192Þ ð.143Þ
Average farm size . . . . . . . . . . . . . . . . . .000 2.011* .003**
ð.001Þ ð.005Þ ð.001Þ
% farms owned . . . . . . . . . . . . . . . . . . . .011 .0251 2.004
ð.008Þ ð.015Þ ð.010Þ
% cotton farmland . . . . . . . . . . . . . . . . .017* 2.002 .007
ð.008Þ ð.016Þ ð.010Þ
No. prior opposite type lynchings . . . . . 2.049 .097 .1301
ð.080Þ ð.099Þ ð.074Þ
% black  lnðtÞ . . . . . . . . . . . . . . . . . . . .034**
ð.011Þ
% black2  lnðtÞ . . . . . . . . . . . . . . . . . . 2.0001
ð.000Þ
NOTE.—Numbers in parentheses are robust SEs clustered on county. Models include an
unreported state dummy variable.
1 P < .10, two-tailed tests.
* P < .05.
** P < .01.

termediate ðits meanÞ, and high ðone SD above its meanÞ levels of Democratic
Party strength on public lynchings in the pre- and post-disfranchisement
periods.15

15
Figure 2 applies one-tailed statistical tests of significance as the direction of the marginal
effect of collective white racial identity on public lynchings is predicted to be positive.

1359

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

F IG . 2.—Marginal effects of white collective identity on public lynchings at high

ðsolid lineÞ, intermediate ðdashed lineÞ, and low ðsquare dotted lineÞ levels of white
collective identity in the pre-Reconstruction and post-Reconstruction periods.

From figure 2 we see that the marginal effect of collective white racial
identity was strongest ðand statistically significantÞ at lower and interme-
diate levels in the pre-disfranchisement period and at higher levels in the
post-disfranchisement period. These results are consistent with our expec-
tations considering the importance of group solidarity in intergroup conflict.
As discussed earlier, in order for one group to dominate another, it cannot
doubt its own cohesion and strength, which are most unequivocally estab-
lished in observable collective group action ðGould 2003; Collins 2004Þ. In-
sofar, the results support the fifth hypothesis that in the pre-disfranchisement
South where white domination had not yet been firmly reestablished, con-
ditions of ambiguous collective white racial identity, captured by low and
intermediate levels of Democratic vote share, influenced whites to enact com-
munity solidarity and empowerment through public lynchings, whereas
whites in settings with a comparatively entrenched collective racial iden-
tity, captured by high levels of Democratic vote share, were not similarly
compelled to assert racial solidarity and power in public lynchings. These

Using the 25th, 50th, and 75th percentiles instead of the mean and values one standard
deviation above and one standard deviation below the mean of Democratic vote share
does not produce results substantively different from the one presented in fig. 2.

1360

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

results conform to Bailey and Snedker’s ð2011Þ argument that contexts in

which the collective cohesion among whites was threatened were condu-
cive to mob violence and suggest that the force of public lynchings in
southern race relations partly lay in agitating a collective racial identity
that let whites create a sense of certainty and power out of perceived un-
certainty and powerlessness before the institutionalization of Jim Crow
ðcf. Garland 2005Þ.
If public lynching had been directly related to competitive electoral pol-
itics, it should arguably have been more common in the absence, not in the
presence, of disfranchisement. In light of the broader view of politics taken
here as involving the alignment of groups in contention and the coercive
power relationships between them, the positive and significant coefficient
for the post-disfranchisement dummy variable, showing that public lynch-
ings were more likely in places with than without formal suffrage restric-
tions, confirms the expectation of hypothesis 6 that disfranchisement was
a “critical event” brightening the interracial social boundary and collective
white racial identity on which Jim Crow rested and altering the circum-
stances of collective white mobilization against blacks. That disfranchise-
ment promoted race-based solidarity and mobilization among whites is
further emphasized by the finding that higher levels of collective white racial
identity, as embodied in higher levels of Democratic Party vote share, had
larger ðand statistically significantÞ marginal effects on public lynching in
the post-disfranchisement period. Confirming hypothesis 7, this finding is in
line with arguments that the implementation of disfranchisement by south-
ern governments legitimized the beliefs, claims, and practices of extremist
white supremacy, particularly reinforcing racial social boundaries and mut-
ing divisions among whites that could otherwise have put their racial unity
and domination at risk ðWoodward 1951; Kousser 1974; Perman 2001;
Redding 2003Þ. Thus, rather than functional substitutes, formal measures
of racial domination such as disfranchisement and informal ones such as
public lynchings complemented each other in enacting and promoting white
group solidarity and power by pressing blacks “into service as a sectional
scapegoat in the reconciliation of estranged white classes and the reunion of
the Solid South” ðWoodward 1974, p. 82; cf. Garland 2005Þ. That model 2
does not indicate any association between the collective white racial identity
channeled through the Democratic Party and public lynching in the period
1916–30 confirms the expectation formulated in hypothesis 8 that public
lynchings figured into the activation and maintenance of racial group bound-
aries and identities most conspicuously during the rise of Jim Crow in the
period 1890–1915 ðas preliminary exploratory analyses showed no indica-
tion of a nonlinear association between the focal independent variable and
the outcome variable, the model includes the former only as a linear pre-

1361

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

dictorÞ. Thus, as both Jim Crow and public lynching served to symbolize as
well as substantialize white domination in the solid South, once the foun-
dational racial categories, boundaries, and identities of the former were
firmly established by the mid-1910s, it was no longer necessary for southern
whites to enact community solidarity and empowerment in communal vio-
lence against a marginalized black population.
That private lynchings bore no statistically discernible association with
the strength of the Democratic Party in the pre-disfranchisement or the
post-disfranchisement period is consistent with hypothesis 9 and the view
that collective identity is a group-level phenomenon that does not affect
individuals or interpersonal dynamics in direct mechanical ways ðagain, as
preliminary exploratory analyses showed no signs of nonlinearity in the
association between the focal independent variable and the outcome, the
model includes the former only as a linear predictorÞ. As such, that private
lynchings were unaffected by the implementation of voting restrictions con-
firms that disfranchisement primarily disempowered blacks as a group and
empowered whites as a group ðPerman 2001Þ and did not represent the type
of cue whites could use to get an edge on blacks on the interpersonal level
ðcf. Gould 2003Þ.
Model 3 provides evidence that a “spillover effect” was in effect from one
type of lethal mob violence to the other in that the number of previous
public lynchings positively and significantly predicted private lynchings in
1890–1915, but not, as indicated by model 1, vice versa. This result is best
viewed in light of the fact that public lynching was the form of violence
that above and beyond everything else caught the attention of white and
black southerners alike. Private lynchings in contrast neither were clearly
set apart from other more mundane forms of interracial violence at the
time nor always met with white community approval. This indicates that
public lynchings created an atmosphere in which whites felt authorized to
seek resolution to primarily personal or familial conflicts with blacks by
lethal interpersonal violence ðcf. Brundage 1993; Garland 2005Þ.
Finally, the results for the control variables are consistent with earlier
studies. The rate of public as well as private lynchings was positively and
nonlinearly associated with relative black population size, an association
showing signs of nonproportionality in the case of private lynchings. ðAs
this association is of secondary importance here, it will not be discussed
further; but for the reader’s information, its temporal development is de-
tailed in table A1 in the appendix.Þ The likelihood of both lynching types
furthermore increased with larger absolute black populations, whereas over-
all local farm ownership conditions were not predictive of either type in the
period 1890–1915, but positively associated with public lynchings in the period
1916–30. The models reveal some noteworthy differences across lynching
types. First, in the period 1890–1915, private lynching grew more likely as

1362

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

average farm size increased, indicating the presence of wealthier planters

with larger landholdings, while public lynchings seem insensitive to such
conditions. This suggests that private lynchings to a greater extent than pub-
lic lynchings directly served as a means of black rural labor force control. This
impression is further strengthened by model 3, showing that public lynchings
grew less likely with increasing average farm sizes in the period 1916–30,
because the agricultural labor shortages resulting from the Great Migration
prompted southern planters, as well as other civic leaders, for example, jour-
nalists, politicians, and industrialists, to publicly speak out against the most
extreme forms of interracial violence. Second, the more cotton farming
dominated the local agricultural economy in the period 1890–1915, the more
common public but not private lynching. This result resonates with the
argument above, as well as that of previous research, that blacks were
particularly vulnerable to interracial mob violence where an interclass com-
munity of interest prevailed among whites. And the more dominant cotton
production, the more white landlords needed to exercise labor control and the
stronger the competition on landless whites from blacks for land and agri-
cultural employment opportunities. Thus, common material stakes in sup-
pressing rural blacks served as a social resource for whites to mobilize in
collective violence against blacks. That cotton farming was no longer asso-
ciated with public lynching after 1915 is therefore unsurprising considering
its diminishing role in southern agriculture as well as the above-mentioned
waning support for mob violence against blacks ðTolnay and Beck 1995Þ.

Collective White Identity from Lynching

In order to bring the analyses full circle and explore to what extent lethal
mob violence against blacks forged collective racial identifications, I fit dy-
namic panel data models assessing the impact of public and private lynch-
ings, respectively, on local white collective racial identifications measured
as the percentage of Democratic votes in presidential elections of 1888–
1916. As discussed earlier, these models include in addition to election year
dummy variables the same control variables as the event-history models
above. However, as the purpose of these models is to investigate whether
collective white racial identity was responsive to different lynching types,
which does not necessitate detailed analyses of the controls, in the interest
of economizing on space, only the estimates for focal independent variables
are reported here ðfull model results are available on requestÞ. If we turn to
table 8, in line with our expectations, as formulated in hypothesis 10, we find
evidence that the number of public lynchings increased collective white
racial identity; it has a positive and statistically significant effect on Dem-
ocratic Party share of votes across all time intervals, an effect that, as in-
dicated by the negative and significant coefficient for its interaction with

1363

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

TABLE 8
Dynamic Panel Data Models Estimates of the Effect of Public and Private
Lynchings on Collective White Racial Identity, 1888–1916

TIME INTERVAL
Four Years Two Years One Year
No. public lynchings . . . . . . . . . . . . . . 30.56** 32.15** 41.21**
ð5.47Þ ð9.76Þ ð12.10Þ
No. public lynchings  collective white
racial identityt21 . . . . . . . . . . . . . . . 2.36** 2.39** 2.53**
ð.07Þ ð.13Þ ð.17Þ
No. private lynchings . . . . . . . . . . . . . 23.50 5.92 5.39
ð10.19Þ ð12.38Þ ð17.78Þ
No. private lynchings  collective white
racial identityt21 . . . . . . . . . . . . . . . .04 2.08 2.14
ð.12Þ ð.15Þ ð.21Þ
NOTE.—Nos. in parentheses are robust SEs clustered on county.
1 P < .10, two-tailed tests.
* P < .05.
** P < .01.

lagged values of the dependent variable, grew weaker at comparatively

higher levels of extant collective white racial identity.
As the size and statistical significance of linear and quadratic coefficients
of a covariate cannot be treated as unconditional effects but need to be
treated as conditional effects and evaluated jointly at relevant values of the
covariate, the constitutive terms of interaction terms do not capture the
unconditional effect of each term ðBrambor et al. 2006Þ. In order to properly
gauge the effect of the number of public lynchings in the dynamic panel
data models presented in table 8, we must calculate its marginal effects
conditional on different one-period-lagged values of collective white racial
identity. Using the same cutoff points to determine low, intermediate, and
high levels of collective white racial identity as in figure 2, figure 3A graph-
ically displays these marginal effects and shows that the effect of public
lynchings was significantly positive at comparatively lower and intermedi-
ate, but not higher, lagged levels of collective white racial identity across
all time intervals. In other words, these results show that public lynchings
enhanced white group solidarity and cohesion in contexts in which collective
white identifications were emerging but less so where they were more well
entrenched. This partly confirms our expectations in hypothesis 10 and thus
suggests that public lynchings not only expressed collective white identifi-
cations but also forged them by shaping understandings and experiences
of the “true” nature of race relations and of “what it meant and how it felt to
be a white Southerner” ðGarland 2005, p. 821Þ. Taken together with above-
reported findings regarding under which conditions of white collective identity

1364

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 F IG . 3.—Marginal effects of number of public and private lynchings on white racial
collective identity, 1888–1916, at high ðsolid lineÞ, intermediate ðdashed lineÞ, and low
ðsquare dotted lineÞ one-period-lagged levels of white collective identity.

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

public lynchings were most common, the following observation by Hale cer-
tainly rings true: “½Public lynching was a ritual that both brought out
and created the white community” ð1998, p. 211; emphasis addedÞ.
Going back to table 8, the results from the models of the effect of pri-
vate lynchings on white collective racial identity appear consistent with ex-
pectations that this type of intergroup violence was not conducive to col-
lective white identity; the terms for its linear coefficient and its interaction
with lagged levels of collective white racial identity do not in any case ap-
proach statistical significance. In fact, as figure 3B shows, the marginal effect
of the number of lynchings occurring in the one-year interval before mea-
sure points is statistically significant and negative at high levels of collective
white racial identity. As I have noted, the Democratic Party was an impor-
tant arena for in-group policing, and not all white violence against blacks
met with community approval but occasionally aroused the ire of other
whites. These findings indicate that private lynchings in some instances
made simmering intrawhite tensions manifest, leading to decreasing levels of
racial cohesion in ways registered by the current measure of collective white
identity. Insofar, the results further reflect the Democratic Party as a central
context wherein and wherethrough the norms of the white community were
established and enforced.

CONCLUDING DISCUSSION
The title of this article, referring to “making race” by “doing violence,”
intends to direct attention to how interracial violence can feed off as well
as into the articulation of symbolic boundaries representing the meanings
of categories such as “white” and “black,” the creation of social boundaries
establishing how those signified by such categories relate and behave to-
ward each other, and the formation of solidary race-based collectives with
common purposes and the capacity for concerted action. The article es-
tablishes in that regard, for one thing, that as dominant white conceptions
of race took an extremist turn around 1890, mob violence linked to inter-
racial sexuality increasingly took the form of public lynchings, serving to
violently construct and police the color line by transforming symbolic rep-
resentations of blacks into a social boundary of stigma and exclusion. For
another, public lynchings but not private lynchings were expressive as well
as generative of the collective white racial identity mobilized through the
discursive and relational context provided by the Democratic Party.
These findings have a number of implications. First, they establish that
the term “lynching” subsumes different forms of intergroup violence into
an undifferentiated concept obscuring the heterogeneity of racial violence
in the post-Reconstruction South. They accordingly strengthen the case for
conceptual disaggregation in the study of racial ðas well as ethnic and na-

1366

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

tionalistÞ violence, that is to say, to treat such violence as reflecting quali-

tatively different processes of genesis and consequence, not as something
varying in degrees alone ðBrubaker and Laitin 1998Þ. Second, the article
marks insofar an important step forward in lynching scholarship because
as foreseen by Creech, Corzine, and Huff-Corzine, “Until more precise dis-
tinctions between lynchings can be made on a systematic basis, the conclu-
sions reached by researchers are tentative at best” ð1989, p. 630Þ. By ele-
vating Brundage’s descriptive lynch mob typology into a heuristic device
for yielding generalizable theoretical and statistical findings as well as con-
firming the main conclusions of previous scholarship, the charge of this
study therefore decidedly has not been to refute previous research but to
suggest lines of analysis entertaining the role of sociocultural as well as
material factors in developing a synthesis of the lynching phenomenon.
To that end, lynching can be viewed as one aspect of the developments
leading up to the system of Jim Crow, which, in turn, can be understood
as a system of durable racial inequality founded on black economic dis-
possession, political disempowerment, and social degradation. Jim Crow
society was based on reestablishing the unequal categorical pair of “white”
and “black,” as well as forging common racial understandings and com-
mitments among whites. And as we have seen, public lynchings were ex-
pressive of as well as conducive to such understandings and commitments.
With their self-conscious communality, ritualism, and symbolism, enacted
not for particular individuals but for, as well as by, whites as a group,
public lynchings gave authority to a particular racial ideology and im-
printed onto the social order the symbolic racial boundaries, categories,
and identities that were to govern southern white-black relations for de-
cades to come.
Like all systems of durable inequality, Jim Crow did not rest solely on
hierarchical categorical distinctions but also on the stable unequal allo-
cation of material resources and opportunities along categorical lines ðTilly
1998; Massey 2007Þ. Recalling that intergroup violence allows for multiple
interpretations across audiences, the social control function at the heart of
accounts of lynching emphasizing its role in economically oppressing rural
blacks may thus be specified in terms of the durable inequality engendering
mechanisms of exploitation and opportunity hoarding. On the one hand,
lynching played into exploitation processes by disciplining the large black
labor force within the southern agricultural system wherein rent-based eco-
nomic advantage obtained monumentally to white landlords, planters, and
merchants at the expense of impoverished black farm tenants and share-
croppers. On the other, it served purposes of opportunity hoarding by re-
ducing competition from rural blacks to lower-to-middling-class whites by
intimidating them to surrender land and tenancy opportunities ðTolnay and
Beck 1995Þ.

1367

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

Third, and finally, by establishing how political institutional conditions

and developments interacted with public lynchings to transform the visions
of extremist white supremacy promoted by the Democratic Party into social
reality in the form of Jim Crow, the article provides insights into the
dynamics of what McVeigh has called “right-wing” mobilization. That is,
the mobilization of “relatively advantaged groups with the goal of pre-
serving, restoring, and expanding ½their rights and privileges . . . ½and
deny½ing similar rights and privileges to other groups in society” ð2009,
p. 32Þ. This work thereby not only sheds further light on a particular aspect
of the violent history of U.S. race relations but adds to a growing body of
research emphasizing how racial inequalities, whether in the past or pres-
ent, cannot be properly understood apart from processes of racial category,
boundary, and identity formation ðfor a review of this literature, see Mc-
Dermott and Samson ½2005Þ. In doing so, it offers clues to the results of a
number of recent studies documenting an empirical link between patterns
of lynching in the decades around 1900 and current oppressive racial prac-
tices on the local ðcountyÞ level. These studies include investigations into
white supremacist hate groups ðDurso and Jacobs 2013; see also Cunning-
ham and Phillips 2007Þ, interracial homicide ðMessner, Baller, and Zeven-
bergen 2005Þ, and resistance to the implementation of federal hate crime
legislation ðKing, Messner, and Baller 2009; see also Jacobs, Carmichael,
and Kent 2005Þ. What this link represents is, however, not very well under-
stood, but glossed with statements that past racial violence has created an
“enduring repressive tradition” ðDurso and Jacobs 2013, p. 129; Jacobs et al.
2005, p. 657Þ or that “racial antagonism dies hard” ðKing et al. 2009, p. 292Þ.
Given the centrality of local settings in the formation and transformation of
exclusion and stratification ðTilly 1998Þ, these traces of lynching in the pres-
ent arguably reflect less the effects of the violence itself than the racial group
boundaries and identities that it helped cement ðcf. Cunningham and Phillips
2007Þ.
In a similar vein, if we consider issues of race and criminal justice in
broader perspective, the present analysis suggests how, when, and why
African-Americans through what Muhammad ð2010Þ calls “the condem-
nation of blackness” were assigned a stigmatized group identity that in part
persists into the present with far-reaching consequences. These conse-
quences include important white-black differentials in criminal justice sys-
tem outcomes such as death sentences ðJacobs and Carmichael 2002Þ, im-
prisonment rates ðJacobs and Carmichael 2001; Western 2006; Muller 2012Þ,
and felon disfranchisement ðBehrens, Uggen, and Manza 2003Þ. There are
insofar striking similarities between the post-Reconstruction image of the
“black beast rapist” and the contemporary “criminal black man” stereotype
that portrays young black men as inherently more sinister, dangerous, and
prone to violence and criminality than white men ðRussell 1998Þ. “The wide

1368

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

diffusion of bestial metaphors in the journalistic and political field ðwhere

mentions of ‘superpredators’, ‘wolf-packs’, ‘animals’ and the like are com-
monplaceÞ” supplies, as Wacquant ð2001, pp. 117–18Þ points out, “a pow-
erful common-sense warrant for . . . . the conflation of blackness and crime
in collective representation and government policy.” It is thus not unrea-
sonable to argue that today’s carceral institutions to a certain extent carry
on the coercive race-making practices of yesteryears. And by proposing that
and how the violent practices of past racial domination impart directionality
to the beliefs and practices through which contemporary racial conflicts are
understood and played out on the local as well as national level, this article
gives truth to the observation that “the past can never be erased and the
ugliest human actions cast the longest shadows” ðMcFeely 1997, p. 318Þ.

APPENDIX

TABLE A1
Time-Varying Effects of Black and Black Squared in Model 3, Table 7

YEAR
1890 1895 1900 1905 1910 1915
ðt 5 .5Þ ðt 5 5.5Þ ðt 5 10.5Þ ðt 5 15.5Þ ðt 5 20.5Þ ðt 5 25.5Þ
% black . . . . . . . . 2.015 .067* .089** .103** .112** .120**
ð.037Þ ð.027Þ ð.028Þ ð.030Þ ð.032Þ ð.033Þ
% black squared . . . .000 2.001** 2.001** 2.001** 2.001** 2.001**
ð.001Þ ð.000Þ ð.000Þ ð.000Þ ð.000Þ ð.000Þ
NOTE.—Nos. in parentheses are robust SEs clustered on county.
1 P < .10, two-tailed tests.
* P < .05.
** P < .01.

REFERENCES
Abrams, Philip. 1982. Historical Sociology. Ithaca, N.Y.: Cornell University Press.
Alba, Richard. 2005. “Bright vs. Blurred Boundaries: Second-Generation Assimilation
and Exclusion in France, Germany, and the United States.” Ethnic and Racial Studies
28 ð1Þ: 20–49.
Arellano, Manuel, and Stephen Bond. 1991. “Some Tests of Specification for Panel Data:
Monte Carlo Evidence and an Application to Employment Equations.” Review of
Economic Studies 58 ð2Þ: 277–97.
Arellano, Manuel, and Olympia Bover. 1995. “Another Look at the Instrumental
Variable Estimation of Error-Components Models.” Journal of Econometrics 68 ð1Þ:
29–51.
Ayers, Edward L. 1984. Vengeance and Justice: Crime and Punishment in the 19th-
Century American South. New York: Oxford University Press.
———. 1992. The Promise of the New South: Life after Reconstruction. Oxford: Oxford
University Press.

1369

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

Bagozzi, Richard P. 1977. “Populism and Lynching in Louisiana.” American Socio-

logical Review 42 ð2Þ: 355–58.
Bail, Christopher A. 2008. “The Configuration of Symbolic Boundaries against Im-
migrants in Europe.” American Sociological Review 73 ð1Þ: 37–59.
———. 2012. “The Fringe Effect: Civil Society Organizations and the Evolution of
Media Discourse about Islam since the September 11th Attacks.” American Socio-
logical Review 77 ð6Þ: 855–79.
Bailey, Amy Kate, and Karen A. Snedker. 2011. “Practicing What They Preach? Lynching
and Religion in the American South, 1890–1929.” American Journal of Sociology 117 ð3Þ:
844–87.
Bartley, Numan V. 1990. The Creation of Modern Georgia. Athens: University of
Georgia Press.
Beck, E. M., James Massey, and Stewart E. Tolnay. 1989. “The Gallows, the Mob, and
the Vote: Lethal Sanctioning of Blacks in North Carolina and Georgia, 1882–1930.”
Law and Society Review 23 ð2Þ: 317–31.
Beck, E. M., and Stewart E. Tolnay. 1990. “The Killing Fields of the Deep South: The
Market for Cotton and the Lynching of Blacks, 1882–1930.” American Sociological
Review 55 ð4Þ: 526–39.
Behrens, Angela, Christopher Uggen, and Jeff Manza. 2003. “Ballot Manipulation and
the ‘Menace of Negro Domination’: Racial Threat and Felon Disfranchisement in the
United States, 1850–2002.” American Journal of Sociology 109 ð3Þ: 559–605.
Berk, Richard A. 1977. “Proof? No. Evidence? No. A Skeptic’s Comment on Inverarity’s
Use of Statistical Inference.” American Sociological Review 42 ð4Þ: 652–56.
Blalock, Hubert M. 1967. Toward a Theory of Minority Group Relations. New York:
Wiley.
Blee, Kathleen M. 2005. “Racial Violence in the United States.” Ethnic and Racial
Studies 28 ð4Þ: 599–619.
Blundell, Richard, and Stephen Bond. 1998. “Initial Conditions and Moment Restric-
tions in Dynamic Panel Data Models.” Journal of Econometrics 87 ð1Þ: 115– 43.
Bohrnstedt, George W. 1977. “Use of the Multiple Indicators–Multiple Causes ðMIMICÞ
Model.” American Sociological Review 42 ð4Þ: 656–63.
Box-Steffensmeier, Janet M., and Bradford S. Jones. 2004. Event History Modeling: A
Guide for Social Scientists. Cambridge: Cambridge University Press.
Box-Steffensmeier, Janet M., and Christopher Zorn. 2002. “Duration Models for
Repeated Events.” Journal of Politics 64 ð4Þ: 1069–94.
Brambor, Thomas, William Roberts Clark, and Matt Golder. 2006. “Understanding
Interaction Models: Improving Empirical Analyses.” Political Analysis 14 ð1Þ: 63–82
Brubaker, Rogers. 2004. Ethnicity without Groups. Cambridge, Mass.: Harvard Uni-
versity Press.
Brubaker, Rogers, and David D. Laitin. 1998. “Ethnic and Nationalist Violence.”
Annual Review of Sociology 24:423–52.
Brundage, W. Fitzhugh. 1993. Lynching in the New South: Georgia and Virginia, 1880–
1930. Urbana: University of Illinois Press.
Burke, Peter J., and Jan E. Stets. 2009. Identity Theory. New York: Oxford University
Press.
Cell, John. 1982. The Highest Stage of White Supremacy: The Origins of Segregation in
South Africa and the American South. Cambridge: Cambridge University Press.
Clubb, Jerome M., William H. Flanigan, and Nancy H. Zingale. 2006. “Electoral Data
for Counties in the United States: Presidential and Congressional Races, 1840–1972”
½computer file. ICPSR 08611-v1. Ann Arbor, Mich.: Inter-University Consortium for
Political and Social Research ½producer and distributor.
Collins, Randall. 1974. “Three Faces of Cruelty: Towards a Comparative Sociology of
Violence.” Theory and Society 1 ð4Þ: 415–40.
———. 2004. Interaction Ritual Chains. Princeton, N.J.: Princeton University Press.

1370

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

Corzine, Jay, Lin Huff-Corzine, and James C. Creech. 1988. “The Tenant Labor Market
and Lynching in the South: A Test of Split Labor Market Theory.” Sociological
Inquiry 58 ð3Þ: 261–78.
Creech, James C., Jay Corzine, and Lin Huff-Corzine. 1989. “Theory Testing and Lynch-
ing: Another Look at the Power Threat Hypothesis.” Social Forces 67 ð3Þ: 626–30.
Cunningham, David. 2013. Klansville, U.S.A.: The Rise and Fall of the Civil Rights–
Era Ku Klux Klan. New York: Oxford University Press.
Cunningham, David, and Benjamin T. Phillips. 2007. “Contexts for Mobilization:
Spatial Settings and Klan Presence in North Carolina, 1964–1966.” American Journal
of Sociology 113 ð3Þ: 781–814.
Dailey, Jane. 2000. Before Jim Crow: The Politics of Race in Postemancipation Virginia.
Chapel Hill: University of North Carolina Press.
Desmond, Matthew and Mustafa Emirbayer. 2009. “What Is Racial Domination?” Du
Bois Review: Social Science Research on Race 6 ð2Þ:335–55.
Downs, Gregory P. 2011. Declarations of Dependence: The Long Reconstruction of Popular
Politics in the South, 1861–1908. Chapel Hill: University of North Carolina Press.
Du Bois, W. E. B. 1935. Black Reconstruction in America, 1860–1880. New York:
Simon & Schuster.
———. ð1903Þ 2003. The Souls of Black Folk. New York: Modern Library.
Durso, Rachel M., and David Jacobs. 2013. “The Determinants of the Number of White
Supremacist Groups: A Pooled Time-Series Analysis.” Social Problems 60 ð1Þ: 128–
44.
Fox, Cybelle, and Thomas A. Guglielmo. 2012. “Defining America’s Racial Boundaries:
Blacks, Mexicans, and European Immigrants, 1890–1945.” American Journal of So-
ciology 118 ð2Þ: 327–79.
Fredrickson, George M. 1971. The Black Image in the White Mind: The Debate on Afro-
American Character and Destiny, 1817–1914. New York: Harper & Row.
Gaines, M. Foster. 1987. Ghosts of the Confederacy: Defeat, the Lost Cause, and the
Emergence of the New South, 1865–1913. New York: Oxford University Press.
Gamson, William A. 1975. The Strategy of Social Protest. Homewood, Ill.: Dorsey.
Garland, David. 2005. “Penal Excess and Surplus Meaning: Public Torture Lynchings
in Twentieth-Century America.” Law and Society Review 39 ð4Þ: 793–833.
Geertz, Clifford. 1973. Interpretation of Cultures: Selected Essays. New York: Basic
Books.
Gould, Roger V. 1995. Insurgent Identities: Class, Community and Protest in Paris from
1848 to the Commune. Chicago: University of Chicago Press.
———. 2003. Collision of Wills: How Ambiguity about Social Rank Breeds Conflict.
Chicago: University of Chicago Press.
Gullickson, Aaron. 2010. “Racial Boundary Formation at the Dawn of Jim Crow: The
Determinants and Effects of Black/Mulatto Occupational Differences in the United
States, 1880.” American Journal of Sociology 116 ð1Þ: 187–231.
Hagen, Ryan, Kinga Makovi, and Peter Bearman. 2013. “The Influence of Political
Dynamics on Southern Lynch Mob Formation and Lethality.” Social Forces 92 ð2Þ:
757–87.
Haines, Michael R., and the Inter-University Consortium for Political and Social Re-
search. 2006. “Historical, Demographic, Economic, and Social Data: The United States,
1790–2000” ½computer file. ICPSR02896-v2. Hamilton, N.Y.: Colgate University/Ann
Arbor: Mich.: Inter-University Consortium for Political and Social Research ½producers.
Ann Arbor, Mich.: Inter-University Consortium for Political and Social Research
½distributor.
Hale, Grace E. 1998. Making Whiteness: The Culture of Segregation in the South, 1890–
1940. New York: Vintage.
Holt, Thomas C. 1995. “Marking: Race, Race-Making, and the Writing of History.”
American Historical Review 100 ð1Þ: 1–20.

1371

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

Inverarity, James. 1976. “Populism and Lynching in Louisiana, 1889–1896: A Test of

Erikson’s Theory of the Relationship between Boundary Crisis and Repressive
Justice.” American Sociological Review 41 ð2Þ: 262–80.
Jacobs, David, and Jason T. Carmichael. 2001. “The Politics of Punishment across Time
and Space: A Pooled Time-Series Analysis of Imprisonment Rates.” Social Forces
80 ð1Þ: 61–89.
———. 2002. “The Political Sociology of the Death Penalty: A Pooled Time-Series
Analysis.” American Sociological Review 67 ð1Þ: 109–31.
Jacobs, David, Jason T. Carmichael, and Stephanie L. Kent. 2005. “Vigilantism, Cur-
rent Racial Threat, and Death Sentences.” American Sociological Review 70 ð4Þ: 656–
77.
Jacobson, Mathew Frye. 1998. Whiteness of a Different Color: European Immigrants
and the Alchemy of Race. Cambridge, Mass.: Harvard University Press.
Kantrowitz, Stephen. 2000. Ben Tillman and the Reconstruction of White Supremacy.
Chapel Hill: University of North Carolina Press.
King, Ryan D., Steven F. Messner, and Robert D. Baller. 2009. “Contemporary Hate
Crimes, Law Enforcement, and the Legacy of Racial Violence.” American Sociolog-
ical Review 74 ð2Þ: 291–315.
Koçak, Özgecan, and Glenn R. Carroll. 2008. “Growing Church Organizations in Diverse
U.S. Communities, 1890–1926.” American Journal of Sociology 113 ð5Þ: 1272–1315.
Kousser, J. Morgan. 1974. The Shaping of Southern Politics: Suffrage Restriction and
the Establishment of the One Party South, 1880–1910. New Haven, Conn.: Yale
University Press.
Lamont, Michèle. 2000. The Dignity of Working Men: Morality and the Boundaries of
Race, Class, and Immigration. Cambridge, Mass.: Harvard University Press.
Lamont, Michèle, and Virag Molnar. 2002. “The Study of Boundaries in the Social
Sciences.” Annual Review of Sociology 28:167–95.
Landis, Richard, and Gary Koch. 1977. “The Measurement of Observer Agreement for
Categorical Data.” Biometrics 33 ð1Þ: 159–74.
Litwack, Leon F. 1998. Trouble in Mind: Black Southerners in the Age of Jim Crow.
New York: Knopf.
Massey, Douglas S. 2007. Categorically Unequal: The American Stratification System.
New York: Russell Sage Foundation.
McAdam, Doug. 1982. Political Process and the Development of Black Insurgency,
1930–1970. Chicago: University of Chicago Press.
McAdam, Doug, Sidney Tarrow, and Charles Tilly. 2001. Dynamics of Contention.
Cambridge: Cambridge University Press.
McCarthy, John D., and Mayer N. Zald. 1977. “Resource Mobilization and Social
Movements: A Partial Theory.” American Journal of Sociology 82 ð6Þ: 1212– 41.
McDermott, Monica. 2006. Working-Class White: The Making and Unmaking of Race
Relations. Berkeley and Los Angeles: University of California Press.
McDermott, Monica, and Frank L. Samson. 2005. “White Racial and Ethnic Identity in
the United States.” Annual Review of Sociology 31:245–61.
McFeely, William S. 1997. “Afterword.” Pp. 318–21 in Under Sentence of Death:
Lynching in the South, edited by W. W. Fitzhugh. Chapel Hill: University of North
Carolina Press.
McVeigh, Rory. 2009. The Rise of the Ku Klux Klan: Right-Wing Movements and
National Politics. Minneapolis: University of Minnesota Press.
McVeigh, Rory, Michael R. Welch, and Thoroddur Bjarnason. 2003. “Hate Crime
Reporting as a Successful Social Movement Outcome.” American Sociological Review
68 ð6Þ: 843–67.
Messner, Steven F., Robert D. Baller, and Matthew P. Zevenbergen. 2005. “The
Legacy of Lynching and Southern Homicide.” American Sociological Review 70 ð4Þ:
633–55.

1372

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
 Doing Violence, Making Race

Moore, Adam. 2011. “The Eventfulness of Social Reproduction.” Sociological Theory

29 ð4Þ: 294–314.
Muhammad, Khalil Gibran. 2010. The Condemnation of Blackness: Race, Crime, and
the Making of Modern Urban America. Cambridge, Mass.: Harvard University Press.
Muller, Christopher. 2012. “Northward Migration and the Rise of Racial Disparity in
American Incarceration, 1880–1950.” American Journal of Sociology 118 ð2Þ: 281–326.
Myrdal, Gunnar. 1944. An American Dilemma: The Negro Problem and Modern Democ-
racy. New York: Harper & Brothers.
Olzak, Susan. 1992. The Dynamics of Ethnic Competition and Conflict. Stanford, Calif.:
Stanford University Press.
Pachucki, Mark, Sabrina Pendergrass, and Michèle Lamont. 2007. “Boundary Processes:
Recent Theoretical Developments and New Contributions.” Poetics 35 ð6Þ: 331–51.
Perman, Michael. 2001. Struggle for Mastery: Disfranchisement in the South, 1888–
1908. Chapel Hill: University of North Carolina Press.
Pfeifer, J. Michael. 2004. Rough Justice: Lynching and American Society, 1874–1947.
Urbana: University of Illinois Press.
Polletta, Francesca. 1998. “ ‘It Was like a Fever . . .’: Narrative and Identity in Social
Protest.” Social Problems 45 ð2Þ: 137–59.
Pope, Whitney, and Charles Ragin. 1977. “Mechanical Solidarity, Repressive Justice,
and Lynchings in Louisiana.” American Sociological Review 42 ð2Þ: 363–69.
Raper, Arthur F. 1933. The Tragedy of Lynching. Chapel Hill: University of North
Carolina Press.
Redding, Kent. 2003. Making Race, Making Power: North Carolina’s Road to Dis-
franchisement. Urbana: University of Illinois Press.
Russell, Katheryn K. 1998. The Color of Crime: Racial Hoaxes, White Fear, Black
Protectionism, Police Harassment, and Other Macroaggressions. New York: New
York University Press.
Saperstein, Aliya, and Aaron Gullickson. 2013. “A ‘Mulatto Escape Hatch’ in the
United States? Examining Evidence of Racial and Social Mobility during the Jim
Crow Era.” Demography 50 ð5Þ: 1921– 42.
Snow, David A., and Robert D. Benford. 1988. “Ideology, Frame Resonance and Par-
ticipant Mobilization.” International Social Movement Research 1:197–218.
Somers, Margaret R. 1994. “The Narrative Constitution of Identity: A Relational and
Network Approach.” Theory and Society 23 ð5Þ: 605– 49.
Sommerville, Diane Miller. 2004. Rape and Race in the Nineteenth-Century South.
Chapel Hill: University of North Carolina Press.
Soule, Sarah A. 1992. “Populism and Black Lynching in Georgia, 1890–1900.” Social
Forces 71 ð2Þ: 431– 49.
Staggenborg, Suzanne. 1993. “Critical Events and the Mobilization of the Pro-Choice
Movement.” Research in Political Sociology 6 ð1Þ: 319– 45.
Stovel, Katherine. 2001. “Local Sequential Patterns: The Structure of Lynchings in the
Deep South, 1882–1930.” Social Forces 79 ð3Þ: 843–80.
Swidler, Ann. 1986. “Culture in Action: Symbols and Strategies.” American Sociological
Review 51 ð2Þ: 273–86.
Tilly, Charles. 1978. From Mobilization to Revolution. Reading, Mass.: Addison-Wesley.
———. 1998. Durable Inequality. Berkeley and Los Angeles: University of California
Press.
———. 2005. Identities, Boundaries, and Social Ties. Boulder, Colo.: Paradigm.
Tolnay, Stewart E., and E. M. Beck. 1995. A Festival of Violence: An Analysis of
Southern Lynchings, 1882–1930. Urbana: University of Illinois Press.
Tolnay, Stewart E., E. M. Beck, and James L. Massey. 1989. “Black Lynchings: The
Power Threat Hypothesis Revisited.” Social Forces 67 ð3Þ: 605–23.
U.S. Bureau of the Census. 1913. Thirteenth Census, 1910, vol. 5. Washington, D.C.:
Government Printing Office.

1373

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms
American Journal of Sociology

———. 1922. Fourteenth Census, 1920, vol. 5. Washington, D.C.: Government Printing
Office.
Wacquant, Loïc. 2001. “Deadly Symbiosis: When Ghetto and Prison Meet and Mesh.”
Punishment and Society 3 ð1Þ: 95–134.
Wasserman, Ira M. 1977. “Southern Violence and the Political Process.” American
Sociological Review 42 ð2Þ: 359–62.
Wawro, Gregory. 2002. “Estimating Dynamic Panel Data Models in Political Science.”
Political Analysis 10 ð1Þ: 25– 48.
Weber, Max. 1978. Economy and Society: An Outline of Interpretive Sociology. Berkeley:
University of California Press.
Western, Bruce. 2006. Punishment and Inequality in America. New York: Russell Sage.
Williamson, Joel. 1984. The Crucible of Race: Black-White Relations in the American
South since Emancipation. New York: Oxford University Press.
Wimmer, Andreas. 2013. Ethic Boundary Making: Institutions, Power, Networks. New
York: Oxford University Press.
Wood, Amy Louise. 2009. Lynching and Spectacle: Witnessing Racial Violence in America,
1890–1940. Chapel Hill: University of North Carolina Press.
Woodward, C. Vann. 1951. Origins of the New South, 1877–1913. Baton Rouge: Louisiana
State University Press.
———. 1974. The Strange Career of Jim Crow, 3d rev. ed. New York: Oxford University
Press.

1374

This content downloaded from

142.150.190.39 on Fri, 31 Jan 2025 20:56:54 UTC
All use subject to https://about.jstor.org/terms