“Let Our Ballots Secure What Our Bullets Have Won”: Union Veterans and the Making of Radical Reconstruction

Abstract

After the Civil War, congressional Republicans used sweeping powers to expand and enforce civil rights for African Americans. Though the electoral benefits of African American suffrage were clear, Republicans had to overcome party divisions and racist voters. This paper argues that the war imbued Northern veterans with the belief that true victory required renewing the Union by abolishing slavery and establishing (imperfect) legal equality. This made veterans more receptive to Radical Reconstruction and ignited activism for it from below. Using difference-in-differences, I show that greater enlistment increased Republican vote share, particularly in pivotal postwar elections. Moreover, “as-if” random exposure to combat deaths increased Republican partisanship among soldiers after the war. Finally, I show that veterans became more likely to vote for African American suffrage. The paper concludes that Union veterans, through their votes and their activism, were a decisive part of the white coalition that backed America’s “Second Revolution.”

Introduction

In the years following the Civil War, the United States was in the midst of a “Second Revolution" (Foner 1988). A Radical congress, controlled by Republicans, reconstructed the South by passing amendments ending slavery, creating rights and protections of national citizenship, and prohibiting the denial of suffrage on the basis of race. To achieve and secure these new rights for African Americans, congressional Republicans maintained a legal state of war against much of the South and then passed sweeping enforcement legislation that greatly expanded the power of the federal government (Downs 2015; Wang 1997). Despite the eventual erosion and elimination of many Reconstruction policies, this “unfinished revolution” meaningfully improved the lives of freed people (Chacon and Jensen 2020; Logan 2020; Stewart and Kitchens 2021; Rogowski 2018) and secured the bedrock upon which the fight for equal civil rights over the following century and a half would be built.

This revolution is all the more remarkable for the fact that, despite widespread racism in the antebellum North, it was achieved through broadly free and fair elections. How did this happen? The dominant explanation suggests that the Republican Party, facing the risk that restored representation to the white South would empower the rival Democratic Party, strategically extended suffrage rights and protections to African Americans in order to build a viable national coalition to retain control of the federal government (Valley 2004). This mirrors the process by which suffrage and civil rights for African Americans were restricted in order to meet the demands of forging the Democratic Party into a successful national coalition that included Southern slaveholders (Bateman 2018). These twin narratives are consistent with arguments from comparative politics suggesting that the logic of electoral competition between rival groups of elites or political parties drives the expansion and contraction of the electorate (Capoccia and Ziblatt 2010; Llavador and Oxoby 2005; Teele 2018). According to the logic of the competition narrative, Republicans pursued these drastic reforms not out of an ideological commitment to equal civil and political rights but out of partisan teamsmanship (Lee 2009) directed at forestalling the return of Democrats to national power. Moreover, some have argued that this instrumental and “thin” support for Reconstruction helps explain its eventual erosion.

Although the electoral interests of the Republican Party were undoubtedly relevant, the competition narrative leaves much unexplained. To begin with, the need for Republicans to expand their coalition in the South was obvious before the Civil War, yet Republicans only came to embrace radical reforms after the war, and then only reluctantly and fitfully (Downs 2015; Wang 1997). The fact is that Republicans were deeply concerned that giving civil and political rights to African Americans would be “political suicide” due to the racism of the Northern electorate (Wang 1997, 6). Their fears were well founded: recent evidence shows that such racial conservatism is durable (Acharya, Blackwell, and Sen 2016) and that racial conservatives abandon parties when they begin to endorse racial liberalism (Kuziemko and Washington 2018). How, then, were white voters convinced to support a Republican Party moving forward with Radical Reconstruction? Second, if Republican elites initiated Reconstruction to build a national coalition, we would expect the push for suffrage expansion to start with national Republican leaders. Yet, it was state-level Republican activists who pushed forward the issue of equal suffrage at the end of the war, to the consternation of party leaders worried about the party’s viability nationwide. Why did these local Republican Party activists come to embrace an aggressive Reconstruction agenda?

To answer these questions and explain how the Republican Party came to pursue Radical Reconstruction, I show that the wartime experiences of the nearly two million Union Army veterans instilled in many a deep sense that the purpose of their sacrifice was to renew the Union by punishing traitors, eradicating slavery, and, to an extent, extending equal civil and political rights. Compared with civilians, Union soldiers’ unique sacrifices during the war, including in combat, reduced willingness to reconcile with the enemy (e.g., Grossman, Manekin, and Miodownik 2015) and intensified their commitment to “win the peace” (Gannon 2011; Janney 2013). Soldiers’ first-hand experiences with slavery convinced many of both the strategic and moral imperative of eliminating the institution, ensuring that emancipation was enshrined, alongside preserving the Union, as one of the twin accomplishments of the war (Manning 2007). Finally, through extensive collaboration with enslaved and freed African Americans in their fight against the Confederacy, many white Union soldiers came to believe that African Americans had earned new rights and protections through their loyalty and sacrifice.

This remarkable “political learning” (Parker 2009) among veterans helps explain the electoral success of the Republican Party in the critical elections from 1864 to 1868 that decided the trajectory of Reconstruction. Although both Democrats and Republicans sought to frame the meaning of the war and the goals of victory (Kalmoe 2020), Republican campaign messaging aligned closely with how veterans and their organizations understood the meaning of the war. Motivated to secure the fruits of their hard-won victory, veterans were particularly likely to find elite messaging from the Radical wing of the Republican Party resonant and persuasive. Given that white Union veterans made up a large share of the Northern electorate after the war, they were a pivotal constituency for Republicans.

Moreover, congressional Republicans were moved to embrace a more expansive Reconstruction agenda in response to mobilization by veterans from below. Veterans had new capacities for organization (Jha and Wilkinson 2012) and became a constituency active in promoting a more radical Reconstruction agenda. Mobilization by veterans organizations during electoral contests and the entrance of veterans into the rank and file positions within the Republican organization likely increased the sway of the radical faction within the party.

I substantiate this argument first by presenting historical evidence for ideological changes in Union veterans over the course of the war. I then empirically test whether soldiers came to back Republicans. Using newly available individual data on nearly all Union Army soldiers and the Full Count 1860 US census, I estimate enlistment rates for counties in eight Northern states as well as townships in Iowa and Wisconsin. Using a continuous difference-in-differences design to identify the effect of enlistment rates on Republican vote share, I find that higher enlistment caused substantial gains for Republicans after the war, particularly in the elections of 1864 through 1868, electing the Congresses that passed the Civil Rights Amendments and critical civil rights legislation. Compared with 1860, the Republican vote share in 1866 increased 4 and 8.5 percentage points more in counties in the third and fourth quartiles of enlistment than in counties in the lowest quartile.

To demonstrate that these ecological estimates reflect effects on soldiers, I examine the intensive effects of wartime sacrifice on individual combatants, focusing on exposure to combat deaths. I expand the work of DeCanio (2007) by digitizing the partisanship of nearly thirty thousand people residing in nine Indiana counties in 1874 and link partisanship data to the service records of more than twenty thousand Union soldiers from those counties. The Union Army deployed men into combat by regiment, which fought in a line, with companies (subunits) arrayed end-to-end. Historical evidence and balance tests demonstrate that variation in casualties across companies from the same regiment was plausibly as-if random. I exploit this natural experiment to show that, within the same regiments, soldiers in companies with higher casualties became more likely to identify as Republicans after the war, reflecting their greater motivation to make their sacrifice meaningful.

I then connect these findings to the broader political struggle over Reconstruction. It is true that some soldiers may have backed Republicans after the war merely because they could not stomach voting for Democrats tainted by their affiliation with secessionists. However, I show that veterans also came to directly support civil rights expansions. Using a difference-in-differences design, townships in Iowa and Wisconsin with greater enlistment saw significantly larger increases in support for Black suffrage in the postwar state referenda. Ecological bounds show these were indeed effects on veterans. This account is bolstered by qualitative and quantitative evidence demonstrating how the discursive and electoral activities of veterans and their organizations put pressure on legislators from below to back Radical Reconstruction measures.

Taken together, the evidence presented below has important implications for how we understand Reconstruction and the political development of race in the United States. For example, focusing on the simultaneous expansion of white suffrage and contraction of Black suffrage in the early Republic, Bateman (2018) argues that transformations in rights and citizenship depend on building winning electoral coalitions that share a vision of what it means to be a nation. This paper extends that logic to Reconstruction. Civil rights expansions became politically possible, not merely due to the strategic opportunity they offered to the Republican Party but also because a new white constituency committed to ending slavery and punishing traitors had been forged in the crucible of the war. Across the longer arc of American history, coalitions built around white supremacy have been easier to cultivate and sustain. Therefore, it is all the more important to explain how the white coalitions that supported making African Americans citizens in the nineteenth and twentieth centuries (see, e.g., Schickler 2016) emerged. Understanding how soldiers acquired a new vision of the place of African American citizens in the Union and how they translated that vision into politics will help us better understand the nature and limits of political support for Reconstruction.

I proceed by first complicating the view that Republicans pursued Radical Reconstruction for electorally strategic reasons, documenting how national leaders feared civil rights and suffrage would divide the party and lose elections. I then argue that military service turned veterans into an important constituency for Republicans. I provide qualitative evidence for political learning among white soldiers and the resonance of Republican messaging with these views. I then describe and report results from statistical tests showing that military service increased support for the Republican Party and for their Reconstruction policies. I conclude by discussing the implications of this paper for understanding the political development of race in the United States, suffrage extensions, and the political legacies of wars.

Background

Conventional accounts of Reconstruction in political science claim that Republicans in Congress, facing the reentry of the South with greater representation, strategically allied with African Americans to preserve their hold on the government. There is undoubtedly some truth to this. Before the war, Republicans won virtually no votes in the states that seceded. The 13th Amendment, freeing millions of enslaved African Americans, had the consequence of increasing seats apportioned to the South. With the seating of Southern delegations, it was clear that Republicans’ hold over the national government would become precarious. These fears were well founded: as Bateman, Katznelson, and Lapinski (2018) document, once Southern Democrats reentered Congress, they decisively shaped federal legislation for decades to come. And although, as early as 1865, some Republicans acknowledged that this dilemma might necessitate the enfranchisement of freedmen in the South (Bonadio 1970; Wang 1997), there are four problems with the partisan competition explanation for the emergence of Radical Reconstruction.

First, despite the obvious incentives, Republicans did not rush to pursue an expansive Reconstruction agenda. Instead, they came to it in fits and starts, avoiding suffrage and focusing initially on securing African Americans civil rights and enforcing the end of slavery (Downs 2015; Wang 1997).

Second, this account ignores the very real possibility that Republicans could have embraced an all-white coalition rather than backing suffrage for African Americans. At its formation barely 10 years earlier, the Republican Party was an uneasy coalition of factions with conflicting reasons for opposing slavery. And although many in the radical wing sought to pair an end to slavery with civil rights, this was a minority position that often divided the party (Bateman 2020); moderate and conservative Republicans opposed slavery for its deleterious effects on whites and hoped the colonization of freed people elsewhere would solve “the race problem” (Foner 1979). By 1865, with the abolition of slavery achieved, rumors of party realignment were rampant and many conservative and moderate Republicans expressed the desire to cut the Radicals loose from the party (Bonadio 1970; Cook 1994; Cox and Cox 1963). This was motivated by disagreement with radicals’ insistence on the full inclusion of African Americans as citizens, which conservatives derided as both a cause of the rebellion and a fatal political liability (Cox and Cox 1963, chap 2).

Conservative Republicans, including sitting US senators, governors, and powerful state party bosses, envisioned an alternative all-white coalition composed of conservative and moderate Republicans, War Democrats, and white Southerners loyal to the Union to the exclusion of Radicals who had “the negro on the brain” (Bonadio 1970). These groups sought to forge this realignment with the help of President Andrew Johnson, a southerner and former Democrat who had opposed slavery and backed the Union during the war. Between 1865 and 1866 this effort was sustained in earnest: Johnson and his conservative allies held numerous meetings and engaged in extensive correspondence that resulted in coordination during state elections, with Democrats endorsing Johnson’s Reconstruction policy and Johnson using patronage jobs to empower conservatives (Cox and Cox 1963). During the 1866 congressional election, these efforts culminated in a National Union Party convention, which attracted Democrats and Republicans to coordinate a campaign against the radical efforts of Republicans in Congress (Riddleberger 1979).

Third, regardless of whether Republicans genuinely accepted the conservative position that restoration of the Union and the formal abolition of slavery had completed the war objectives, moderate Republicans were deeply concerned about what policies a racist Northern public would support. Although majorities of Republicans in two states voted in favor of Black suffrage in prewar referenda (Bateman 2020), the majority of voters rejected suffrage, and, in most states, Republicans did not dare to put the issue on the ballot (Dykstra 1993). After the war started, Illinoisans voted overwhelmingly to restrict Blacks from entering the state and forbade them to vote (Allardice 2011, 101), and even in 1865, referenda in Connecticut and Wisconsin failed. All the while, Democrats made explicit racist appeals to voters, accusing Republicans of plotting to invert the “natural” supremacy of whites by setting African Americans as civic and “social” equals (implying interracial relationships; Dykstra 1993; Kalmoe 2020), and that, by denying whites the right to determine locally who could vote and enjoy the protections of citizenship, “subjugate” white men to the interests of “the negro” (Field 1982). When they debated the extent of the rights embedded in the 14th Amendment and the subsequent enforcement acts, Republicans explicitly voiced concerns about backlash from the Northern public and the failure of suffrage referenda in particular. They waited until after the convincing victory in the 1866 elections to start a gradual push to extend suffrage to African Americans (Wang 1997).

Finally, when Republicans began publicly pushing for Radical Reconstruction measures, including suffrage, it came from activists in the state and local parties. Nationally minded party leaders sought to steer a more moderate course. In the summer of 1865, state party leaders and congressional representatives in Ohio, Wisconsin, and Iowa worked to keep African American suffrage off of the party platform in the fall elections, fearing that it would drive a wedge between the party and President Johnson and cost the party votes. But local party activists in Iowa voted the issue onto the platform, and despite maneuvers to prevent this in Wisconsin and Ohio, these states saw substantial mobilization by radicals to force the issue into the campaign (Bonadio 1970; Cook 1994; Fishel 1963; McManus 1998). When Congress began legislating Reconstruction policy in early 1866, many congressional Republicans were upset by Johnson’s veto of the Freedmen’s Bureau Act and the Civil Rights Act. However, they were also concerned about party infighting just before the upcoming congressional elections, and so they strove to publicly downplay any break with the president and hoped to salvage a compromise with him (Bonadio 1970; Cox and Cox 1963). By contrast, immediately after Johnson vetoed the Freedmen’s Bureau Act, Republican state legislatures passed resolutions that explicitly called on Congress to override the veto, openly criticized Johnson, and in the case of Wisconsin, censured one of the state’s Republican senators for siding with Johnson (Cook 1994). Furthermore, that summer, Republican House members faced challengers with more radical bona fides in nominating conventions, pushing them to more openly challenge Johnson and his policies (Bonadio 1970).

All of this points to the period of 1865–1866 as an important “counterfactual node” (Bateman and Teele 2020)—a point at which the course of Reconstruction could have taken a more conservative direction. Had moderate Republicans embraced a coalition with Johnson or had voters, persuaded by racist appeals, handed Republicans a minority or reduced majority of seats in the fall of 1866, Reconstruction may well have ended without even the ratification of the 14th Amendment. Yet this did not come to pass. Instead, between 1865 and 1866, moderate Republicans rejected a conservative coalition and embraced more radical positions when Johnson’s conservative plan yielded Southern statehouses filled with former secessionists, the return to de facto slavery through “Black Codes,” ¹ and mass violence against freed people in Memphis and New Orleans (Foner 1988; Wang 1997). And despite this increasingly radical trajectory, Republicans won elections by large margins that fall.

Argument

Missing from these accounts is the key role played by Union veterans. I argue that experiences in the war imbued veterans with a commitment to a vision of the Union that extirpated slavery, and to an extent, incorporated African Americans as citizens. This made white Union veterans both more receptive to Republican arguments in favor of congressional Reconstruction and more likely to mobilize for Reconstruction from below.

Wartime Experiences

There is broad historical consensus that enlistment in the Union Army was not driven by commitments to end slavery. Soldiers routinely explicitly averred that they were not abolitionists. Instead, historians note that men enlisting in the Union Army were motivated by a sense of honor, duty, and a commitment to preserving the United States as a beacon of (white) democratic self-government (Gallagher 2011; Manning 2007).

Yet, military service was a life-altering event for Union veterans (Costa and Kahn 2008). It removed them from their homes, embedded them in a hierarchical disciplinary organization, and exposed them to new people and places, even before they experienced combat. Recent work shows that smaller life changes such as moving, getting divorced, or being exposed to the threat of military service can cause durable changes in attitudes and partisanship (Erikson and Stoker 2011; Hobbs 2019). Removed from their prewar environs and thrust into new and intense experiences, soldiers were primed for political transformation.

Sacrifice

Serving in the Union Army transformed soldiers in three ways that made them more likely to be a constituency for Reconstruction. First, compared with other Northern citizens, Union soldiers made much greater sacrifices in the war. Whereas Faust (2008) argues that the scale of death in the Civil War was felt at home, Marshall (2014) shows that the change in aggregate death rates during the war, even among young men, was not a radical departure from the high baseline mortality of the time. By contrast, soldiers personally experienced immense amounts of death, suffering, and terror during and after combat. Experiments in social psychology show that making sacrifices for a cause can intensify commitments to that cause, whereas military service (Koenig 2020) and combat experience in particular (Grossman, Manekin, and Miodownik 2015) have been shown to produce intense antipathy against former enemies. Thus, Union veterans had more at stake in ensuring that their sacrifices during the war were meaningful, and in particular, that their hard-won victory over the South was secure. And for decades after the war, veterans worked to remind an increasingly disinterested public of their sacrifice in the war (Cook 2021; Janney 2013).

Meaning of the War

As one soldier asked, “If all this untold expense of blood and treasure, of toil and suffering, of want and sacrifice, of grief and mourning is … to result in no greater good than the restoration of the Union as it was, what will it amount to?” (Manning 2007, 84). Through their service, many soldiers had come to understand that, alongside preserving the Union, the abolition of slavery was the central achievement of the war and gave their sacrifice meaning (Gannon 2011; Janney 2013). Prior to the war, few Union soldiers had first-hand experiences with slavery. Evidence from diaries and letters suggests that when soldiers saw slaves in person, it elicited moral outrage—among other reasons, many were convinced that slavery eroded the virtues and civic institutions necessary for republican government—and roused the strategic recognition that slavery would have to be destroyed to win the war and prevent future conflict (Manning 2007, chaps. 2–4). Thus we find soldiers putting to pen sentiments like this:

the rebellion is abolitionizing the whole army. You have no idea of the changes that have taken place in the minds of the soldiers in the last two months; indeed, men of all parties seem unanimous in the belief that to permanently establish the Union, is to first wipe [out] the institution [of slavery]. (Manning 2007, 45)

As the human cost of the war mounted, soldiers found meaning for the bloodletting in the Christian salvation of the Union through what Lincoln called “a new birth of freedom” brought about emancipation (Gannon 2011; Hunt 2010; Janney 2013; McConnell 1992).

Some historians have questioned whether soldiers sincerely came to support emancipation during the war (Gallagher 2011; White 2014). Yet, there is considerable evidence that this support was present just after the war in the campaign slogans of veterans’ groups during the elections of 1866 and 1868: “We’ll wipe treason out as we wiped slavery’s stain; For traitors and slaves we’ve no place in our land,” “For God and the Union, for Freedom and Right / Let our ballots secure what our bullets have won” (Dearing 1952, 166). Veterans retained this interpretation of the war—“the Won Cause” (Gannon 2011)—decades later: At “Blue and Gray” reunions, Union veterans never relinquished the moral supremacy of their cause, and they ardently disputed the Southern “Lost Cause” narrative of the war, which denied the moral achievement of emancipation and the Civil Rights Amendments (Cook 2021; Gannon 2011; Janney 2013; McConnell 1992). For white Union veterans, “winning the peace” included more than restoring the Union: it also entailed bringing an actual end to slavery.

Collaboration

Third, Union soldiers actively collaborated with enslaved and freed African Americans in their fight against the Confederacy, leading some veterans to acquire new attitudes about race and civil rights (see, e.g., White 2016). Although there is limited evidence that “contact” as such reduces racial prejudice, its effects are stronger when contact is prolonged, socially condoned, and toward a shared purpose (see, e.g., Mo and Conn 2018), as in the case of collaboration between Union soldiers and African Americans against the Confederacy.

Most white Northerners had limited or no interaction with African Americans, whereas Union soldiers who spent time in the South frequently met enslaved and freed African Americans. They presided over “contraband camps” (Hahn 2003), depended on the labor of nearly two hundred thousand freed people working for the Union Army (McPherson 2008, 145), and received vital intelligence on Confederate troop movements from African Americans who risked their lives to help (Hunt 2010). During the war, more than 180,000 African American men served in United State Colored Troops (USCT) combat units (McPherson 2008). White Union soldiers, therefore, either knew of or fought alongside African American regiments. Though some initially opposed the formation of the USCT, this resistance faded (Manning 2007), and some Union veterans came to believe that, through their loyalty on and off the battlefield, African Americans had earned status as citizens (Gannon 2011). One officer exclaimed that “seeing 115,000 colored … soldiers fighting equally … for our common country” made him believe that “the colored man” should be “ELEVATED” (Manning 2007, 192), while another soldier wrote that “the slaves have been our only friends,” which “entitles them to their freedom, or whatever they desire” (Hunt 2010, 95).

These beliefs in equality were persistent, if limited. For decades after the war, the Grand Army of the Republic (GAR)—the largest Union veterans’ organization—in contrast to almost all other social organizations of the time, was racially integrated both nationally and within local posts. During Memorial Day ceremonies, white and Black veterans paraded and attended church together (Gannon 2011). When posts denied admission to Black veterans, they faced censure. Against prevailing segregation, Black and white servicemen were buried together in GAR cemeteries. This inclusion was justified by appeals to the wartime service of African American soldiers. Still, it would be wrong to paint to rosy a picture of this postwar integration: African American members held only symbolic offices (McConnell 1992), and pleas from African American veterans in the South for the GAR to take a stand against the rise of lynching and Jim Crow laws went unanswered (Gannon 2011).

Electoral Consequences

More than 2 million men served in the Union Army during the Civil War, proportionally, the largest mobilization in US history other than the Second World War. Because women were denied suffrage and most Southern states were denied representation in the first peace-time elections, Union veterans constituted an outsized share of the postwar electorate. Under conservative assumptions (see Section E.1 in the Supplementary Appendix), I estimate that, by 1870, nearly 24% of eligible voters in the North were veterans. Because veterans’ understanding of the war resonated with Republican framing of Reconstruction, they proved to be a pivotal constituency for Republicans in postwar elections.

Both Republicans and Democrats sought to win soldiers to their side. At the height of the war, Democratic newspapers framed the conflict as a partisan effort by Republicans that wasted the lives of thousands of soldiers. They made extensive appeals to white supremacy, accusing Republicans of making emancipation the prime goal of the war and arguing that preserving slavery could obtain peace and restore the Union with no further loss of life. By contrast, Republican newspapers railed against rebels and the treason of their alleged Northern allies and portrayed the war as a sacrifice in the service of a noble cause. Nearly all Republican papers endorsed emancipation—though few actually suggested it was the central objective of the war—and most came to embrace military service by African Americans (Kalmoe 2020, chap. 5). In short, Republican framing closely paralleled the understanding of the war that soldiers had acquired, whereas Democratic frames alienated even Democrats in the ranks (Manning 2007, 150–3). Whether Republican frames informed soldiers’ self-understanding or the other way around, Republican campaign messages were far more likely to resonate with soldiers.

In the aftermath of the war, Democrats adopted an electoral strategy of courting veterans by splitting support for the Union from civil rights. They put forward as candidates “War Democrats” who had joined Republicans in 1864 and backed slates of ex-officers candidates in so-called soldiers’ parties, with names like the “Union Anti-Negro Suffrage Party” and platforms that called for payments to veterans and praised white soldiers for saving the Union while denouncing civil rights reforms (Dearing 1952, 66; Dykstra 1993; Field 1982).

By contrast, Republicans in this period earned a reputation for “waving the bloody shirt.” Dearing (1952) recounts numerous cases in which Republican candidates and former military commanders alerted veterans in campaign speeches to the threats posed by their former enemies and called upon them to vote as if they were still an army on the field of battle. Republican newspapers and campaign speeches drew attention to the de facto reimposition of slavery through the “Black Codes,” the election of former Confederates to political office, and violent resistance by Southerners to federal occupation and Reconstruction as evidence of the resurgent political power of the former Confederacy (Riddleberger 1979). Although Republicans, who shared the Democrats’ belief that soldiers harbored racial animus, initially sought to downplay issues of African American rights; this did not last. Civil rights, including the 14th Amendment, were justified to the public as morally right, necessary to ensure a de facto end to slavery and as a way to prevent overrepresentation of the South (Wang 1997).

A sample of 20 Republican newspapers in Wisconsin (Section D.5) shows that, even when national and state leaders kept support for Black suffrage off the party platform in 1865, 90% published overwhelmingly pro-suffrage content and 85% expressed that suffrage had been earned by African Americans through their loyalty, an idea found in 35% of pro-suffrage coverage. Fifty percent of papers justified suffrage as a means of securing war objectives like ending slavery and punishing Southern treason. By contrast, only 35% of Republican papers and 5% of pro-suffrage articles framed suffrage as a strategic necessity to prevent rebels from returning to power.

Moreover, electoral evidence supports the idea that Republican campaign messaging resonated with Union vets. Union Army soldiers voting in 1864 supported Lincoln over his Democratic rival by a margin of 78 to 22, breaking toward Republicans by 25 points more than voters at home (White 2014). Meanwhile, Democatic “Soldiers’ Parties” lost by wide margins in 1865 (Dykstra 1993). Thus, even though Republicans pursued policies that risked alienating many Northern voters, they likely picked up pivotal votes from returning veterans.

Party Activists

Veterans not only came to support the Republican Party but also played important roles shaping the party from within. Over the course of the war, many Democratic office holders who served in the Army became Republicans, and after the war, Union Army veterans filled party caucuses. Out of the 97 legislators elected to the Iowa General Assembly in the fall of 1865, 37 were veterans, 33 of them had never held office before, and 34 were Republicans. But more importantly, veterans’ organizations became a vital part of Republican grassroots mobilization.

Given the repugnance veterans had for Democrats, voting for the Republican Party might have been a choice for the “lesser of two evils.” Yet, veterans’ decisions to join social organizations, where they had more choice, revealed strong commitments to Radical policies. Although many soldiers joined apolitical veterans’ groups, the largest and most important—the Boys in Blue, the Soldiers’ and Sailors’ National Union League, GAR, and the White Boys in Blue—took explicitly political positions on the issues of the war and Reconstruction (Dearing 1952, 80–123). All but one of these major veterans’ organizations backed Radical Republicans.

Of the Union veterans organizations, the GAR was the largest, longest lasting, and most important. Every Republican president from Grant to McKinley was a member, and, at its height in the 1890s, its membership reached 500,000 (McConnell 1992). Founders and leaders of the GAR understood that political mobilization to protect wartime gains was the heart of the organization (McConnell 1992), leading them to rally behind Radical Republicans. This appears to have animated grassroots membership as well: the GAR grew rapidly despite its sharply political reputation; and its leaders and historians attributed the rapid atrophy of the organization after 1868 to flagging motivation after the achievement of Radical Reconstruction (Dearing 1952; McConnell 1992).

During the elections of 1866 and 1868, the GAR and other veterans’ organizations were an integral part of Republican voter mobilization: they canvassed through parades and mass meetings, attended conventions, and maintained campaign clubs (Dearing 1952; McConnell 1992). In September, veterans’ groups from across the North sent 15–25,000 delegates to a convention in Pittsburgh that attacked President Johnson as a traitor, endorsed the 14th Amendment and congressional power over the South, and called for the protection of freed men and women (Cashdollar 1965; Dearing 1952). Newspapers and returning delegates carried this message to voters. James Blaine—a leading House Republican—credited the convention with “consolidat[ing] almost en masse the soldier vote of the country in support of the Republican party” and the ratification of the 14th Amendment (Cashdollar 1965). When the GAR professed partisan neutrality in 1867, its commander—House Republican John Logan—was explicit in private communications: “The organization of the GAR has been and is being run in the interest of the Republican party” (Dearing 1952, 176). And in the 1868 campaign, there was active coordination at the national level between the Republican Party and veterans’ organizations, including the GAR (Dearing 1952).

Veterans’ organizations generated and disseminated discursive frames that memorialized the sacrifices made by soldiers, gave meaning to that sacrifice, and warned of the threat posed by their former enemies. It was the GAR that first popularized the observance and rituals of Memorial Day. During elections, veterans’ organizations warned their members that former Rebels and their Copperhead allies who “yesterday were using the bullet to overthrow the government, [and] to-day … are using the ballot to control it” (Dearing 1952, 150). They called upon their members to win “another victory at the ballot box” to avert this “calamity” (Dearing 1952, 117).

In summary, though a radical faction that embraced abolition and civil rights for African Americans existed in the Republican Party before the war (Bateman 2020), it was not obvious that they could win the support of moderate Republicans or voters. But large numbers of veterans exited the war with a new vision of the Union: where slavery was abolished, traitors punished, and those loyal to the Union, including African Americans, endowed with equal rights of citizenship. Following a logic similar to the process of African American disenfranchisement outlined by Bateman (2018), these veterans and their organizations comprised a new constituency within the Republican Party that made a political coalition for Radical Reconstruction electorally and politically viable.

Testing the Argument

Despite compelling qualitative evidence, there are reasons to take pause. Historical explanations for the electoral success of Republicans in the key elections of 1866 and 1868 make almost no mention of veterans (Foner 1988; Riddleberger 1979). Some historians dispute the claim that soldiers’ views on slavery and race changed during the war (Gallagher 2011), and others argue that the 1864 soldiers’ vote is evidence of partisan selection into service rather than of growing support for Republicans and emancipation (White 2014). Moreover, statistical analyses show that enlistment was higher in states and counties with greater prewar Republican vote share (Costa and Kahn 2008, 52–3; Kalmoe 2020). At the same time, data on the individual attitudes and votes of soldiers and citizens are not available.

To address these concerns, I triangulate evidence from statistical tests of different implications of my argument. First, I examine whether veterans became more likely to vote for the Republican Party after the war at both an ecological and individual level. Then, I explore whether veteran support for Republicans actually reflected agreement on the issues of Reconstruction. Finally, I examine whether constituency pressure by veterans drove Republican legislators to embrace more radical legislation.

Design

Enlistment Rates and Republican Voteshare

I first show that, at an ecological level, higher enlistment rates in the Union Army caused an increase in voting for Republicans. I identify this effect using a continuous difference-in-differences (DD) design (Angrist and Pischke 2008, 234–5). Here, the “treatment” is wartime enlistment rates in a county. The first difference compares counties before the war (when no one was enlisted) and after the war starts (when enlistment happened and the “treatment” occurred). Unlike binary DD estimators, all counties had some enlistment and enlistment rates are continuous. Thus, I estimate the effect of differing “intensities” of enlistment on the within-county change in support for Republicans, addressing concerns of partisan selection into service. This estimator is given in Equation 1.

(1) $$ {\displaystyle \begin{array}{l}\mathrm{Republican}\;{\mathrm{voteshare}}_{ie}\\ {}\hskip1em ={\displaystyle \begin{array}{l}{\alpha}_i+{\alpha}_e+\beta \mathrm{Enlistment}\;{\mathrm{Rate}}_i{\operatorname{}}^{\ast}\mathrm{Civil}\;{\mathrm{War}}_e+{\varepsilon}_i+{\varepsilon}_y.\end{array}}\end{array}} $$

In this equation, the subscript i denotes the county, y the year, and e the state-election (for instance, the Massachusetts congressional elections of 1860). Civil War _e is a dummy variable that is 1 if the election occurs in 1861 or later and 0 otherwise. Enlistment Rate _i is the fraction of military-aged males in a county that served in the Civil War (between 0 and 1), α_i is a county-level fixed effect, and α_e is a state-election fixed effect, which imposes the assumption of parallel trends within states. Enlistment Rate _i is constant within counties, and Civil War _e is constant within state-elections, so only the interaction remains in the equation and is captured by the parameter β (Angrist and Pischke 2008, 233–4). Errors are clustered by both county and year. I estimate this equation using elections between 1854 and 1880, the first year in which the Republican Party contested elections, and the first presidential election after the conventional “end” of Reconstruction.

In SA Section A, I show that under very similar assumptions as the familiar binary DD estimator, this estimates the least squares linear approximation of the (potentially nonlinear) average causal response function of Republican vote share across different levels of enlistment. These assumptions are (1) parallel trends in Republican vote share among counties within the same state across different levels of enlistment and either (2a) within states, the effect of enlistment rates is not heterogeneous or that heterogeneity is independent of enlistment rates or (2b) there is no confounding of the selection into enlistment rates.

Combat Experience and Individual Partisanship

These ecological effects could be consistent with increased Republican voting among nonveterans in high-enlistment areas rather than among veterans. ² To demonstrate that the war caused soldiers to vote Republican, I also examine the effects of service on individual partisanship.

It would be natural to compare the partisanship of those who served versus those who did not, but this is fraught with problems: (i) The effect of enlistment on partisanship is likely to be confounded, and data on conditioning variables for this period are limited and noisy. (ii) Although there were “random” draft lotteries during the war, few men were actually drafted, there was extensive two-sided noncompliance, and uncovering the pool randomized into service would be extremely difficult. (iii) Wartime service drastically increased postwar mobility, producing differential attrition. Solutions for this attrition would involve conditioning, negating the advantages of the lottery. ³

Instead, I examine the intensive effects of wartime sacrifice on soldiers. Not only does this permit estimating effects on all, rather than a small subset, of soldiers; it tests a primary mechanism in my argument: soldiers who made greater sacrifices should have been more motivated to secure the victory won in the war, making them more receptive to Republican framing of Reconstruction and thus more likely to vote Republican.

I operationalize wartime sacrifice as exposure to combat casualties and identify its causal effect on the postwar partisanship of individual soldiers using a natural experiment. Exposure to combat was driven by selection at several levels. First, men chose when to enlist, their term of enlistment, the kind of unit (infantry, cavalry, or artillery), and their regiment (Costa and Kahn 2008, 52–7). Second, once in their units, soldiers could desert, seek transfers, and decide whether to reenlist. Finally, Army commanders could assign units to different theaters of operation and to more or less risky duties depending on their experience, morale, and reliability. At each stage, soldiers’ partisanship may have driven selection (Kalmoe 2020).

I address this selection problem by exploiting variation in exposure to combat deaths among men who joined the same infantry regiments at the same time. The vast majority of men serving in the Union Army were in the infantry. The basic unit in which infantrymen were mobilized, maneuvered, and went into combat was the regiment. When first organized, regiments consisted of approximately 1,000 men, but by the middle of the war, regimental strength was considerably lower (Hess 2015). Regiments were further subdivided into 10 companies of equal size. Men in the same company were mustered in together and then trained, lived, worked, and fought directly alongside each other (Costa and Kahn 2008). Although the deaths of any man in their regiment likely affected soldiers, deaths of men within the same company were undoubtedly more meaningful: they were likely to have known the fallen personally, possibly even before the war, and to have witnessed their deaths.

Variation in company-level combat casualties for men serving in the same infantry regiment was plausibly random. Men in the same regiment made the same decision on when and how they would serve and traveled through and fought in the same places. In battle, infantry regiments typically formed up in a line of men, two deep, horizontally arranged by company, approximately 140 yards across (see Figure B1; Hess 2015). In the chaos and smoke of battle, which companies in that line received more casualties was effectively arbitrary. ⁴

Nevertheless, the number of combat casualties seen by a soldier was also determined by how long they stayed with their unit. This could be affected by choices a soldier made to transfer or desert or that led to promotion, injury, or death; all of which could be driven by differences in partisanship. Thus, I construct an “intent-to-treat” measure of exposure to combat deaths.

(2) $$ \mathrm{Company}\;{\mathrm{Casualties}}_i = \sum \limits_j^n{KIA}_j\cdot \left(i\ne j\right)\cdot \left({t}_i^{\ast}\cap {t}_j\ne \varnothing \right). $$

I construct the company-level casualties treatment for soldier i as follows. Soldiers j through n–1 are all other men who ever served in the company c to which soldier i was first assigned. The variable t_j is the set containing all dates that soldier j served in company c, based on muster records. KIA_j is an indicator for whether soldier j died at the hands of the enemy: in battle, of battle wounds, or as a prisoner of war. To remove bias induced by the selection process determining whether soldier i stayed or left the company, $ {t}_i^{\ast } $ is the set of dates soldier i should have served in the company based on the date of his muster and the term of enlistment. Thus, Company Casualties _i is the number of men who soldier i was assigned to serve with who died as a result of combat.

I estimate the effect of company casualties using the following regression model, where i is a soldier, α_r is a fixed effect for men joining the same regiment in the same year, and ε_c is an error shared by men who joined the same company in the same year. In robustness checks, X _i is a set of covariates for soldier i drawn from their military records and the 1860 US Census.

(3) $$ \mathrm{Partisanship}\;{1874}_i={\alpha}_r+\mathrm{Company}\;{\mathrm{Casualties}}_i+{\mathbf{X}}_{\mathbf{i}}+{\varepsilon}_c. $$

Although military records and the US Census provide many characteristics on which to check balance, they lack partisanship. To fill the gap, I trained and validated a machine-learning classifier that predicts the baseline partisanship of soldiers based on demographic characteristics of name, birth year, and birthplace (See Section B.4). I show that companies with greater casualties are balanced in mean predicted partisanship and individual treatment is balanced on predicted partisanship, demographic features, education, household composition, and property ownership (See Section B.5).

Data

Enlistment Rates

I measure enlistment rates using a novel database of Civil War soldiers: the American Civil War Research Database (ACWRD). Drawing on unit histories and official records, it links data on individual soldiers, military units, and engagements. For seven states (Illinois, Iowa, Wisconsin, Massachusetts, Vermont, Maine, and Connecticut), it is possible to link more than 90% of soldiers to their residence at the time of enlistment. ⁵ I then match these residences to counties in 1860. I compute the county-level enlistment rate by dividing the number of soldiers in a county by the number of military-aged males (between the ages of 10 and 39) present in the county in the 1860 Census. For a more detailed exposition and validation of this data, see Section D.3 and Kalmoe (2020).

Elections: Republicans: I measure support for the Republican Party using congressional and presidential election returns. I collapse county-level returns from United States Historical Election Returns, 1824-1968 to 1860 boundaries, using areal interpolation (ICPSR 1999), and then calculate Republican vote share.

Elections: Black Suffrage: I measure direct support for Reconstruction policies using votes in state referenda to remove the word “white” from the qualifications for suffrage in the state constitutions in Iowa (1857 and 1868) and Wisconsin (1857 and 1865). Other states held similar referenda, but either lack pre- and postwar votes or residences for soldiers. I draw on data compiled by Dykstra (1993) and McManus (1998) to calculate pro-suffrage votes in clusters of townships for which the boundaries were stable and election results are available in both pre- and postwar referenda. (See Section D.2)

Covariates: I collect a battery of economic, demographic, and political covariates for counties from the 1860 Census and ICPSR (1999). Please see Section D.1 for a full list of these measures.

Service and Individual Partisanship

First, I measure individual postwar partisanship using county directories for nine counties in Indiana produced in 1874 (see Section B.3.1 and [DeCanio 2007]). These directories contain the names, ages, birthplaces, year of move into a county, and party affiliations of most adult men residing in those counties. I measure partisanship as Democrat = 1, Republican = 1, and the partisan swing (Republican – Democrat / 2). Second, I used the ACWRD to identify 21,301 soldiers from Indiana regiments who listed residences in those counties. I then linked these soldiers to the first company in which they served and calculated Company Casualties _i as defined above. Third, I linked these soldiers to the 1860 Census using the fastLink algorithm (Enamorado, Fifield, and Imai 2019), blocking on county of residence at enlistment and matching on name and birth year. This enables me to include prewar covariates and focus my analyses on a set of 10,358 soldiers that are “findable” pretreatment. Finally, I linked the soldiers to the 1874 county directories, matching on county, name, and birth year. I was able to find 3,914 soldiers after the war, 3,264 of whom were located in the 1860 Census. For greater detail on these data and the matching procedures, see Section B.3.

Veterans and Voting Republican

Enlistment Rates and Republican Gains

Did counties with higher enlistment rates vote more strongly for Republicans during Reconstruction? Table 1 reports the estimates of the difference-in-differences from Equation 1. Column (1) shows the main result: the shift toward Republicans was greater after the Civil War in places with more enlistment. The effect size suggests that a 10-percentage-point increase in enlistment would yield a 4.2-percentage-point (p < 0.0001) increase in Republican vote share. ⁶ This estimate is robust to alternately excluding all counties in which Republicans ever failed to contest an election (Column 2) or including a dummy for county-elections that Republicans did not contest (Column 3), measuring enlistment as either total enlistment or surviving veterans (Table A2), and dropping individual states (not shown). Furthermore, the effects of enlistment were higher in counties where the average soldier experienced higher regimental casualties, which is consistent with my argument that sacrifice made soldiers more receptive to Republican messaging (Figure A6).

Table 1. Difference-in-Differences Estimate of Effect of County Enlistment Rate on Republican Vote Share

Note: Data from congressional and presidential elections across 384 counties between 1854 and 1880. Standard errors clustered by county and election year. Counties with election cycles in which the GOP does not contest an election cycle either treated as a 0, the election is marked with a dummy, or the all observations for that county are dropped. *p < 0.05, ^** p < 0.01, ^*** p < 0.001.

To interpret this as the causal effect of enlistment rates, we must believe two of three assumptions. First, counties with different rates of enlistment should have parallel trends before the war. Figure 1 shows the raw Republican vote share in elections between 1856 and 1880 averaged by enlistment quartiles. Across the three election cycles prior to 1861, all four quartiles had parallel trends. Figure A1 formally tests the slope on enlistment rates and the difference in GOP vote share between 1860 and every other election from 1854 to 1920. In prewar elections, these differences are not significantly different from 0.

Figure 1. Republican Voteshare Trends by Enlistment Quartile

Note: Panel A shows the unadjusted trend in county-level Republican vote share averaged by within-state enlistment quartile. Trends start in 1856, the first year in which Republicans contested elections in all eight states. Panel B shows the same data, subtracted from the quartile average in 1860.

Second, it could be that baseline differences between high and low enlistment counties affected how they responded to the war, producing time-varying confounding. Figure A2 shows the within-state relationship between enlistment and 32 prewar demographic, economic, and political covariates. Demographically, enlistment was higher in counties with smaller populations, fewer white people, more southern-born, and more men. Economically, enlistment was higher where agricultural output was less and more people worked in manufacturing. Politically, enlistment was higher where the Republican Party and its antecedents had performed worse. Yet, the results are substantively the same when including interactions between each covariate and the postwar indicator (see Figure A3 and Table A1).

Third, and alternatively, if the effects of enlistment were heterogeneous across counties with different levels of enlistment, estimates might be biased. I address this concern by flexibly modeling heterogeneity in the DD effect of enlistment for each county, weighting cases based on similarity in prewar covariates. The average partial effect of enlistment across all counties is substantively the same (See Section A.4).

Combat Deaths and Individual Partisanship

Table 2 reports the effects of exposure to company-level combat deaths for soldiers who were linked to the 1860 Census. In baseline models (columns 1–3), which include only regiment fixed effects, I estimate that one additional casualty in a company ⁷ increased support for Republicans by 0.9 percentage points (p = 0.025), decreased support for Democrats by 1 percentage point (p = 0.005), and swung the margin toward Republicans by 1.9 percentage points (p = 0.008). These results remain virtually unchanged when including covariates drawn from enlistment records (columns 4–6) and when further adding Census covariates (columns 7–9). ⁸ Consistent with the idea that support for Republicans was tied to viewing emancipation as an achievement of the war, the effects of casualties were larger for soldiers who enlisted after the Emancipation Proclamation (See Figure B14).

Table 2. Effect of Company Casualties on Postwar Partisanship (Census-Linked, Best Matches)

Note: Sample includes men serving in Indiana Regiments who were matched to the 1860 Census and their best match, if any, in the 1874 People’s Guides. Baseline and control models, respectively, include data on 3,105 individual soldiers, serving in 538 companies, across 211 regiments. Regiment fixed effects includes a dummy for each group of soldiers who joined a regiment in the same year. Standard errors are clustered by company. *p < 0.05, ^** p < 0.01, ^*** p < 0.001.

These results are robust to including all soldiers or only those located in the Census, using all plausible matches or only the best matches in 1874, and measuring the treatment as company combat deaths or combat death rates (See Section B.6). Moreover, although the majority of soldiers cannot be located postwar, differences in attrition across levels of treatment are small and unrelated to predicted prewar partisanship and treatment effects are unchanged when reweighting soldiers based on their inverse probability of being found in 1874 (See Section B.7).

These effects are remarkable given that (i) the dependent variable is measured with error, increasing standard errors; (ii) this identification strategy nets out variation in regimental combat experiences and exploits only “intended” exposure; and (iii) these effects persist in 1874, which was nearly 10 years after war and a year of tremendous electoral losses for Republicans.

Interpretation

Taken together, these analyses, alongside the historical evidence, demonstrate that wartime service turned Union soldiers from Democrats into Republicans. These wartime changes resulted in large and sustained increases in support for Republicans at the polls, particularly in vital elections that determined the fate of Reconstruction.

It is difficult to dismiss these results as the result of selection by Republicans or racial liberals into service. There is no reason to believe there were large numbers of racial liberals outside of the Republican Party who could have flocked to the party afterward. Nor can it be that racial liberals enlisted at higher rates and remained loyal to the party, whereas moderates and conservatives defected over Radical policy changes. Republican vote shares in 1864 through 1868 increased over 1860, but they increased more in places with higher enlistment.

Even the individual effects are large. Back-of-the-envelope calculations imply that decreasing or increasing combat deaths experienced by Indiana soldiers by one (within-regiment) standard deviation could have shifted the results of the 1874 congressional elections in Indiana (in reality, an 8-5 Democratic majority) to a 10-3 or a 7-6 Democratic majority (see Section E.2).

Between the changes in veterans themselves and their campaign mobilization, they provided Republicans the votes they needed to pursue and secure their Reconstruction agenda.

A Constituency for Reconstruction

Even if veterans voted Republican, does that mean they were a constituency for Reconstruction? Or did they vote Republican for other reasons?

Patronage? Support for the Republican Party may have been driven by patronage to soldiers and officers that overrode racial prejudice and opposition to legal equality. Skocpol (1993) argues that, starting in the 1880s and 1890s, the first major social welfare program in the United States transferred vast sums of money to Union Army pensioners and was managed by Republican appointees as a vote-buying scheme. But in the era I discuss, only a tiny fraction of veterans received pensions, the amounts paid were smaller, and the partisan divide over pensions did not arise until decades later.

Lesser Evil? It could have been the case that Union veterans had no option but to vote Republican, despite repugnance toward African Americans and expanding civil rights. Democrats may have been tainted by association with Southern Democrats who led secession and Northern “Copperheads” who had opposed the war and pushed for a negotiated peace with “traitors” (White 2014). Yet, as the failure of “soldier’s parties” to attract support and the overwhelmingly Radical hue of veterans’ organizations suggest, veterans were interested in securing many of the same goals as Republicans.

Voting for Suffrage

The extension of suffrage to African Americans was one of the most radical reforms adopted during Reconstruction. This makes suffrage a strong litmus test of increased veteran support for the policy agenda of Radical Reconstruction. I estimate the effect of military service on voting for suffrage in pre- and postwar referenda in Iowa and Wisconsin. Table 3 reports the result of difference-in-differences and lagged-dependent variables ecological regressions. Both designs yield virtually identical results: veteran support for suffrage increased by 31 and 32 percentage points, compared with those who remained home (p < 0.001). These results persist when conditioning on the fraction of people eligible to vote in the prewar election and restricting the sample to townships with smaller population increases between 1860 and 1870. In Table C1, I restrict the analyses to Wisconsin. Even when controlling for the fraction of Republicans who did not support suffrage in 1857 and adding county fixed effects, townships with more enlistment saw greater increases in support for suffrage.

Table 3. Effect of Enlistment on Support for Black Suffrage (Iowa and Wisconsin Township Returns)

Note: Enlistment rate is number of men serving over those eligible to vote in 1865 (WI) or 1868 (IA). Suffrage vote totals come from state constitutional referenda in stable clusters of townships/counties in Iowa and Wisconsin. All models include state fixed effects. Lagged dependent variables are allowed different slopes by state. Control variables include (i) the fraction eligible to vote in 1857 over those eligible to vote in the postwar referenda. Townships are weighted by number of white men. The restricted sample includes only townships where the population eligible to vote in 1865/1868 changed by less than 50% between 1860 and 1870. Standard errors are robust. Full results are reported in the Supplementary Tables. *p < 0.05, ^** p < 0.01, ^*** p < 0.001.

Although ecological regression is notorious for its limitations, plausible conclusions can be reached with care (Section C). The central problem with ecological regression is that there may be “contextual” effects: in our case, the change in support for suffrage among soldiers and civilians may differ across areas with different enlistment rates. Aggregation makes it impossible to identify this pattern, potentially biasing ecological estimates. I follow Jiang et al. (2020) and partially identify contextual effects and the individual effects of service. For the effect on veterans to be nonpositive, the contextual effects must have been very large, and focusing on townships in which prewar support for suffrage was close to zero, the effect of service on veterans is mathematically bounded above zero (Section C.3).

Nevertheless, it could be that veterans were different in ways other than military service that made them more likely to become supportive of suffrage after the war. The most plausible source of this bias would be partisanship. Republicans may have been both more likely to enlist and, either because party elites endorsed suffrage or ongoing activism by pro-suffrage radicals within the party, more likely to become pro-suffrage. There are a few reasons to doubt this. First, effects of enlistment on suffrage are robust to conditioning on prewar suffrage support and pro-Republican/antisuffrage vote share in 1857, as well as restricting the analysis to townships in which Republicans and suffrage received support from less than 1% of voters in 1857 (Figure C3). Second, the Wisconsin referendum took place in November 1865, well before national Republicans endorsed suffrage. In Wisconsin, the suffrage question divided Republicans. Led by one of the state’s Republican Senators, the party convention did not endorse suffrage; the leading Republican paper stated suffrage was a “minor issue” and support for it was not the “standard of party orthodoxy,” and statements to the contrary appeared in only 1% of Republican coverage of the issue. The Republican gubernatorial candidate repeatedly refused to take any public stance on the suffrage referendum (Fishel 1963; McManus 1998). Instead, as I report above, Republican newspapers, like many veterans, argued for suffrage based on African American loyalty during the war.

Mobilizing for Suffrage

Yet veterans did not just passively receive messaging from party activists. There is preliminary evidence that veterans as constituents pushed Republican legislators toward more expansive Reconstruction goals.

In March 1866, legislators in the Iowa General Assembly voted whether to remove racial qualifications on rights, including suffrage, from the state constitution. Although Republicans had won the preceding fall on a platform that included Black suffrage, they had considerable trepidation about following through: party leaders did not want to appear out of step with President Johnson, who was against suffrage (Cook 1994). Iowa legislators had the choice to support qualified suffrage for African Americans, universal suffrage for men, and whether to eliminate all racial restrictions, including those on holding office. Even conditioning on prewar support for Republicans and Black suffrage in their constituencies, Republican legislators who had more veteran constituents were significantly more likely to endorse universal over limited suffrage for African American men and, though not significant, more likely to endorse the most radical position of eliminating restrictions on office holding (See Table C2).

Conclusion

Taken together, the evidence presented above tells a clear story. Reconstruction has often been portrayed as an archetypal case of suffrage extension driven by the strategic needs of political parties (Valelly 2004). Yet, despite clear incentives, opposition from white voters and party factionalism could have rendered Reconstruction stillborn. It was therefore decisive that Union veterans exited the war with a sense that the Union must be remade, not just restored. They placed the destruction of slavery at the center of their wartime victory and became receptive to a vision of national membership that incorporated, in limited ways, African Americans. As a consequence, they emerged as an important new constituency that provided pivotal support as both voters and activists for Reconstruction.

This has clear and important implications for the history of the American Civil War and Reconstruction. In congressional elections of 1866, Republicans won a supermajority that empowered them to pursue equal suffrage laws and enabled them to override Johnson’s veto. Historians documenting this consequential election make little to no mention of veterans (Foner 1988; Riddleberger 1979). Yet, I show that Republicans benefited immensely from veterans’ votes. The pivotal seats needed to secure their supermajority were won by margins of 4 points, which could have been more than achieved by a one-standard-deviation increase in enlistment. And because enlistment was higher where Republicans had performed poorly before the war, these gains likely translated into more seats.

Historians have fiercely debated whether diaries show that Union soldiers came to support emancipation and the Republican Party (Manning 2007) or that their politics were unchanged (Gallagher 2011; White 2014). This paper presents decisive systematic evidence of veterans’ political transformation.

In political science, this finding contributes to a growing body of work that takes up Mayhew’s (2005) charge to delineate the political consequences of the American Civil War (e.g., Kalmoe 2020). War changed white Northern men who served, and in so doing, it shored up a coalition for Radical Reconstruction. This also addresses comparative research on the effects of war on the politics of combatants (Grossman, Manekin, and Miodownik 2015; Jha and Wilkinson 2012; Koenig 2020). In a departure from previous findings (though see White 2016), I’ve shown that wartime sacrifice and antipathy toward a shared enemy can induce combatants to support policies to include allies belonging to a racial out-group, even in the face of widespread racism.

More fundamentally, this paper adds to a recent reevaluation of the political development of race and civil rights in the United States. In conjunction with Bateman (2020), evidence of ideological transformation among white Union veterans and its political consequences provides an important corrective to elite-driven competition explanations of rights extensions during Reconstruction. The importance of bottom-up mobilization, alongside elite party interests, to the passage of Reconstruction parallels recent work showing that it was the local-level incorporation of African American voters and mobilization by labor unions that pushed Northern Democrats to embrace civil rights in the twentieth century (Schickler 2016). In both Reconstruction and the Civil Rights era a century later, African Americans’ advocacy and activism were indispensable (Valelly 2004). Yet this activism depended on a political opportunity structure constrained by the nature of the white electoral coalitions that supported their claims, in limited and flawed ways, to equal citizenship.

More work can be done to systematically probe the nature of veterans’ support for Reconstruction and to understand how it shaped the legislation that was passed. Which Reconstruction issues most animated veterans and their organizations? Did legislators with veteran constituents endorse more expansive civil rights and enforcement legislation, or limited reforms? And in light of the reversal of Reconstruction in the subsequent decades (Foner 1988), how durable was veterans’ support for civil rights? Despite the obstruction of Southern congressional delegations, Republican efforts to secure civil rights did not end in 1877 (Bateman, Katznelson, and Lapinski 2018; Wang 1997). Although some of these efforts were sparked by the electoral needs of the party, Republicans in several Northern state legislatures passed equal accommodation laws after the Supreme Court struck down much of the 1875 Civil Rights Act (Johnson 1919).

Were these efforts animated by veterans? And was Republican abandonment of African American voting rights in the 1890s enabled by growing disinterest among veterans? Or might it have been that aging veterans no longer provided pivotal votes, even as thousands of them joined an early civil rights organization and expressed their anger and dismay that Republican failures were “responsible for the condition that the colored citizen [is] in the South to-day” (Cook 2021, 18)? These questions deserve investigation from the starting point of the perspective, provided here, that the transformation of white constituencies, whether veteran or otherwise, cannot be explained with instrumental party-competition narratives alone. An earnest understanding of the racial history of the United States requires that we take seriously the ideological changes of tide that have fed in and out of the pivotal moments, such as the Civil War and Radical Reconstruction, and that still define our politics today.