The Civil War

The Anglo-Irish Treaty signed on December 6, 1921 created the Irish Free State, constitutionally a British dominion, but possessing considerable legislative powers. The Northern Ireland parliament was given the choice of opting out of the new state: It did so on December 7, 1922. To protreaty nationalists, the treaty brought an end to the Act of Union between Britain and Ireland (1800–1921) and would provide a basis for evolution. Antitreaty republicans saw in the treaty the disestablishment of the 32-county Irish Republic proclaimed during the Easter Rising of 1916 and accused those who signed it of betrayal. The treaty was narrowly approved by the Irish parliament, Dáil Éireann, on January 7, 1922. The independence movement had split behind two dominant leaders, protreaty Michael Collins, and Éamon de Valera, President of the Republic proclaimed in 1916. Yet the public was in favor of peace. The occupation by the antitreaty IRA (henceforth “the IRA”) of a number of buildings, including the Four Courts, in Dublin on March 12, 1922 posed a clear challenge to the provisional government and to majority rule. After an election on June 16—and an ultimatum from the British government—they attacked the Four Courts on June 28. What ensued was a war between protreaty and antitreaty republicans for de facto control over the new state. Yet it was also a war about national legitimacy, about which side could define themselves as most Irish.

Figure 1 shows the border between Northern Ireland and the Free State. Because of partition in 1920 Northern Ireland’s security forces had been suppressing the IRA, so the focus of the IRA was southerly. After their defeat in Dublin in early July it retreated south, where it tried to defend fixed positions behind the line on figure 1 running from Limerick in the west to Waterford in the southeast. Yet figure 1 also shows that between July 24 and August 8 government troops landed in Clew Bay in county Mayo, Fenit and Tarbert in Kerry, and Youghal and Passage West in Cork. These naval landings established entry points behind the Limerick–Waterford line and would separate IRA brigades in the southwest, west, and northwest. The line collapsed. On August 19, IRA Chief of Staff Liam Lynch told his men to organize themselves into “active service units” and adopt guerrilla tactics.

Figure 1.

The Limerick–Waterford line and places of naval landings, July 24, 1922 to August 8, 1922.

The war had become more diffuse in two ways: first in traveling to a new location after the surrender of the Four Courts, and second in terms of expanding in scope with the collapse of the Limerick–Waterford line (Schulte and Weidmann Reference Schutte and Weidmann2011: 144, 152). As in conventional civil wars, the conflict first became more diffuse by being relocated. With the guerrilla war, it did so by escalating. As the IRA retreated to its most familiar environs, this led to attacks behind their lines and more localized executions. IRA Commander Tom Maguire later reflected that “I was concerned very much by what you termed fragmentation, by the effort to travel around, make contact, and hold our groups together” (Ó Gadhra Reference Ó Gadhra1999: 139). All the executions were carried out in the second half of the war (November 17, 1922 to May 30, 1923). They took place in 17 counties of men from 21 of the 32 Irish counties. Both in terms of the methods used to win it and the IRA’s campaign of destruction, the conflict escalated as it became irregular.

The island’s geography had traditionally posed problems for consolidating central authority. Yet the IRA had made a fundamental mistake by leaving the capital in the hands of their enemies, allowing the provisional government to broadcast themselves to the outside world as the lawful government in overall control of the situation (Blake Reference Blake1986: 36). With Dublin as its epicenter, power was projected north, south, and west as it had been in all previous efforts to subjugate the island. But this was not a half conquest; the writ of the state soon extended into every region of the new state. The Free State would intern more than double the number of people the British had done in the earlier conflict (1919–21). Patrick Mangan was a railway worker executed at the Curragh army base on December 19, 1922. His father went insane as a response and was committed to an asylum where he died two years later. His brother Thomas, who also worked on the railways, was interned for more than a year, while his sister, a member of Cumann na mBan (the IRA’s sister organization), lost her job in a local Christian Brothers School.Footnote ¹

Civil wars need not become protracted if the early fighting is for the capital, and if external resources are placed at the state’s disposal early on (Herbst Reference Herbst2004: 367). By September 1922, Britain had supplied 27,400 rifles, 6,606 revolvers, and 246 Lewis guns to the provisional government (Hopkinson Reference Hopkinson1988: 125). Moreover, the public became alienated from the IRA when its guerrilla tactics subverted the economic life of the country by destroying roads, bridges, and railway lines. An army report in March 1923 stated that armed opposition could only be found in a few places, and where IRA columns existed it was mainly due to the mountainous terrain.Footnote ² The IRA declared a unilateral ceasefire on April 30, 1923. They buried their arms and returned home: without a formal surrender.

The Spread of the Executions

As a deterrent the executions prevented the further assassination of members of the parliament. As punishment they allowed the army to punish local IRA men for their actions or those of their comrades. However, the executions also demonstrated (symbolically) the commitment of the treatyite elite on the standing of authority in Ireland and broadcast its “frame of reference” or “code of power” (Parsons Reference Parsons and Eckstein1964: 35). For the provisional government they were “an extreme but morally justifiable response to an IRA campaign that sought to overthrow a lawfully constituted state” (Gavin Foster Reference Foster2015: 156). Antitreatyites countered that the Free State bowed to force majeure. One republican poster—“The British Empire Unlimited”— compared the position of two Irish ministers, Kevin O’ Higgins and Richard Mulcahy, in a downward chain of authority, to the roles Generals Sir Hamar Greenwood and Nevil MacReady played in the executions after the 1916 Rising.Footnote ³

The executions are no longer a taboo subject: Most studies of the Irish revolution discuss them. There are two legal histories (Cambell Reference Cambell1994; Enright Reference Enright2019) and one PhD thesis (Breen-Murphy Reference Breen-Murphy2010) on the policy. The civil war saw the loss of leaders on both sides, such as Michael Collins and Liam Lynch and historians of the executions stress elite psychology. They speak of “bitterness” (Valiulis Reference Valiulis1992: 184), “hatred” (Coogan Reference Coogan2006: 141), “ruthlessness” (Ferriter Reference Ferriter2015: 281), “fratricidal violence” (Townshend Reference Townshend2014: 451), “savagery” (Neeson Reference Neeson1989: 277), and “terror” (Coogan and Morrison Reference Coogan and Morrison1998: 4). Some ask whether the restraint necessary for the existence of civilized life had also loosened (Roy Foster Reference Foster2015: 281; Garvin Reference Garvin1996: 93; Townshend Reference Townshend2014: 441).

Less attention has been paid to geography. The executions formed part of “a war of reconquest” (Keogh Reference Keogh1994: 15–17) and the state’s territorial imperative explains more about their diffusion than elite politics. Indeed mapping their spread in 1922–23 is a way of tracing the expansion of state power. The first phase was from October to December 1922, when the executions were decided at the national level. The strategic hub of any state-building project was first Dublin and the Central Lowland lying north of a line from Galway Bay to Dublin and south of the drumlin belt (north of which lay Northern Ireland). The first step in the consolidation of state authority was to control this historic “eastern triangle”: the geographical nucleus “from which men have their best chance of commanding the whole country” (Andrews Reference Andrews, Moody and Martin1967: 21). The first 29 executions of the civil war took place within this eastern triangle. Figure 2 shows their relatively central location.

Figure 2.

Places and numbers of executions, October to December 1922.

Sources: Executions by Provisional or Saorstát Éireann Governments, 1922–24 (TSCH S 1884 A). National Archives Dublin; (P150/1657), Eamon de Valera Papers, UCD Archives, Dublin; (TCD 7808/324) Erskine Childers Papers, Archives and Manuscripts, Trinity College Dublin.

The second phase began on January 11, 1923 when the decision was taken to decentralize responsibility to committees of officers at the battalion level, resulting in more localized executions. Twenty-one men were executed in 1922, a further 60 in 1923. Figure 3 shows their territorial reach. Between November 17, 1922 and April 26, 1923, 35 executions took place in the eastern command area, 20 in the western command area, 19 in the southern command area, and 7 in the Curragh.Footnote ⁴ Indeed the new state executed many more IRA prisoners in a year than the British Army had done between 1916 and 1921. The latter were concentrated in Dublin. The Free State was more determined to dominate rural areas. When the IRA split over the treaty, the First Western division was the only one along the western seaboard not to go antitreaty: protreaty strength was concentrated on the east coast, the midlands, and the border counties. Before January 11, no executions had taken place in the western counties of Clare, Galway, Kerry Limerick, and Mayo: Subsequently, there were 20.

Figure 3.

Places and number of executions, January to May 1923.

Sources: Executions by Provisional or Saorstát Éireann Governments, 1922–24 (TSCH S 1884 A). National Archives Dublin; (P150/1657), Eamon de Valera Papers, UCD Archives, Dublin; (TCD 7808/324) Erskine Childers Papers, Archives and Manuscripts, Trinity College Dublin.

One can also identify a third stage when local factors influenced their location within counties. When General Paddy O’Daly became commanding officer of the National Army in Kerry he decided to have four men from different areas of the county executed as a deterrent against further IRA activity. The men chosen to face the firing squad on January 20, 1923 at Ballymullen Gaol Tralee were from Ballycarbery, Castleisland, Causeway, and Knockeendubh (near Killarney) (Horgan Reference Horgan2015: 291). In county Tipperary of the same month, four IRA men who had been sentenced to death were transferred from Templemore, where the people were reportedly angry, to the protreaty town of Roscrea. They were executed at Roscrea Castle on January 15.

The executions on December 8 had broadcast state power territorially; Richard Barrett was from Munster, Liam Mellowes (T.D.) represented Connaught, Joe McKelvey was from Ulster, and Rory O’Connor was from Leinster. However, at the decisive meeting of the cabinet and the army council on January 11, Minister for Home Affairs Kevin O’Higgins complained that the psychological effect of an execution in Dublin is very slight in Wexford, Galway, or Waterford (O’Halpin Reference O’Halpin1999: 32). By the end of the civil war, executions had taken place in the four corners of the state: counties Donegal, Kerry, Louth, and Wexford. Republican historian Dorothy Macardle (Reference Macardle1937: 863–4) first noted the territorial logic. January 1923 saw executions take place in every direction, north, south, east, and west: 3 in Dundalk on the 13th, 4 in Roscrea and one in Carlow on the 15th, and 11 (2 in Limerick, 4 in Tralee, and 5 in Athlone) on the 20th. She claimed that victims were chosen from the center of each district where the IRA was operating with success.

Broadcasting state power over territory has been a problem for many new states, especially those with huge territories, dispersed populations, and culturally distinct groups (Herbst Reference Herbst2004: 36). In Ireland, the obstacles were the local power of the IRA and the public’s alleged lack of civic virtue in not supporting the authorities. The executions broadcast power across three types of boundary. The territorial boundary refers to the way they were spread throughout the territory of the state, making the presence of the state felt at the local level. In terms of social boundaries, those killed were mainly young working-class men, adding credence to the view that the republican cause was “back to the men of no property.”Footnote ⁵ In terms of organizational boundaries, most protagonists had been members of the same organizations when the civil war started. The executions constituted a point of “no return” because they demonstrated that the government was willing to back its own propaganda with ruthless measures against its former comrades.

Given the bitterness of the treaty split, it is not hard to explain why the executions were introduced: Explaining their spatial diffusion is the purpose here. I first trace the origins of the policy to elite politics. Then I consider what guerrilla war conditions explain about their diffusion. Finally, because commanders on the ground had such a say in 1923, I ask how much their changing relationship with Dublin also tells us.

“Fratricidal Flipping,” Centralization, and Elite Politics

Staniland (Reference Staniland2012) suggests that “fratricidal flipping” may begin when, during a peace process, some leaders of an insurgent organization decide that the centralization of authority is necessary to achieve their goals. He relates “fratricidal flipping” to the process of “ethnic defection” from an insurgency to a state. By “ethnic defection” he means that some insurgents go over to the state’s side and “flip” in their allegiances by turning against former comrades. The decision to sign the treaty was not ethnic defection—the protreatyites were nationalists who valued full Irish independence—but their acceptance of the treaty raised similar accusations “of betrayal, of selling out, or becoming puppets in the hands of the state (Staniland Reference Staniland2012: 18) .” The choice of some IRA men to turn their guns on their former comrades, and their possession of intelligence later had a decisive impact militarily.

Staniland land outlines three scenarios: (1) ex ante political disagreements will drive defection by moderates, particularly during peace processes; (2) defection is the result of deliberate state policies of manipulation and repression; and (3) insurgent forces use force against part of their own movement when it serves strategic goals, such as the consolidation of central authority and army discipline (ibid.: 19–21). Both ex ante political differences and British policies of divide and rule explain the transition from treaty to civil war in 1922. Yet the protreatyite desire to consolidate power in the face of a renewed IRA insurgency explains the origins of the execution policy. The policy gained momentum in the context of the reorganization of institutions and the process of legitimation, both part of the process of centralizing power (Janos Reference Janos and Eckstein1964: 130–40).

The executions became the instrument of a small group of men, insecure in their power, and skeptical of the civic virtues of the population they ruled. They departed from the firm but conciliatory policies of Michael Collins, Commander in Chief of the National Army. He remained against executions up to his death in an ambush on August 22, 1922. Collins had occupied a dual role: head of a democratically elected government, and president of the Irish Republic Brotherhood, formed in 1858. During the summer of 1922 he had seemed bent on entrusting key posts within the Free State to other republican brothers (Regan Reference Regan2007: 339–40). The most dramatic sign that this revolutionary fraternity would not last was the execution of four leaders of the Brotherhood on December 8, 1922.

The new civilian leaders had less tolerance for the shady interconnections between governmental and nongovernmental organizations. On September 27, 1922 the provisional government passed a Special Powers Resolution that allowed the military to detain persons taken in arms who had been caught participating in arson, looting, attacking on private property, or attacking Free State military forces. They were to be tried secretly by military tribunals with the authority to pass death sentences. The accused had a right to representation by legal staff within the army. The first batch took place in Dublin on November 17 when James Fisher, Peter Cassidy, Richard Twohig, and John Gaffney were executed in Kilmainham Jail. The men were very young; Fisher was only 18.

The official statement on the executions, which took place on December 8, said that the four leaders had been killed as “a reprisal” for the assassination of Seán Hales the previous day, and as a “solemn warning” against others. In late September 1922, the IRA Chief of Staff, Liam Lynch, had decided that those who voted for the executions and “aggressive Free State supporters” were “legitimate targets.” After Hales’s assassination, senior army officers had requested permission to take out and execute four leaders from the Four Courts garrison: The government authorized them to do so the next morning, fearing that their own supporters would cave in as the IRA began to act on Lynch’s orders (Blythe Reference Blythe1954).Footnote ⁶ The killing of Hales had forced the provisional government’s hand. William Cosgrave, now president of the Executive Council, claimed that his government had been willing to extend leniency to all remaining prisoners subject to military trials, but then Hales’s assassination changed things (Knirck Reference Knirck2014: 76).

The provisional government then adopted an official hostage policy, under which prisoners would be sentenced to death, but received a stay of execution pending the improvement of order in the areas where they had been active (Breen-Murphy Reference Breen-Murphy2010: 3). This policy was adopted in Kerry on December 13, 1922, and later extended to some other counties. In Kerry, General Murphy put up a notice confirming that four of those sentenced to death in the county had had their sentences commuted. Reports from the local commanding officer had commented on the favorable condition of the county, leading Richard Mulcahy, Army Chief of Staff, to order that the men be used as hostages. However, if after Thursday December 21, there were ambushes of troops or interference with railways or private property “the sentences will be carried out forthwith” (Irish Times, December 1, 1922).

The final step in the escalation of the policy was the decision on January 11 to devolve responsibility to local commanders, which resulted in executions in counties Louth, Tipperary, Carlow, Kerry, Tipperary, Westmeath, Waterford, Offaly, Laois, Wexford, Donegal, Galway, and Clare in 1923. At the decisive meeting of the cabinet and the army council, Patrick Hogan, the Minister for Agriculture, warned of a land war (Regan Reference Regan1999: 120). Those wanting more severe methods got their way: Committees of officers were established at battalion level throughout the country to pass sentences on IRA prisoners. The policy gained a fresh momentum: From early December cases were almost exclusively dispatched by such committees (Breen-Murphy Reference Breen-Murphy2010: 190). Units of army officers, many not ranked higher than captain, were formed as mobile military tribunals, traveling around rural Ireland to conduct the trials of prisoners that had been apprehended under the September resolution (Walsh Reference Walsh2015: 387). Even sympathizers caught carrying republican propaganda could now find themselves facing a firing squad.

Staniland (Reference Staniland2012: 21) suggests that fratricidal conflict can occur either in a top-down fashion, when an insurgent faction decides that victory can best be achieved by the consolidation of power under its aegis, or from below, when local feuds escalate into a more general military conflict. The Irish independence movement disintegrated from the top down: the cabinet, then the Dáil, and then the IRA split over the treaty. Yet local military power was crucial. After the July 1921 truce with the Crown forces local IRA brigades had continued to recruit, prepare for further warfare, and take over bases from the departing British troops. When each brigade voted on the treaty three-quarters went antitreaty, controlling two-thirds of the new state’s territory (Hopkinson Reference Hopkinson1988: Map 1). On July 1, 1922 the First Southern Division, (12 brigades covering Cork, Kerry, Waterford, and the western part of Limerick) had 27,277 men on its roles. The Second Southern (five brigades covering Limerick, Tipperary, and Kilkenny) had 8,955 men. Most of these 35,000-plus men were free to take part in the guerrilla war in August.Footnote ⁷

Because the executions constituted “a point of no return” within the nationalist movement each step in the implementation of the policy constituted an escalation of the conflict. Most historians back the protreatyites: “[A]t some point or other the government had to meet force with greater force and IRA terror with state terror” (O’Halpin Reference O’Halpin1999: 37). The executions worked as a deterrent: After Hales’s death no other member of the parliament would be killed. They also killed off the possibilities for reconciliation by an elite committed to the norm of civilian supremacy (Regan Reference Regan1999: 125). Yet elite politics do not explain why the policy gained momentum when it spread territorially or the absence of elite purges. By September 1923 exactly half of the antitreaty candidates who were elected to the Dáil in June 1922 were in custody (Irish Times, September 29, 1923). Most had survived the civil war. There were no further executions of republican elites after the executions in Mountjoy jail on December 8.

Guerrilla War

Ministers considered the guerrilla war “a movement of anarchy,” “a criminal conspiracy against the state,” and “a criminal” rather than a civil war (Knirck Reference Knirck2014: 34–55). Cahir Davitt, First Advocate General, argued that prior to the lapse of the amnesty offer on October 15, 1922, the IRA had tacitly been accorded belligerent rights. After that date they lost those rights, especially as they wore no uniform, carried no military emblems, and did not openly carry their arms (Davitt Reference Davitt1955: 81–82). The transgression model of civil war violence suggests that those who do not recognize the legal claims of the sovereign place themselves beyond the protection of the laws of regular warfare and can be treated as criminals (Kalyvas Reference Kalyvas2006: 62–63). Legality was important because the pretruce IRA had also been considered illegal by the British authorities. And the Catholic hierarchy had also condemned them. The Catholic bishops’ joint pastoral published on October 11, 1922 declared it a matter of “divine law” that the “legitimate authority in Ireland just now” was the provisional government, that “the guerrilla warfare now being carried on by the Irregulars was without moral sanction; and therefore that the killing of National Soldiers was murder before God” (Murray Reference Murray2000: 87). In 1922 the provisional government insisted that the press call the antitreaty IRA “irregulars.” Besides bearing unauthorized arms, the offenses during which people were arrested for trial, such as aiding the “irregulars,” disrupting railway lines, alleged robbery of a mail van, attempted burglary, and theft and armed robbery, were also “irregular.”

Balcells (Reference Balcells2010) poses the question of whether violence is a product of prewar political differences or of wartime conditions. In 1923 executions were administered not as a response to national politics but to shape local political and military realities. While those carried out in 1922 had taken place in or near Dublin, in 1923 they mainly took place outside of Leinster (Gavin Foster Reference Foster2015: 155–56). January 1923, when a total of 34 prisoners were executed, was critical. The increase was a response to a campaign of wholesale destruction in December after which some took matters into their own hands (Keogh Reference Keogh1994: 15–16). Finally, after a lull in February 1923, their resumption in March was calculated to achieve “a satisfactory geographical spread in the interests of deterrence” (Enright Reference Enright2019: 97).

The local power of the IRA came to matter. In Wexford, because of a lack of support from Waterford Command, inefficiency, and an uncooperative population, IRA activity had increased well into the new year.Footnote ⁸ The South Wexford Brigade had 1,367 men on the rolls in July 1922, four hundred more than it had had a year earlier. The columns of Thomas O’Sullivan in the New Ross area and of Bob Lambert around Wexford town controlled large areas of the countryside in the spring of 1923. One of the most active units was the Kyle flying column in Central Wexford: The three men executed in Wexford town on March 13 were from this unit. Kalyvas (Reference Kalyvas2006: 333–36) suggests that the denser the social contacts at the local level the more the audience for such acts of violence that broadcast the power of the state. On April 10, 1923 six men were taken aside in Galway jail and told they would be executed in Tuam the next morning. They were all part of the second western division under Tom Maguire. Tuam was nearer Headford, where most of the men were from, and their execution was a reprisal for the attack on the army barracks there on April 8 (Newall Reference Newall2015: 28–29).

Kalyvas (Reference Kalyvas2006: 196) argues that in the early stages of civil war—when situations of contested sovereignty exist—indiscriminate violence, including the targeting of civilians, is more likely. Selective state violence is more effective after a government has achieved territorial control. More discriminate forms of violence become possible when the population cooperates in terms of providing information, and because proper policing is established in local areas. In Ireland executions had not been needed to establish control; they were a means of consolidating it at the local level. Most took place in rural areas where state control was predominant, not absolute. Due to many of its men being part of the same units before 1921, the army was well informed about its enemies, their hiding places, and their tactics. It could more easily distinguish them from the general population than the Crown forces had been able to do.

The executions were selective in a number of ways. The Army Emergency Powers Resolution was to apply only to those found in arms. Of the 81, the number of those sentenced to death for the possession of ammunition, arms, or explosives is 63. Thirteen of these faced another charge, while seven sentences were for taking part in attacks in which soldiers died, for conspiracy to murder or for murder. For bank robberies only two were executed (Cambell Reference Cambell1994: 360–72). Those guilty of other offenses were generally not put before military courts, despite the Minister for Agriculture Patrick Hogan wanting “immediate and drastic action against people who seized other peoples’ land.”Footnote ⁹ Secondly, while the army held many IRA prisoners as hostages, most were spared. Seán Prendergast, an IRA prisoner in Gormonstown internment camp, recalled that the response of the prisoners to the executions on December 8 was that the state wanted to spread fear and terror. Anyone could be next (Prendergast Reference Prendergast1952). Yet the executions had ceased by June 1923, when more than 12,000 republicans were in jail or internment camps. Most cases brought before the courts (or the ancillary committee system) did not result in executions (Yeates Reference Yeates2015: 258). Moreover, geographic variation was pronounced. A total of 40 men were killed while in army custody in county Kerry (Ryle Dwyer Reference Ryle Dwyer1996: 364). Yet only one execution took place in neighboring Cork, the most violent county during the earlier War of Independence. Lastly, 645 women were interned for “actively assisting” the IRA (Matthews Reference Matthews2012: Appendix 1). Their execution was not considered. During the entire war, no woman was charged with a capital offense.

Four civilians were executed. Balcells (Reference Balcells2010), Gutiérez-Sanin and Wood (Reference Gutiérrez-Sanin and Wood2017), and Steele (Reference Steele2017) consider how political divisions influence violence against civilians during wartime. High levels of prior mobilization, strong collective identities, and the existence of compact populations with clear loyalties can make civilian support a critical resource. In Ireland, civilians were generally not targeted. Men were selected for execution not by virtue of their membership of particular groups but after allegations of individual behavior. In such a nonethnic (perhaps nonideological) civil war, such as that fought over the Treaty, while local political opinions were hard to observe the army knew who remained active in the IRA. Lines of division were also formed late. Tom Maguire reflected that initially “you might be in touch with personalities on your side today, and tomorrow you could be told that they had gone over to the Free State side” (Ó Gadhra Reference Ó Gadhra1999: 139). A number who “flipped” sides ended up being executed. The five executed for treachery at Portobello Barracks on January 8 were Free State soldiers who chose to aid the IRA. Seven former British soldiers were executed after taking the antitreaty side.

Those executed were usually young IRA volunteers. The ranks of the IRA had swelled after the truce with Britain. Gavin Foster (Reference Foster2015: 35) stresses that prior to 1914 the young had occupied the bottom rung of the social ladder; they were politically nonentities. The Sinn Féin independence movement that swept the polls in 1918 then idealized Irish youth in contrast to the corruption of the older “Home Rule” parliamentary tradition. The civil war saw a return to the prewar status quo. The worst example was when the bodies of three young men, aged 16 to 17, were found in Dublin having been picked up by detectives the previous day for distributing republican leaflets. Most of those executed were under 25; a significant minority were also teenagers. A contrast with 1916 suggests itself. The 14 leaders that were executed after the Easter Rising were mainly well-known figures in the cultural life of the city; both class and race might have suggested they be spared the final punishment for treason. When they were executed they were instantly turned into national martyrs. While those executed in 1922–23 also saw themselves as stepping into this revolutionary tradition of sacrifice, there was no comparable public reaction to their deaths.

T.D.’s lives were usually spared. Tom Maguire, a member of the second Dáil, survived after he had been captured by the National Army and sentenced to death. Yet all five of the men executed in Tuam, county Galway, on April 11 (including his teenage brother), were part of his 2nd Western Brigade. Four of the five men executed in Athlone on January 20 were also members of the western division. Maguire speculated that the authorities found it easier to “make an example of younger brothers” (Ó Gadhra Reference Ó Gadhra1999: 141), and government propaganda blamed youthful irresponsibility for prolonging the IRA campaign (Foster 2015: 32). Were such youthful targets an alternative to elite purges? The state was willing to execute leaders. When Senator Bagwell, Chair of the Great Northern Railway, was abducted (and his house burnt down), Commandant General Dan Hogan—with a certain amount of publicity—gathered 40 of the most prominent IRA prisoners into Mountjoy jail, suggesting that were Bagwell killed, the number of executions would be far greater than on December 8. Bagwell was allowed to escape (Blythe Reference Blythe1954). The IRA’s own lack of ruthlessness in not implementing Lynch’s orders against parliamentarians explains the rarity of elite executions. Minister Ernest Blythe recalled the case of T.D. Frank Bulfin, who had attempted to resign from the Dáil because he was subject to intimidation by republicans. Bulfin was persuaded to remain, but the incident made Blythe feel that if there had been a big scare following one or two deaths of a T.D. or senator, “it would be our duty to apply sufficient terror to counteract it” (ibid.: 174).

Regan (Reference Regan2013) argues that its association with Bolshevism accounts “for the ferocity with which the rebellion was put down.” Most executions were of ordinary young men from working-class backgrounds. Of the five leaders executed, Erskine Childers was a well-known novelist, Rory O’Connor was a trained engineer, and Joseph McKelvy had some accountancy training. Of the 76 victims who were not leaders, only John Larkin (son of a magistrate in Derry) was from a middle-class family. The remaining 75 were from ordinary, mainly rural backgrounds. Although not all had work experience, a common occupation was laboring: Quite a few were farmers’ sons. Condemning the first executions (of four young Dubliners) on November 17, the Labour Leader Thomas Johnson argued that their disadvantages of youth, social standing, and education ought to have been taken into account in the sentencing.Footnote ¹⁰

Yet guerrilla war conditions explain more than class. Fear of social disorder had influenced the January 11 meeting. Patrick Hogan thought the land situation so ungovernable that it was up to the army to take actions against people who seized land (Regan Reference Regan1999: 120). In February, the minister of defence established the Special Infantry Corps specifically to tackle agrarian disorder. Yet no one was executed for agrarian offenses or labor agitation in 1923; or for any of the offenses the Corps was established to deal with. The Corp’s role was to provide aid to the civil power: Executions did not fall within its powers. Contrast this approach with the policy on the railways where the state’s military interests were at stake. The IRA attacks involved burning carriages and destroying goods in transit and were on a larger scale than during the war of independence. Four railway workers were executed in Kildare on December 19, while two more men were executed in Limerick jail on January 20 for their part in the destruction of the railways in the Tuam area.

The authorities tried to depersonalize the executions. The criteria for selecting men remained secret. The notice relatives received was usually in the form of “remains of ______ coffined and buried” with the name of the dead man inserted in the blank. Remains were interred at the place of execution. The move away from public executions the previous century had largely been driven by a fear of crowds, and the potential to call the legitimacy of the state into question. For similar reasons, the executions were carried out behind closed doors. This strategy nonetheless failed. The shooting of four Kerry-born IRA men in Drumboe Castle county Donegal on March 14, 1923 after the killing of a Free State soldier is an example. Anecdotal evidence suggests that a team of former British soldiers were brought in to carry out the execution. Yet when the firing squad failed to kill one of the men, Charlie Daly, the officer who oversaw the court-martial and the execution stepped forward to deliver the coup de grace with a revolver. This officer was none other than Joe Sweeney, who had been a close student friend of Daly (Griffith and O’Grady Reference Griffith and O’Grady1998: 306). Selective violence should not imply that there was nothing personal about the executions.

Nonetheless, seeing the spread of the executions as a response to guerrilla war conditions explains more about their diffusion than elite politics or class conflict. Moreover, the policy remained selective. The military courts and committees had tried almost 1,200 men, and of these more than 400 were sentenced to death (Enright Reference Enright2019: 6). The majority of this 400 survived. Diffusion can imply a process of contagion “in which individuals function outside normative constraint and their actions are propelled by high levels of emotional and psychological strain” (Soule and Roggeband Reference Soule, Roggeband, Snow, Soule, Kriesi and McCammon2004: 244). This approach to the executions is common among historians. An execution policy “born in bloodshed and hatred” (Coogan Reference Coogan2006: 138) created a spiral of bitterness and revenge such that the civil war acquired the character of “a vendetta on a national scale” (Hopkinson Reference Hopkinson1988: 190). Selective violence suggests a more “organised, rational and goal oriented” model of diffusion (Soule and Roggeband Reference Soule, Roggeband, Snow, Soule, Kriesi and McCammon2004: 294, 311).

This diffusion was shaped by the two related dimensions of state power: centralization and the localization of violence (Mann Reference Mann1986: 521). A rapid reorganization of administrative power had taken place at the center, especially as the influence of the civilian leaders grew. At the same time, recognizing the psychological distance between Dublin and rural Ireland, these leaders devolved the power of execution to army officers at the local command level, feeling that the authority of the state was insufficiently felt there. Garvin (Reference Garvin1996: 103) shows that the Irish army wanted to break the hold the local IRA had on shaping people’s perceptions of the center. The next section discusses the relationship between this devolution of initiative and the spread of the executions. Centralization and the localization of violence now came in tension. The result was a lack of control.

Principal Agent Problems

Once the decision to devolve responsibility was taken in January, the government (“the principal”) had ceded the initiative to the commanders on the ground (“the agents”). The execution policy produced “a principal-agent problem” that became acute in 1923. Such “problems” revolve on how a principal controls a self-interested agent in possession of superior information (Mitchell Reference Mitchell2004: 5). The army’s legal staff played no role in selecting those to be executed. The approach predicts problems down the chain of command when commanding officers and their subordinates come to have different preferences as to application of military force.

IRA Commander Tom Maguire reflected: “It is said that the Free State ‘provincialized’ its killings, both official and unofficial, by having the majority of them carried out away from Dublin, in contrast to the British who had all theirs, except one in Dublin and Cork. It is my opinion that their objective was to involve all their senior officers in this policy, so that there would be no denying it afterward. Joe Sweeny carried out executions in Drumboe in Donegal. Dan Hogan had them in Dundalk. Michael McCormick had them in Portlaoise, Birr, and Roscrea. Joseph Cummins had them in Wexford. Liam Stack had one in Carlow. Seán MacEoin had them in Athlone. And Michael Brennan had them in Tuam, Limerick, and Ennis” (Ó Gadhra Reference Ó Gadhra1999: 139–42).

While all executions still had to be approved by the Army Council, local actors increasingly gained the initiative. In Kerry on January 11, 1923, an army poster appeared in Tralee saying that four men had been found guilty of the possession of firearms and sentenced to death. It also warned that the IRA activity in the county had to stop, or the stay on their execution would be removed. In response, the local IRA commander made it known that were the men killed, eight named supporters of the Free State would be shot. The men were never executed. However, on January 20, 1923 four different local men were executed. The night before, Richard Mulcahy—now commander in chief—had authorized the execution by signing the back of a brown envelope. This was a response to a request for a deterrent because of the renewed IRA activity in the county; the local commander had simply telegraphed Mulcahy for his approval (Breen-Murphy Reference Breen-Murphy2010: 3).

The executions that took place in Tuam, Tralee, and Ennis in April 1923 were also a response to events determined by the local authorities (Enright Reference Enright2019: 106–11). An IRA column led by Michael Pierse in north Kerry had agreed to lay down arms in return for captured men from his column not being executed. When the authorities did execute one of those men, James Hanlon, two of the released men, Reginald Hathaway and Edward Greaney, decided to join a column led by Timothy (Aero) Lyons. This column had several men who had rejected Pierse’s peace move. When seven of them were found hiding in the Clashmealcon Caves near Causeway, after one private had been killed in the shooting that ensued, Lyons negotiated a surrender for himself. When he was climbing to safety on April 16 the army either severed the rope or it snapped, and he fell to his death. Two other men had drowned while swimming to freedom. The three surviving men (including Hathaway and Greaney) were executed on April 25.

With the principal agent approach we shift from the language of control to that of revenge. In Kerry, outside the towns, the center, south, and east of the county remained dominated by IRA units well into 1923, especially in mountainous areas. On March 6, 1923 three officers and two privates in the National Army were killed by a booby trap in the village of Knocknagoshel. The two privates, Laurence O’Connor (Causeway) and Michael Galvin (Killarney), as well as Lieutenant Patrick O’Connor (Castleisland), were from the county (The Kerryman, August 16, 2017). The following day the local commanding officer authorized the use of prisoners to clear mines. This may have been interpreted as a sign that the authorities wanted revenge: Over the following four weeks 23 prisoners were killed, and 5 more were legally executed.

The IRA had become divided into “the hunters and the hunted.” Macardle (Reference Macardle1991: 6) comments that the National Army was “fighting for a cause they hated against the cause they loved, hunting and killing their own comrades and leaders.” She recounts the story of Bertie Murphy from Kerry who was 17 when killed by the Free State soldiers in September 1922. He had become captain of his local Na Fíanna (the IRA’s youth wing) after one of its officers joined the National Army. This officer (named Healy) then searched for him for four months and was part of the army guard at a local hotel after Murphy had been arrested. The soldiers first used Murphy as a hostage to remove IRA barricades, and eventually shot him in captivity after a group of Free State troops had been ambushed (ibid.: 99–100).

One could not distinguish the official executions from other forms of state violence. Kerry Republicans claimed that prisoners were murdered on the roads or in bed, fired on indiscriminately while in prison compounds, or died as a consequence of the mistreatment of the wounded in prison.Footnote ¹¹ Principal-agent problems generally emerge when units of the army exploit the distance between them and the central command to carry out acts of violence for private or opportunistic ends. An example was the unlawful killing of men after their capture or surrender. O’Halpin (Reference O’Halpin1999: 35) suggests that as many as 150 IRA men were killed in this way: Enright (Reference Enright2019: 6) gives a figure of 125. This was not “policy” but was tolerated as a “practice” (Wood Reference Wood2014: 470). One example was Brian MacNeill, son of Minister of Education Eoin MacNeill, who appears to have been killed in September 1922 by Free State soldiers after surrendering in arms in Donegal. The minister chose to believe that this account of the death of his son, an IRA officer in a remote county, was republican propaganda.

When the initiative passed to officers on the ground, the army had knowledge of the prior activities of those it sentenced. In January, O’Daly wired Mulcahy for his approval to execute “three exceptionally bad cases.” After a lull in February, Mulcahy decided that “bad cases” should be prepared for execution when the process resumed (Breen-Murphy Reference Breen-Murphy2010: 193, 214). At the same time, those further down the ranks could also spare lives. Arrested in Dublin after an attack on Free State troops, with four men who were executed on November 30, one young man had his death sentence commuted to penal servitude by the authorities because he was the son of Michael Mallin who had been executed in 1916 (Davitt Reference Davitt1955: 43, 55). Typically, when the army apprehended a group of IRA men, only some were executed. Mick Kennedy had been arrested and convicted of the possession of arms and ammunition with Patrick McNamara, who was executed in Roscrea on January 15, 1923. A Free State soldier had intervened and arranged a pardon for Kennedy (O’Dwyer Reference O’Dwyer2006: 186). On January 26, the army executed three young men in county Offaly after a series of robberies. A fourth teenager from the same area was pardoned because of his youth, the alleged intervention of the Catholic Church, and also the possible status of his family who were substantial farmers (McConway Reference McConway2007).

The principal agent approach assumes an inverse relationship between the degree of accountability within an army and the territorial distance between the principal and its agents. The appointment of Dubliner O’Daly (formerly a member of Collins’s intelligence “squad”) as commanding officer for Kerry in the southwest resulted in a series of brutal extrajudicial killings by officers within his inner circle, including the massacre of nine IRA prisoners by tying them to a mine at Ballyseedy on March 7, 1923. When a court of inquiry was established by the army to investigate Ballyseedy, the court president, with two other senior officers of the National army, was O’Daly. The result was a whitewash; O’Daly retained his command (Price Reference Price2018).

If state violence had an ideal form; it would be top down, authorized for strategic purposes by an elected government, and carried out within a legal framework by state agents.Footnote ¹² The principal-agent approach questions whether these criteria can survive the decentralization of initiative in civil war. At the 2016 commemorative event for Michael Collins, current President M. D. Higgins called for both sides’ “atrocities to be recognised as cruel, vicious uncontrolled, and at times, informed by vengeance rather than any compassion” (Irish Times, August 21, 2016). Elements on both sides were “out of control.” Because the origins of the policy can be traced to Collins’s assassination, the executions formed part of a chain of reprisals. Dan Breen of the 3rd Tipperary IRA Brigade recalled the situation in Tipperary:

“A policy of reprisal and counterreprisal now held sway and continued in ever-growing intensity until the end of the war. Executions on one side were followed in many cases by executions on the other side, while mansions and houses of prominent supporters of the Irish Free State were given to the flames” (Breen Reference Breen1959: 137).

Among the senators (many of whom were Protestant), more than 50 percent (21) lost their country homes. In the year after the treaty was signed (up to December 5, 1922), a total of 89 mansions or large country houses were burned down; in the 15 weeks following December 6, 1922 a total of 103 were destroyed (Clarke Reference Clarke2014: 7). This sectarian turn indicated what could have happened had the war continued (Roy Foster Reference Foster2015: 281).

The principal-agent model focuses on how violence changes in character as actors respond to the decentralization of authority during civil war. The diffusion of the executions was a top-down process in the sense that local actors “emulated” the actions of those in higher positions of authority. The executions on December 8 had signaled to those further down the ranks that the lives of those in captivity were forfeit. They were a clear declaration of intent (Regan Reference Regan1999: 112). And when 34 men were executed for the illegal possession of arms in January, hundreds knew that this charge was equally applicable to them (Macardle Reference Macardle1937: 863). At the same time, the agents also took advantage of the lack of accountability to the center. The principal-agent model suggests that state violence became more extensive because of the greater number of actors that could initiate executions, the greater variety of local factors that could justify such acts, and the loosening of central control when the policy became decentralized. It also became harder to differentiate official executions from other acts of state violence, suggesting that the policy was less “rational” and “goal oriented” than “selective violence” might imply. Ultimately, what brought them to an end was not the resolution of the principal agent problems but the IRA ceasefire.