Force structure and counterinsurgency outcome: the case of the Cyprus Emergency (1955-1959)

ABSTRACT This article examines the Cyprus Emergency (1955–1959) to test the force structure thesis. According to the thesis, armies that deploy more manpower per armored vehicle would succeed in counterinsurgency because they could win people’s trust, secure intelligence from civilians, and use force selectively. Using the congruence method of within-case analysis, I show that the causal process and logic of the force structure argument are not confirmed in the favorable case of Cyprus. Despite its preference for infantry and police units, the British garrison failed to win over the people and persuade civilians to share information about the insurgent; besides, Britain’s intelligence breakthroughs and selective violence did not result from the logic of the force structure thesis, nor did the conflict end as the argument would predict. Political conditions, instead, played a greater role. Ultimately, the case of Cyprus warns against the reassuring belief, inherent in the force structure thesis, that military organizational adjustments in favor of manpower can pave the way to victory against irregular opponents.


Introduction
After prioritizing counterinsurgency for a decade and a half following the 11 September terrorist attacks, the United States is now focusing again on mechanized conventional warfare (National Defense Strategy of the United States of America 2018; Osborn 2016aOsborn , 2016b. However, some warn that the renewed emphasis on mechanized warfare may dangerously erode the ability of land forces to fight insurgents (Vrolyk 2019) in an international security environment where the threat of Islamist uprisings is still alive and backed by regional powers (see Ostovar 2019;Kapur 2017).
Concerns about the tradeoffs between mechanization and victory in counterinsurgency have a rigorous social scientific foundation in the force structure thesis. The thesis holds that military organizations that deploy more manpower per armored vehicle would be likely to defeat insurgents (Lyall and Wilson III 2009). The present article will test this argument to assess the above-mentioned concerns.
The force structure thesis is relevant in the population-centric paradigm of counterinsurgency. The paradigm emerged from the Cold War era and became conventional wisdom during the US-led campaigns in Afghanistan and Iraq as it promised solutions to the difficulties met in those conflicts (Michaels and Ford 2011). 1 As part of the dominant school of thought, the thesis influenced US counterinsurgency to the extent that General McChrystal's command in Afghanistan mentioned Lyall and Wilson's scholarship as a source of guidance (Barnes 2009). Even if practitioners' enthusiasm about counterinsurgency gradually subsided, the thesis remains prominent in the academic literature. 2 The force structure thesis accepts the population-centric assumption that counterinsurgency is a contest over good governance. Victory would require political reforms to address the grievances behind the rebellion and persuade civilians to cooperate with the government. In this way, the counterinsurgent could secure information about the enemy from civilians and destroy the insurgent without harming and alienating noncombatants (Lyall and Wilson III 2009). 3 Yet, mechanized armies could not deploy enough manpower to interact with the people and build mutual trust; besides, they would be supplied in ways that cause self-isolation, thus running into defeat (Lyall and Wilson III 2009). The force structure thesis, therefore, suggests that organizational adjustments in favor of manpower over military machines could boost states' ability to defeat insurgencies: the more soldiers per armored vehicle, the higher the chances to win.
Considering that shaping armies along these lines may reduce their readiness to face mechanized opponents, it is important to test the force structure thesis. Is it true that more manpower leads to more popular support, better intelligence from civilians, and victory over insurgents? If not, decreasing mechanization may involve a higher risk of defeat in conventional maneuver warfare, but no benefits in counterinsurgency.
Testing the argument also matters for the academic debate on counterinsurgency outcomes. According to some critics, counterinsurgency success became less likely in the twentieth-century because of changing political conditions. Specifically, the rise of the self-determination principle in the state system enhanced insurgents' international legitimacy, made local elites reluctant to collaborate, and inhibited coercion against insurgents (Smith and Toronto 2010, 523;MacDonald 2013). Lyall and Wilson acknowledged these obstacles, but they insisted that armies relying on manpower are still more likely to win (2009: 87). As recent scholarship shows mixed evidence about the thesis (Enterline et al. 2013, 189;Caverley and Sechser 2017: 706, 712), further tests are needed.
Empirically, the thesis rests on statistical analysis of 286 insurgencies between 1800 and 2005. While quantitative methods can highlight correlations between different factors across cases, they cannot show causation and prove that a given outcome actually came through the causal pathway posited by the theory under scrutiny; the case study method, instead, is designed to do that (George and Bennett 2005, 21-22). In fact, supporters of the force structure thesis performed limited qualitative analysis of two United States divisions in Iraq with different levels of mechanization 4 but only to provide suggestive evidence of the posited causal process (Lyall and Wilson III 2009, 94-101). Yet, research on the impact of mechanization on counterinsurgency outcomes has overlooked within-case analysis to test the causal logic of the thesis. 5 Interestingly, even critics retained a predominantly quantitative approach (Friedman 2011;MacDonald 2013) or performed no case studies (Smith and Toronto 2010). This paper, therefore, will test the causal logic and process of the force structure argument in the case of Britain's campaign against the National Organization of Cypriot Fighters (EOKA) between April 1955 and March 1959. The Cyprus Emergency would be a particularly relevant case because the British garrison in Cyprus displayed a low level of mechanization; consequently, the case provides favorable conditions for the force structure thesis to demonstrate its causal logic about the advantages of manpower and explain the outcome of the campaign by way of the posited causal process. If, instead, the causal process of the force structure argument fails to manifest itself and explain the Cyprus Emergency, the logic of the thesis can hardly be expected to account for other cases presenting less favorable conditions and the argument should be at least reconsidered (see Eckstein 1975;George and Bennett 2005, 121-122).
Using primary and secondary sources, I will rely on the congruence method of withincase analysis to assess whether the British campaign in Cyprus unfolded according to the causal logic and process of the force structure thesis. After specifying the hypotheses that constitute the argument's causal mechanism, I will observe the level of mechanization of the British garrison in Cyprus to deduce the predictions of the force structure thesis about Britain's campaign against EOKA; finally, I will assess the extent to which withincase observations are congruent with the predictions of the thesis. When predictions and observations correspond, I will check whether such congruence actually results from the main causal factor and logic of the argument.
I find that the British campaign against EOKA does not confirm the force structure argument and invites caution about the idea that organizational adjustments in favor of manpower would foster victory in counterinsurgency. Such adjustments may compromise states' ability to confront modern mechanized armies, without increasing ability to defeat insurgents. The case of Cyprus would also imply that mechanization may not be responsible for counterinsurgency failure, therefore the current shift in the United States defense strategy may not be as risky as some observers believe. Ultimately, the case suggests that counterinsurgency outcomes are shaped by political conditions, not force structure.
In the remainder, I will set out the force structure argument and spot three testable hypotheses with observable implications about its causal mechanism before specifying the predictions of the thesis about the Cyprus Emergency. After that, I will give a factual account of the British campaign and I will analyze it to test the argument. Finally, I will reflect on the implications of my analysis.

The force structure argument: hypotheses about the causal mechanism
The major causal factor in the force structure thesis is the level of mechanization, or the number of soldiers per armored vehicle including battle tanks, armored personnel carriers, armored fighting vehicles, self-propelled artillery, and scout cars (Lyall and Wilson III 2009, 83). The higher the number of soldiers per vehicle, the lower the level of mechanization and vice versa.
The causal explanation of the force structure argument holds that mechanization would reduce the number of infantrymen in relation to armored vehicles. Given the technical complexity of modern military machines, a mechanized military can only be supplied by specialized industries outside the war zone through a logistical system that includes military bases located far away from populated areas (Lyall and Wilson III 2009, 77). In this way, mechanization would drastically reduce the frequency of interaction between the counterinsurgent and the local population and would show that soldiers do not want to expose themselves to hostile firepower to protect civilians from insurgent violence (Lyall and Wilson III 2009, 77). Consequently, mechanization would exacerbate people's mistrust and hinder the recruitment of local collaborators that may provide accurate information about insurgents and their supporters. Confronted with information starvation, mechanized armies would end up using force indiscriminately, thus alienating the populace and inducing civilians to side with the insurgent for protection or revenge (Lyall and Wilson III 2009, 77-78). Instead, military organizations that rely on manpower over machines could deploy more troops to mingle with the population, build trust and recruit informers, secure intelligence about the opponent, and use force selectively without losing popular support, thus attaining victory. 6 The causal process of the force structure argument can be presented as a sequence of three interconnected hypotheses, or conjectured causal relations, with observable implications.

Hypothesis 1 (H1):
The higher the number of soldiers per armored vehicle, the easier for the counterinsurgent to win over the local population and obtain information about the insurgency from civilians, and vice versa.

Hypothesis 2 (H2):
Information about the identity and location of insurgents will cause the counterinsurgent to use force selectively and avoid harming civilians. Limited information about the insurgency, instead, will cause the counterinsurgent to use indiscriminate violence.

Hypothesis 3 (H3):
Selective violence will destroy the insurgent organization without alienating the people and will cause victory in counterinsurgency. Indiscriminate violence, instead, will increase people's resentment against the counterinsurgent and cause defeat. Importantly, each of the above-mentioned hypotheses is a stage logically following from the previous one in the force structure causal chain. The force structure argument would be confirmed only if all of the three interconnected hypotheses hold true in the case studied in this paper. If only some hypotheses are confirmed, then the posited causal chain breaks at some point, which justifies doubts about its validity.
What should we expect to observe if each of the three hypotheses hold true, then? And what evidence would falsify those hypotheses, instead? H1 would be confirmed if we see that a counterinsurgent that is mostly deploying infantry and police units to counter an insurgency is able to mingle with the local population, protect the people from insurgent violence, and gradually win the trust of civilians. We should see at least small groups of civilians having enough confidence to come forward and overtly side with the government, organize larger groups of civilians in the society to resist the insurgency, and share information about the insurgent with the security forces without being coerced to do so. H1 would be falsified if the counterinsurgent relies on infantry, police, and other units with a low number of armored vehicles and still fails to reassure the population from the insurgent threat and get information about rebels from civilians. Besides, intelligence gains that result from coercion, intimidation, and use of force against the insurgent organization or civilians would not be consistent with H1 and its emphasis on persuasion.
H2 would be confirmed if a counterinsurgent that has secured accurate intelligence about the enemy avoids measures that are likely to harm non-combatants and takes precautions to minimize the loss of civilian lives. For example, we should see the counterinsurgent avoiding measures like torture, scorched-earth, and mass deportation, the creation of free-fire zones without evacuating civilians, indiscriminate shootings and summary executions. H2 would also be confirmed, or at least not clearly contradicted, if the counterinsurgent adopts measures that limit the freedom and property rights of civilians but are not designed to jeopardize their survival. If, instead, the counterinsurgent has reliable intelligence about the enemy but still targets civilians intentionally, then H2 would be falsified. Likewise, if the counterinsurgent is self-restrained even when it has limited information or for reasons that are not related to the amount and quality of information about the enemy, H2 will not be confirmed.
Finally, H3 would be confirmed by evidence that a counterinsurgent deploying infantry and police achieved its declared political goals after weakening the insurgent organization without having harmed non-combatants. We should see insurgent leaders conceding victory to the government after accepting that the military damage suffered during counterinsurgency operations is too extensive to recover from. If, instead, the counterinsurgent weakens the insurgent organization through selective violence but still fails to break the insurgent's will and ability to fight, then H3 is not confirmed. H3 would be equally falsified if the counterinsurgent attains its declared goals because of factors that are unrelated to the damage inflicted on the insurgent organization through selective violence.

Predictions of the force structure argument about Britain's campaign in Cyprus
As mentioned above, the Cyprus Emergency is a case that the force structure thesis should be able to explain easily. Indeed, the major causal factor of the argument -the level of mechanization of the British garrison in Cyprus during the conflict -displays values that should facilitate rather than hinder the causal process and outcome predicted by the thesis.
Importantly, one cannot simply take data about the mechanization level of the British garrison in Cyprus from current databases built for the purpose of quantitative analysis. Indeed, such data do not refer to the specific garrisons that actually confronted insurgents; instead, those data would refer to the soldiers-per-vehicle ratio in the entire army of the counterinsurgent in the year prior to the campaign (Lyall and Wilson III 2009, 84). For example, when Lyall and Wilson classify the British Army as having the second highest level of mechanization during the Cyprus Emergency, they included all the British soldiers and vehicles deployed outside Cyprus too. Therefore, a closer look at the British garrison is essential to get a clearer idea of the combination of manpower and machines that Britain used against EOKA.
The British garrison in Cyprus mostly relied on infantry and police, not armored units. After the declaration of the state of emergency in November 1955, the Army would deploy 17,000 men against EOKA by , 22,500 by June 1956, and almost 31,000 by December 1956 (French 2015: 122, 145). The regular and auxiliary police of Cyprus as well as the Special Constabulary would join the counterinsurgency campaign with a strength of 4,764 men all ranks (French 2015, 119). In January 1958 there was still a member of the security forces for each Greek Cypriot household on the island (Walton 2013, 308).
The British did not deploy any battle tanks, armored personnel carriers, armored fighting vehicles, or self-propelled artillery against EOKA, with the only exception of three cavalry squadrons from the Royal Horse Guards (RHG) that arrived in Cyprus in February 1956. The RHG squadrons in Cyprus were equipped with Ferret scout cars, a lightly armored wheeled vehicle that carried one light machine gun and was operated by a crew of two men. Overall, the RHG would deploy no more than 50 to 60 Ferret scout cars during the Cyprus Emergency. 7 Interestingly, the RHG often replaced scout cars with mules or horses in mountainous areas to reduce the risk of alerting EOKA sentinels and village groups. Other units that were specialized in gunning roles, like the Royal Artillery (RA) regiment, arrived without artillery or other armored vehicles and relieved exhausted infantry units in dismounted patrolling and internal security operations. 8 Considering the low level of mechanization of the British garrison in Cyprus, the force structure thesis would predict Britain to be able to win over the local population and secure information about EOKA from Greek Cypriot civilians (H1), use force selectively based on intelligence about the insurgency without harming non-combatants (H2), and defeat EOKA (H3) with all the corresponding observable implications set out in the previous paragraph. In the next paragraph, I will provide a factual account of the conflict before analyzing the campaign and testing the predictions of the thesis.

The opening phase (April 1955 -November 1955)
The EOKA insurgency derived from the nationalist aspiration of the Greek Cypriot community -80% of the population of Cyprus -to achieve "Enosis", or union with Greece. Yet, Cyprus was also home to a Turkish minority -17% of the population -that opposed Enosis out of fear of marginalization (Gates 2013).
Britain took over the administration of Cyprus in 1878 and annexed the island as a colony in 1914 bashing Greek Cypriot nationalism. After the Second World War, the Orthodox Church, led by Archbishop Makarios III since 1950, rallied Greek Cypriot nationalist organizations under the Ethnarchy Council and engaged in a diplomatic campaign for self-determination with the support of the Greek government (Holland 1998, 26-28). Faced with Britain's refusal to negotiate, the Ethnarchy began to plan an insurgency and appointed colonel Georgios Grivas as military commander.
Grivas created the National Organization of Cypriot Fighters or EOKA. Trained fighters, numbering almost 300, were divided into mountain gangs and town groups. The former would ambush the security forces in village areas, while the latter would target soldiers, policemen, state officials, and government informers in the cities; most EOKA members, however, joined village groups that included approximately 750 people responsible for providing food, shelter, manpower, communications, and information about the enemy (French 2015: 56-57, 64).
Grivas aimed to undermine Britain's will to fight, instead of defeating the British militarily. EOKA would resort to sabotage of government and military infrastructures as well as hit-and-run tactics to provoke, frustrate, and expose British repression (French 2015, 48). At the same time, the Orthodox Church would keep the Greek Cypriot community ideologically committed to the nationalist cause, while Greece would apply diplomatic pressure on Britain through international organizations to achieve selfdetermination (see Averoff-Tossizza 1986; Klapsis 2019; Novo 2013).
On 1 April 1955 insurgents started their sabotage campaign. Yet, between May and August 1955, EOKA concentrated its offensive on the Cyprus police whose manpower was mostly from the Greek community (Novo 2012, 422). Grivas could infiltrate an intelligence unit in the police and administration, which enabled EOKA to single out Greek police officers and members of the Special Branch for assassination (Anderson 1994, 184-185;Slack 2019, 97-99). EOKA's intelligence-driven offensive overwhelmed the police. Faced with limited popular support and a high risk of being exposed by EOKA spies and murdered, police officers would tender mass resignations or stop investigating: by the end of Summer 1955 the colonial government admitted that the police was too demoralized, intimidated, and colluded with EOKA to maintain law and order (Anderson 1994, 185;Novo 2012, 421-422).
As Greek police agents became unreliable, the colonial government established the Mobile Reserve Unit and the Auxiliary Police Force recruiting from the Turkish minority (Anderson 1994, 191). The new recruitment policy, however, would gradually connote the police as a Turkish organization and indeed 60% of police manpower came from the Turkish minority by 1956 (Novo 2012, 422). Few of them spoke Greek or had connections with the Greek community. Consequently, Britain's ability to obtain information about EOKA from the Greek community -the only group that could have provided itwas seriously affected.
On the diplomatic level, in August 1955 Britain called a conference with Greece and Turkey on the future status of Cyprus, but only to propose a tripartite administration that would involve Turkey too as a partner to hinder Enosis. The ensuing riots in Cyprus led to the appointment of Field-Marshal Sir John Harding as Governor. As EOKA escalated its terrorist campaign, Harding declared the state of emergency on 26 November 1955.

Harding's campaign (December 1955 -October 1957)
Both the Cabinet in London and the Governor of Cyprus believed that Britain could not achieve victory over EOKA in purely military terms: it was popular support for Enosis that had to be eliminated. 9 According to Harding, victory would correspond to "a political situation in Cyprus which will ensure that when self-determination is applied the outcome will be a decision to remain within the Commonwealth, with Enosis finally rejected". 10 Importantly, Harding was confident that Britain could win over the Greek Cypriot community as he believed that civilians were willing to side with Britain but were simply too intimidated by EOKA and influenced by the Orthodox Church to support the government (French 2015, 93). Harding, therefore, set out the conditions to achieve victory, including the destruction of EOKA "beyond all hope of recovery", the imposition of government control over the education system, and the neutralization of the Church as a political actor that indoctrinated Cypriot youths with Greek nationalism. 11 Accordingly, Harding had some of the most prominent Greek Orthodox clergymen arrested and expelled from Cyprus, including Makarios who was deported to the Seychelles in March 1956. After that, Harding would focus on the elimination of EOKA guerrillas and the punishment of their supporters among the people. Since December 1955 Harding had imposed collective fines to rebuild any property damaged or destroyed after EOKA attacks, sabotage, or riots. 12 Curfews would be an additional measure to force civilians to pay collective fines or prevent riots and ambushes. 13 Occasionally, the security forces would also evict suspect EOKA sympathizers from their houses and destroy civilians' property when used by insurgents. 14 Collective punishment would neither curb people's support for EOKA nor induce the Greek community to share information about the insurgency. 15 The government, however, expected the destruction of EOKA forces to change the attitude of the population. Specifically, Harding considered the elimination of EOKA mountain gangs an essential goal as they exerted disproportionate influence over villagers. 16 The government pursued this objective with large-scale operations consisting in the prolonged cordoning of mountain areas and the continuous search for EOKA gangs and village group members, hideouts, arms depots, transportation and communication lines.
Between November 1955 and February 1957 the security forces carried out a dozen large-scale cordon-and-search operations, but the ones mounted between April and July 1956 -Kennett, Pepperpot, Lucky Alphonse, and Spread Eagle -were among the most decisive. The British destroyed, captured, or expelled some of the largest mountain gangs, arrested suspects among EOKA village groups, seized arms and ammunition, and captured Grivas' diary that would become a major source of information about the insurgency (French 2015, 135-136;Robbins 2012, 726-727;Dimitrakis 2008, 383).
After the British offensive, EOKA called a unilateral truce that lasted only two weeks. Between September and November 1956 EOKA town groups resumed the fight with an unprecedented spate of incidents and assassinations (Holland 1998, 154). Britain responded by passing regulations that made the death penalty mandatory for anybody in possession of firearms or explosives. The mandatory death penalty would improve intelligence by compelling prisoners to disclose information about EOKA to avoid execution. 17 Prisoners would also join the British side as informers or spies to approach EOKA village groups and identify more affiliates. Between December 1956 and March 1957 the security forces built on their intelligence gains and targeted EOKA town groups forcing Grivas to call a truce in March 1957 (French 2015, 149-150).
EOKA had been contained, but insurgents could still control the local population through terrorism and the ideological appeal of Greek nationalism. 18 It was on the crucial issue of population control that Harding's campaign failed completely and EOKA could rebuild itself during the truce. The government correctly considered the Orthodox Church as a powerful obstacle to population control as the clergy actively promoted Enosis through the education system. 19 Harding, therefore, tried to establish an alternative system of inter-communal schools under direct government control, but his policy failed. While the Church condemned Harding's reforms as an attempt to destroy the Hellenic identity of Cyprus, EOKA successfully enforced a boycott against inter-communal institutes (French 2015, 182).
Harding's position deteriorated further as Britain announced in April 1957 the end of the National Service, which made a massive deployment of manpower in Cyprus possible only for few more years. Meanwhile, EOKA had recovered and in October 1957 insurgents resumed their assassination and sabotage campaign. With no political solution in sight and limited popular support for the government (Holland 1998, 206-208), Harding left in October 1957 to be replaced by Sir Hugh Foot.

Communal conflict and the end of the Emergency (November 1957 -February 1959)
During Foot's governorship, Cyprus plunged into communal conflict. EOKA had initially spared the Turkish community from attacks; yet, as the mountain gangs came under pressure, Grivas targeted the Turks to start communal riots in urban areas and force Britain to withdraw its troops from the mountains back into the cities (French 2015, 258). EOKA's tactics, however, contributed to militarize the Turkish minority. By November 1957, Turkish Cypriots had created a paramilitary organization -Turk Mudya Eskilat (TMT) or Turkish Resistance Organization -with the support of the Turkish government.
Turkey had helped Britain to block Enosis at the beginning of the emergency, but Ankara demanded partition in return (Hatzivassiliou 1991). In June 1958 the Turkish government decided to coerce Britain and ordered TMT to attack and expel Greek Cypriots from those areas where the Turks were the majority in an attempt to impose partition. EOKA would retaliate against the Turkish community, thus causing a spiral of attacks and reprisals that killed over one hundred people by July 1958. Communal conflict marked Britain's complete loss of population control in Cyprus but Turkey eventually restrained TMT and ended violence as the Pan-Arab revolution in Iraq in July 1958 reminded Ankara that cooperation with Britain was vital to protect the Turkish borders (Holland 1998, 269).
Despite the opposition of Greece, Britain and Turkey eventually adopted the Macmillan government's plan on the status Cyprus based on administrative partition. EOKA waged its final offensive to prevent its implementation. Between September and December 1958 insurgents were responsible for over 430 incidents, including sabotage and ambushes against soldiers and British civilians (French 2015, 274). The British counteroffensive used dismounted patrols, pseudo-gangs, and cordon-and-search operations. Based on information from prisoners, between November and December 1958 the security forces captured or killed about 150 EOKA members (French 2015, 274-280).
EOKA started rebuilding itself but Grivas could not prevent Greece and Turkey from resuming bilateral negotiations after the Berlin Crisis in November 1958 made the two governments willing to solve the Cyprus dispute and focus on the Soviet threat. They reached an agreement in February 1959. Cyprus would become independent; Britain, Greece and Turkey would guarantee its sovereignty and the British would maintain their military bases (Holland 1998, Ch. 11). Makarios accepted the compromise and Grivas ordered EOKA to cease hostilities.

Case analysis
Do the events presented above confirm the three interlocking hypotheses of the force structure causal mechanism? This section answers in the negative and demonstrates how the force structure argument failed a relatively easy test.

Assessing H1: from manpower to popular support and intelligence gains?
The Cyprus Emergency case would not confirm the first part (H1) of the force structure causal process, according to which states that fight insurgents with manpower rather than machines will be able to win over the population and secure intelligence from civilians. Despite its reliance on infantry and police units in Cyprus, Britain never won over the Greek Cypriot majority; besides, Britain's intelligence gains during the emergency did not come from organized groups of Greek civilians trusting the government and turning against EOKA, but from captured documents and prisoners who talked out of fear.
Overall, interactions between members of the security forces and Greek Cypriot civilians were marked by reciprocal mistrust. A combination of genuine Greek nationalism and EOKA's use of provocative as well as retributive violence against the security forces and their potential or actual collaborators kept the Greek Cypriot community behind the insurgency.
The Greek irredentism that propelled Enosis was an issue that no amount of manpower proved able to handle. The Church and EOKA sustained a rigidly exclusive rightwing nationalist ideology that linked Cyprus to a Hellenic and Christian cultural universe where there was no place for other groups, especially the Turkish-speaking Sunni Muslim minority (Newsinger 2002, 91;Novo 2013). When observed through the lens of Pan-Hellenic nationalism, Turkish Cypriots appeared as Asian barbarians with no political or civil rights and a potential demographic threat to Greek civilization on the island (Heraclides 2012, 120-121;Kitromilides 1990, 13). Greek nationalism would be a major obstacle preventing interaction between the Greek majority and the police, especially after EOKA's initial offensive forced the colonial government to recruit police manpower from the Turkish minority. As the police grew into a Turkish organization, it lost its legitimacy and attracted the spurn that the Greek nationalist ideology had always had in store for ethnic minorities, thus keeping Greek civilians from sharing information.
The relationship between the Greek Cypriot community and British military personnel was not significantly different. Even if the Greek population did not despise British soldiers as much as they abhorred the Turks, the army was still the most obvious expression of Britain's opposition to Enosis and EOKA used provocative violence to ensure that the gap between the Greeks and the manpower of the British garrison would grow as wide as possible.
EOKA's frustrating hit-and-run tactics against the security forces and Greek Cypriots' reluctance to report about EOKA crimes exacerbated British soldiers' mistrust and resentment against civilians. British soldiers considered EOKA fighters as "terrorists", "thugs", or "gangsters" who shot their victims in the back (French 2011, 71). As most Greek civilians were perceived to cover up the crimes of EOKA "gangsters", they would be vilified as willing accomplices of terrorism too. Following EOKA attacks against soldiers and civilians, the British propaganda would give details of Greek people's allegedly sadistic attitude on the murder scenes. After EOKA shot two British servicemen in Limassol in August 1958, for example, the government radio would describe how a Greek bystander, ostensibly delighted at the view of British blood, had refused to help staunch the wounds of the victims (Holland 1998, 268). EOKA's murders, therefore, were taken as unmistakable evidence of "the moral delinquency" of Greek Cypriots as a whole (Holland 1998, 288).
EOKA attacks were designed to exacerbate anti-Greek feelings, outrage the security forces and induce British soldiers to ill-treat non-combatants and further alienate Greek Cypriots from the government. The assassination of British civilians was one of the most successful provocative tactics by Grivas. For example, in the hours following the assassination of a soldier's wife in Famagusta in October 1958, British soldiers vandalized Greek property, rounded up and arrested every Greek male in the vicinity of the murder scene, beat up suspects publicly and injured at least one hundred of them. 20 Rioting soldiers also caused the death of two men and a teenager girl before discipline could be restored (French 2015, 209), thus giving EOKA a major propaganda victory in the struggle over population control.
The mistrust between the Greek population and the British army would grow even further during the period of communal violence. EOKA's retaliations on the Turks increased insurgents' legitimacy as the protector of the Greek community from the TMT threat, while forcing Britain to escalate repression. As the British tried to stop the spiral of attacks and reprisals by arresting large groups of suspects indiscriminately, Greek detainees would greatly outnumber Turkish detainees and EOKA propaganda would inflate mass arrests and other incidents as evidence that Britain had sided with the Muslim minority. 21 EOKA, however, did not only use the appeal of the Greek nationalist ideology and provocative violence to drive a wedge between civilians and the security forces. In fact, insurgents made extensive use of retributive violence to deter cooperation with the British. EOKA's definition of "traitor" encompassed fighters who had talked (or were deemed to have talked) during interrogations, government and police informers, civil servants, communists and left-wing trade unionists, unsympathetic journalists and editors, teachers, students, and even the parents of pupils who refused to adhere to demonstrations, boycotts, and school riots (French 2015, 158-167). With only 106 police stations in 639 villages and towns, the colonial government could not provide individual police protection to such a broad group of potential victims, therefore EOKA could assault, intimidate, or assassinate hostile individuals and discourage collaboration effectively (French 2015, 169-170). Overall, spontaneous cooperation from Greek Cypriot civilians was extremely limited, as is suggested by the very low number of Greek collaborators that had to be relocated to Britain to save them from nationalists' revenge after the emergency (French 2017).
In sum, while H1 posits that reliance on manpower makes the counterinsurgent more likely to win over the population, that did not happen in the case of Cyprus. Britain preferred manpower over machines and still failed to convince the Greeks to cooperate against EOKA, which contradicts H1.
Interestingly, the British did secure significant intelligence gains at different stages during the campaign even without popular support. The British obtained information about EOKA not from groups of Greek civilians turning against the insurgency and supporting the colonial government, but from captured documents and prisoners talking under threat and out of fear (Dimitrakis 2008, 383). That runs counter to the first part (H1) of the force structure argument according to which civilians should share information out of trust towards a counterinsurgent that prefers manpower over armored vehicles.
The security forces seized EOKA documents during surveillance and cordon-and-search operations based on information from detainees. 22 The most authoritative source of information about EOKA the British captured during cordon-and-search operations was Grivas' own diary, a document of 250,000 words describing the structure, organization, funding system, routine practices, and tactics of EOKA in detail. The diary confirmed that EOKA's communication system depended almost entirely on typewritten messages carried by civilians serving as couriers. Couriers secretly dropped their letters at venues used as post boxes, like shops, bus and petrol stations, or pharmacies where documents would be collected before reaching their destination (Dimitrakis 2008, 383-384). Tracking the couriers in surveillance operations, the British would raid post boxes or spot the hideouts of insurgent gangs where they would arrest EOKA affiliates and seize documents containing orders and instructions from Grivas, reports from Grivas' subordinates to their leader, plans of attack and even name lists of EOKA killers and bombers (Slack 2019, 102).
Once arrested, EOKA members would become an important source of intelligence, but for reasons that have little to do with the posited logic of the force structure thesis. Prisoners revealed information because the tempo and success of British of operations had lowered their morale, as was the case of fighters captured during the 1956 offensive against EOKA mountain gangs. 23 More often, EOKA prisoners talked under threat of mandatory capital punishment for possession of firearms and explosives. The threat of execution was all the more effective due to the young age of many EOKA members, including town group killers, that quickly implicated others to avoid capital punishment. 24 At times, prisoners shared information because of physical violence during interrogations, but torture never became a routine practice during the British campaign in Cyprus (French 2015, 233-235). Finally, some EOKA prisoners collaborated because of death threats from their own organization. Grivas' merciless policy to execute actual or potential traitors even on the basis of mere suspicions compelled individuals who felt already condemned by EOKA to talk or collaborate in order to be eligible for police protection (French 2017, 74).
In conclusion, the causal logic of the first part (H1) of the force structure thesis is not confirmed: Britain faced people's enduring hostility despite its reliance on manpower and achieved intelligence gains in spite of people's mistrust and reluctance to cooperate.

Assessing H2: from intelligence gains to selective violence?
The British campaign against EOKA shows some evidence apparently confirming H2: indeed, Britain tended to use force selectively and British self-restraint correlates with intelligence breakthroughs, even if information about EOKA did not come through the causal process and logic of the force structure thesis, as shown above. Consistently with the observable implications of H2, the British never used counter-terror measures like scorched-earth, indiscriminate shootings or bombings as a reprisal against civilians after insurgents' attacks, nor did Britain rely on torture or capital punishment on the same scale as it did in Kenya. However, the apparent congruence between thesis predictions and case observations hides a weaker causal connection than it may appear at first.
Indeed, Britain was self-restrained before and after beginning to secure major intelligence gains in mid 1956. In other words, within-case variations in the amount of intelligence information available to the British do not correspond to any variation in the level of self-restraint of the colonial government during the campaign. That would question the causal nexus between information about the insurgency and selective violence as specified in the second part (H2) of the force structure thesis.
Importantly, evidence suggests that Britain's caution about indiscriminate violence was determined by different factors, like the risk of media exposure of abuses to the international public opinion and the influence of third-party states rather than just the amount of intelligence available. In Kenya and Malaya the presence of free media that could report about violence against civilians was very limited; in addition to that, the Malayan Communists and Mau Mau rebels did not have any state patron to support them diplomatically and protest about violations of human rights (Bennett 2013;Ucko 2019); therefore, targeting civilians was a safer policy. EOKA, instead, could count on the diplomatic sponsorship of Greece, a NATO member state that was ready to raise the issue of human rights violations in support of EOKA propaganda at the United Nations and other international fora, like the Council of Europe. 25 Furthermore, the colonial government in Cyprus was under the watchful eye of a high number of journalists ready to report about stories of brutality against civilians (French 2015, 168-169).
Consequently, the British were fully aware of the potential political backlashes of indiscriminate violence and considered self-restraint a more convenient course of action. For example, Harding resorted to collective punishment but, under the supervision of the Colonial Office, he took precautions to limit the level of harm on civilians: fines would be proportionate to individual wealth so as to avoid financial ruin; house evictions would be used only against those individuals who had alternative accommodation in order to avoid homelessness and hunger; property and crop destruction could be followed by monetary compensation depending on the specific circumstances. 26 The colonial government took this kind of precautions since December 1955 when Britain was still struggling to get information about EOKA. Self-restraint in spite of limited information about the enemy is inconsistent with H2 and its observable implications.
Likewise, Harding's policy to deport Orthodox clergymen from Cyprus was subject to government controls and restrictions since the very beginning of the counterinsurgency campaign. The Cabinet was concerned that deportation of Church leaders would be bad publicity and could have made it difficult for Greece to encourage Makarios to resume negotiations; the government, therefore, recommended self-restraint despite the "continuing provocation" of the Church and the prominent role of the Orthodox clergy in supporting the rebellion financially and ideologically. 27 Moreover, regardless of the amount of information they had about EOKA, during the entire emergency British leaders never considered or even discussed indiscriminate reprisals or mass deportation of Greek civilians to concentration camps. When British leaders explicitly discussed measures that may have resulted in the death of a large number of Greeks -like the extension of the mandatory death penalty -the Cabinet's concerns about negative publicity and the ensuing diplomatic problems came up and restrained the colonial government (Holland 1998, 172-175). While the British government granted Harding's requests on the capital punishment regulations, the Cabinet also sought and obtained assurance from the Governor that the prerogative of mercy would be applied systematically and executions would be actually allowed only under exceptional circumstances. 28 The same type of concern was behind self-restraint during military operations against EOKA. Harding, for example, acknowledged that air bombings could destroy insurgents' morale, as they did in Malaya and Kenya;yet, in June 1956 the Governor would let the Commander in Chief of the Middle East Land Forces know that aggressive air operations should be restricted to exceptional circumstances in Cyprus due to "important political considerations", that is the risk of hostile international political propaganda in case of civilian fatalities. 29 Importantly, concerns about media exposure and the involvement of third-party states in the conflict were constant, exactly like Britain's self-restraint; therefore, one may surmise that those factors may have had a more consistent causal relation to Britain's preference for selective violence than intelligence availability. In conclusion, while intelligence breakthroughs after Summer 1956 contributed to cause self-restraint, H2 is not decisively confirmed because other factors co-determined Britain's preference for selective violence and, arguably, they did so to a larger extent.

Assessing H3: from selective violence to victory?
The final part of the force structure causal process (H3) holds that intelligence-driven selective violence in counterinsurgency leads to victory because it destroys the insurgent organization without alienating the people. The Cyprus Emergency does not fully confirm this hypothesis: Britain did use violence selectively and successfully contained EOKA after inflicting important damages on Grivas' organization, but the outcome of the campaign was not a clear victory over the insurgency. Britain attained some of its political goals, but mostly because of its alliance with Turkey rather than selective violence.
As shown above, Harding identified the conditions of victory with the destruction of EOKA beyond all hopes of recovery and the spontaneous rejection of Enosis by the Greek Cypriot community. By the end of the campaign, neither condition had been met and Britain had lost its colony. At the same time, however, Britain had averted Enosis and maintained its military bases. The outcome of the campaign was not a victory for either sides.
The survival of EOKA and its ability to rebuild itself due to population control would challenge the contention that selective violence is key to victory. In the case of the Cyprus Emergency, selective violence was certainly not decisive. Despite the success of Britain's offensives against EOKA, in late 1958 the insurgent still retained a formidable arsenal, looted large amounts of dynamite from mines to resume its sabotage campaign, and was going to recruit more fighters and receive more arms from abroad. 30 The reason Grivas accepted to end the insurgency depended on the shifting position of the Greek government and Makarios with regard to Enosis since Summer 1958, rather than the damage that Britain had inflicted upon EOKA. The Greek government abandoned the goal of Enosis and accepted independence as the least of evils after it became clear that Britain and Turkey would implement the Macmillan plan imposing a de facto administrative partition, which would possibly prelude to territorial partition (Assos 2011, 114). Greece had always been concerned that a prolonged terrorist campaign in Cyprus may have resulted in that kind of outcome, but it still supported EOKA to control and restrain Grivas (Klapsis 2019). As the risk of territorial partition became too high to bear, Greece decided it would no longer support the armed struggle. Since EOKA's strategy depended on the support of the Greek government, Grivas acknowledged the insurgency would be internationally isolated once Greece had withdrawn its sponsorship. 31 Makarios himself followed the Greek government, which further isolated Grivas and the die-hard nationalist members of the Orthodox Church and the Ethnarchy Council. 32 Turkey played a key role in shifting the position of Greece and Makarios by militarizing the Turkish minority and creating the conditions that made the threat of partition credible. Indeed, during the communal conflict in Summer 1958, Turkey and the TMT demonstrated for the first time to the Greek majority their resolve and power to split the island. Turkish paramilitaries, under the direct control of Ankara, established separate municipalities and began to rule portions of the Cypriot territory with the compliance of Turkish civil servants and police officers. 33 The Greek government and Makarios understood that the territorial gains of the Cypriot Turks could not be reversed anymore, therefore continuing the insurgency even after the Anglo-Turkish agreement on the Macmillan plan would only increase the risk of two separate states on the island or even war with Turkey (Assos 2011, 115). As Turkey showed interest in a final settlement with Greece after the Berlin crisis, Athens seized the opportunity. Consequently, Grivas was only left with the options of fighting alone or accepting a compromise short of union with Greece.
In conclusion, the use of selective violence definitely placed Britain in a position of advantage over the insurgency, but it did not break EOKA's ability and will to fight. Britain's partnership with Turkey and Ankara's indirect military intervention in the campaign eventually caused the insurgency to end on terms that were not a victory for Britain. H3, therefore, would not be clearly confirmed.

Explaining Britain's failure to win
This article aimed to test the force structure argument and the case of Cyprus was selected for that specific purpose. However, this section will briefly address a question arising from the case itself: if the force structure thesis lacks analytical leverage, which argument would explain the outcome of the Cyprus Emergency?
The case corroborates the thesis that the rise of the self-determination principle in the state system during the twentieth century defeated counterinsurgency in three ways: it enhanced insurgents' external support, hindered the recruitment of local collaborators, and constrained the use of brute force (MacDonald 2013; Smith and Toronto 2010). In Cyprus, ethnic nationalism and external intervention supporting the self-determination claims of the Turkish and Greek Cypriot communities made local elites unavailable to support Britain's political agenda. At the same time, the diplomatic involvement of Greece and Turkey discouraged Britain from using the coercive measures that had yielded victory in Malaya and Kenya.
Britain could build no durable partnership with the leaders of the two communities on the island. In their meetings with Lennox-Boyd in July 1955, for example, both the leftwing and the right-wing mayors of Cyprus stated to the Colonial Secretary they would only support self-determination and union with Greece (Holland 1998, 62) with Greek diplomatic support. Britain eventually relinquished all hopes to find cooperative groups in the Greek Cypriot majority (Holland 1998, 221-222) and turned to the Turks. After a short period of collaboration, however, Cypriot Turks would turn against the British colonial government. Turkish Cypriot leaders aimed at self-determination and made it clear to Britain that they would only follow the Turkish government's instructions to attain partition (Holland 1998, 219, 252). As the descent into communal conflict would prove, the counterinsurgent simply had no collaborators to support continuing British rule over Cyprus. Before the Greek-Turkish final agreement, the British accepted a de facto administrative partition with the Macmillan Plan. Yet, as the Colonial Office acknowledged, this policy meant that Britain would follow and implement Cypriot Turks' political agenda, not the other way around (Holland 1998, 276-277).
Britain could not compensate for the absence of collaborators by escalating coercion and compel the people to support the government. One may speculate that Britain, after containing EOKA, might have destroyed Grivas' organization by deporting the Greek population. Yet, as explained in the case analysis, mass deportation would trigger diplomatic backlashes Britain was unprepared to withstand. Nor could Britain resort to coercion when Turkish Cypriot leaders turned rebellious. Indeed, violent repression of Turkish Cypriot nationalism would accelerate Ankara's indirect military intervention and territorial partition (French 2015: 263, 267). Besides, as the Cyprus police depended on the Turks, the option of police repression against Turkish Cypriot self-determination was seriously compromised.
Left without local collaborators and the coercive options that had curbed the insurgencies in Malaya and Kenya, Britain could only contain EOKA and leave it to the external powers that had supported the nationalist aspirations of the two communities on the island to find a political solution. In sum, the case suggests that counterinsurgency success is not about choosing the most appropriate force structure or simply adopting sophisticated and intelligent military strategies: counterinsurgency outcome is shapedeven if not in a deterministic fashion -by political conditions that are likely to be unfavorable, especially during campaigns against nationalist insurgencies abroad.

Conclusion
This article tested the force structure argument against the Cyprus Emergency, a case that should have facilitated the causal process and logic of the thesis. Yet, the thesis did not pass the test. The case outcome resulted from political context rather than the British garrison's mechanization level. The failure of the force structure argument to explain a relatively easy case would justify skepticism about the chances of the thesis to pass more difficult tests and account for cases in which the dominance of manpower over machines is less prominent.
On the policy level, the test performed in this article would caution against the belief that it is necessary to lower the mechanization level of the armed forces in order to defeat insurgencies. The British campaign against EOKA suggests that the dominance of manpower over machines may have little to do with population control, intelligence information gains, and war outcome. Lowering the degree of mechanization may reduce states' chances to win modern conventional wars without necessarily improving states' performance in counterinsurgency. On the other hand, mechanized forces have provided essential force protection in counterinsurgency, especially against insurgents that can rely on heavy firepower or shift from guerrilla to conventional warfare (Smith and Toronto 2010;Oliver 2011). All that may be good news for defense planners and policy-makers, as it means that preparations for high-speed maneuver warfare will not make armies unprepared to defeat insurgencies. Consequently, the United States' shift towards mechanized warfare may not necessarily be a missed opportunity to succeed at counterinsurgency.
The case study in this article would also have implications for the academic literature. Firstly, the case would question the population-centric argument that civilians are an indispensable source of intelligence and, therefore, the government should not alienate the populace. The Cyprus Emergency shows an alternative pathway to intelligence gains that is relatively independent from people's consent. Interestingly, the way Britain gained intelligence does not exactly correspond to the enemy-centric pathway based on mass terror and bribery of local elites either (Hazelton 2017). Indeed, Britain gained information through surveillance operations to seize documents, attrition against EOKA to take prisoners, and interrogation of captives under threat of capital punishment. The case, in sum, highlights a middle ground between coercion and persuasion, which confirms how different paradigms are simply ideal types coexisting in practice (see Paul et al. 2016). Even if counterinsurgency success does not depend exclusively on intelligence, information about the enemy matters; therefore, future research should keep delving into the existence, relative importance, and viability of different pathways to intelligence gains.
Secondly, the Cyprus Emergency would invite a deeper look into the role of coethnics in counterinsurgency. Some scholars hold that coethnics are effective due to their knowledge of local society, which would help undermine support for the insurgency (Lyall 2010b;Souleimanov and Aliyev 2015). Others express skepticism arguing that coethnics' excessive brutality and corruption may invite resistance (Šmíd and Mareš 2015). The Cyprus Emergency corroborates the skeptical view, but for additional reasons. Indeed, the case analysis suggests that coethnics may be ineffective when a third-party state can influence their allegiance and the insurgent has successfully infiltrated them. Greek Cypriot police agents were sympathetic towards union with Greece and their ranks hosted EOKA affiliates that facilitated insurgents' offensive. Turkish Cypriot police officers, on the other hand, were reliable against EOKA, but shifted their allegiance to Turkey and the TMT when Ankara and its proxies tried to impose partition. Future research should explore whether and how coethnics' loyalty can be guaranteed under those conditions. Current studies are insightful, but they privilege cases in which the influence of external powers on coethnics is not strong (Souleimanov et al. 2018).
Finally, this article is a reminder that the accumulation of knowledge on counterinsurgency requires the specification and rigorous assessment of causal logic and mechanisms. As Mearsheimer and Walt (2013: 438, 443-445) pointed out, Lyall and Wilson's scholarship reflects a tendency to privilege "simplistic hypothesis testing", that is the use of social scientific methods to measure covariations among variables at the expense of theorydriven efforts to verify why and how causal relations may obtain. Without increasing such efforts, however, research on counterinsurgency may produce a growing body of empirical findings not unified in coherent explanations. While this paper did test a hypothesis, it did so to demonstrate the enduring validity of this argument.