I. Setting the scene: the 3 January 2020 incident

It is well accepted that for a non-international armed conflict (NIAC) to begin, a minimum threshold of intensity of the fighting is needed.Footnote ² Following this approach, the legality of targeted killing operations against the members of non-state armed groups (NSAG) in the past two decades has been analysed either within the context of an ongoing armed conflict under international humanitarian law (IHL) or outside the context of armed conflict and, therefore, under international human rights law (IHRL).Footnote ³ In other words, the beginning of a NIAC based on the occurrence of a single lethal attack against a member of an NSAG has never been a central focus of the legal deliberations surrounding such operations.

On the contrary, due to the dominance of the famous first shot theoryFootnote ⁴ in the sphere of international armed conflicts (IAC) classification, it seems acceptable to argue that an initial use of military force by a state against military personnel of another state, per se, will trigger the beginning of an IAC between the two states. This theory was introduced by Jean Pictet in interpreting Article 2(1) common to the four Geneva Conventions (GCs).Footnote ⁵ However, the targeted killing of an Iranian commander by United States forces on 3 January 2020, due to its occurrence on the soil of a third state, Iraq, raised serious doubts, as will be discussed below, about the appropriateness of applying the first shot theory in these kinds of incidents.

Inspired by this concern, this paper seeks to reconsider and rediscuss the beginning of an IAC in the view of targeted killing in a third state with a focus on rationale, as well as the background and humanitarian object and purpose of the Geneva Conventions. The paper argues that, regardless of the lawfulness of the targeted killing of military personnel in the territory of a third state under jus ad bellum or other rules of international law, such an operation does not automatically trigger an IAC between the targeting state and the state of the targeted individual. By examining the elements that constitute an IAC, the paper proposes a revision to the “first shot” formula and argues that this new interpretation aligns more closely with the object and purpose of the GCs. In this regard, the details of the 3 January 2020 operation are referred to solely for the purpose of illustrating the issue at stake, and not to comment on its lawfulness, which has already been reviewed extensively, as discussed below.

On 3 January 2020, Major General Soleimani, the commander of the Iran Quds Force responsible for Guard Corps foreign operations,Footnote ⁶ was killed by missiles fired from a United States (US) MQ-9 Reaper drone.Footnote ⁷ The attack took place in Baghdad, the capital city of Iraq, where both Iranian and American forces were present to support the Iraqi government in its ongoing internal security issues. Iraq condemned this operation as an aggression against the state, the government, and the people of Iraq.Footnote ⁸ The Iraqi prime minister also stated that he had been expecting to meet with General Soleimani on the day he was killed to receive a message from Iran for Saudi Arabia, as part of Iraq’s mediation between the two countries.Footnote ⁹ On the same day, the president of the United States announced officially that this attack was conducted under his direction.Footnote ¹⁰

The novelty of this case – the use of military force against a member of the armed forces of one state in the territory of another state, outside the context of an armed conflict - has prompted many scholars to analyse the lawfulness of this operation from various perspectives. Anthony Dworkin, for example, discussed how the Soleimani strike marked a novel shift in targeted killings. He compared this occurrence with the most similar incident, which was the US operation in 1943 against Isoroku Yamamoto, the Japanese naval commander responsible for the attack on Pearl Harbour; highlighting that the attack against Yamamoto took place at a time when the two countries were engaged in an all-out war, “[b]y contrast, the United States targeted Maj. Gen. Soleimani … at a time when there were no direct conventional hostilities underway between the United States and Iran”.Footnote ¹¹ Other commentators have compared the attack to operations carried out for assassinating foreign government officials during peacetime, a policy practiced by certain states.Footnote ¹² However, there are significant differences between these operations and the 3 January operation. Most of the former operations were conducted covertly by intelligence forces on the soil of the targeted person’s territorial state. Quite the opposite, the operation of 3 January 2020 was conducted in the territory of a third state, implemented by military forces, and immediately recounted in detail for the public by the US president.Footnote ¹³

Some scholars have examined the operation from the perspective of jus ad bellum, assessing whether the United States could lawfully invoke its right to self-defence, and commented on the legality of this operation based on this assessment.Footnote ¹⁴ However, overall, very few analyses were conducted for the assessment of such an attack under jus in bello.Footnote ¹⁵ For example, the United Nations (UN) ex-Special Rapporteur on extra-judicial, summary, or arbitrary executions, Agnès Callamard, while admitting that the US strike against General Soleimani did trigger an IAC,Footnote ¹⁶ argues for the application of human rights law.Footnote ¹⁷ Drawing on comparisons with extra-territorial targeted killings in NIAC, the Special Rapporteur raises the question of whether non-belligerent circumstances should be considered a decisive factor in legal assessment of an attack against a state official on the territory of a third state.Footnote ¹⁸ Unfortunately, the ex-Special Rapporteur did not follow this line of argument, rather contending that even if the beginning of IAC is to be accepted, IHL cannot be considered as the best “fit” for the situation due to the presence of people unaware of and unprepared for an IAC.Footnote ¹⁹ This partly policy-based argument,Footnote ²⁰ however, cannot by itself provide a definitive answer to the question of whether IHL applies to such killings of military personnel in the territory of a third state.Footnote ²¹ In general, the classification of the resort to armed forces by state A against state C in the territory of state B, as also acknowledged by Djemila Carron, has not received much attention in the literature on jus in bello. Footnote ²²

In this paper, we posit that the “first shot” theory cannot be applied in situations where the first strike targets a member of the armed forces of another state on the territory of a third state. Such an application would result in adverse consequences contrary to the humanitarian object and purpose of the GCs. As Eric Marin, the president of the International Committee of the Red Cross (ICRC) at the time of the adoption of additional protocols to the GCs, aptly stated, the purpose of these conventions is “not to serve individual or national interests … but to protect human life in the interests of the entire international community”.Footnote ²³ By extending the “first shot” formula to these types of operations, we risk undermining the very foundation of the GCs and the broader framework of IHL, which seeks to limit the impact of armed conflict on individuals. We argue that applying IHL norms to operations like the one on 3 January may inadvertently violate the human right to life. This could set a dangerous precedent, encouraging states to conduct extra-judicial killings of military officials from other states wherever they are found. Such practices would not only violate a fundamental human right but also undermine the protective mechanisms established by the GCs, leading to a more permissive environment for military operations focused solely on killing.

In light of this humanitarian concern and the novelty of the subject matter, we propose a different criterion for the beginning of an IAC in the scenario of targeted killing of the state’s military officials in the territory of a third state, without prior armed conflict. To achieve this, the paper applies the rules of treaty interpretation, as formulated in Articles 31 and 32 of the Vienna Convention on the Law of Treaties (VCLT).Footnote ²⁴ These rules were adopted after the adoption of the GCs, yet, as described by the European Court of Human Rights (ECtHR), they “enunciate in essence generally accepted principles of international law”, which permitted the Court to apply them retrospectively to a treaty that was entered into force before the VCLT.Footnote ²⁵ Today, these rules are considered to reflect the customary rules of international law, and as such have been applied to all treaties, irrespective of the adoption date, subject matter, objective, or number of states party to the treaty.Footnote ²⁶

In developing our arguments and for the sake of clarity, we assume the following scenario: state A through a unilateral use of lethal force kills a member or several members of the armed forces of state C in the territory of state B without any prior armed conflict between states A and C. The paper focuses on the classification of the situation between state A (the targeting state) and state C (whose member of the armed forces is killed in the territory of state B).

It argues that no IAC begins between states A and C due to the act of targeting military personnel in the territory of a third state. For this purpose, since common Article 2 does not encompass a definition of an armed conflict, the paper starts with reviewing the dominant interpretation of this Article, the well-known “first shot” formula, as well as other major views regarding the IAC classification, and discusses why none of these interpretations can be fully upheld. Considering this and drawing on the elements of Article 31 of the VCLT, part three of the paper argues that the beginning of an IAC necessarily requires a military intervention in the form of unconsented-to invasion or military operation in the territory of the attacked state. On this basis, it argues that in the case of the unilateral use of force by state A against a member or a number of members of the armed forces of state C occurring in the territory of state B, without any prior armed conflict, another criterion is needed for the beginning of an IAC between states A and C. The final part of the paper is a confirmation of this interpretation, as stipulated in Article 32 of the VCLT, by the review of the historical context that led to the adoption of the common Article 2(1), including the rationale behind such formulation, which remained valid even during the adoption of the first 1977 Additional Protocol.Footnote ²⁷

II. First shot theory and its opponents

In the absence of a definition of “armed conflict” in common Article 2 of the GCs, scholars diverge on what acts, under what conditions, trigger the beginning of an IAC. According to the majority opinion, when state A uses force against state B, an IAC begins.Footnote ²⁸ This opinion has originated from the so-called first shot theory reflected in Pictet’s 1952 ICRC Commentary on the first Geneva Convention, which states that “the Convention becomes applicable as from the actual opening of hostilities … It makes no difference how long the conflict lasts, or how much slaughter takes place”.Footnote ²⁹

While this doctrine remained for years as the prominent interpretation, new developments in ways of resorting to armed force by states led some scholars to challenge this doctrine. Remarkably, the International Law Association (ILA) contends that no resort to armed force by states amounts to an IAC; rather, its existence depends on a threshold of intensity.Footnote ³⁰ The proponents of this view argue that a customary norm has emerged to include intensity as the second criterion for the beginning of an IAC and therefore maintain that isolated incidents, such as border clashes and naval incidents, are not classified as armed conflict.Footnote ³¹ In this regard, what is important and relevant to our discussion is that, according to this approach, the sought intensity can never result from a one-sided unilateral strike. Hence, while in the ILA’s view numerous factors, such as the number of fighters, the duration and territorial extent of fighting, the number of casualties, the extent of the destruction of property, the displacement of the population, the type and quantity of weapons used, and the involvement of the Security Council or other actors to broker cease-fire efforts are measured to assess the intensity of confrontations,Footnote ³² these should be always considered in the context of fighting between the two sides. The ILA report emphasises this point and states that “an armed attack that is not part of intense armed fighting is not part of an armed conflict”.Footnote ³³ In this way, a significant one-sided armed attack without any response does not amount to an act that can trigger the beginning of an IAC.Footnote ³⁴

Although this result is in accordance with the main tenet of this paper, we cannot adhere to the criterion of intensity. As already mentioned by many scholars, an approach that subjects the determination of an IAC to the fulfilment of such conditions weakens the protection regime of the GCs.Footnote ³⁵ Pejic in her insightful article states that:

It seems that the ILA was also aware of this concern and, to compensate, advocated for separating the application of IHL from the beginning of armed conflict, as it stated: “it appears possible to allay the concern about the application of IHL. It could be applicable to military operations even in the absence of armed conflict without redefining ‘intense fighting’ to include significant one-sided first strikes”.Footnote ³⁷ In other words, such an attack, in the ILA understanding, is governed by IHL, although it does not trigger the beginning of an IAC. Yet, this approach seems confusing because the primary goal for determining the beginning of an armed conflict is to establish whether the special legal regime governing the situations of armed hostilities applies. Moreover, it remains unclear whether this approach would result in a conclusion that outlaws the killing of armed forces, which is the main concern of this paper.

Recently, an attempt has been made to merge the two theories of the first shot and intensity in order to “maximize the advantages of both without including their respective drawbacks or creating new negative consequences”.Footnote ³⁸ This new approach, developed by T.D. Gill, is based on separating the applicability of at least some humanitarian law provisions from the existence of an armed conflict, “reserving the full applicability of the law of armed conflict to situations that cross a certain threshold of intensity and/or duration and constitute an armed conflict”.Footnote ³⁹ In this way, Gill advocates for the application of protective measures of Geneva law in the case of minor hostilities. For example, he argues that in armed incidents below the required threshold, PoWs cannot be held until the end of active hostilities as stipulated under GCIII; rather, they would have to be repatriated as soon as the incident is over.Footnote ⁴⁰ In his view, “targeting in the context of such armed incidents would be restricted to those persons and objects that were directly involved in the armed incident and only to the extent they posed a threat of death or serious injury”.Footnote ⁴¹ He continues: “[e]ven then, use of force would be prohibited if there were other non-forcible alternatives available and adequate to address the situation”.Footnote ⁴² Gill argues that, in the context of his theory, the jus ad bellum principles of proportionality and necessity, as applied in the case of self-defence, are identically or analogously applicable as restrictions on targeting in situations of low threshold hostilities.

This is a very interesting insight. Nevertheless, in our view, aside from the inherent complexity in the relationship between jus ad bellum and jus in bello, this approach allows the initial attack by the attacking state to completely disregard the jus ad bellum principles of proportionality and necessity. This is because, unlike the situation of self-defence, there is no prior act in this scenario to enable any comparison or assessment. As such, this theory cannot provide much insight into the issue of a unilateral first strike in the territory of a third state.

Another opposite view raised against the first shot theory is an approach that can be described as classification according to the “target” of the attack,Footnote ⁴³ developed extensively by Carron. In her approach, it is argued that if state A uses force on the territory of state B against state C, such an act by itself does not trigger an IAC between state A and the territorial state B; rather, there will be an IAC between states A and C.Footnote ⁴⁴

Carron acknowledges that the history of the adoption of common Article 2 tends towards considering territory as a decisive element in triggering an IAC.Footnote ⁴⁵ She, however, deviates from this finding by emphasising that the ordinary meaning of “High Contracting Parties” denotes their identities as states and not their territories. Moreover, she argues that the object and purpose of the GCs’ call for rejecting the territorial criterion because it is aimed to cover all kinds of hostilities between states, and for this purpose, the territory is irrelevant, except for the case of occupation.Footnote ⁴⁶ She further explains that all attacks by state A against “external manifestations” of state C in the territory of state B trigger an IAC only between states A and C and not between A and B (the territorial state). Based on the literature on jus ad bellum, she considers the state military representations composed of, in particular, military bases and armed forces present in the territory of another state, as the first and most important external manifestation of states.Footnote ⁴⁷ She qualifies the situation between states A and B as an IAC only when state B as the territorial state resorts to armed force against state A.Footnote ⁴⁸

The classification of armed conflict between states A and B falls outside the scope of this paper; however, in our view, as will be discussed below, in the classification of IACs, the territory is and should remain a decisive factor in the beginning of an IAC. Meaning that from our perspective, in Carron’s scenario, an IAC begins between states A and B by applying the first shot theory and no armed conflict begins between states A and C by merely targeting a member or several members of the armed forces of state C by state A in the territory of state B.

III. The beginning of IAC and the necessity of military intervention in the territory of another state

The starting point of our interpretation, in accordance with Article 31 of the VCLT, is the ordinary meaning of terms used in common Article 2, considered in its context and in light of the object and purpose of the GCs. The subsequent practice and any other relevant rules of international law will be also considered together with the context. In this regard, it is important to add that “relevant rules of intentional law” allow the consideration of not only the formal sources of international lawFootnote ⁴⁹ but also the legal tradition, which includes judicial practice, soft law, and legal doctrines and expert bodies in particular when the matter is not expressly regulated.Footnote ⁵⁰ In this regard, the interpretations of the GCs by the ICRC, as the guardian of IHL with its more than 150 years of expertise in this field, may have special weight. This matter is evidenced by the fact that ICRC opinions are frequently referenced by States, courts, and international bodies,Footnote ⁵¹ and its work directly influences the development of IHL.Footnote ⁵² In this regard, it is also worth mentioning that the International Law Commission (ILC) Special Rapporteur on the issue of identification of customary international law, in his Fourth Report, has referred to the work of the ICRC and stated that: “…although the conduct of ‘other actors’ is not directly creative, or expressive, of customary international law, it may very well have an important (albeit indirect) role in the development and identification of customary international law”.Footnote ⁵³ The ILC Special Rapporteur on the issue of subsidiary means for the determination of rules of international law in a more direct way has stated that: “it would be hard to deny the influence of the teachings produced by the International Law Association, Harvard Research in International Law, ICRC, or the experts in State-empowered UN treaty bodies in certain areas of international law”.Footnote ⁵⁴ He also emphasised that the ICRC has played an important role in the development of the cotemporary international humanitarian law.Footnote ⁵⁵

The relevant parts of common Article 2 of the GCs stipulate that:

The wording of common Article 2(1) guarantees that irrespective of the declaration or recognition of the state of affairs by the parties, any situation of real armed hostility between two or more states is covered by the GCs, and consequently, the maximum protections are available to those protected under these Conventions. This will be only achieved if the term “armed conflict” is defined independent from the desire of the belligerent sides so that it covers all forms of “real” or “actual” military confrontation between states. Several states have incorporated the same concept in their military manuals. For example, in the US Department of Defense Law of War Manual 2023, it is mentioned that: “the United States has interpreted ‘armed conflict’ in Common Article 2 of the 1949 Geneva Conventions to include any situation in which there is hostile action between the armed forces of two parties, regardless of the duration, intensity or scope of the fighting”.Footnote ⁵⁶ The UK Joint Service Manual of the Law of Armed Conflict notes that the GCs and API do not contain any definition of armed conflict; however, in accordance with this manual:

The Military Manual of Denmark declares that: “it is the factual circumstances that determine whether an armed conflict is in existence. IHL is applicable even if one of the parties to the conflict does not recognise that an armed conflict exists”.Footnote ⁵⁸ The Law of Armed Conflict Manual of Germany stipulates that: “an international armed conflict triggering the applicability of Law of Armed Conflict (LOAC) exists if one State Party to a conflict uses armed force against another State. It is irrelevant, however, whether the Parties to the conflict consider themselves to be at war with each other, and how they denote this conflict”.Footnote ⁵⁹ The Australia Military Manual declares that: “the term ‘international armed conflict’ refers to conflict between nations in which at least one party has resorted to the use of armed force to achieve its aim”.Footnote ⁶⁰

However, in our view, the “real” or “actual” confrontations or hostilities between the belligerents do not necessarily mean actual fighting between the two. Rather, the underlying logic of “real” or “actual” hostilities is clearly seen in the second part of Article 2(2), which emphasises the application of the GCs in situations of partial or total occupation, even when there is no armed resistance.

Importantly, the 2016 ICRC Commentary explains that “[t]he hostile nature of an occupation derives from the unconsented-to invasion or presence of a State’s armed forces in the territory of another State”.Footnote ⁶¹ This explanation is without prejudice to the established definition of occupation as mentioned in the Hague Conventions providing that a “territory is considered occupied when it is actually placed under the authority of the hostile army. The occupation extends only to the territory where such authority has been established and can be exercised”.Footnote ⁶² In other words, the Commentary does not aim to define the occupation; instead, it is emphasising on the element of unconsented-to invasion or presence in the territory of another state as the determining factor for the realisation of an occupation.

Is it possible to use the same logic to state that even without explicit reference in Article 2(1), the IAC begins when the unconsented-to invasion or presence in the territory takes place? To put it another way, while there is no doubt that the use of armed forces is a necessary element in the beginning of an IAC, does it matter where and against what kind of target this military force is deployed? As mentioned earlier, the current literature on IAC does not delve much into determining the elements of armed conflict. Comparatively, some points can be inferred from the 2016 Commentary, though it still remains ambiguous on this point.

The Commentary, from one side, chooses the same wording used to identify the key factors of occupation to determine the conditions of the occurrence of an IAC and provides that:

It furthermore stipulates that: “[a]ny unconsented-to military operations by one State in the territory of another State should be interpreted as an armed interference in the latter’s sphere of sovereignty and thus may be an international armed conflict under Article 2(1)”.Footnote ⁶⁴

Based on our understanding, these phrases of the Commentary suggest that, as in the case of occupation, the ICRC cosniders an IAC also begins when an unconsented-to invasion or presence occurs in the territory of another state.

On the other hand, however, the Commentary states that: “[t]he existence of an international armed conflict is determined by the occurrence of hostilities against the population, armed forces or territory of another State …”.Footnote ⁶⁵ It further explains that: “… the use of armed force not directed against the enemy’s armed forces but only against the enemy’s territory, its civilian population, and/or civilian objects, including (but not limited to) infrastructure, constitutes an international armed conflict for the purposes of Article 2(1)”.Footnote ⁶⁶ It is worth mentioning here that the Commentary, in accordance with established rules of international responsibility of states, acknowledges that the use of force by state officials or persons qualified as “agents” of a state may suffice for the beginning of an IAC.Footnote ⁶⁷

As is evident in these quotations from the 2016 Commentary, the ICRC, in interpreting common Article 2(1), at times focuses on the element of invasion against the territory as the sole component of the beginning of an IAC. In other instances, however, the use of the conjunction “or” gives the impression that the existence of an IAC is determined by the occurrence of hostilities against the civilian population/objects/infrastructures, armed forces, or territory of another state. These ambiguities in the Commentary raise serious questions. For example, if we take the conjunction “or” seriously, we may legitimately argue that when state A’s agents or military forces capture a civilian citizen of state C on their own soil or in the territory of state B, an IAC begins between states A and C. In the same vein, we may contend that if the armed forces of state A destroy or confiscate a civilian infrastructure owned by civilian nationals of state C on the territory of a third state, an IAC begins to exist between states A and C. There is no need to argue how awkward these conclusions are. In other words, if this reading prevails, all cases of expropriation of foreign assets belonging to foreign investors by the forces or agents of the host state will be classified as IACs!

On this basis, we argue that the conjunction “or” is used to emphasise that when the unconsented-to invasion takes place against the territory, an IAC begins, and further attacks against the civilian population/objects/infrastructures or armed forces of the targeted state are not necessarily required to classify the situation as an IAC. In other words, the sole invasion or unconsented-to presence in the territory of another state, with or without further attacks against the civilian population/objects/infrastructures or armed forces of the invaded party, will result in the beginning of an IAC between the two states. The converse reasoning leads to absurd results as demonstrated above.

The state practice also confirms this conclusion. For example, in the situation of attacks by armed forces belonging to the receiving state against diplomatic missions or when the missions were attacked by the agents of the third states, as was the case in the 1999 bombardment of the Chinese embassy in BelgradeFootnote ⁶⁸ and the missile attack against an Iranian consular building in Damascus by Israeli forces on 1 April 2024,Footnote ⁶⁹ the involved states have generally condemned these attacks based on diplomatic and consular legal regimes, without any discussion of the beginning of an armed conflict and application of IHL.Footnote ⁷⁰ Even in its application before the International Court of Justice (ICJ) in the Diplomatic and Consular Staff Case in Tehran, the United States referred to the Vienna Conventions on Diplomatic and Consular Relations, the Convention on the Prevention and Punishment of Crimes against Internationally Protected Persons, the bilateral Treaty of Amity between two states, and the UN Charter, but it did not invoke the GCs at all.Footnote ⁷¹

Following the above, it can be argued that in cases of unilateral attack, only the use of force against the territorial integrity of another state triggers the application of the GCs. In this discussion it is important to recall that as vessels and aircraft on the high seas are exclusively subject to flag state “jurisdiction”, if state A attacks the aerial or naval forces of state C in or above the high seas, there exists an IAC between state A and state C.Footnote ⁷²

This means that the mere fact that state A resorts to unilateral use of force against a member of the armed forces of state C located in the territory of state B does not by itself begin an IAC between states A and C because no unconsented-to presence in or invasion of the territory of state C has occurred.Footnote ⁷³ This crucial point entails that in the case of the unilateral use of force by state A against a member or a number of members of the armed forces of state C occurring in the territory of a third state, without any prior armed conflict between states A and C, the attack, notwithstanding its size or lethality, does not by itself result in the beginning of an IAC between states A and C.

Hence, when the first strike occurs in the territory of a third state, another criterion is needed for the onset of an IAC between states A and C. We maintain that in such circumstances, an IAC begins only upon an explicit declaration of war by the state whose armed forces member(s) is shot outside its territory, as provided for in Article 2(1), or when targeted state subsequently resorts to armed force within a time frame that can be objectively related to the initial attack and forms part of the same operational context. As a result, when there is no declaration of war by state C or no military reaction from its side as a response, no IAC exists between state A as targeting state and state C, as the state whose member of the armed forces has been targeted outside its territory.

In our view, such an interpretation is compatible with the object and purpose of the GCs. The very purpose of the first IHL principle, that is, the principle of humanity, is to decrease the negative impacts of armed conflicts on civilians, those hors de combat and civilian objects. The principle of distinction, in line with the principle of humanity, allows for targeting combatants and military objectives during an armed conflict. These basic and fundamental rules are for the purpose of humanising the armed conflicts, but in no way can the legitimacy of targeting combatants under IHL be interpreted as permitting “unlawful killing” in situations outside of an armed conflict. As noted by Geiss,

Such an interpretation is even more compelling as the application of IHL in this scenario provides no “real” protection. Even if the attack does not result in death but leads to the detention of the person in uniform, granting him/her PoW status serves no practical purpose. The most important advantage of being considered a PoW is immunity from punishment for merely taking part in hostilities. In our scenario, however, there is no room for applying such a privilege, because there existed no prior hostility for the person captured to take part in. Even if that person kills or injures the attacker, this will amount to an act of self-defence recognised under all criminal legal systems leading to their acquittal. As mentioned by Clapham,

This interpretation also accords with the interpretation in good faith as it refrains from treating IHL solely as a permissive set of rules that provides no protection, but instead serves the purpose of “assassination-laundry” where states might seek to disguise extra-judicial killings as lawful military actions. We believe that by adhering to this good faith interpretation, we preserve the integrity of the Geneva Conventions, ensuring that they remain a legal framework designed to safeguard humanity and to save life.

The historical background of the adoption of common Article 2 also demonstrates that territory was an integral element in the definition of armed conflict at the time of the adoption of the GCs. This topic, which will be discussed in the next section, will confirm our interpretation resulting from the application of Article 31 of the VCLT.Footnote ⁷⁶

IV. Beginning of international armed conflict in preparatory work of Article 2(1)

Common Article 2 of the 1949 GCs, when drafted, was a complete novelty as the 1929 Conventions had no provision indicating the circumstances of their application, because “presumably, at that time, these [treaties] were considered as war conventions and the word ‘war’ was not thought to need a definition”.Footnote ⁷⁷ The inter-war period demonstrated the wrongness of this presumption since, from the formalistic perspective of the third 1907 Hague Convention,Footnote ⁷⁸ the applicability of the GCs could have been contested in the case of non-recognition or undeclared wars.Footnote ⁷⁹ During this time, on only one occasion did the straightforward application of these instruments occur, namely in the Chaco war (1932–5) when Paraguay declared war against Bolivia during their ongoing armed conflict.Footnote ⁸⁰ On the contrary, during the war of Ethiopia of 1935–6, started by the invasion of Italian troops of Abyssinia (today Ethiopia), in which there was no declaration of war by either side,Footnote ⁸¹ the ICRC delegates sent to Ethiopia “would have to stand by helplessly and watch as mustard gas was released and hospitals were bombed by the Italian air force.”Footnote ⁸² Similarly, during the Second Sino–Japanese War in 1937, without formal declarations by either side,Footnote ⁸³ the ICRC, with a few exceptions, had no presence in the territories occupied by the Japanese army.Footnote ⁸⁴

The question of the applicability of the GCs to situations other than declared wars was raised for the first time in 1934. The fifteenth Conference of the International Red Cross, based on the ICRC report regarding the problems arising from the too-literal interpretation of the 1929 GCs, expressed “the desire that these Conventions, which are adopted for the case of declared wars, also be applied by analogy in the event of armed conflict between States without there having been a declaration of war”.Footnote ⁸⁵

In 1937, the International Commission of Experts, convened by the ICRC, decided to replace the subjective criteria of animus belligerandi (in the form of declaration of war) with an objective criterion because the former could allow states “to circumvent the restrictions on war by the blissfully simple device of proclaiming their action to be, say, reprisal or self-defence rather than war”.Footnote ⁸⁶ The Commission believed that it would be more proper to replace the word “war” with the expressions “armed conflict between States”, “hostilities”, or “recourse to force”. This was derived from the unanimous decision of the Commission that the Convention must apply to all armed conflicts between states, while noting that preliminary declarations for waging war would be more and more frequently omitted in the future.Footnote ⁸⁷ On this basis, the Commission proposed the following draft new article: “[t]he present Convention is applicable between the High Contracting Parties from the moment hostilities have actually broken out, even if no declaration of war has been made and whatever the form that such armed intervention may take”.Footnote ⁸⁸

This passage and its clear reference to armed intervention indicates that what was in the minds of the members of the Commission was that such a use of armed forces necessarily entails a kind of intervention. The outbreak of the Second World War put a halt to these attempts. Immediately upon the close of the war, the ICRC resumed the task and thus took up once more the 1937 draft. The Preliminary Conference of National Red Cross Societies held in 1946 adopted the above draft with reference to armed intervention for the scope of application of the Geneva Convention on Wounded and Sick.Footnote ⁸⁹ The draft was also confirmed by the first Commission of the 1947 Conference of Government Experts, dealing with the revision of the Convention on Wounded and Sick.Footnote ⁹⁰ However, the third Commission of Government Experts Conference, established to study the condition and protection of civilians in the time of war, submitted a slightly different text: “[t]he present Convention is applicable between the Contracting Parties from the outset of any armed conflict, whether it has, or has not, been recognised as a state of war by the parties concerned”.Footnote ⁹¹ The Government Experts emphasised that the following general observations have to be borne in mind when reading the above recommendations:

In other words, in referring to the actual state of war, the governmental experts had in mind those cases of aggression where the aggressor state evades its humanitarian obligation by using terms like “‘legitimate self-defence’, ‘penetration’, ‘protection’, ‘necessity for the maintenance of internal security’…” in avoiding the use of the term war.Footnote ⁹³ These terms need to be understood against the background of the League of Nations’ approach to limiting war but leaving unregulated other forms of the use of force such as interventions, military reprisals, and self-defence.Footnote ⁹⁴ Even the Secretary-General of the League, in analysing the legal status of collective enforcement measures, such as naval blockade, taken under Article 16 of the Covenant,Footnote ⁹⁵ stated that:

The Bureau of the 1947 Conference of Government Experts, reviewing the recommendations of different Commissions, adopted by consensus the text proposed by the third Commission with no reference to the notion of “declared war” or the phrase “armed intervention”, as follows: “[t]he present Convention is applicable between the Contracting Parties from the outset of any armed conflict, whether it has, or has not, been recognised as a state of war by the parties concerned”.Footnote ⁹⁷ The chairman of the Bureau emphasised that such a decision is only for the humanitarian aspect of these Conventions: “sans nous occuper des répercussions politiques que cela pourrait entraîner”.Footnote ⁹⁸ Based on the feedback received, the ICRC submitted the following text for the common Article 2 to the XVIIth International Red Cross Conference:

The ICRC remarks under this Article state that “the general observations made by the Government Experts concerning the Civilian Convention … and Article 1 of the Committee’s Draft, have led the ICRC to adopt a wording which differs materially from previous drafts”; however, “the clause appearing at the beginning of the Article, ‘beyond the stipulations to be implemented in peace time’ was inserted in obedience to a suggestion made by the June Experts”.Footnote ¹⁰⁰

The ICRC, however, did not explain why it took up again the reference to the notion of declared war, nor did it explain why the phrase “have not been recognised as a state of war by the parties concerned” was replaced with the new one providing that “even should the state of war not be recognized by one of them”, the literal interpretation of which could lead to a conclusion contrary to what the Article aimed to serve. The XVII Stockholm International Red Cross Conference adopted the text with some editorial changes and submitted it to the Diplomatic Conference of 1949, which likewise adopted the text with almost no debate as following:

Such consensus, however, did not necessarily mean that the participants were completely aware of the constituting elements of the definition provided in the article. For example, the UK representative stated that:

Furthermore, the delegate from China requested an explanation for the phrase “even should the state of war not be recognized by one of them”.Footnote ¹⁰³ These ambiguities for the two permanent members of the UN Security Council remained unanswered, probably because at the time these questions were raised, the ongoing discussions were focused on the third paragraph of the draft Article 2 stipulating the application of the Conventions in NIACs, later adopted as the current common Article 3. Nonetheless, it is worth mentioning that this ambiguity persisted even during the 1977 diplomatic conference for the adoption of the additional protocols to the GCs. For example, the delegate from Italy stated that the phrase “even if the state of war is not recognized by one of them” in common Article 2 must be construed liberally, in the sense that the GCs and, in consequence, the Protocol applied in the case of any armed conflict that might arise between two or more of the High Contracting Parties, whether or not the state of war were recognised by one, several, or all the Parties.Footnote ¹⁰⁴

It is important to note that the 1952 and 1960 ICRC commentaries respectively, on the first and second GCs, say nothing about the phrase “even should the state of war not be recognized by one of them”. The following explanation, however, is included in the 1960 and 1958 commentaries, respectively, on the third and fourth Conventions:

The Commentary on Article 2(1) of the third Geneva Convention (GCIII) further contains this extra sentence that “[e]ven if the existence of a state of war is disputed, Article 3 can be applied”.Footnote ¹⁰⁶

How can such a difference between these commentaries be explained, given that they were all written around the same time by Pictet and other ICRC legal advisers? Does referring to Article 3 instead of the whole GCIII in ICRC Commentary imply that there are situations in which military personnel might get captured by the adversary, but such a situation does not necessarily amount to an IAC? Of course, in this context, it is not logical to assume that the above-quoted explanation in the 1960 Commentary on GCIII is referring to the situation as a NIAC. Rather, invoking the common Article 3 is to refer to its minimal protective provisions as “the elementary considerations of humanity”,Footnote ¹⁰⁷ which, in the words of the ICJ, are “even more exacting in peace than in war”.Footnote ¹⁰⁸ As will be demonstrated below, in our view, the authors of the ICRC commentaries at the time, who were engaged in the negotiations, had a conception of armed conflict in mind, according to which any armed conflict entailed a violation of the territorial integrity in the form of either occupation or armed intervention. This is the justification for their referral to the term “armed intervention” in defining “armed conflict”,Footnote ¹⁰⁹ meaning that military operations without any military intervention will not necessarily trigger application of the Conventions.

Although the adopted text of Article 2(1), contrary to its draft versions, does not include any reference to the notion of intervention, the Pictet commentaries, in line with the understanding that existed among drafters about the meaning of the factual state of war, again refer to the first formulations of this article and define armed conflict as “[a]ny difference arising between two States and leading to the intervention of members of the armed forces”Footnote ¹¹⁰. This intervention, as demonstrated above, was always related to the aggression by one state against the territorial integrity of another state. The necessity of territorial elements for the beginning of an IAC in the history of the GCs, to use the words of Carron, derived from the intention of drafters for having a factual and objective application of the IHL.Footnote ¹¹¹

This interpretation remained intact even during the consultation for the adoption of the 1977 Additional Protocol I (API). The historical background leading to the adoption of this instrument indicates that the inclusion of common Article 2(1), and also its commentary,Footnote ¹¹² as a field of application of API did not raise any debate.Footnote ¹¹³ In this regard, it is important to note that the ICRC in its first draft provisions of API relies again on the concept of military intervention, as it explains that:

For the ICRC, and the experts consulted, this self-explanatory nature of the definition of the armed conflict was due to the “the humanitarian character of the laws and customs considered”.Footnote ¹¹⁵ In other words, although the whole merge of the rules on conduct of hostilities with the protection rules of Geneva Law was based on this fundamental concept that “military necessities must in certain cases give way to those of humanity”,Footnote ¹¹⁶ the humanitarian object and purpose of the common Article 2(1) remained unchanged.

This review confirms our interpretation in the previous section that the concept of IAC under Common Article 2(1) of the GCs depends on the nature of the engagement with the territory of the state that is the victim of the initial attack. While IHL is generally conceived as a “carefully thought-out balance between the principles of military necessity and humanity”,Footnote ¹¹⁷ this does not change the fact that IHL is, by its nature, restrictive. Anne Quintin in her seminal work on the nature of IHL reviews extensively the historical background and preparatory works of different IHL treaties and demonstrates that “treaty IHL evolved primarily as a restrictive regime, placing obligations and prohibitions on States”.Footnote ¹¹⁸ The expansion of the scope of the GCs by the inclusion of Article 2(1) in 1949 was to break out of the straitjacket of declared wars to ensure that no actual armed conflict evades the protection of these instruments. In particular, enlarging the scope of the application of the Conventions was meant to ensure that actual situations of military intervention are covered by these Conventions, but by no means was such an expansion intended to characterise any type of resort to military force as an armed conflict.

V. Final remarks

The main tenet of this paper is that the “first shot” theory should not extend beyond its domain of application – namely incidents involving the use of force between states that begin with an unconsented-to invasion or presence in the territory of the other state. This, however, does not imply that in cases of invasion of territory, or what is assumed in international law to be territory, we believe that the application of the first shot theory automatically, and in all cases, leads to satisfactory results. As noted by its critics, this formula leads to the application of the whole regime of IHL, meaning inter alia that even in minor incidents where states use a limited force against one another, the killing or capturing of combatants, as well as imposition of collateral damages to civilians and civilian objects, would be deemed lawful, just as in real full-fledged wars. We did not adhere to the critical views on the first shot theory because they fail to resolve the problem without introducing additional difficulties, like opening the door for leverage of state discretion in the classification of the IACs. As discussed in the fourth part of this paper, this kind of discretion was exactly what the drafters of the common Article 2(1) sought to avoid. However, this rejection does not mean that we keep a close eye on the weak points of the first shot formula. Exactly the opposite. It is the acknowledgement of these shortcomings that leads us to argue that the first shot formula in no case should be extended to unilateral attacks in the territory of a third non-belligerent state, in particular, in the cases of targeted killing of military personnel. In this way, while we agree that the notion of “armed conflict” in common Article 2(1) evolves in response to the changing environment, we believe that its meaning can never be extended to encompass what it is intended to prevent.