Safe areas for the protection of civilians. An overview of existing research and scholarship

Safe areas for the protection of civilians remain a possible instrument in the foreign policy-making toolbox. Lately, this was highlighted by calls for a safe area in Syria by German chancellor Angela Merkel and defence minister Annegret Kramp-Karrenbauer. Yet, the subsequent debates demonstrated that many policy-makers, pundits and scholars lack a detailed knowledge about the concept. They further showed that various different perspectives on the idea of safe areas exist. This article therefore provides a comprehensive and inclusive account of the academic engagement with the topic of safe areas. It demonstrates that some arguments focus on the (geo)political dynamics of safe areas, whereas others emphasise the potential to protect civilians. It further presents the legal codification of consent-based safe areas in the Geneva Conventions as well as debates about the legality of enforced, post-Cold War safe area cases. Regarding the research on displacement and safe areas, the majority tends to focus on the containment effects, although some highlight the capacity of safe areas as an alternative to flight. Overall, however, the article concludes that recent and conceptual academic engagement with safe areas remains rare and that a safe area perspective is missing in relevant current empirical cases. More research that contributes to the conceptual development of safe areas or to the assessment of current empirical examples, for instance in South Sudan, is necessary.


Introduction
In March 2020, German Chancellor Angela Merkel called for a safe area in Idlib in North-Western Syria (Reuters 2020). The call followed on an earlier proposal for a safe area in Northern Syria made by German defence minister Annegret Kramp-Karrenbauer in October 2019 (Tagesschau 2019). Kramp-Karrenbauer's proposal triggered an extensive debate in Germany, in which at least three things became quite clear. First, it showed that safe areas remain a possible instrument in the foreign policy toolbox vis-à-vis armed conflicts. After the genocide in Srebrenica in 1995, the term had largely disappeared from policy debates, at least until the Syrian war (Orchard 2014, p. 55 f.). Yet, the phenomenon of safe spaces for the protection of civilians continued to exist on the ground. This is highlighted, for instance, by the so-called "Protection of Civilians-Sites" in South Sudan (e.g. Keen 2017, p. 30 ff.;Lilly 2014;Bosha 2014;Briggs 2017) or certain "Internally Displaced Persons (IDP)-camps" in the Central African Republic (CAR) and the Democratic Republic of the Congo (DRC; e.g. IRIN 2014; Schütte 2015, p. 197;Recchia 2018, p. 385). Second, the debate following Kramp-Karrenbauer's proposal demonstrated K that many policy-makers, pundits and scholars lack detailed knowledge and understanding regarding the safe area concept. Third, the proposal revealed that many conceptual uncertainties and very different perspectives exist regarding safe areas. The defence minister herself showed a lack of conceptual precision, as she seemed to use the terms safe area ("Schutzzone") and security zone ("Sicherheitszone") interchangeably (Tagesschau 2019;Deutsche Welle 2019). 1 Yet, she is not the only one being confused; many different perspectives on the safe area phenomenon exist. For some, as this article will show, safe areas are a legal concept to provide a neutral space for innocent civilians in wartimes. Others tend to see safe areas as a tool in the context of "humanitarian intervention" or as a means to control or prevent migration. In sum, all three insights from the recent safe area debate reiterate a demand made already by the editors of the German Peace Report 2017 ("Friedensgutachten") and others (Schoch et al. 2017, p. 9; also e.g. Heinemann-Grüder 2019, p. 66): There is an urgent need for more systematic and detailed engagement with the topic of safe areas in policy-making and academia.
This article therefore intends to provide a comprehensive and inclusive account of the academic engagement with the topic of safe areas. This is done by identifying and presenting the most prominent perspectives and argument of the existing literature. The article's structure is neither completely argument-nor author-based, but intends to reflect both, the most essential arguments as well as the broad spectrum of different scholars and their most essential contributions. Exemplary authors are presented for specific arguments, which, however, does not imply that other authors have not made similar arguments or that a specific argument is the only contribution of a referenced scholar. Furthermore, the focus of this article lies on scholarly literature that deals with safe areas conceptually. Therefore, it takes academic publications (books, peerreviewed journals etc.) from various disciplines into account, but largely excludes (historical) literature that only deals with empirical manifestations of safe areas, without contributing to an analysis of the concept as such. Furthermore, policy papers and "grey literature" are largely excluded as they rarely provide scientifically valid, conceptual contributions. It should however be mentioned, that at least since the beginning of the Syrian war, quite a few publications of this kind exist (an exhaustive list would exceed the scope of this article, see e.g. Adar (2020), Atallah and Mahdi (2017), Eichhorst (2015), Keck (2017), Meininghaus and Heinemann-Grüder (2017), The Washington Institute 2016).
Below, the article continues with a brief definition of safe areas, as the academic discussion still lacks a clear, broad and accepted understanding of the term. Then, a first essential cleavage between arguments that focus on the (geo)politics surrounding the creation, existence or fate of safe areas and arguments that focus on the ability to protect endangered civilians is presented. Subsequently, the contributions of international law, which has dealt to some extent with safe areas, are introduced. Here, on the one hand, the idea of a consensual safe space for vulnerable or civilian groups in times of conflict has influenced several legal concepts of the Geneva Conventions framework. On the other hand, legal scholars have debated the legality of militarily enforced safe areas after the Cold War that went beyond the ideas of international humanitarian law (IHL). Eventually, a crucial cleavage also exists among scholars that focus on the relationship between safe areas and displacement. Many tend to see safe areas as a tool to contain refugees and to restrict the right to asylum. Others, however, make arguments that safe areas could also provide a valuable alternative to flight for would-be refugees. Lastly, the focus on displacement further demonstrates that some of today's protection situations, for instance in the CAR or South Sudan, are not viewed through a safe area lens. Some other relevant insights might come, for instance, from military planning (e.g. Sewall et al. 2010, p. 77ff.;Beadle and Kjeksrud 2014), humanitarian perspectives (e.g. Briggs 2017) or foreign policy specialists (see following paragraph). However, these approaches are largely omitted here due to reasons of scope and because many respective writings are rather practice-or policy-focused than academic. This article concludes by stating that the relevance of safe areas extends over different perspectives and disciplines and that no all-inclusive accounts of the topic exist. Fundamental research gaps remain regarding many facets of the safe area phenomenon. It is further confirmed that current academic engagement with the topic is rare, despite several recent empirical examples that should be investigated through a safe area lens.

Defining the phenomenon of safe areas
Before we begin with the different arguments and perspectives of the existing scholarship, we have to have a brief look on what actually constitutes the phenomenon. A safe area can be defined as a confined geographical space within a conflict zone in which at least one external actor or all belligerent parties effectively guarantee protection for civilians from a threat of conflict-related physical violence. This definition is developed in detail elsewhere by the author and is therefore only briefly explained here (see Hering 2018, p. 17). Other existing definitions usually come with different shortcomings. Posen (1996), for instance, understands safe areas only as a military tool, whereas Orchard (2014, p. 55) sees them merely as "operations undertaken by international actors". Both therefore exclude safe areas that are essentially based on the belligerents' consent, such as the 1936 Madrid zone, the 1937 Shanghai zone or other examples (Sandoz 1995, p. 904 ff.;Ristaino 2008; see also below). Further shortcomings often exist e.g. regarding a safe area's geographical limitation and size, its target group, its location, the kind of provided protection, its legal basis or the question of neutrality. Some of these issues are picked up in different sections below and a detailed discussion can be found in Hering (2018). Furthermore, a precise definition is necessary, as one can find many different terms in policy-making and the academic literature. Yamashita (2004, p. 4) provides a non-exhaustive list of seventeen different terms. Yet, it is difficult to discover a consistent use as the same empirical event is often described by different, interchangeable terms or the same term is used for somewhat different phenomena. In 1993, for instance, the events in Srebrenica and five other Bosnian cities have been labelled "safe areas" by the United Nations (UN) Security Council (1993aCouncil ( , 1993b, while others described the same event, as "safety zones" (Landgren 1995, p. 437), "safe spaces" (Hyndman 2003, p. 167), "safe cities" (ibid.) or "safe havens" (Arulanantham 2000, p. 21). In addition, many other terms such as "de-escalation zones", "buffer zones", "safe corridors" or "no-fly zones" (just to name a few) exist. These are similar to, but on their own usually not covered by this article's safe area definition. In this article, the term "safe area" is used as an umbrella term for all phenomena and conceptions as long as they are covered by the aforementioned definition. It is however argued in Hering (2018) that within the "safe area" definition a distinction between the underlying logic of protection (belligerent's consent, credible force and international legitimacy, see also Orchard 2014) as well as between different sizes (large-scale "safe zones" and small-scale "safe havens", see also Posen 1996, p. 77 f.) can account for several other existing safe area variations (e.g. regarding conflict environment, military presence, legal basis, type of provided protection).
A few more elaborations on the aforementioned definition shall be made. The definition regards safe areas as a tool for the protection of primarily civilians. It is sufficient that a safe area provides only minimal physical security for the protected, although in some cases the provided protection has been far more comprehensive. Obviously, a safe area can only be described accordingly as long as its effect is the provision of this kind of protection. In other words, only the rhetorical declaration of a safe area or the mere intention to create one is not sufficient. Over time, at least minimal physical security must be the on-the-ground effect. Nonetheless, this is no "success" criterion, meaning that a safe area can also come to an end e.g. by collapse. A safe area has to be located inside an active zone of conflict, for instance ongoing inter-state war, non-international armed conflict or one-sided mass atrocities. A safe area far away from the threat of violence would be neither necessary nor possible to identify. 2 Furthermore, a safe area can be based on the logic of the belligerents' consent or on the logic of external third party enforcement. In case of the former, it is usually demilitarised and based on a formalized, written and ideally pre-conflict agreement, whereas the latter typically requires military protection. Apart from these differences in the underlying logic, the definition also allows for different sizes in terms of geography and population. In the past, dimensions have varied between the size of a building or camp (e.g. temporarily during the Rwandan genocide; McQueen 2005, p. 97;Dallaire 2003, p. 243, 419) and entire regions of a country (e.g. in Northern Iraq after 1991; Sandoz 1995, p. 916). Lastly, the definition has no limitations regarding whether a phenomenon is actually labelled as a "safe area", similar or completely different.
2 For instance, an IDP camp hundreds of kilometres away from the violence cannot be a safe area, whereas an IDP camp inside the zone of active conflict could be one (if the other conditions are fulfilled!). This distinction regarding geographical location has very important implications, e.g. regarding protection requirements, access etc. Generally, however, it is not argued here that all IDP camps inside of conflict zones automatically constitute safe areas. Nevertheless, some so-called "IDP camps" that are under military protection can be understood accordingly (see section 10). Furthermore, belligerents might specifically agree (e.g. in a written agreement) not to attack a certain IDP camp, which then could potentially be a safe area based on consent. It is noteworthy, however, that the mere legal restriction of attacks against IDP camps in international law as well as some form of general restraint by the belligerents is not sufficient.

A focus on (geo)politics
With this definition and the relevant conceptual clarifications, we can now turn to the assessment of the existing scholarship. Here, a first general group of arguments can be identified, which tends to focus on the (geo)political dynamics surrounding a safe area's creation, existence or fate. This section presents the relevant arguments of several exemplary scholars. In addition, arguments regarding safe areas as a tool for the containment of refugees, which are presented in the respective section below, are closely related.
That safe areas are geopolitically charged argues, for instance, Jennifer Hyndman (2003, p. 182). According to her, the creation of safe areas is rather about the geopolitical interests of powerful states than about the alleviation of civilian suffering (ibid.). Thus, she stresses the need for consent, as safe areas inside conflict zones would never be neutral as they are always politicised (ibid.: p. 183). Accordingly, she is very sceptical about the success chances of enforced safe areas by third party involvement without relying on ground-level negotiations (ibid.: p. 167). She bases her findings on case studies of Bosnia, the Open Relief Centres in Sri Lanka during the 1990s as well as on a so-called "preventive zone" in Somalia in 1992. 3 Charles Ingrao (2013, p. 224) similarly claims that self-interested state behaviour can explain the rise and fall of safe areas, at least in the cases of Srebrenica and Zepa in Bosnia. In his book (ibid.: p. 203 ff.), he investigates comprehensively five of the six Bosnian safe areas (only Tuzla is left out). Based on their creation, existence and eventual collapse, he argues that a "simple corollary of raison d'état explains the career of the safe areas (...) of Srebrenica and Zepa" (ibid.: p. 224). Nevertheless, he states that the massacre of Srebrenica shamed the great powers to give in to the "humanitarian impulse" to do something (ibid.).
Similarly, Stuart D. Gordon (2001, p. 213) assesses the safe areas in Bosnia and concludes that the Srebrenica aftermath influenced further international engagement in the conflict. According to him, the collapse of the safe area in Srebrenica immediately decreased the credibility of great powers and involved organisations such as NATO or the UN with its responsible peacekeeping mission (ibid.). However, the consequences of the failed safe areas also led to international negotiations and resulted in the Dayton Accords (ibid.). Furthermore, he highlights the problem that safe areas can affect the ethnic dimensions of a conflict as they might favour a certain ethnicity, increase the tensions by countering ethnic cleansing or even contributing to it by concentrating an ethnicity in a small location. This could lead to a number of "unacceptable choices" (ibid.: p. 227).
In contrast, Carol McQueen's (2005) argument focuses more on the interests of individual states than of global power dynamics. In her book, she provides a comprehensive account of the safe areas in Northern Iraq, in Bosnia and in Rwanda during as well as after the genocide. She concludes that states do not act uniquely in a "purely strategic, material environment" or in a "normative and legal one" (ibid. 149). Neither of these motivations could explain the considered cases of safe areas. Instead, she argues, that states are influenced by both motivations simultaneously, which leads her to promote a "broadened conception of interest" (ibid.). Beyond the motivational factors, she also identifies three elements that make a safe area's success more likely. First, it has to be implemented decisively, by a single state or multinational force as a clear interim measure. Second, it has to be included into a broader conflict resolution, and third, it has to be backed-up by sufficiently strong will and deterrence (ibid.: p. 161).
Stefano Recchia's (2018, p. 362) more recent piece on safe areas that are "established by powerful states" focuses on the protected civilians and applies a strict consequentialist rights-based logic (ibid.: p. 367). Yet, his conclusions emphasise (geo)political consequences of safe areas for the wider conflict. Despite potential positive short-term civilian protection effects, he argues that "they may worsen the plight of vulnerable civilians over the medium term" (ibid.: p. 362). Large-scale safe areas may prolong and escalate ethnic conflicts by limiting the protected groups' willingness to compromise, by being abused as base for military offensives or even by encouraging them to seek unilateral secession (ibid.). Nonetheless, although he regards safe areas rather critically, he does not discard them completely (ibid.: p. 366).
Lastly, by building on approaches of political economy and Michel Foucault, David Keen's (2017, p. 36) take on safe areas similarly argues that they depend on "complex and shifting goals of in-country and international actors". What matters, according to him, are the motives, functions and wider strategies a safe area is integrated into (ibid.: p. 36). Based on an analysis of these elements in six cases, he appears to be very sceptical about safe areas, although he as well does not seem to reject the concept completely (ibid.). As potential problems he lists, inter alia, that safe areas are not always safe, that they may undermine the right to asylum or that local and international actors might manipulate safe areas (ibid.: p. 36 ff.).

A focus on the protection of civilians
A second general string of safe area arguments focuses less on the (geo)political dynamics, but rather emphasises discussions regarding their effect on the protection of civilians. Many, but not all, arguments can be placed in a tradition of humanitarian interventionism. Selected research and authors are presented in this section.
Phil Orchard (2014Orchard ( , p. 55, 2018, for instance, places safe areas directly in a context of on-going debates regarding the protection of civilians and the responsibility to protect (R2P; also Jacob 2018 and Birnie and Welsh 2018, see section on migration studies, could be named here). Orchard also highlights that safe areas largely disappeared after Rwanda and Bosnia in the mid-1990s (ibid.). Yet, he believes that the idea is still relevant today as a "safe area has the potential to protect people (...)" (ibid.: p. 57). In his article, he develops a typology of three different safe area models based on the tactical consent of the belligerent parties, the presence of an credible international military force or the absence of both (these issues have also been raised by this article's definition and appear again in the following section on international law) (ibid. 59 ff.). Similarly to McQueen, Orchard (2014, p. 69 f.) also concludes that safe areas could be successful without the consent of all belligerents, if there would be a credible, long-term military presence, long-term humanitarian assistance and a decisive mandate to protect civilians. Without consent, however, a safe area today would have to be established within the framework of R2P (ibid.). In a more recent article, Orchard (2018) argues that the emergence of militarily enforced safe areas in the 1990s was only possible due to an increasing UN focus on civilian protection and the interpretation of forced migration as a threat to international peace and security.
Quentin Outram (1997) approaches safe areas specifically from the angle of humanitarian aid. He is one of the very few authors, who investigates the engagement of the Economic Organisation of West African States in Liberia between 1990 and 1996 from a safe area perspective. 4 He argues that the organisation's peacekeepers established a more or less stable safe area around the capital Monrovia, which provided physical security (ibid.: p. 194, 200). Furthermore, it provided the ground for humanitarian access to many IDPs, which is why Outram concludes that even without all parties' consent, safe areas can be beneficial in humanitarian emergencies (ibid.: p. 202).
Astrid Stuth Cevallos and Bryan Frederick (2018, p. 174) provide a study on lessons-learned regarding safe areas. Similar to their colleagues mentioned above, they argue that safe areas can be an effective tool to protect civilians. However, based on failures of the past, they conclude that this requires substantial resolve and capabilities by the intervener, as "[e]fforts to create safe areas 'on the cheap' have often resulted in disaster" (ibid.). Hikaru Yamashita's (2004, p. 193) research also concludes that political support for human rights protection and humanitarian relief is essential for the future of safe areas. In general, his work is one of the most comprehensive accounts on safe areas and their theoretical conception. For him, "safe areas, as practices, are human actions", which is why he analyses spatial practices and sovereign space in international politics (ibid.: p. 193, 8 ff.). Based on the case studies of Iraq, Rwanda and Bosnia, he derives a typology that includes three models. First, the conventional model of humanitarian space, which is based on consent, neutrality as well as demilitarisation and is compatible with states' territorial practices of sovereign space (ibid.: p. 17 ff., 24). Second, the shelter model, which does not require the belligerents' consent, follows the principle of impartiality and allows for the use of force for human rights protection. It does not violate territorial spaces, but conflicts with sovereign practises of states (ibid.: p. 20 ff., 24). Third, the homeland model, which actively supports one party in the conflict and thus reshapes territorial spatial practises (ibid.).
Lastly, it should be mentioned that some literature exists, which emphasises the local self-protection agency of civilians, for instance, in so-called "zones of peace" (e.g. Hancock and Mitchell 2007;Masullo 2015;Kaplan 2017;Bosi et al. 2015). Baines and Paddon (2012) as well as Paddon Rhoads and Sutton (2020) even assess civilian self-protection in relation to government-provided safe spaces in Uganda and the UN safe areas in South Sudan, respectively. This literature might be relevant for future safe area analyses, yet most existing studies do not seem to fit under the umbrella of this article's safe area understanding, which is further underlined by the frequent reference of very different case studies.

First attempts of a legal codification
The idea of protecting civilians in certain designated areas is not at all a new one. Historical examples of sparing civilian populations are reported, for instance, from ancient Greece or from Islamic traditions (Graf von Rittberg 1969, p. 3 ff.; Sandoz 1995, p. 900). However, as generally most codifications of international law are more recent, also attempts to codify safe areas legally are to be found within the last 150 years. Henry Dunant's (unsuccessful) suggestions from the 1870s, to declare certain towns in war zones neutral territories in order to shelter wounded combatants, is generally regarded as the first modern example (e.g. ibid.; McQueen 2005, p. 2). It took several theoretical and practical approaches and two destructive world wars before Dunant's ideas on "safe towns" became legally codified in the 1949 Geneva Conventions 80 years later. Yves Sandoz (1995, p. 901

Consent-based safe areas of the Geneva Conventions
Out of these earlier legal debates and empirical manifestations five different types of legal concepts developed, that are still valid today. All five of them are codified in the 1949 1st and 4th Geneva Conventions (ICRC 1949a(ICRC , 1949b or in its 1977 1st Additional Protocol (ICRC 1977). As such, these legal rules formally only apply in international armed conflicts, a fact that might turn out to be problematic consider-ing certain empirical examples and possible future applications in non-international armed conflicts.
The 1st Geneva Convention (Art. 23) describes hospital zones and localities that can be created on a party's own territory as well as in occupied areas. In addition to the wounded and sick, it only allows the inclusion of medical and administrative personnel, resulting in a de facto demilitarisation of this area. Accordingly, Article 23 stipulates a mutual, and potentially pre-conflict, recognition of the parties' hospital zones and localities (ICRC 1949a). However, the 1st Geneva Convention only addresses the fate of the wounded and sick in the armed forces. Although the affected people are combatants hors de combat, this article focuses on safe areas for the protection of civilians. Yet, it is included here, as Article 23 sets an important precedent for other forms of geographical locations excluded from the effects of war by mutual agreement.
One of these is Article 14 of the 4th Geneva Convention, which codifies hospital and safety zones and localities and generally follows the provisions of Article 23. However, this time, the addressed group are civilian wounded and sick as well as, more extensively, aged persons, children under fifteen, expectant mothers and mothers of children under seven (ICRC 1949b). According to Sandoz (1995, p. 907), the inclusion of the later groups follows the logic that those persons likewise do only pose a neglectable military threat. Due to the inclusion of civilians without the need of immediate medical attention, the term "safety zones" was added (ibid.: p. 907 f.). 5 The following Article 15 of the 4th Geneva Convention specifies the term neutralised zones. This approach is arguably the closest to the political safe area discourse as it does not only include wounded and sick (regardless of whether they are combatants), but also allows for basically every civilian as long as he or she does not perform "work of a military character" (ICRC 1949b). Moreover, it is limited to "regions where fighting is taking place", a necessary condition for this article's definition of safe areas (ibid). Article 15 requires a written agreement of the belligerent parties, which obviously cannot predate the conflict due to the location near the on-going fighting (see also Sandoz 1995, p. 908). Said agreement is supposed to include provisions regarding the administration, food supply and supervision of the neutralised zone, indicating that the Geneva Conventions' architects were already well aware of certain challenges of safe areas. According to Sandoz (ibid.) the empirical events of the Jacquinot zone in 1937 and Jerusalem in 1948 directly influenced this article.
Almost 30 years later, in 1977, the 1st Additional Protocol to the 1949 Geneva Conventions was established. 6 Its Article 59 further specifies non-defended localities, which are geographically precisely defined inhabited locations that are evacuated of all military personnel and mobile equipment. The remaining population must not use immobile military installations, attack the adversary party or support 5 It can be argued that "normal" hospitals can function as safe areas in conflicts. Although, this might be sometimes the case, Gordon and Perugini (2019) demonstrate that hospitals are not necessarily always exempted from attacks. 6 As most of the Geneva Conventions, it is again only directed at international armed conflicts. In addition, a notable number of states, including the US, Israel or Turkey did not ratify the 1st Additional Protocol.
K any military operations. In turn, it is prohibited to attack the non-defended localities "by any means whatsoever" (ICRC 1977). This is included here, as in reality nondefended locations might function as a safe sanctuary for civilians during a conflict, although a mere military withdrawal is not sufficient to count as a safe area for this article.
Lastly, the following Article 60 of the 1st Additional Protocol further establishes demilitarised zones that largely follow the idea of non-defended localities, but are to be created and maintained in a more orderly fashion and are supposed to be further away from the conflict zone (ICRC 1977;Sandoz 1995, p. 912). As such, they are in sharp conflict with one of this article's core criteria.
These five legal concepts can currently provide a specific legal basis for safe areas within the IHL framework. Despite their differences, they have three essential things in common that distinguish them from many empirical applications of safe areas. They must have a non-military character, they must be based on (preferably formalised) consent of the belligerent parties and they are specifically designed for international armed conflicts (Sandoz 1995, p. 920 ff.;Landgren 1995, p. 441).
Notably and more recently, in 2000 the International Law Association (2000) called for the establishment of safe areas for the protection of IDPs and the facilitation of humanitarian assistance. In general, Emanuela-Chiara Gillard (2017) provides a current and comprehensive overview of the legal basis of safe areas in IHL and compares it to other legal concepts relevant for safe areas (e.g. Security Council Chapter VII action or international refugee law). However, since the end of the Cold War, most empirical safe area cases did not follow the consent-based, demilitarised approach of the Geneva Conventions framework.

Legality of post-Cold War safe areas
Most recent legal writings on safe areas deal with the legality of empirical cases after the Cold War. Almost all of them argue that the post-1990 cases of safe areas were distinct from earlier applications, as they were not based on the consent of all belligerent parties. Thus, one might conclude that a new type of safe areas emerged after the Cold War. Moreover, several of the legal writings referenced below, debate dynamics of refugees movements and internal displacement. However, these issues are primarily discussed in the following sections. In general, however, current engagement of international law scholars with safe areas seems to be relatively low and the topic hardly influences current debates. 7 Yves Sandoz' (1995) work was already displayed extensively as it is one of the very few, if not only, comprehensive accounts on the development of safe areas in IHL, culminating in the presented five articles of the Geneva Conventions framework. In addition, he is one of the few authors analysing possible pre-1990 empirical cases of safe areas, for instance, in Cyprus (1974), in Saigon (1975), in Phnom Penh (1975 or during the Falklands-Malvinas conflict (1982). His focus, however, lies on an assessment of the post-1990cases in Northern Iraq (1991 and in the former Yugoslavia (Croatia andBosnia, 1991-1995). Following his investigation of the Iraqi case, where US-led forces imposed a safe area consisting of large parts of the Kurdish populated Northern Iraq, he argues that this safe area was outside of the framework provided by IHL (ibid.: p. 916 ff.). Furthermore, although he does not necessarily condemn the operation on moral grounds, he argues that its fundamental intent was political rather than humanitarian (ibid.: p. 919). Regarding the case of the former Yugoslavia, Sandoz (ibid.: p. 924) argues that the "United Nations Protected Areas" in today's Croatia were equally not covered by the IHL framework and rather a "peace-keeping measure" due to the high level of "UN soldiers" involved. With regard to Srebrenica and the five other safe areas in Bosnia, he similarly argues that the lack of consent, the subsequent need for military backing as well as the ultimate failure to provide protection demonstrate that these examples were outside of the legal framework of IHL (ibid.: p. 924 f.).
In the same volume containing the Sandoz (1995) chapter, three other authors assess safe areas from a legal point of view. Based on an analysis of the post-1990 cases of Iraq, the former Yugoslavia, the French post-genocide safe area in Rwanda as well as partly of the "Open Relief Centres" in Sri Lanka, B.S. Chimni (1995) distinguishes his understanding of "safety zones" from the Geneva Convention approaches (ibid.: p. 825 ff.). For him safe areas are largely about (internal) displacement, refugee movements and asylum dynamics. They might serve as a tool to insulate a safe space in conflict, facilitate the orderly movement of potential refugees or to reduce flight to neighbouring states (ibid.: p. 826). Furthermore, he identifies certain difficulties regarding the compatibility of safe areas with human rights, refugee and humanitarian law, for instance, denying the right to asylum or forcible movements (ibid.: p. 832). Also in said volume, Eugene Cotran (1995) specifically investigates the 1991 safe area in Northern Iraq. He argues that it was legal, however referring not to IHL, but simply to an assumed UN Security Council authorisation (ibid.: p. 866 f.). Lastly, Leonardo Franco (1995) looks at safe areas specifically concerning IDPs and from the perspectives of international human rights law, IHL and international refugee law. He argues that safe areas might be a useful tool to avoid refugee movements by creating safety, protection and humanitarian assistance in the country of origin (ibid.: p. 896). However, he, too, points out the potential danger of denying refugees a right to flee. 8 Beyond the 1995 volume that contained the aforementioned chapters, very few legal publications deal with safe areas specifically. A notable exception is Annette Simon's (2005) German work on safe areas as a means of protection for persecuted persons. She concludes that "UN safety zones", understood as internationally enforced safe areas inside a conflict zone, are a necessary and adequate addition to the international protection framework (ibid.: p. 267). In today's identity driven conflicts, the consent-based approaches of the Geneva Conventions would be inadequate (ibid.). Furthermore, she argues that enforced safe areas are generally a legally valid instrument, as long as they are backed by UN Security Council resolutions, fundamental rights are guaranteed within the safe area and someone in the international community assumes responsibility for maintaining these rights and the safe area itself (ibid.: p. 267 f., 282).
Simon's approach is similar to the analysis of Mohamed S. Elewa (2001). He specifically assess the six Bosnian safe areas from both, the legal perspectives of the Geneva Conventions framework as an example of jus in bello and of UN Security Council resolutions as an example of jus ad bellum (ibid.: p. 431). He concludes that UN safe areas will leave civilians concentrated at a specific location and even more vulnerable, if certain provisions are not met (ibid.: p. 461 f.). To prevent this, he demands the adoption of a "United Nations Convention on the Establishment of Safe Areas" (ibid.: p. 462).
Nils Geißler (1999, p. 283 ff.) analyses the legal dimensions of safe areas as a tool to protect IDPs. As the previous authors, he distinguishes between IHL safe areas and enforced safe areas and especially highlights the problems of the latter from a humanitarian perspective, e.g. the lack of neutrality (ibid.: p. 287).
Another example is provided by Surya P. Subedi (1999). Following the creation of non-consent-based safe areas after the Cold War, he asks whether the international community has the legal right to do so (ibid.: p. 24). He is neither specifically interested in the political motivation of great powers nor in the effectiveness of safe areas (ibid.: p. 23). Instead, he argues that the post-1990 cases showed that the international community does not have the right to intervention, but a right to humanitarian assistance, which can also justify the creation of safe areas (ibid.: p. 33). Thus, his line of argumentation is similar to the, then not yet codified idea of R2P. He claimed already in 1999 that the international community has a legal right to enforce humanitarian assistance without the affected state's consent. However, as the R2P (in both its 2001 and 2005 interpretation) legally relies on the UN Security Council's Chapter VII authority, Subedi's argument of a more general "right to humanitarian assistance" would certainly be disputed by many scholars.
Eventually, one of the most current legal writings on safe areas comes from Lokman B. Çentikaya (2017). Analysing the situation in Syria and possible Turkish intervention, he sees safe areas as an effective tool to provide protection as well as to prevent refugee movements (ibid.: p. 61). Furthermore, he gives the somewhat dubious argument that an enforced safe area "does not constitute an undue intervention (...) [w]hen such a State (sic!) internationalises its internal affairs by initiating and failing to prevent large outflows of its people" (ibid.). In contrast to many other authors referenced here, he apparently sees refugee movements not just analytically as a political motivation, but also a potential jus ad bellum. Compared to e.g. Subedi (1999), he rather seems to seek legal justification for an intervention in the fate of neighbouring states than in the human rights and protection needs of affected civilians.

Containing refugees
As mentioned before, several scholars see safe areas not so much in the light of protecting civilians but rather as a tool for refugee containment. In some cases, this is in line with politicians' rhetoric justifying a safe area with the prevention of refugee movements (e.g. Turkish president Erdogan's call for a safe area in Northern Syria, see The Guardian 2019). In other cases, authors presume that refugee containment is the hidden motivation behind third parties' safe area interventions. Moreover, some scholars analyse the prevention of flight as an unintended effect of safe areas. The respective containment arguments of Chimni (1995), Franco (1995) and Çentikaya (2017) have already been presented above. Another, in the safe area literature frequently cited article, comes from Karin Landgren (1995). She as well analyses the safe areas in Iraq, Bosnia, post-genocide Rwanda and the Open Relief Centres in Sri Lanka. Based on these case studies, she argues that the consent of the warring parties is paramount, as is the civilian and neutral nature of a safe area (453 f., 457 f.). Furthermore, she underlines that safe areas can only be a short-term measure and therefore cannot provide an alternative to asylum (ibid.: p. 456), an argument that is reiterated by many practitioners and in the migration grey literature (see e.g. Gilbert and Rüsch 2017). In the past, Landgren (1995, p. 456) argues, safe areas have been established to interfere with the right to seek asylum and in sharp contrast to the non-refoulement principle.
Barry R. Posen (1996) equally understands safe areas as a refugee containment tool. In an article titled "Military Responses to Refugee Disasters", he lists largescale "safe zones" and closely "circumscribed safe havens" as two of five possible military options (ibid.: p. 77 f.). 9 He realises their temporary nature, as they do not address the original cause of refugee movements, but still regards them as a military solution to prevent these "disasters" at least short-term. 10 Similarly, Bill Frelick (1993Frelick ( , 1997 argues that after the Cold War, safe areas were used to contain refugee movements. He states that safe areas could "keep the refugees close to home and keep [up] (...) the pressure on human rights violators", yet that they also were in line with "xenophobic tendencies then growing in Europe and emerging in North America" (ibid.: p. 40). Thus, he is highly critical of safe areas for the alleged sake of endangered civilians as he calls this line of argumentation a "humanitarian masquerade" (ibid.: p. 69). Recently, Frelick (2017) placed Donald Trump's call for a safe area in Syria in the context of refugee containment. Correspondingly, Cécile Dubernet (2001, p. 38) concludes in an extensive study on internal displacement that IDP protection is an instrument of containment.
Katy Long (2012) assesses refugee containment through a combination of safe areas and border closures. She, too, understands such policies as interest-based attempts to undermined states' obligations under the asylum framework (ibid.: p. 472). According to her, safe areas with their provision of minimal physical security and humanitarian relief inside the conflict country often function to replace the wider responsibilities of refugee protection in a host state (ibid.). Regarding the empirical support of her argument, however, she turns largely to the already mentioned examples of the 1990s, ignoring possible newer cases of de facto safe areas. These more recent potential cases are briefly presented below.

An alternative to flight
In contrast to the arguments that regard safe areas as a tool to prevent refugee movements, some authors argue that safe areas could actually constitute a valuable alternative for would-be refugees. Ahilan T. Arulanatham (2000), for instance, understands safe areas positively as they would provide an alternative to flight (a similar argument is found in Tiso 1994). In order to create functioning safe areas he argues that three conditions have to be met. First, safe areas must be in the interest of the affected state, second, the non-refoulement principle has to be maintained, and third, refugee (and IDP protection) must be human rights based (Arulanatham 2000, p. 55). What is more, he wants to restructure the Office of the UN High Commissioner for Refugees (UNHCR) as responsible organisation for the implementation of these safe areas (ibid.: p. 44 ff.). Despite his rather positive take on safe areas, Arulanatham also demonstrates what happens when neither consent nor military security protect a safe area. He reports on a situation, where the Sri Lankan government abducted IDPs from a camp administered by the UNHCR. Many of them had stayed within the country, relying on the UNHCR's domestic protection as an alternative to flight (ibid.: p. 1 f.). Yet, the UNHCR could not provide tangible physical security and even failed to report on those incidents (ibid.).
Similar to Arulanatham, Daniel Jacob (2018) argues more recently that safe areas should be seen as a complement, not as an alternative to the right to asylum and longterm conflict resolution (ibid.: p. 324, 326). However, to be successful, (enforced) safe areas require a clear legal mandate and effective military deployment (ibid.: p. 321 ff., 324 ff.). Also Rutger Birnie and Jennifer Welsh (2018) largely seem to agree with Jacob as they make a normative case for safe areas as a response to mass flight and, especially, internal displacement (ibid.: p. 348). Next to a "right to stay" (ibid.: p. 349), they further highlight safe areas' potential to facilitate refugees' return (ibid.: p. 348).

Today's absence of a safe area perspective
A final insight from the literature and current case studies is that, today, the term "safe area" is not applied to appropriate situations, especially in the context of intrastate protection and displacement. As demonstrated above, many scholars, especially during the 1990s, saw a close connection between the use of safe areas and sheltering endangered displaced within their country of origin. These dynamics did not disappear. On the contrary, the global number of IDPs that benefit from UNHCR (2020) protection is several times higher than in the mid-1990s. Yet, despite the fact that many IDPs live in camp-like structures (confined geographically spaces) and are protected by international actors, today few scholars or practitioners refer to safe areas. Relevant cases are rather covered by the large umbrella term "IDP camp". Although no clear definition exists, this concept lays the focus on displacement and can generally be understood as a transfer of the "refugee camp" concept into the domestic (see, for instance, Global Protection Cluster 2010, p. 9 for the protection needs of IDPs). While this is undoubtedly a suitable approach to the majority of internal displacement settlements, in some cases a safe area perspective would be more expedient.
For instance, this would have been the case in late 2013 when more than 100,000 protection-seeking civilians sought the proximity of French forces at the airport of the CAR's capital Bangui. Despite initial reluctance, French and African Union forces eventually decided to provide minimal physical protection in form of patrols and tank deployment (Bourgois 2013;Ngoupana 2013;UN Children's Fund 2014). This situation has been widely labelled as "IDP camp". Yet, it was not an accumulation of persons displaced from all over the county, but of endangered Bangui residents, who were if anything "displaced" within their hometown. Accordingly, one might ask whether a safe area perspective might be more suitable than an "IDP camp"-label. One can even go further and argue that a displacement framing is dangerous as international actors might then fail to address the situation as to what it actually is: a protection crisis, where people need security within their hometown.
Furthermore, many UN peacekeeping missions are now tasked with the protection of civilians or IDPs in particular. Protection does not necessarily entail the creation of protected confined geographical spaces. Yet, in several cases, e.g. in the DRC, (de facto) safe areas are deliberately created or emerge unintentionally (IRIN 2014;Schütte 2015, p. 197;Recchia 2018, p. 385). A very specific case is South Sudan, where the mission even went away from the "IDP camp" label and created so-called "Protection of Civilians-Sites". 11 These sites are almost perfectly described by the safe area definition. Nonetheless, in South Sudan, as in the other cases, only very few practitioners or scholars use a safe areas perspective for policy-making or as an analytical lens (for notable exceptions regarding South Sudan see e.g. Keen 2017, p. 30 ff.;Lilly 2014;Bosha 2014;Briggs 2017). Consequently, the lessons from the existing debates and arguments presented in this article, do not influence current (de facto) safe area situations or their analysis.

Conclusion
Safe areas are a phenomenon, which touches upon a variety of different subjects such as the protection of civilians, global ethics and responsibilities, just war and humanitarian intervention, UN politics and peacekeeping, refugee movements, geopolitics, state interests or IHL. Scholars from very different academic backgrounds have dealt with the phenomenon. Despite or because of this, no single approach exits that would explain all of these dynamics or all empirical cases. The majority of the existing literature focuses in depth on a specific conceptual aspect or on specific cases, but fails to see safe areas comprehensively. Safe areas are a phenomenon with a wide range of empirical examples over the last 100 years. It is of concern for scholars and practitioners of international politics, peace and conflict studies, international law as well as of migration studies (and one might add the military, humanitarian actors and foreign policy specialists). Therefore, more conceptual and comprehensive research of safe areas is necessary.
The article presented different dominant arguments and perspectives within the existing scholarship on safe areas. Some arguments highlight the (geo)political dynamics of safe areas, while others focus on the ability to protect civilians. Here, it would be interesting to add studies on the emergence of unintended (de facto) safe areas, for instance in the CAR or South Sudan. Due to their unintended nature they cannot be explained by top-down (geo)political or protection of civilians motivations. From a legal perspective, the most prominent contributions are the jus in bello, consent-based safe areas codifications in the Geneva Conventions framework. Beyond this, several studies exist on the jus ad bellum legality of militarily enforced safe areas after the Cold War. However, gaps remain regarding the legal standing of safe areas in the context of increasing practices of protection of civilians, for instance in UN peacekeeping. It should also be asked whether the consent-based Geneva Conventions safe areas for international armed conflicts are adequate in a time of intra-state, identity-driven conflicts that intentionally target civilians. Regarding perspectives of migration and displacement, it became clear that differences exist on whether safe areas prevent flight and whether this is normatively good or bad (see also Hering 2019). Here, scholars should look deeper into the role safe areas do, could or should play in the current situation of large-scale global displacement. All of these open issues-global-displacement, protection of civilians by UN peacekeepers and times of identity-driven conflicts-are current, global challenges. The debates presented in this article as well as past and current empirical examples show, on the one hand, that safe areas could play an important role in these contexts. On the other hand, they also demonstrate that many pitfalls and hazards exists when safe areas are used without proper strategies, intentions and careful consideration. More, up-to-date research that assesses safe areas systematically and in the light of today's challenges is urgently needed, not only because safe areas de facto continue to exist empirically.
However, despite the continued relevance of safe areas, this article showed that systematic and current academic literature remains rare. The Syrian war has created some notable exceptions (e.g. a special issue of "Global Responsibility to Protect" containing Birnie and Welsh 2018;Jacob 2018;Orchard 2018;Recchia 2018), but in general only a handful of recent publications could be found that go beyond "grey" and policy literature. These then appear to focus on very specific aspects of the phenomenon (e.g. Çentikaya 2017;Gillard 2017;Long 2012). 12 In addition, almost all post-Cold War publications analyse the same case studies from the 1990s.
Although, the political labelling of operations as "safe areas" disappeared after the Srebrenica genocide, this should not blur the analytical judgement. The "Protection of Civilians-Sites" in South Sudan, the airport "IDP camp" in the CAR as well as certain "IDP camps" in the DRC and probably elsewhere are too similar to the history of safe areas as that they could be ignored by this analytical lens. Therefore, it requires more research assessing these cases, connecting them with the existing safe area literature and identifying similarities and differences. These current examples further demonstrate that Orchard (2014, p. 57) is right when he argues that safe areas can potentially protect people. Hence, there additionally remains a moral case for further investigations of the safe area concept.
Funding Open Access funding enabled and organized by Projekt DEAL.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4. 0/.