Changes in Treaty Interpretation: The ICRC’s Updated Commentaries to the Geneva Conventions

Last Saturday (12 August) marked the 74^th anniversary of the adoption of the 1949 Geneva Conventions on the protections to be accorded to victims of armed conflict. With 196 states parties, the four Geneva Conventions are (together with the United Nations Convention on the Rights of the Child) the most widely ratified of all treaties, with more states having ratified or acceded to them than the UN Charter (which has 193 parties).

A few years ago, the International Committee of the Red Cross (ICRC) began the important project of updating its Commentaries to the 1949 Geneva Conventions, and the 1977 Additional Protocols to those Conventions. The original commentaries to the 1949 Conventions were largely prepared in the 1950s, and the need for an update to those commentaries is obvious given the developments since then. So far, the ICRC has published updated commentaries to the first three Geneva Conventions.

When the ICRC published its Customary International Humanitarian Law Study nearly two decades ago, that work raised interesting and fundamental questions about the formation and identification of customary international law. That study provided a concrete context for scholars (see for example this volume) and states (see this US response to the ICRC) to address issues such as:  what exactly is it that constitutes state practice (operational practice only or verbal/written practices also)?; to what extent may treaties, and practice under treaties contribute to the formation of customary international law?; should the practice of states count equally?

The ICRC’s updated Commentaries on the Geneva Conventions (GCs) raise similarly interesting questions about the other main source of international law – treaty law. This post seeks to explore questions of treaty interpretation that are raised by the updated Commentaries. Issue that arise include: the use of a later treaty for the interpretation of an earlier treaty, in cases in which not all parties to the earlier treaty are parties to the later treaty; and how to conceptualise the use of judicial decisions as a means of treaty interpretation, and how such use relates to the methods of treaty interpretation set out in the Vienna Convention on the Law of Treaties.

Presumably the point of a new edition of any work is to take changes in the field being studied into account, and to make changes to the original version of the work. In the case of the updated Commentary, the Introduction (para. 5 of the updated Commentary to GC I) states that “[t]he objective is to ensure that the new editions reflect contemporary practice and legal interpretation” and that “[t]he new Commentaries reflect the ICRC’s current interpretations of the law, where they exist” (emphasis mine).  It therefore should come as no surprise that in revising the commentaries the ICRC has, on occasion, changed its view as to what the law is. It is inevitable that on some occasions, the view will be taken that contemporary developments will mean that current interpretations of the law are different to what they were when the commentaries were first written back in the 1950s. However, given that the text of the Geneva Conventions have not changed, changes in interpretation of the treaty raise the familiar question of whether it is legitimate to say that a treaty means something today which it was not thought to mean in the past. Even if such a change can happen legitimately, how does treaty interpretation change or evolve over time?  What are the processes by which changes in treaty interpretation can take place?This post will:

• first, point to three instances where the ICRC, in its revised Commentaries takes a position as to the interpretation of Geneva Convention, which is at variance with the view of the law that the ICRC itself previously expressed in the original version of the Commentaries.
• Second, explore how the ICRC comes to the conclusion that the law is not what the ICRC previously thought the law was. What types of justifications are advanced for the differences of view?
• Third, consider how the ICRC’s justifications for a new interpretation might (or might not) fit into established methods of treaty interpretation
• Fourth, explore how the ICRC expresses its new position.

Three Differences in Interpretation

i. Definition of International Armed Conflict

One area where the ICRC’s Commentaries take a position that is different from the position apparently taken in the earlier version of the commentaries relates to the test that is to be applied in determining that an international armed conflict exists. An international armed conflict is a conflict between two states. But what happens when a state uses force on the territory of another state, against a non-state actor present there, without the consent of the state on whose territory force is used. There might well be a non-international armed conflict between the state using force and the non-state armed group but is there also an international armed conflict between the state using force and the state on whose territory the force is being used? A good example is the use of force by the United States and other states against the so-called Islamic State in Syria. This action, unlike the one in Iraq, across the border took place without the consent of the territorial state – the Syrian government. Was there an international armed conflict between the US and Syria? This depends on how one interprets Art. 2 common to the four Geneva conventions when it speaks of “armed conflict which may arise between two or more of the High Contracting Parties, even if the state of war is not recognized by one of them.”

The previous version of the commentary to Common Article 2 appeared to suggest that for there to be an international armed conflict between two states there needed to be the involvement of two armed forces. The 1958 Commentary to the Fourth Geneva Convention refers to “[a]ny difference arising between two States and leading to the intervention of members of the armed forces’ (p. 20, emphasis added).” However, the updated commentaries (para. 222 of the 2016 GCI commentary; para. 255 of the 2021 GCII commentary) state that the position taken in 1958 “would mean that for an armed conflict to exist in the sense of Article 2(1), the simultaneous involvement of at least two opposing States through their armed forces is required. That interpretation is too narrow.”

According to the updated Commentary, an international armed conflict arises whenever one state unilaterally uses force against another state. Furthermore, uses of force on the territory of another state, without the consent of the latter state amount to a use of force against the latter state, bringing into effect an international armed conflict between the two states. This would still be the case, even if the reason for the use of force was to engage a non-state actor on the territory of the latter state. In taking this view, the updated commentaries adopts (and cites) a position that I have advocated for (see this chapter and an updated version here).

ii. Definition of Occupation

A second example of a place where the ICRC’s new Commentaries take a position that is different from a position they previously took relates to the definition of occupation. Under Common Article 2, the Geneva Conventions apply in their entirety to situations of occupation. The previous version of the commentaries (namely the commentary to Art. 6 of the fourth Geneva Convention) suggested that the definition of occupation for the purposes of Fourth Convention was different from the definition of occupation contained in Art. 42 the Hague Regulations of 1907.  According to that provision, “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.’” It was stated in the 1958 Commentary that:

“the word ‘occupation’, as used in the Article, has a wider meaning than it has in Article 42 of the Regulations annexed to the Fourth Hague Convention of 1907. So far as individuals are concerned, the application of the Fourth Geneva Convention does not depend upon the existence of a state of occupation within the meaning of the Article 42 referred to above. The relations between the civilian population of a territory and troops advancing into that territory, whether fighting or not, are governed by the present Convention. There is no intermediate period between what might be termed the invasion phase and the inauguration of a stable regime of occupation. Even a patrol which penetrates into enemy territory without any intention of staying there must respect the Conventions in its dealings with the civilians it meets.” (p. 60)

However, the updated Commentaries say that “Subsequent treaties, including the Geneva Conventions, have not altered the definition of occupation in the Hague Regulations” (para. 294 of GCI commentary). It goes on to state that “the Fourth Convention builds on the Hague Regulations but does not replace them for the purposes of defining the notion of occupation” (para. 296) and that the definition in Hague Regulations is the exclusive definition  of occupation in IHL and applies to the Geneva Conventions too (para 298)

iii. Whose consent is required for the implementation of humanitarian services?

A third example of a change of position by the ICRC relates to the interpretation of Common Art. 9 of GCI (also Arts 9/9/10 of GCs II, III & IV). That Article provides that the ICRC and impartial humanitarian organizations can offer their services to parties to the armed conflict. However, the provision goes on to state that the right to undertake these humanitarian services is “subject to the consent of the Parties to the conflict concerned”. The question that arises is which parties are “concerned” such that their consent is required. Think for example of a humanitarian relief operation to supply food in the context of an armed conflict. If that food has to come from territory outside the state, whose consent is required? Just the party on whose territory the relief action is to be carried out, or all those parties through which the relief operation must pass. The 1952 Commentary to GC 1 say that:

“[t]he ‘Parties concerned’ must be taken to mean those upon which the possibility of carrying out the action contemplated depends. For example, when consignments of relief are forwarded, it is necessary to obtain the consent not only of the state to which they are being sent, but also of the state from which they come, of the countries through which they pass in transit and, if they have to pass through a blockade, of the Powers which control the blockade.” (p.111)

However, the updated commentary to GCI, Common Art. 9  reverses this position and takes the view that states through which the goods must transit are not parties “concerned” for the purposes of this provision. Thus, according to the ICRC, the consent of such a transit state is not required as a matter of this provision (though the ICRC accepts that such consent will be required by general international law). The new Commentary (para. 1166) states that party would only be “concerned”:

“because the proposed humanitarian activities will take place on its territory or in an area under its control, as is the case for an Occupying Power. Thus, the opposing Party to the international armed conflict does not need to be asked for its consent on the basis of Article 9. This is the case even when, for example, the proposed relief convoys need to pass through the opposing Party’s territory. In that instance – while a different set of rules regulate its position with regard to its obligation to allow and facilitate ‘passage’ (see para. 1168) – that Party is not, for the purpose of Article 9, ‘concerned’ because the proposed humanitarian activities are not going to take place on its territory. However, its consent will need to be sought on the basis of general international law.”

How are the New Positions Justified?

As stated out the outset, the ICRC’s updated Commentaries provides opportunity for revisiting, in a concrete context, important questions of treaty law, and of treaty interpretation in particular. Where the ICRC has taken the view that a treaty text does not, today, mean what the ICRC thought it to mean previously, the question is immediately raised as to why the ICRC thinks the meaning of the text has changed, and as to how that change may have occurred. However, it must be noted that the expression of a different view does not necessarily mean that the ICRC thinks that the interpretation has changed over time. It may well be that the ICRC is of the view that the position originally expressed was wrong. If that were the case, one would not need to find a theory of evolutive interpretation of treaties to justify the change.

In the first example, dealing with the changing definition of international armed conflict, the change of position does not appear to be based on anything that has happened between publication of the previous commentaries and now. Instead, the change appears simply to be an attempt to interpret the text of the treaty in the light of its objects and purpose, and in the light of other areas of international law. In this particular case, the ICRC seems to be suggesting that it got the interpretation wrong back in 1958. The ICRC does not refer to any new developments which led to a changing meaning. However, it is entirely possible that new factual developments (the fact that there have been a number of uses of force on the territory of a state without its consent) might have alerted highlighted to an issue it wasn’t previously alive to and to the ICRC the consequences of the old interpretation.

In the second case, the new definition of occupation, the authors of the revised commentary refer back to the drafting of the Convention in 1949, and to the presumed intention of the drafters, but appear to rely primarily on case law of the International Tribunal for the former Yugoslavia (ICTY) and the International Court of Justice (ICJ) in justifying their conclusion that the interpretation of the treaty means something different. In this case, the argument does not seem to be one that relies on an evolutionary interpretation of the treaty as traditionally understood. In particular, there is no suggestion that the meaning has changed as a result of adopting the object and purposes of the Convention to new factual circumstances, nor is the change accounted for by the subsequent practice of the parties or based on their subsequent agreement. The change here is due to a re-reading of the presumed intention of the drafters, but critically on case-law that has arisen since the first commentaries were produced.  

The question that is raised by the second example is what role should decisions of international tribunals play in assessing treaty interpretation. When one looks at the rules of treaty interpretation in Articles 31-33 of the Vienna Convention on the Law of Treaties (VCLT), one does not see use of case law as a tool of interpretation. However, the list of sources of international law in Art. 38 of the Statute of the ICJ does include judicial decisions as subsidiary means for the determination of rules of law. These decisions help us to identify what the law is. Although international law is not a common law system, it does still give significance to decisions of international tribunals, and the ICRC is right to recognise this point when it deals with treaty interpretation. There is nothing in Article 38(1)(d), that indicates that, when it speaks of judicial decisions as subsidiary means for the determination of rules of law, the “law” there excludes treaty law. Judicial decisions are as much a means for the identification of treaty law as they are means for identification of any of the other rules of law. However, the question is how one fits this into a theory of treaty interpretation and into the recognised tools for treaty interpretation under the law of treaties. Hopefully, the judicial decision that is being referred to will have itself reasoned on the basis of the methods of treaty interpretation to be found in the VCLT. However, what the ICRC’s updated commentary show is that an interpreter can (and often will) short circuit the VCLT methodology simply by referring to the judicial decisions as a subsidiary means for the identification of content of treaty law.

Parenthetically, I will say that though the revised Commentary is right to pay attention to the judicial decisions in reassessing its interpretation of the meaning of occupation under the Geneva Conventions, the ICRC somehow failed to cite case law that supports its original position, and which was contrary to the new position. In particular it failed to refer to an ICTY decision that quoted the previous commentary with approval [Trial Chamber in Prosecutor v Rajić, IT-95-12, Review of the Indictment (Trial Chamber), para. 41 13 September 1996.]

It may also be suggested that the change in interpretation would have been better justified if explicit reference could have been made to the materials which the VCLT provides for in relation to interpretation of treaties. However, it could be argued that the ICRC was implicitly referring to the notion of “systemic integration” (interpreting treaty obligations in the light of other international law obligations) as set out in Article 31(3)(c) of the VCLT but didn’t quite say so.

In the third case, the ICRC seems to justify the change in their interpretation based on subsequent developments in the law, particularly the effect of Additional Protocol 1 (API) of 1977. The starting point on this issue is that in interpreting a treaty, the interpreter must take into account a subsequent agreement between the parties to the treaty regarding its interpretation, and subsequent practice of the parties to the treaty “which establishes the agreement of the parties regarding its interpretation” (Art. 33(1)(a) & (b) VCLT). The question here is the extent to which one can use subsequent treaty practice in order to justify a new interpretation of the 1949 Conventions. This question is particularly sensitive here because not all the parties to the 1949 Geneva Conventions have become parties to the Additional Protocols (APs) 0f 1977.

The Additional Protocols cannot be considered as a subsequent agreement regarding the interpretation or application of the Geneva Conventions within the meaning of Art. 33(1)(a) of the VCLT because not all the parties to the GCs are parties to the APs. It might also seem at first sight that, for the same reason, the APs cannot constitute relevant subsequent practice under Art. 33(1)(b). However, it is in fact possible to use practice concerning a subsequent treaty in order to interpret an earlier treaty even if not all the the parties to the earlier treaty are parties to the later one. This is because the practice of the parties in negotiating and adopting the text of the subsequent treaty may be practice in the interpretation of the earlier treaty. Even if not all of them go on to ratify the later treaty, what they said in negotiating and adopting that later treaty may be practice in the interpretation of the earlier treaty. There are some provisions of AP I (for example Art. 45(3)) that were negotiated and adopted on the basis of a particular interpretation of a provision of the GCs. In other words, the negotiating history of the later treaty, in and of itself, is subsequent practice for the earlier one, and that negotiating history may establish the agreement of the parties.

However, in the particular case of common Articles 9/9/9/10 of the GCCs it seems to me to be problematic to rely on the later provisions of API to justify a new interpretation. In the updated commentaries, the ICRC uses Art. 70 of API, which deals with humanitarian relief operations, to provide a particular interpretation of the common articles. Recall, the issue is whose consent is required for the implementation of humanitarian relief operations. In the earlier commentary, the ICRC said Parties concerned means all parties upon whom implementation of the relief action depends, i.e any state through whose territory the operation will pass. In the new commentary, the ICRC says only the party in whose territory the relief action will be carried out needs to consent under Common Article 9. They base this, in part, on the fact that in their view transit states have already consented under Art. 70 AP I. Even if that were correct, it would seem wrong to say that this changes the meaning of Common Articles 9/9/9/10 because the question that would then arise is what the position of non-parties to API would be. The position taken would seem to suggest that CA9 means one thing for parties to API and another thing for non-parties.

How are the Changes of Interpretation Expressed?

While, it is not surprising to see changes to the ICRC position on a number of occasions, the different ways in which the changes of position are expressed calls for comment. In the first case referred to above, relating to the definition of an international conflict in cases of unconsented uses of force on the territory of another state, the difference between the previous Commentary and the new Commentary is clearly exposed by the authors of the present commentary in the body of the text. In this case, the passage from the previous Commentary is quoted in the text of the updated Commentary, the difference of view in the new Commentary is set out and the reasons for the difference are explained.

In the second case, relating to the revised position on the definition of occupation, the difference between the previous Commentary and the new one is indicated in the new commentary but only in a footnote. While the question of whether the concept of occupation in the 1949 Geneva Conventions is the same as that in Article 42 of the Hague Regulations is discussed at some length in the new Commentary, the body of the text fails to indicate that it is taking a position contrary to that previously taken by the ICRC. Instead, one only finds this difference of view acknowledged in a footnote (see n. 154 of the 2019 commentaries to GCI and n. 160 of the 2021 commentaries to GCIII ) that reads: “It has been argued that two distinct definitions of occupation exist, one drawing on the Hague Regulations and the other on the Fourth Geneva Convention” and then a reference to old commentary  – “See Pictet (ed.), Commentary on the Fourth Geneva Convention, ICRC, 1958, p. 60” and three scholarly works (including one by yours truly!)  

In the third case, relating to the meaning of parties “concerned” in Common Art. 9, the new commentaries do not even allude to the fact that they are taking a different approach from that of the old commentary. Despite the fact that there is no indication in the text or footnote that a change is being adopted, the new commentary does seek to justify the position adopted in 2016.

In my view, the way in which the ICRC Commentary deals with the first change of position is preferable. It is better to acknowledge the changes in position having provided reasons for them. That way, the conflicting positions and the reasons for the change can be better assessed. Even with the second example above, the discussion of the differing views does it make it easier for the reader to take a view on the new position expressed by the ICRC as the conflicting arguments are presented in the text.  However, what may not be clear to the reader is that the position expressed is a new one for the ICRC (as compared to the previous Commentary). To the contrary, in the third example, the ICRC hardly even sets out that there is a contrary point of view, let alone identifying that the organization, in the previous edition of the Commentary, had taken that contrary view.